diff options
Diffstat (limited to 'drivers/block')
45 files changed, 5545 insertions, 1976 deletions
diff --git a/drivers/block/DAC960.c b/drivers/block/DAC960.c index 1f286ab..e086fbb 100644 --- a/drivers/block/DAC960.c +++ b/drivers/block/DAC960.c @@ -140,13 +140,14 @@ static int DAC960_getgeo(struct block_device *bdev, struct hd_geometry *geo) return 0; } -static int DAC960_media_changed(struct gendisk *disk) +static unsigned int DAC960_check_events(struct gendisk *disk, + unsigned int clearing) { DAC960_Controller_T *p = disk->queue->queuedata; int drive_nr = (long)disk->private_data; if (!p->LogicalDriveInitiallyAccessible[drive_nr]) - return 1; + return DISK_EVENT_MEDIA_CHANGE; return 0; } @@ -163,7 +164,7 @@ static const struct block_device_operations DAC960_BlockDeviceOperations = { .owner = THIS_MODULE, .open = DAC960_open, .getgeo = DAC960_getgeo, - .media_changed = DAC960_media_changed, + .check_events = DAC960_check_events, .revalidate_disk = DAC960_revalidate_disk, }; @@ -1789,7 +1790,7 @@ static bool DAC960_V2_ReadControllerConfiguration(DAC960_Controller_T unsigned short LogicalDeviceNumber = 0; int ModelNameLength; - /* Get data into dma-able area, then copy into permanant location */ + /* Get data into dma-able area, then copy into permanent location */ if (!DAC960_V2_NewControllerInfo(Controller)) return DAC960_Failure(Controller, "GET CONTROLLER INFO"); memcpy(ControllerInfo, Controller->V2.NewControllerInformation, diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig index 83c32cb..717d6e4 100644 --- a/drivers/block/Kconfig +++ b/drivers/block/Kconfig @@ -470,6 +470,27 @@ config XEN_BLKDEV_FRONTEND block device driver. It communicates with a back-end driver in another domain which drives the actual block device. +config XEN_BLKDEV_BACKEND + tristate "Block-device backend driver" + depends on XEN_BACKEND + help + The block-device backend driver allows the kernel to export its + block devices to other guests via a high-performance shared-memory + interface. + + The corresponding Linux frontend driver is enabled by the + CONFIG_XEN_BLKDEV_FRONTEND configuration option. + + The backend driver attaches itself to a any block device specified + in the XenBus configuration. There are no limits to what the block + device as long as it has a major and minor. + + If you are compiling a kernel to run in a Xen block backend driver + domain (often this is domain 0) you should say Y here. To + compile this driver as a module, chose M here: the module + will be called xen-blkback. + + config VIRTIO_BLK tristate "Virtio block driver (EXPERIMENTAL)" depends on EXPERIMENTAL && VIRTIO diff --git a/drivers/block/Makefile b/drivers/block/Makefile index 40528ba..76646e9 100644 --- a/drivers/block/Makefile +++ b/drivers/block/Makefile @@ -36,6 +36,7 @@ obj-$(CONFIG_BLK_DEV_UB) += ub.o obj-$(CONFIG_BLK_DEV_HD) += hd.o obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o +obj-$(CONFIG_XEN_BLKDEV_BACKEND) += xen-blkback/ obj-$(CONFIG_BLK_DEV_DRBD) += drbd/ obj-$(CONFIG_BLK_DEV_RBD) += rbd.o diff --git a/drivers/block/amiflop.c b/drivers/block/amiflop.c index 7888501..8eba86b 100644 --- a/drivers/block/amiflop.c +++ b/drivers/block/amiflop.c @@ -1658,12 +1658,12 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) } /* - * floppy-change is never called from an interrupt, so we can relax a bit + * check_events is never called from an interrupt, so we can relax a bit * here, sleep etc. Note that floppy-on tries to set current_DOR to point * to the desired drive, but it will probably not survive the sleep if * several floppies are used at the same time: thus the loop. */ -static int amiga_floppy_change(struct gendisk *disk) +static unsigned amiga_check_events(struct gendisk *disk, unsigned int clearing) { struct amiga_floppy_struct *p = disk->private_data; int drive = p - unit; @@ -1686,7 +1686,7 @@ static int amiga_floppy_change(struct gendisk *disk) p->dirty = 0; writepending = 0; /* if this was true before, too bad! */ writefromint = 0; - return 1; + return DISK_EVENT_MEDIA_CHANGE; } return 0; } @@ -1697,7 +1697,7 @@ static const struct block_device_operations floppy_fops = { .release = floppy_release, .ioctl = fd_ioctl, .getgeo = fd_getgeo, - .media_changed = amiga_floppy_change, + .check_events = amiga_check_events, }; static int __init fd_probe_drives(void) @@ -1768,8 +1768,8 @@ static int __init amiga_floppy_probe(struct platform_device *pdev) return -EBUSY; ret = -ENOMEM; - if ((raw_buf = (char *)amiga_chip_alloc (RAW_BUF_SIZE, "Floppy")) == - NULL) { + raw_buf = amiga_chip_alloc(RAW_BUF_SIZE, "Floppy"); + if (!raw_buf) { printk("fd: cannot get chip mem buffer\n"); goto out_blkdev; } diff --git a/drivers/block/ataflop.c b/drivers/block/ataflop.c index 605a67e..ede16c6 100644 --- a/drivers/block/ataflop.c +++ b/drivers/block/ataflop.c @@ -1324,23 +1324,24 @@ static void finish_fdc_done( int dummy ) * due to unrecognised disk changes. */ -static int check_floppy_change(struct gendisk *disk) +static unsigned int floppy_check_events(struct gendisk *disk, + unsigned int clearing) { struct atari_floppy_struct *p = disk->private_data; unsigned int drive = p - unit; if (test_bit (drive, &fake_change)) { /* simulated change (e.g. after formatting) */ - return 1; + return DISK_EVENT_MEDIA_CHANGE; } if (test_bit (drive, &changed_floppies)) { /* surely changed (the WP signal changed at least once) */ - return 1; + return DISK_EVENT_MEDIA_CHANGE; } if (UD.wpstat) { /* WP is on -> could be changed: to be sure, buffers should be * invalidated... */ - return 1; + return DISK_EVENT_MEDIA_CHANGE; } return 0; @@ -1570,7 +1571,7 @@ static int fd_locked_ioctl(struct block_device *bdev, fmode_t mode, * or the next access will revalidate - and clear UDT :-( */ - if (check_floppy_change(disk)) + if (floppy_check_events(disk, 0)) floppy_revalidate(disk); if (UD.flags & FTD_MSG) @@ -1904,7 +1905,7 @@ static const struct block_device_operations floppy_fops = { .open = floppy_unlocked_open, .release = floppy_release, .ioctl = fd_ioctl, - .media_changed = check_floppy_change, + .check_events = floppy_check_events, .revalidate_disk= floppy_revalidate, }; diff --git a/drivers/block/brd.c b/drivers/block/brd.c index b7f51e4..dba1c32 100644 --- a/drivers/block/brd.c +++ b/drivers/block/brd.c @@ -35,10 +35,6 @@ */ struct brd_device { int brd_number; - int brd_refcnt; - loff_t brd_offset; - loff_t brd_sizelimit; - unsigned brd_blocksize; struct request_queue *brd_queue; struct gendisk *brd_disk; @@ -440,11 +436,11 @@ static int rd_nr; int rd_size = CONFIG_BLK_DEV_RAM_SIZE; static int max_part; static int part_shift; -module_param(rd_nr, int, 0); +module_param(rd_nr, int, S_IRUGO); MODULE_PARM_DESC(rd_nr, "Maximum number of brd devices"); -module_param(rd_size, int, 0); +module_param(rd_size, int, S_IRUGO); MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes."); -module_param(max_part, int, 0); +module_param(max_part, int, S_IRUGO); MODULE_PARM_DESC(max_part, "Maximum number of partitions per RAM disk"); MODULE_LICENSE("GPL"); MODULE_ALIAS_BLOCKDEV_MAJOR(RAMDISK_MAJOR); @@ -552,7 +548,7 @@ static struct kobject *brd_probe(dev_t dev, int *part, void *data) struct kobject *kobj; mutex_lock(&brd_devices_mutex); - brd = brd_init_one(dev & MINORMASK); + brd = brd_init_one(MINOR(dev) >> part_shift); kobj = brd ? get_disk(brd->brd_disk) : ERR_PTR(-ENOMEM); mutex_unlock(&brd_devices_mutex); @@ -575,25 +571,39 @@ static int __init brd_init(void) * * (1) if rd_nr is specified, create that many upfront, and this * also becomes a hard limit. - * (2) if rd_nr is not specified, create 1 rd device on module - * load, user can further extend brd device by create dev node - * themselves and have kernel automatically instantiate actual - * device on-demand. + * (2) if rd_nr is not specified, create CONFIG_BLK_DEV_RAM_COUNT + * (default 16) rd device on module load, user can further + * extend brd device by create dev node themselves and have + * kernel automatically instantiate actual device on-demand. */ part_shift = 0; - if (max_part > 0) + if (max_part > 0) { part_shift = fls(max_part); + /* + * Adjust max_part according to part_shift as it is exported + * to user space so that user can decide correct minor number + * if [s]he want to create more devices. + * + * Note that -1 is required because partition 0 is reserved + * for the whole disk. + */ + max_part = (1UL << part_shift) - 1; + } + + if ((1UL << part_shift) > DISK_MAX_PARTS) + return -EINVAL; + if (rd_nr > 1UL << (MINORBITS - part_shift)) return -EINVAL; if (rd_nr) { nr = rd_nr; - range = rd_nr; + range = rd_nr << part_shift; } else { nr = CONFIG_BLK_DEV_RAM_COUNT; - range = 1UL << (MINORBITS - part_shift); + range = 1UL << MINORBITS; } if (register_blkdev(RAMDISK_MAJOR, "ramdisk")) @@ -632,7 +642,7 @@ static void __exit brd_exit(void) unsigned long range; struct brd_device *brd, *next; - range = rd_nr ? rd_nr : 1UL << (MINORBITS - part_shift); + range = rd_nr ? rd_nr << part_shift : 1UL << MINORBITS; list_for_each_entry_safe(brd, next, &brd_devices, brd_list) brd_del_one(brd); diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c index 9279272..8f4ef65 100644 --- a/drivers/block/cciss.c +++ b/drivers/block/cciss.c @@ -64,6 +64,10 @@ MODULE_DESCRIPTION("Driver for HP Smart Array Controllers"); MODULE_SUPPORTED_DEVICE("HP Smart Array Controllers"); MODULE_VERSION("3.6.26"); MODULE_LICENSE("GPL"); +static int cciss_tape_cmds = 6; +module_param(cciss_tape_cmds, int, 0644); +MODULE_PARM_DESC(cciss_tape_cmds, + "number of commands to allocate for tape devices (default: 6)"); static DEFINE_MUTEX(cciss_mutex); static struct proc_dir_entry *proc_cciss; @@ -193,7 +197,9 @@ static int __devinit cciss_find_cfg_addrs(struct pci_dev *pdev, u64 *cfg_offset); static int __devinit cciss_pci_find_memory_BAR(struct pci_dev *pdev, unsigned long *memory_bar); - +static inline u32 cciss_tag_discard_error_bits(ctlr_info_t *h, u32 tag); +static __devinit int write_driver_ver_to_cfgtable( + CfgTable_struct __iomem *cfgtable); /* performant mode helper functions */ static void calc_bucket_map(int *bucket, int num_buckets, int nsgs, @@ -231,7 +237,7 @@ static const struct block_device_operations cciss_fops = { */ static void set_performant_mode(ctlr_info_t *h, CommandList_struct *c) { - if (likely(h->transMethod == CFGTBL_Trans_Performant)) + if (likely(h->transMethod & CFGTBL_Trans_Performant)) c->busaddr |= 1 | (h->blockFetchTable[c->Header.SGList] << 1); } @@ -556,6 +562,66 @@ static void __devinit cciss_procinit(ctlr_info_t *h) #define to_hba(n) container_of(n, struct ctlr_info, dev) #define to_drv(n) container_of(n, drive_info_struct, dev) +/* List of controllers which cannot be hard reset on kexec with reset_devices */ +static u32 unresettable_controller[] = { + 0x324a103C, /* Smart Array P712m */ + 0x324b103C, /* SmartArray P711m */ + 0x3223103C, /* Smart Array P800 */ + 0x3234103C, /* Smart Array P400 */ + 0x3235103C, /* Smart Array P400i */ + 0x3211103C, /* Smart Array E200i */ + 0x3212103C, /* Smart Array E200 */ + 0x3213103C, /* Smart Array E200i */ + 0x3214103C, /* Smart Array E200i */ + 0x3215103C, /* Smart Array E200i */ + 0x3237103C, /* Smart Array E500 */ + 0x323D103C, /* Smart Array P700m */ + 0x409C0E11, /* Smart Array 6400 */ + 0x409D0E11, /* Smart Array 6400 EM */ +}; + +/* List of controllers which cannot even be soft reset */ +static u32 soft_unresettable_controller[] = { + 0x409C0E11, /* Smart Array 6400 */ + 0x409D0E11, /* Smart Array 6400 EM */ +}; + +static int ctlr_is_hard_resettable(u32 board_id) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(unresettable_controller); i++) + if (unresettable_controller[i] == board_id) + return 0; + return 1; +} + +static int ctlr_is_soft_resettable(u32 board_id) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(soft_unresettable_controller); i++) + if (soft_unresettable_controller[i] == board_id) + return 0; + return 1; +} + +static int ctlr_is_resettable(u32 board_id) +{ + return ctlr_is_hard_resettable(board_id) || + ctlr_is_soft_resettable(board_id); +} + +static ssize_t host_show_resettable(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct ctlr_info *h = to_hba(dev); + + return snprintf(buf, 20, "%d\n", ctlr_is_resettable(h->board_id)); +} +static DEVICE_ATTR(resettable, S_IRUGO, host_show_resettable, NULL); + static ssize_t host_store_rescan(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) @@ -741,6 +807,7 @@ static DEVICE_ATTR(usage_count, S_IRUGO, cciss_show_usage_count, NULL); static struct attribute *cciss_host_attrs[] = { &dev_attr_rescan.attr, + &dev_attr_resettable.attr, NULL }; @@ -973,8 +1040,8 @@ static void cmd_special_free(ctlr_info_t *h, CommandList_struct *c) temp64.val32.upper = c->ErrDesc.Addr.upper; pci_free_consistent(h->pdev, sizeof(ErrorInfo_struct), c->err_info, (dma_addr_t) temp64.val); - pci_free_consistent(h->pdev, sizeof(CommandList_struct), - c, (dma_addr_t) c->busaddr); + pci_free_consistent(h->pdev, sizeof(CommandList_struct), c, + (dma_addr_t) cciss_tag_discard_error_bits(h, (u32) c->busaddr)); } static inline ctlr_info_t *get_host(struct gendisk *disk) @@ -1490,8 +1557,7 @@ static int cciss_bigpassthru(ctlr_info_t *h, void __user *argp) return -EINVAL; if (!capable(CAP_SYS_RAWIO)) return -EPERM; - ioc = (BIG_IOCTL_Command_struct *) - kmalloc(sizeof(*ioc), GFP_KERNEL); + ioc = kmalloc(sizeof(*ioc), GFP_KERNEL); if (!ioc) { status = -ENOMEM; goto cleanup1; @@ -2529,7 +2595,7 @@ static int fill_cmd(ctlr_info_t *h, CommandList_struct *c, __u8 cmd, void *buff, } } else if (cmd_type == TYPE_MSG) { switch (cmd) { - case 0: /* ABORT message */ + case CCISS_ABORT_MSG: c->Request.CDBLen = 12; c->Request.Type.Attribute = ATTR_SIMPLE; c->Request.Type.Direction = XFER_WRITE; @@ -2539,16 +2605,16 @@ static int fill_cmd(ctlr_info_t *h, CommandList_struct *c, __u8 cmd, void *buff, /* buff contains the tag of the command to abort */ memcpy(&c->Request.CDB[4], buff, 8); break; - case 1: /* RESET message */ + case CCISS_RESET_MSG: c->Request.CDBLen = 16; c->Request.Type.Attribute = ATTR_SIMPLE; c->Request.Type.Direction = XFER_NONE; c->Request.Timeout = 0; memset(&c->Request.CDB[0], 0, sizeof(c->Request.CDB)); c->Request.CDB[0] = cmd; /* reset */ - c->Request.CDB[1] = 0x03; /* reset a target */ + c->Request.CDB[1] = CCISS_RESET_TYPE_TARGET; break; - case 3: /* No-Op message */ + case CCISS_NOOP_MSG: c->Request.CDBLen = 1; c->Request.Type.Attribute = ATTR_SIMPLE; c->Request.Type.Direction = XFER_WRITE; @@ -2577,6 +2643,31 @@ static int fill_cmd(ctlr_info_t *h, CommandList_struct *c, __u8 cmd, void *buff, return status; } +static int __devinit cciss_send_reset(ctlr_info_t *h, unsigned char *scsi3addr, + u8 reset_type) +{ + CommandList_struct *c; + int return_status; + + c = cmd_alloc(h); + if (!c) + return -ENOMEM; + return_status = fill_cmd(h, c, CCISS_RESET_MSG, NULL, 0, 0, + CTLR_LUNID, TYPE_MSG); + c->Request.CDB[1] = reset_type; /* fill_cmd defaults to target reset */ + if (return_status != IO_OK) { + cmd_special_free(h, c); + return return_status; + } + c->waiting = NULL; + enqueue_cmd_and_start_io(h, c); + /* Don't wait for completion, the reset won't complete. Don't free + * the command either. This is the last command we will send before + * re-initializing everything, so it doesn't matter and won't leak. + */ + return 0; +} + static int check_target_status(ctlr_info_t *h, CommandList_struct *c) { switch (c->err_info->ScsiStatus) { @@ -2653,6 +2744,10 @@ static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c) c->Request.CDB[0]); return_status = IO_NEEDS_RETRY; break; + case CMD_UNABORTABLE: + dev_warn(&h->pdev->dev, "cmd unabortable\n"); + return_status = IO_ERROR; + break; default: dev_warn(&h->pdev->dev, "cmd 0x%02x returned " "unknown status %x\n", c->Request.CDB[0], @@ -3103,6 +3198,13 @@ static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd, (cmd->rq->cmd_type == REQ_TYPE_BLOCK_PC) ? DID_PASSTHROUGH : DID_ERROR); break; + case CMD_UNABORTABLE: + dev_warn(&h->pdev->dev, "cmd %p unabortable\n", cmd); + rq->errors = make_status_bytes(SAM_STAT_GOOD, + cmd->err_info->CommandStatus, DRIVER_OK, + cmd->rq->cmd_type == REQ_TYPE_BLOCK_PC ? + DID_PASSTHROUGH : DID_ERROR); + break; default: dev_warn(&h->pdev->dev, "cmd %p returned " "unknown status %x\n", cmd, @@ -3136,10 +3238,13 @@ static inline u32 cciss_tag_to_index(u32 tag) return tag >> DIRECT_LOOKUP_SHIFT; } -static inline u32 cciss_tag_discard_error_bits(u32 tag) +static inline u32 cciss_tag_discard_error_bits(ctlr_info_t *h, u32 tag) { -#define CCISS_ERROR_BITS 0x03 - return tag & ~CCISS_ERROR_BITS; +#define CCISS_PERF_ERROR_BITS ((1 << DIRECT_LOOKUP_SHIFT) - 1) +#define CCISS_SIMPLE_ERROR_BITS 0x03 + if (likely(h->transMethod & CFGTBL_Trans_Performant)) + return tag & ~CCISS_PERF_ERROR_BITS; + return tag & ~CCISS_SIMPLE_ERROR_BITS; } static inline void cciss_mark_tag_indexed(u32 *tag) @@ -3170,12 +3275,6 @@ static void do_cciss_request(struct request_queue *q) int sg_index = 0; int chained = 0; - /* We call start_io here in case there is a command waiting on the - * queue that has not been sent. - */ - if (blk_queue_plugged(q)) - goto startio; - queue: creq = blk_peek_request(q); if (!creq) @@ -3365,7 +3464,7 @@ static inline u32 next_command(ctlr_info_t *h) { u32 a; - if (unlikely(h->transMethod != CFGTBL_Trans_Performant)) + if (unlikely(!(h->transMethod & CFGTBL_Trans_Performant))) return h->access.command_completed(h); if ((*(h->reply_pool_head) & 1) == (h->reply_pool_wraparound)) { @@ -3400,14 +3499,12 @@ static inline u32 process_indexed_cmd(ctlr_info_t *h, u32 raw_tag) /* process completion of a non-indexed command */ static inline u32 process_nonindexed_cmd(ctlr_info_t *h, u32 raw_tag) { - u32 tag; CommandList_struct *c = NULL; __u32 busaddr_masked, tag_masked; - tag = cciss_tag_discard_error_bits(raw_tag); + tag_masked = cciss_tag_discard_error_bits(h, raw_tag); list_for_each_entry(c, &h->cmpQ, list) { - busaddr_masked = cciss_tag_discard_error_bits(c->busaddr); - tag_masked = cciss_tag_discard_error_bits(tag); + busaddr_masked = cciss_tag_discard_error_bits(h, c->busaddr); if (busaddr_masked == tag_masked) { finish_cmd(h, c, raw_tag); return next_command(h); @@ -3417,6 +3514,63 @@ static inline u32 process_nonindexed_cmd(ctlr_info_t *h, u32 raw_tag) return next_command(h); } +/* Some controllers, like p400, will give us one interrupt + * after a soft reset, even if we turned interrupts off. + * Only need to check for this in the cciss_xxx_discard_completions + * functions. + */ +static int ignore_bogus_interrupt(ctlr_info_t *h) +{ + if (likely(!reset_devices)) + return 0; + + if (likely(h->interrupts_enabled)) + return 0; + + dev_info(&h->pdev->dev, "Received interrupt while interrupts disabled " + "(known firmware bug.) Ignoring.\n"); + + return 1; +} + +static irqreturn_t cciss_intx_discard_completions(int irq, void *dev_id) +{ + ctlr_info_t *h = dev_id; + unsigned long flags; + u32 raw_tag; + + if (ignore_bogus_interrupt(h)) + return IRQ_NONE; + + if (interrupt_not_for_us(h)) + return IRQ_NONE; + spin_lock_irqsave(&h->lock, flags); + while (interrupt_pending(h)) { + raw_tag = get_next_completion(h); + while (raw_tag != FIFO_EMPTY) + raw_tag = next_command(h); + } + spin_unlock_irqrestore(&h->lock, flags); + return IRQ_HANDLED; +} + +static irqreturn_t cciss_msix_discard_completions(int irq, void *dev_id) +{ + ctlr_info_t *h = dev_id; + unsigned long flags; + u32 raw_tag; + + if (ignore_bogus_interrupt(h)) + return IRQ_NONE; + + spin_lock_irqsave(&h->lock, flags); + raw_tag = get_next_completion(h); + while (raw_tag != FIFO_EMPTY) + raw_tag = next_command(h); + spin_unlock_irqrestore(&h->lock, flags); + return IRQ_HANDLED; +} + static irqreturn_t do_cciss_intx(int irq, void *dev_id) { ctlr_info_t *h = dev_id; @@ -3759,7 +3913,8 @@ static void __devinit cciss_wait_for_mode_change_ack(ctlr_info_t *h) } } -static __devinit void cciss_enter_performant_mode(ctlr_info_t *h) +static __devinit void cciss_enter_performant_mode(ctlr_info_t *h, + u32 use_short_tags) { /* This is a bit complicated. There are 8 registers on * the controller which we write to to tell it 8 different @@ -3814,7 +3969,7 @@ static __devinit void cciss_enter_performant_mode(ctlr_info_t *h) writel(0, &h->transtable->RepQCtrAddrHigh32); writel(h->reply_pool_dhandle, &h->transtable->RepQAddr0Low32); writel(0, &h->transtable->RepQAddr0High32); - writel(CFGTBL_Trans_Performant, + writel(CFGTBL_Trans_Performant | use_short_tags, &(h->cfgtable->HostWrite.TransportRequest)); writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); @@ -3861,7 +4016,8 @@ static void __devinit cciss_put_controller_into_performant_mode(ctlr_info_t *h) if ((h->reply_pool == NULL) || (h->blockFetchTable == NULL)) goto clean_up; - cciss_enter_performant_mode(h); + cciss_enter_performant_mode(h, + trans_support & CFGTBL_Trans_use_short_tags); /* Change the access methods to the performant access methods */ h->access = SA5_performant_access; @@ -4032,6 +4188,9 @@ static int __devinit cciss_find_cfgtables(ctlr_info_t *h) cfg_base_addr_index) + cfg_offset, sizeof(h->cfgtable)); if (!h->cfgtable) return -ENOMEM; + rc = write_driver_ver_to_cfgtable(h->cfgtable); + if (rc) + return rc; /* Find performant mode table. */ trans_offset = readl(&h->cfgtable->TransMethodOffset); h->transtable = remap_pci_mem(pci_resource_start(h->pdev, @@ -4066,7 +4225,7 @@ static void __devinit cciss_get_max_perf_mode_cmds(struct ctlr_info *h) static void __devinit cciss_find_board_params(ctlr_info_t *h) { cciss_get_max_perf_mode_cmds(h); - h->nr_cmds = h->max_commands - 4; /* Allow room for some ioctls */ + h->nr_cmds = h->max_commands - 4 - cciss_tape_cmds; h->maxsgentries = readl(&(h->cfgtable->MaxSGElements)); /* * Limit in-command s/g elements to 32 save dma'able memory. @@ -4302,7 +4461,7 @@ static __devinit int cciss_message(struct pci_dev *pdev, unsigned char opcode, u tag = readl(vaddr + SA5_REPLY_PORT_OFFSET); if ((tag & ~3) == paddr32) break; - schedule_timeout_uninterruptible(HZ); + msleep(CCISS_POST_RESET_NOOP_TIMEOUT_MSECS); } iounmap(vaddr); @@ -4329,11 +4488,10 @@ static __devinit int cciss_message(struct pci_dev *pdev, unsigned char opcode, u return 0; } -#define cciss_soft_reset_controller(p) cciss_message(p, 1, 0) #define cciss_noop(p) cciss_message(p, 3, 0) static int cciss_controller_hard_reset(struct pci_dev *pdev, - void * __iomem vaddr, bool use_doorbell) + void * __iomem vaddr, u32 use_doorbell) { u16 pmcsr; int pos; @@ -4344,8 +4502,7 @@ static int cciss_controller_hard_reset(struct pci_dev *pdev, * other way using the doorbell register. */ dev_info(&pdev->dev, "using doorbell to reset controller\n"); - writel(DOORBELL_CTLR_RESET, vaddr + SA5_DOORBELL); - msleep(1000); + writel(use_doorbell, vaddr + SA5_DOORBELL); } else { /* Try to do it the PCI power state way */ /* Quoting from the Open CISS Specification: "The Power @@ -4376,12 +4533,64 @@ static int cciss_controller_hard_reset(struct pci_dev *pdev, pmcsr &= ~PCI_PM_CTRL_STATE_MASK; pmcsr |= PCI_D0; pci_write_config_word(pdev, pos + PCI_PM_CTRL, pmcsr); - - msleep(500); } return 0; } +static __devinit void init_driver_version(char *driver_version, int len) +{ + memset(driver_version, 0, len); + strncpy(driver_version, "cciss " DRIVER_NAME, len - 1); +} + +static __devinit int write_driver_ver_to_cfgtable( + CfgTable_struct __iomem *cfgtable) +{ + char *driver_version; + int i, size = sizeof(cfgtable->driver_version); + + driver_version = kmalloc(size, GFP_KERNEL); + if (!driver_version) + return -ENOMEM; + + init_driver_version(driver_version, size); + for (i = 0; i < size; i++) + writeb(driver_version[i], &cfgtable->driver_version[i]); + kfree(driver_version); + return 0; +} + +static __devinit void read_driver_ver_from_cfgtable( + CfgTable_struct __iomem *cfgtable, unsigned char *driver_ver) +{ + int i; + + for (i = 0; i < sizeof(cfgtable->driver_version); i++) + driver_ver[i] = readb(&cfgtable->driver_version[i]); +} + +static __devinit int controller_reset_failed( + CfgTable_struct __iomem *cfgtable) +{ + + char *driver_ver, *old_driver_ver; + int rc, size = sizeof(cfgtable->driver_version); + + old_driver_ver = kmalloc(2 * size, GFP_KERNEL); + if (!old_driver_ver) + return -ENOMEM; + driver_ver = old_driver_ver + size; + + /* After a reset, the 32 bytes of "driver version" in the cfgtable + * should have been changed, otherwise we know the reset failed. + */ + init_driver_version(old_driver_ver, size); + read_driver_ver_from_cfgtable(cfgtable, driver_ver); + rc = !memcmp(driver_ver, old_driver_ver, size); + kfree(old_driver_ver); + return rc; +} + /* This does a hard reset of the controller using PCI power management * states or using the doorbell register. */ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) @@ -4391,10 +4600,10 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) u64 cfg_base_addr_index; void __iomem *vaddr; unsigned long paddr; - u32 misc_fw_support, active_transport; + u32 misc_fw_support; int rc; CfgTable_struct __iomem *cfgtable; - bool use_doorbell; + u32 use_doorbell; u32 board_id; u16 command_register; @@ -4418,12 +4627,16 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) * likely not be happy. Just forbid resetting this conjoined mess. */ cciss_lookup_board_id(pdev, &board_id); - if (board_id == 0x409C0E11 || board_id == 0x409D0E11) { + if (!ctlr_is_resettable(board_id)) { dev_warn(&pdev->dev, "Cannot reset Smart Array 640x " "due to shared cache module."); return -ENODEV; } + /* if controller is soft- but not hard resettable... */ + if (!ctlr_is_hard_resettable(board_id)) + return -ENOTSUPP; /* try soft reset later. */ + /* Save the PCI command register */ pci_read_config_word(pdev, 4, &command_register); /* Turn the board off. This is so that later pci_restore_state() @@ -4451,16 +4664,28 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) rc = -ENOMEM; goto unmap_vaddr; } + rc = write_driver_ver_to_cfgtable(cfgtable); + if (rc) + goto unmap_vaddr; - /* If reset via doorbell register is supported, use that. */ - misc_fw_support = readl(&cfgtable->misc_fw_support); - use_doorbell = misc_fw_support & MISC_FW_DOORBELL_RESET; - - /* The doorbell reset seems to cause lockups on some Smart - * Arrays (e.g. P410, P410i, maybe others). Until this is - * fixed or at least isolated, avoid the doorbell reset. + /* If reset via doorbell register is supported, use that. + * There are two such methods. Favor the newest method. */ - use_doorbell = 0; + misc_fw_support = readl(&cfgtable->misc_fw_support); + use_doorbell = misc_fw_support & MISC_FW_DOORBELL_RESET2; + if (use_doorbell) { + use_doorbell = DOORBELL_CTLR_RESET2; + } else { + use_doorbell = misc_fw_support & MISC_FW_DOORBELL_RESET; + if (use_doorbell) { + dev_warn(&pdev->dev, "Controller claims that " + "'Bit 2 doorbell reset' is " + "supported, but not 'bit 5 doorbell reset'. " + "Firmware update is recommended.\n"); + rc = -ENOTSUPP; /* use the soft reset */ + goto unmap_cfgtable; + } + } rc = cciss_controller_hard_reset(pdev, vaddr, use_doorbell); if (rc) @@ -4478,30 +4703,31 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) msleep(CCISS_POST_RESET_PAUSE_MSECS); /* Wait for board to become not ready, then ready. */ - dev_info(&pdev->dev, "Waiting for board to become ready.\n"); + dev_info(&pdev->dev, "Waiting for board to reset.\n"); rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_NOT_READY); - if (rc) /* Don't bail, might be E500, etc. which can't be reset */ - dev_warn(&pdev->dev, - "failed waiting for board to become not ready\n"); + if (rc) { + dev_warn(&pdev->dev, "Failed waiting for board to hard reset." + " Will try soft reset.\n"); + rc = -ENOTSUPP; /* Not expected, but try soft reset later */ + goto unmap_cfgtable; + } rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_READY); if (rc) { dev_warn(&pdev->dev, - "failed waiting for board to become ready\n"); + "failed waiting for board to become ready " + "after hard reset\n"); goto unmap_cfgtable; } - dev_info(&pdev->dev, "board ready.\n"); - /* Controller should be in simple mode at this point. If it's not, - * It means we're on one of those controllers which doesn't support - * the doorbell reset method and on which the PCI power management reset - * method doesn't work (P800, for example.) - * In those cases, don't try to proceed, as it generally doesn't work. - */ - active_transport = readl(&cfgtable->TransportActive); - if (active_transport & PERFORMANT_MODE) { - dev_warn(&pdev->dev, "Unable to successfully reset controller," - " Ignoring controller.\n"); - rc = -ENODEV; + rc = controller_reset_failed(vaddr); + if (rc < 0) + goto unmap_cfgtable; + if (rc) { + dev_warn(&pdev->dev, "Unable to successfully hard reset " + "controller. Will try soft reset.\n"); + rc = -ENOTSUPP; /* Not expected, but try soft reset later */ + } else { + dev_info(&pdev->dev, "Board ready after hard reset.\n"); } unmap_cfgtable: @@ -4528,11 +4754,12 @@ static __devinit int cciss_init_reset_devices(struct pci_dev *pdev) * due to concerns about shared bbwc between 6402/6404 pair. */ if (rc == -ENOTSUPP) - return 0; /* just try to do the kdump anyhow. */ + return rc; /* just try to do the kdump anyhow. */ if (rc) return -ENODEV; /* Now try to get the controller to respond to a no-op */ + dev_warn(&pdev->dev, "Waiting for controller to respond to no-op\n"); for (i = 0; i < CCISS_POST_RESET_NOOP_RETRIES; i++) { if (cciss_noop(pdev) == 0) break; @@ -4545,6 +4772,148 @@ static __devinit int cciss_init_reset_devices(struct pci_dev *pdev) return 0; } +static __devinit int cciss_allocate_cmd_pool(ctlr_info_t *h) +{ + h->cmd_pool_bits = kmalloc( + DIV_ROUND_UP(h->nr_cmds, BITS_PER_LONG) * + sizeof(unsigned long), GFP_KERNEL); + h->cmd_pool = pci_alloc_consistent(h->pdev, + h->nr_cmds * sizeof(CommandList_struct), + &(h->cmd_pool_dhandle)); + h->errinfo_pool = pci_alloc_consistent(h->pdev, + h->nr_cmds * sizeof(ErrorInfo_struct), + &(h->errinfo_pool_dhandle)); + if ((h->cmd_pool_bits == NULL) + || (h->cmd_pool == NULL) + || (h->errinfo_pool == NULL)) { + dev_err(&h->pdev->dev, "out of memory"); + return -ENOMEM; + } + return 0; +} + +static __devinit int cciss_allocate_scatterlists(ctlr_info_t *h) +{ + int i; + + /* zero it, so that on free we need not know how many were alloc'ed */ + h->scatter_list = kzalloc(h->max_commands * + sizeof(struct scatterlist *), GFP_KERNEL); + if (!h->scatter_list) + return -ENOMEM; + + for (i = 0; i < h->nr_cmds; i++) { + h->scatter_list[i] = kmalloc(sizeof(struct scatterlist) * + h->maxsgentries, GFP_KERNEL); + if (h->scatter_list[i] == NULL) { + dev_err(&h->pdev->dev, "could not allocate " + "s/g lists\n"); + return -ENOMEM; + } + } + return 0; +} + +static void cciss_free_scatterlists(ctlr_info_t *h) +{ + int i; + + if (h->scatter_list) { + for (i = 0; i < h->nr_cmds; i++) + kfree(h->scatter_list[i]); + kfree(h->scatter_list); + } +} + +static void cciss_free_cmd_pool(ctlr_info_t *h) +{ + kfree(h->cmd_pool_bits); + if (h->cmd_pool) + pci_free_consistent(h->pdev, + h->nr_cmds * sizeof(CommandList_struct), + h->cmd_pool, h->cmd_pool_dhandle); + if (h->errinfo_pool) + pci_free_consistent(h->pdev, + h->nr_cmds * sizeof(ErrorInfo_struct), + h->errinfo_pool, h->errinfo_pool_dhandle); +} + +static int cciss_request_irq(ctlr_info_t *h, + irqreturn_t (*msixhandler)(int, void *), + irqreturn_t (*intxhandler)(int, void *)) +{ + if (h->msix_vector || h->msi_vector) { + if (!request_irq(h->intr[PERF_MODE_INT], msixhandler, + IRQF_DISABLED, h->devname, h)) + return 0; + dev_err(&h->pdev->dev, "Unable to get msi irq %d" + " for %s\n", h->intr[PERF_MODE_INT], + h->devname); + return -1; + } + + if (!request_irq(h->intr[PERF_MODE_INT], intxhandler, + IRQF_DISABLED, h->devname, h)) + return 0; + dev_err(&h->pdev->dev, "Unable to get irq %d for %s\n", + h->intr[PERF_MODE_INT], h->devname); + return -1; +} + +static int __devinit cciss_kdump_soft_reset(ctlr_info_t *h) +{ + if (cciss_send_reset(h, CTLR_LUNID, CCISS_RESET_TYPE_CONTROLLER)) { + dev_warn(&h->pdev->dev, "Resetting array controller failed.\n"); + return -EIO; + } + + dev_info(&h->pdev->dev, "Waiting for board to soft reset.\n"); + if (cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_NOT_READY)) { + dev_warn(&h->pdev->dev, "Soft reset had no effect.\n"); + return -1; + } + + dev_info(&h->pdev->dev, "Board reset, awaiting READY status.\n"); + if (cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_READY)) { + dev_warn(&h->pdev->dev, "Board failed to become ready " + "after soft reset.\n"); + return -1; + } + + return 0; +} + +static void cciss_undo_allocations_after_kdump_soft_reset(ctlr_info_t *h) +{ + int ctlr = h->ctlr; + + free_irq(h->intr[PERF_MODE_INT], h); +#ifdef CONFIG_PCI_MSI + if (h->msix_vector) + pci_disable_msix(h->pdev); + else if (h->msi_vector) + pci_disable_msi(h->pdev); +#endif /* CONFIG_PCI_MSI */ + cciss_free_sg_chain_blocks(h->cmd_sg_list, h->nr_cmds); + cciss_free_scatterlists(h); + cciss_free_cmd_pool(h); + kfree(h->blockFetchTable); + if (h->reply_pool) + pci_free_consistent(h->pdev, h->max_commands * sizeof(__u64), + h->reply_pool, h->reply_pool_dhandle); + if (h->transtable) + iounmap(h->transtable); + if (h->cfgtable) + iounmap(h->cfgtable); + if (h->vaddr) + iounmap(h->vaddr); + unregister_blkdev(h->major, h->devname); + cciss_destroy_hba_sysfs_entry(h); + pci_release_regions(h->pdev); + kfree(h); + hba[ctlr] = NULL; +} + /* * This is it. Find all the controllers and register them. I really hate * stealing all these major device numbers. @@ -4555,15 +4924,28 @@ static int __devinit cciss_init_one(struct pci_dev *pdev, { int i; int j = 0; - int k = 0; int rc; + int try_soft_reset = 0; int dac, return_code; InquiryData_struct *inq_buff; ctlr_info_t *h; + unsigned long flags; rc = cciss_init_reset_devices(pdev); - if (rc) - return rc; + if (rc) { + if (rc != -ENOTSUPP) + return rc; + /* If the reset fails in a particular way (it has no way to do + * a proper hard reset, so returns -ENOTSUPP) we can try to do + * a soft reset once we get the controller configured up to the + * point that it can accept a command. + */ + try_soft_reset = 1; + rc = 0; + } + +reinit_after_soft_reset: + i = alloc_cciss_hba(pdev); if (i < 0) return -1; @@ -4581,6 +4963,11 @@ static int __devinit cciss_init_one(struct pci_dev *pdev, sprintf(h->devname, "cciss%d", i); h->ctlr = i; + if (cciss_tape_cmds < 2) + cciss_tape_cmds = 2; + if (cciss_tape_cmds > 16) + cciss_tape_cmds = 16; + init_completion(&h->scan_wait); if (cciss_create_hba_sysfs_entry(h)) @@ -4616,62 +5003,20 @@ static int __devinit cciss_init_one(struct pci_dev *pdev, /* make sure the board interrupts are off */ h->access.set_intr_mask(h, CCISS_INTR_OFF); - if (h->msi_vector || h->msix_vector) { - if (request_irq(h->intr[PERF_MODE_INT], - do_cciss_msix_intr, - IRQF_DISABLED, h->devname, h)) { - dev_err(&h->pdev->dev, "Unable to get irq %d for %s\n", - h->intr[PERF_MODE_INT], h->devname); - goto clean2; - } - } else { - if (request_irq(h->intr[PERF_MODE_INT], do_cciss_intx, - IRQF_DISABLED, h->devname, h)) { - dev_err(&h->pdev->dev, "Unable to get irq %d for %s\n", - h->intr[PERF_MODE_INT], h->devname); - goto clean2; - } - } + rc = cciss_request_irq(h, do_cciss_msix_intr, do_cciss_intx); + if (rc) + goto clean2; dev_info(&h->pdev->dev, "%s: <0x%x> at PCI %s IRQ %d%s using DAC\n", h->devname, pdev->device, pci_name(pdev), h->intr[PERF_MODE_INT], dac ? "" : " not"); - h->cmd_pool_bits = - kmalloc(DIV_ROUND_UP(h->nr_cmds, BITS_PER_LONG) - * sizeof(unsigned long), GFP_KERNEL); - h->cmd_pool = (CommandList_struct *) - pci_alloc_consistent(h->pdev, - h->nr_cmds * sizeof(CommandList_struct), - &(h->cmd_pool_dhandle)); - h->errinfo_pool = (ErrorInfo_struct *) - pci_alloc_consistent(h->pdev, - h->nr_cmds * sizeof(ErrorInfo_struct), - &(h->errinfo_pool_dhandle)); - if ((h->cmd_pool_bits == NULL) - || (h->cmd_pool == NULL) - || (h->errinfo_pool == NULL)) { - dev_err(&h->pdev->dev, "out of memory"); + if (cciss_allocate_cmd_pool(h)) goto clean4; - } - /* Need space for temp scatter list */ - h->scatter_list = kmalloc(h->max_commands * - sizeof(struct scatterlist *), - GFP_KERNEL); - if (!h->scatter_list) + if (cciss_allocate_scatterlists(h)) goto clean4; - for (k = 0; k < h->nr_cmds; k++) { - h->scatter_list[k] = kmalloc(sizeof(struct scatterlist) * - h->maxsgentries, - GFP_KERNEL); - if (h->scatter_list[k] == NULL) { - dev_err(&h->pdev->dev, - "could not allocate s/g lists\n"); - goto clean4; - } - } h->cmd_sg_list = cciss_allocate_sg_chain_blocks(h, h->chainsize, h->nr_cmds); if (!h->cmd_sg_list && h->chainsize > 0) @@ -4695,6 +5040,62 @@ static int __devinit cciss_init_one(struct pci_dev *pdev, h->gendisk[j] = NULL; } + /* At this point, the controller is ready to take commands. + * Now, if reset_devices and the hard reset didn't work, try + * the soft reset and see if that works. + */ + if (try_soft_reset) { + + /* This is kind of gross. We may or may not get a completion + * from the soft reset command, and if we do, then the value + * from the fifo may or may not be valid. So, we wait 10 secs + * after the reset throwing away any completions we get during + * that time. Unregister the interrupt handler and register + * fake ones to scoop up any residual completions. + */ + spin_lock_irqsave(&h->lock, flags); + h->access.set_intr_mask(h, CCISS_INTR_OFF); + spin_unlock_irqrestore(&h->lock, flags); + free_irq(h->intr[PERF_MODE_INT], h); + rc = cciss_request_irq(h, cciss_msix_discard_completions, + cciss_intx_discard_completions); + if (rc) { + dev_warn(&h->pdev->dev, "Failed to request_irq after " + "soft reset.\n"); + goto clean4; + } + + rc = cciss_kdump_soft_reset(h); + if (rc) { + dev_warn(&h->pdev->dev, "Soft reset failed.\n"); + goto clean4; + } + + dev_info(&h->pdev->dev, "Board READY.\n"); + dev_info(&h->pdev->dev, + "Waiting for stale completions to drain.\n"); + h->access.set_intr_mask(h, CCISS_INTR_ON); + msleep(10000); + h->access.set_intr_mask(h, CCISS_INTR_OFF); + + rc = controller_reset_failed(h->cfgtable); + if (rc) + dev_info(&h->pdev->dev, + "Soft reset appears to have failed.\n"); + + /* since the controller's reset, we have to go back and re-init + * everything. Easiest to just forget what we've done and do it + * all over again. + */ + cciss_undo_allocations_after_kdump_soft_reset(h); + try_soft_reset = 0; + if (rc) + /* don't go to clean4, we already unallocated */ + return -ENODEV; + + goto reinit_after_soft_reset; + } + cciss_scsi_setup(h); /* Turn the interrupts on so we can service requests */ @@ -4729,21 +5130,9 @@ static int __devinit cciss_init_one(struct pci_dev *pdev, return 1; clean4: - kfree(h->cmd_pool_bits); - /* Free up sg elements */ - for (k-- ; k >= 0; k--) - kfree(h->scatter_list[k]); - kfree(h->scatter_list); + cciss_free_cmd_pool(h); + cciss_free_scatterlists(h); cciss_free_sg_chain_blocks(h->cmd_sg_list, h->nr_cmds); - if (h->cmd_pool) - pci_free_consistent(h->pdev, - h->nr_cmds * sizeof(CommandList_struct), - h->cmd_pool, h->cmd_pool_dhandle); - if (h->errinfo_pool) - pci_free_consistent(h->pdev, - h->nr_cmds * sizeof(ErrorInfo_struct), - h->errinfo_pool, - h->errinfo_pool_dhandle); free_irq(h->intr[PERF_MODE_INT], h); clean2: unregister_blkdev(h->major, h->devname); @@ -4841,16 +5230,16 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev) iounmap(h->cfgtable); iounmap(h->vaddr); - pci_free_consistent(h->pdev, h->nr_cmds * sizeof(CommandList_struct), - h->cmd_pool, h->cmd_pool_dhandle); - pci_free_consistent(h->pdev, h->nr_cmds * sizeof(ErrorInfo_struct), - h->errinfo_pool, h->errinfo_pool_dhandle); - kfree(h->cmd_pool_bits); + cciss_free_cmd_pool(h); /* Free up sg elements */ for (j = 0; j < h->nr_cmds; j++) kfree(h->scatter_list[j]); kfree(h->scatter_list); cciss_free_sg_chain_blocks(h->cmd_sg_list, h->nr_cmds); + kfree(h->blockFetchTable); + if (h->reply_pool) + pci_free_consistent(h->pdev, h->max_commands * sizeof(__u64), + h->reply_pool, h->reply_pool_dhandle); /* * Deliberately omit pci_disable_device(): it does something nasty to * Smart Array controllers that pci_enable_device does not undo diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h index 579f749..16b4d58 100644 --- a/drivers/block/cciss.h +++ b/drivers/block/cciss.h @@ -200,7 +200,7 @@ struct ctlr_info * the above. */ #define CCISS_BOARD_READY_WAIT_SECS (120) -#define CCISS_BOARD_NOT_READY_WAIT_SECS (10) +#define CCISS_BOARD_NOT_READY_WAIT_SECS (100) #define CCISS_BOARD_READY_POLL_INTERVAL_MSECS (100) #define CCISS_BOARD_READY_ITERATIONS \ ((CCISS_BOARD_READY_WAIT_SECS * 1000) / \ @@ -209,8 +209,9 @@ struct ctlr_info ((CCISS_BOARD_NOT_READY_WAIT_SECS * 1000) / \ CCISS_BOARD_READY_POLL_INTERVAL_MSECS) #define CCISS_POST_RESET_PAUSE_MSECS (3000) -#define CCISS_POST_RESET_NOOP_INTERVAL_MSECS (1000) +#define CCISS_POST_RESET_NOOP_INTERVAL_MSECS (4000) #define CCISS_POST_RESET_NOOP_RETRIES (12) +#define CCISS_POST_RESET_NOOP_TIMEOUT_MSECS (10000) /* Send the command to the hardware @@ -222,6 +223,7 @@ static void SA5_submit_command( ctlr_info_t *h, CommandList_struct *c) h->ctlr, c->busaddr); #endif /* CCISS_DEBUG */ writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET); + readl(h->vaddr + SA5_REQUEST_PORT_OFFSET); h->commands_outstanding++; if ( h->commands_outstanding > h->max_outstanding) h->max_outstanding = h->commands_outstanding; @@ -238,11 +240,13 @@ static void SA5_intr_mask(ctlr_info_t *h, unsigned long val) { /* Turn interrupts on */ h->interrupts_enabled = 1; writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); } else /* Turn them off */ { h->interrupts_enabled = 0; writel( SA5_INTR_OFF, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); } } /* @@ -256,11 +260,13 @@ static void SA5B_intr_mask(ctlr_info_t *h, unsigned long val) { /* Turn interrupts on */ h->interrupts_enabled = 1; writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); } else /* Turn them off */ { h->interrupts_enabled = 0; writel( SA5B_INTR_OFF, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); } } @@ -270,10 +276,12 @@ static void SA5_performant_intr_mask(ctlr_info_t *h, unsigned long val) if (val) { /* turn on interrupts */ h->interrupts_enabled = 1; writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); } else { h->interrupts_enabled = 0; writel(SA5_PERF_INTR_OFF, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); + (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); } } diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h index 35463d2..d9be6b4 100644 --- a/drivers/block/cciss_cmd.h +++ b/drivers/block/cciss_cmd.h @@ -53,9 +53,11 @@ #define CFGTBL_ChangeReq 0x00000001l #define CFGTBL_AccCmds 0x00000001l #define DOORBELL_CTLR_RESET 0x00000004l +#define DOORBELL_CTLR_RESET2 0x00000020l #define CFGTBL_Trans_Simple 0x00000002l #define CFGTBL_Trans_Performant 0x00000004l +#define CFGTBL_Trans_use_short_tags 0x20000000l #define CFGTBL_BusType_Ultra2 0x00000001l #define CFGTBL_BusType_Ultra3 0x00000002l @@ -141,6 +143,14 @@ typedef struct _ReadCapdata_struct_16 #define BMIC_CACHE_FLUSH 0xc2 #define CCISS_CACHE_FLUSH 0x01 /* C2 was already being used by CCISS */ +#define CCISS_ABORT_MSG 0x00 +#define CCISS_RESET_MSG 0x01 +#define CCISS_RESET_TYPE_CONTROLLER 0x00 +#define CCISS_RESET_TYPE_BUS 0x01 +#define CCISS_RESET_TYPE_TARGET 0x03 +#define CCISS_RESET_TYPE_LUN 0x04 +#define CCISS_NOOP_MSG 0x03 + /* Command List Structure */ #define CTLR_LUNID "\0\0\0\0\0\0\0\0" @@ -234,6 +244,8 @@ typedef struct _CfgTable_struct { u8 reserved[0x78 - 0x58]; u32 misc_fw_support; /* offset 0x78 */ #define MISC_FW_DOORBELL_RESET (0x02) +#define MISC_FW_DOORBELL_RESET2 (0x10) + u8 driver_version[32]; } CfgTable_struct; struct TransTable_struct { diff --git a/drivers/block/cciss_scsi.c b/drivers/block/cciss_scsi.c index 727d022..6961002 100644 --- a/drivers/block/cciss_scsi.c +++ b/drivers/block/cciss_scsi.c @@ -84,7 +84,6 @@ static struct scsi_host_template cciss_driver_template = { .proc_name = "cciss", .proc_info = cciss_scsi_proc_info, .queuecommand = cciss_scsi_queue_command, - .can_queue = SCSI_CCISS_CAN_QUEUE, .this_id = 7, .cmd_per_lun = 1, .use_clustering = DISABLE_CLUSTERING, @@ -108,16 +107,13 @@ struct cciss_scsi_cmd_stack_elem_t { #pragma pack() -#define CMD_STACK_SIZE (SCSI_CCISS_CAN_QUEUE * \ - CCISS_MAX_SCSI_DEVS_PER_HBA + 2) - // plus two for init time usage - #pragma pack(1) struct cciss_scsi_cmd_stack_t { struct cciss_scsi_cmd_stack_elem_t *pool; - struct cciss_scsi_cmd_stack_elem_t *elem[CMD_STACK_SIZE]; + struct cciss_scsi_cmd_stack_elem_t **elem; dma_addr_t cmd_pool_handle; int top; + int nelems; }; #pragma pack() @@ -191,7 +187,7 @@ scsi_cmd_free(ctlr_info_t *h, CommandList_struct *c) sa = h->scsi_ctlr; stk = &sa->cmd_stack; stk->top++; - if (stk->top >= CMD_STACK_SIZE) { + if (stk->top >= stk->nelems) { dev_err(&h->pdev->dev, "scsi_cmd_free called too many times.\n"); BUG(); @@ -206,13 +202,14 @@ scsi_cmd_stack_setup(ctlr_info_t *h, struct cciss_scsi_adapter_data_t *sa) struct cciss_scsi_cmd_stack_t *stk; size_t size; + stk = &sa->cmd_stack; + stk->nelems = cciss_tape_cmds + 2; sa->cmd_sg_list = cciss_allocate_sg_chain_blocks(h, - h->chainsize, CMD_STACK_SIZE); + h->chainsize, stk->nelems); if (!sa->cmd_sg_list && h->chainsize > 0) return -ENOMEM; - stk = &sa->cmd_stack; - size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * CMD_STACK_SIZE; + size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * stk->nelems; /* Check alignment, see cciss_cmd.h near CommandList_struct def. */ BUILD_BUG_ON((sizeof(*stk->pool) % COMMANDLIST_ALIGNMENT) != 0); @@ -221,18 +218,23 @@ scsi_cmd_stack_setup(ctlr_info_t *h, struct cciss_scsi_adapter_data_t *sa) pci_alloc_consistent(h->pdev, size, &stk->cmd_pool_handle); if (stk->pool == NULL) { - cciss_free_sg_chain_blocks(sa->cmd_sg_list, CMD_STACK_SIZE); + cciss_free_sg_chain_blocks(sa->cmd_sg_list, stk->nelems); sa->cmd_sg_list = NULL; return -ENOMEM; } - - for (i=0; i<CMD_STACK_SIZE; i++) { + stk->elem = kmalloc(sizeof(stk->elem[0]) * stk->nelems, GFP_KERNEL); + if (!stk->elem) { + pci_free_consistent(h->pdev, size, stk->pool, + stk->cmd_pool_handle); + return -1; + } + for (i = 0; i < stk->nelems; i++) { stk->elem[i] = &stk->pool[i]; stk->elem[i]->busaddr = (__u32) (stk->cmd_pool_handle + (sizeof(struct cciss_scsi_cmd_stack_elem_t) * i)); stk->elem[i]->cmdindex = i; } - stk->top = CMD_STACK_SIZE-1; + stk->top = stk->nelems-1; return 0; } @@ -245,16 +247,18 @@ scsi_cmd_stack_free(ctlr_info_t *h) sa = h->scsi_ctlr; stk = &sa->cmd_stack; - if (stk->top != CMD_STACK_SIZE-1) { + if (stk->top != stk->nelems-1) { dev_warn(&h->pdev->dev, "bug: %d scsi commands are still outstanding.\n", - CMD_STACK_SIZE - stk->top); + stk->nelems - stk->top); } - size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * CMD_STACK_SIZE; + size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * stk->nelems; pci_free_consistent(h->pdev, size, stk->pool, stk->cmd_pool_handle); stk->pool = NULL; - cciss_free_sg_chain_blocks(sa->cmd_sg_list, CMD_STACK_SIZE); + cciss_free_sg_chain_blocks(sa->cmd_sg_list, stk->nelems); + kfree(stk->elem); + stk->elem = NULL; } #if 0 @@ -824,13 +828,18 @@ static void complete_scsi_command(CommandList_struct *c, int timeout, break; case CMD_UNSOLICITED_ABORT: cmd->result = DID_ABORT << 16; - dev_warn(&h->pdev->dev, "%p aborted do to an " + dev_warn(&h->pdev->dev, "%p aborted due to an " "unsolicited abort\n", c); break; case CMD_TIMEOUT: cmd->result = DID_TIME_OUT << 16; dev_warn(&h->pdev->dev, "%p timedout\n", c); break; + case CMD_UNABORTABLE: + cmd->result = DID_ERROR << 16; + dev_warn(&h->pdev->dev, "c %p command " + "unabortable\n", c); + break; default: cmd->result = DID_ERROR << 16; dev_warn(&h->pdev->dev, @@ -854,6 +863,7 @@ cciss_scsi_detect(ctlr_info_t *h) sh->io_port = 0; // good enough? FIXME, sh->n_io_port = 0; // I don't think we use these two... sh->this_id = SELF_SCSI_ID; + sh->can_queue = cciss_tape_cmds; sh->sg_tablesize = h->maxsgentries; sh->max_cmd_len = MAX_COMMAND_SIZE; @@ -1007,11 +1017,15 @@ cciss_scsi_interpret_error(ctlr_info_t *h, CommandList_struct *c) break; case CMD_UNSOLICITED_ABORT: dev_warn(&h->pdev->dev, - "%p aborted do to an unsolicited abort\n", c); + "%p aborted due to an unsolicited abort\n", c); break; case CMD_TIMEOUT: dev_warn(&h->pdev->dev, "%p timedout\n", c); break; + case CMD_UNABORTABLE: + dev_warn(&h->pdev->dev, + "%p unabortable\n", c); + break; default: dev_warn(&h->pdev->dev, "%p returned unknown status %x\n", diff --git a/drivers/block/cciss_scsi.h b/drivers/block/cciss_scsi.h index 6d5822f..e71d986 100644 --- a/drivers/block/cciss_scsi.h +++ b/drivers/block/cciss_scsi.h @@ -36,13 +36,9 @@ addressible natively, and may in fact turn out to be not scsi at all. */ -#define SCSI_CCISS_CAN_QUEUE 2 /* -Note, cmd_per_lun could give us some trouble, so I'm setting it very low. -Likewise, SCSI_CCISS_CAN_QUEUE is set very conservatively. - If the upper scsi layer tries to track how many commands we have outstanding, it will be operating under the misapprehension that it is the only one sending us requests. We also have the block interface, diff --git a/drivers/block/cpqarray.c b/drivers/block/cpqarray.c index 946dad4..b2fceb5 100644 --- a/drivers/block/cpqarray.c +++ b/drivers/block/cpqarray.c @@ -911,9 +911,6 @@ static void do_ida_request(struct request_queue *q) struct scatterlist tmp_sg[SG_MAX]; int i, dir, seg; - if (blk_queue_plugged(q)) - goto startio; - queue_next: creq = blk_peek_request(q); if (!creq) diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index ba95cba..09ef9a8 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -28,9 +28,9 @@ #include "drbd_int.h" #include "drbd_wrappers.h" -/* We maintain a trivial check sum in our on disk activity log. +/* We maintain a trivial checksum in our on disk activity log. * With that we can ensure correct operation even when the storage - * device might do a partial (last) sector write while loosing power. + * device might do a partial (last) sector write while losing power. */ struct __packed al_transaction { u32 magic; @@ -80,7 +80,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev, if ((rw & WRITE) && !test_bit(MD_NO_FUA, &mdev->flags)) rw |= REQ_FUA; - rw |= REQ_UNPLUG | REQ_SYNC; + rw |= REQ_SYNC; bio = bio_alloc(GFP_NOIO, 1); bio->bi_bdev = bdev->md_bdev; @@ -92,7 +92,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev, bio->bi_end_io = drbd_md_io_complete; bio->bi_rw = rw; - if (FAULT_ACTIVE(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD)) + if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD)) bio_endio(bio, -EIO); else submit_bio(rw, bio); @@ -176,13 +176,17 @@ static struct lc_element *_al_get(struct drbd_conf *mdev, unsigned int enr) struct lc_element *al_ext; struct lc_element *tmp; unsigned long al_flags = 0; + int wake; spin_lock_irq(&mdev->al_lock); tmp = lc_find(mdev->resync, enr/AL_EXT_PER_BM_SECT); if (unlikely(tmp != NULL)) { struct bm_extent *bm_ext = lc_entry(tmp, struct bm_extent, lce); if (test_bit(BME_NO_WRITES, &bm_ext->flags)) { + wake = !test_and_set_bit(BME_PRIORITY, &bm_ext->flags); spin_unlock_irq(&mdev->al_lock); + if (wake) + wake_up(&mdev->al_wait); return NULL; } } @@ -258,6 +262,33 @@ void drbd_al_complete_io(struct drbd_conf *mdev, sector_t sector) spin_unlock_irqrestore(&mdev->al_lock, flags); } +#if (PAGE_SHIFT + 3) < (AL_EXTENT_SHIFT - BM_BLOCK_SHIFT) +/* Currently BM_BLOCK_SHIFT, BM_EXT_SHIFT and AL_EXTENT_SHIFT + * are still coupled, or assume too much about their relation. + * Code below will not work if this is violated. + * Will be cleaned up with some followup patch. + */ +# error FIXME +#endif + +static unsigned int al_extent_to_bm_page(unsigned int al_enr) +{ + return al_enr >> + /* bit to page */ + ((PAGE_SHIFT + 3) - + /* al extent number to bit */ + (AL_EXTENT_SHIFT - BM_BLOCK_SHIFT)); +} + +static unsigned int rs_extent_to_bm_page(unsigned int rs_enr) +{ + return rs_enr >> + /* bit to page */ + ((PAGE_SHIFT + 3) - + /* al extent number to bit */ + (BM_EXT_SHIFT - BM_BLOCK_SHIFT)); +} + int w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) { @@ -285,7 +316,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) * For now, we must not write the transaction, * if we cannot write out the bitmap of the evicted extent. */ if (mdev->state.conn < C_CONNECTED && evicted != LC_FREE) - drbd_bm_write_sect(mdev, evicted/AL_EXT_PER_BM_SECT); + drbd_bm_write_page(mdev, al_extent_to_bm_page(evicted)); /* The bitmap write may have failed, causing a state change. */ if (mdev->state.disk < D_INCONSISTENT) { @@ -334,7 +365,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) + mdev->ldev->md.al_offset + mdev->al_tr_pos; if (!drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) - drbd_chk_io_error(mdev, 1, TRUE); + drbd_chk_io_error(mdev, 1, true); if (++mdev->al_tr_pos > div_ceil(mdev->act_log->nr_elements, AL_EXTENTS_PT)) @@ -511,227 +542,6 @@ cancel: return 1; } -static void atodb_endio(struct bio *bio, int error) -{ - struct drbd_atodb_wait *wc = bio->bi_private; - struct drbd_conf *mdev = wc->mdev; - struct page *page; - int uptodate = bio_flagged(bio, BIO_UPTODATE); - - /* strange behavior of some lower level drivers... - * fail the request by clearing the uptodate flag, - * but do not return any error?! */ - if (!error && !uptodate) - error = -EIO; - - drbd_chk_io_error(mdev, error, TRUE); - if (error && wc->error == 0) - wc->error = error; - - if (atomic_dec_and_test(&wc->count)) - complete(&wc->io_done); - - page = bio->bi_io_vec[0].bv_page; - put_page(page); - bio_put(bio); - mdev->bm_writ_cnt++; - put_ldev(mdev); -} - -/* sector to word */ -#define S2W(s) ((s)<<(BM_EXT_SHIFT-BM_BLOCK_SHIFT-LN2_BPL)) - -/* activity log to on disk bitmap -- prepare bio unless that sector - * is already covered by previously prepared bios */ -static int atodb_prepare_unless_covered(struct drbd_conf *mdev, - struct bio **bios, - unsigned int enr, - struct drbd_atodb_wait *wc) __must_hold(local) -{ - struct bio *bio; - struct page *page; - sector_t on_disk_sector; - unsigned int page_offset = PAGE_SIZE; - int offset; - int i = 0; - int err = -ENOMEM; - - /* We always write aligned, full 4k blocks, - * so we can ignore the logical_block_size (for now) */ - enr &= ~7U; - on_disk_sector = enr + mdev->ldev->md.md_offset - + mdev->ldev->md.bm_offset; - - D_ASSERT(!(on_disk_sector & 7U)); - - /* Check if that enr is already covered by an already created bio. - * Caution, bios[] is not NULL terminated, - * but only initialized to all NULL. - * For completely scattered activity log, - * the last invocation iterates over all bios, - * and finds the last NULL entry. - */ - while ((bio = bios[i])) { - if (bio->bi_sector == on_disk_sector) - return 0; - i++; - } - /* bios[i] == NULL, the next not yet used slot */ - - /* GFP_KERNEL, we are not in the write-out path */ - bio = bio_alloc(GFP_KERNEL, 1); - if (bio == NULL) - return -ENOMEM; - - if (i > 0) { - const struct bio_vec *prev_bv = bios[i-1]->bi_io_vec; - page_offset = prev_bv->bv_offset + prev_bv->bv_len; - page = prev_bv->bv_page; - } - if (page_offset == PAGE_SIZE) { - page = alloc_page(__GFP_HIGHMEM); - if (page == NULL) - goto out_bio_put; - page_offset = 0; - } else { - get_page(page); - } - - offset = S2W(enr); - drbd_bm_get_lel(mdev, offset, - min_t(size_t, S2W(8), drbd_bm_words(mdev) - offset), - kmap(page) + page_offset); - kunmap(page); - - bio->bi_private = wc; - bio->bi_end_io = atodb_endio; - bio->bi_bdev = mdev->ldev->md_bdev; - bio->bi_sector = on_disk_sector; - - if (bio_add_page(bio, page, 4096, page_offset) != 4096) - goto out_put_page; - - atomic_inc(&wc->count); - /* we already know that we may do this... - * get_ldev_if_state(mdev,D_ATTACHING); - * just get the extra reference, so that the local_cnt reflects - * the number of pending IO requests DRBD at its backing device. - */ - atomic_inc(&mdev->local_cnt); - - bios[i] = bio; - - return 0; - -out_put_page: - err = -EINVAL; - put_page(page); -out_bio_put: - bio_put(bio); - return err; -} - -/** - * drbd_al_to_on_disk_bm() - * Writes bitmap parts covered by active AL extents - * @mdev: DRBD device. - * - * Called when we detach (unconfigure) local storage, - * or when we go from R_PRIMARY to R_SECONDARY role. - */ -void drbd_al_to_on_disk_bm(struct drbd_conf *mdev) -{ - int i, nr_elements; - unsigned int enr; - struct bio **bios; - struct drbd_atodb_wait wc; - - ERR_IF (!get_ldev_if_state(mdev, D_ATTACHING)) - return; /* sorry, I don't have any act_log etc... */ - - wait_event(mdev->al_wait, lc_try_lock(mdev->act_log)); - - nr_elements = mdev->act_log->nr_elements; - - /* GFP_KERNEL, we are not in anyone's write-out path */ - bios = kzalloc(sizeof(struct bio *) * nr_elements, GFP_KERNEL); - if (!bios) - goto submit_one_by_one; - - atomic_set(&wc.count, 0); - init_completion(&wc.io_done); - wc.mdev = mdev; - wc.error = 0; - - for (i = 0; i < nr_elements; i++) { - enr = lc_element_by_index(mdev->act_log, i)->lc_number; - if (enr == LC_FREE) - continue; - /* next statement also does atomic_inc wc.count and local_cnt */ - if (atodb_prepare_unless_covered(mdev, bios, - enr/AL_EXT_PER_BM_SECT, - &wc)) - goto free_bios_submit_one_by_one; - } - - /* unnecessary optimization? */ - lc_unlock(mdev->act_log); - wake_up(&mdev->al_wait); - - /* all prepared, submit them */ - for (i = 0; i < nr_elements; i++) { - if (bios[i] == NULL) - break; - if (FAULT_ACTIVE(mdev, DRBD_FAULT_MD_WR)) { - bios[i]->bi_rw = WRITE; - bio_endio(bios[i], -EIO); - } else { - submit_bio(WRITE, bios[i]); - } - } - - drbd_blk_run_queue(bdev_get_queue(mdev->ldev->md_bdev)); - - /* always (try to) flush bitmap to stable storage */ - drbd_md_flush(mdev); - - /* In case we did not submit a single IO do not wait for - * them to complete. ( Because we would wait forever here. ) - * - * In case we had IOs and they are already complete, there - * is not point in waiting anyways. - * Therefore this if () ... */ - if (atomic_read(&wc.count)) - wait_for_completion(&wc.io_done); - - put_ldev(mdev); - - kfree(bios); - return; - - free_bios_submit_one_by_one: - /* free everything by calling the endio callback directly. */ - for (i = 0; i < nr_elements && bios[i]; i++) - bio_endio(bios[i], 0); - - kfree(bios); - - submit_one_by_one: - dev_warn(DEV, "Using the slow drbd_al_to_on_disk_bm()\n"); - - for (i = 0; i < mdev->act_log->nr_elements; i++) { - enr = lc_element_by_index(mdev->act_log, i)->lc_number; - if (enr == LC_FREE) - continue; - /* Really slow: if we have al-extents 16..19 active, - * sector 4 will be written four times! Synchronous! */ - drbd_bm_write_sect(mdev, enr/AL_EXT_PER_BM_SECT); - } - - lc_unlock(mdev->act_log); - wake_up(&mdev->al_wait); - put_ldev(mdev); -} - /** * drbd_al_apply_to_bm() - Sets the bitmap to diry(1) where covered ba active AL extents * @mdev: DRBD device. @@ -811,7 +621,7 @@ static int w_update_odbm(struct drbd_conf *mdev, struct drbd_work *w, int unused return 1; } - drbd_bm_write_sect(mdev, udw->enr); + drbd_bm_write_page(mdev, rs_extent_to_bm_page(udw->enr)); put_ldev(mdev); kfree(udw); @@ -891,7 +701,6 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector, dev_warn(DEV, "Kicking resync_lru element enr=%u " "out with rs_failed=%d\n", ext->lce.lc_number, ext->rs_failed); - set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags); } ext->rs_left = rs_left; ext->rs_failed = success ? 0 : count; @@ -910,7 +719,6 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector, drbd_queue_work_front(&mdev->data.work, &udw->w); } else { dev_warn(DEV, "Could not kmalloc an udw\n"); - set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags); } } } else { @@ -921,6 +729,22 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector, } } +void drbd_advance_rs_marks(struct drbd_conf *mdev, unsigned long still_to_go) +{ + unsigned long now = jiffies; + unsigned long last = mdev->rs_mark_time[mdev->rs_last_mark]; + int next = (mdev->rs_last_mark + 1) % DRBD_SYNC_MARKS; + if (time_after_eq(now, last + DRBD_SYNC_MARK_STEP)) { + if (mdev->rs_mark_left[mdev->rs_last_mark] != still_to_go && + mdev->state.conn != C_PAUSED_SYNC_T && + mdev->state.conn != C_PAUSED_SYNC_S) { + mdev->rs_mark_time[next] = now; + mdev->rs_mark_left[next] = still_to_go; + mdev->rs_last_mark = next; + } + } +} + /* clear the bit corresponding to the piece of storage in question: * size byte of data starting from sector. Only clear a bits of the affected * one ore more _aligned_ BM_BLOCK_SIZE blocks. @@ -938,7 +762,7 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, int wake_up = 0; unsigned long flags; - if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) { + if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) { dev_err(DEV, "drbd_set_in_sync: sector=%llus size=%d nonsense!\n", (unsigned long long)sector, size); return; @@ -971,21 +795,9 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, */ count = drbd_bm_clear_bits(mdev, sbnr, ebnr); if (count && get_ldev(mdev)) { - unsigned long now = jiffies; - unsigned long last = mdev->rs_mark_time[mdev->rs_last_mark]; - int next = (mdev->rs_last_mark + 1) % DRBD_SYNC_MARKS; - if (time_after_eq(now, last + DRBD_SYNC_MARK_STEP)) { - unsigned long tw = drbd_bm_total_weight(mdev); - if (mdev->rs_mark_left[mdev->rs_last_mark] != tw && - mdev->state.conn != C_PAUSED_SYNC_T && - mdev->state.conn != C_PAUSED_SYNC_S) { - mdev->rs_mark_time[next] = now; - mdev->rs_mark_left[next] = tw; - mdev->rs_last_mark = next; - } - } + drbd_advance_rs_marks(mdev, drbd_bm_total_weight(mdev)); spin_lock_irqsave(&mdev->al_lock, flags); - drbd_try_clear_on_disk_bm(mdev, sector, count, TRUE); + drbd_try_clear_on_disk_bm(mdev, sector, count, true); spin_unlock_irqrestore(&mdev->al_lock, flags); /* just wake_up unconditional now, various lc_chaged(), @@ -1000,27 +812,27 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, /* * this is intended to set one request worth of data out of sync. * affects at least 1 bit, - * and at most 1+DRBD_MAX_SEGMENT_SIZE/BM_BLOCK_SIZE bits. + * and at most 1+DRBD_MAX_BIO_SIZE/BM_BLOCK_SIZE bits. * * called by tl_clear and drbd_send_dblock (==drbd_make_request). * so this can be _any_ process. */ -void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, +int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line) { unsigned long sbnr, ebnr, lbnr, flags; sector_t esector, nr_sectors; - unsigned int enr, count; + unsigned int enr, count = 0; struct lc_element *e; - if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) { + if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) { dev_err(DEV, "sector: %llus, size: %d\n", (unsigned long long)sector, size); - return; + return 0; } if (!get_ldev(mdev)) - return; /* no disk, no metadata, no bitmap to set bits in */ + return 0; /* no disk, no metadata, no bitmap to set bits in */ nr_sectors = drbd_get_capacity(mdev->this_bdev); esector = sector + (size >> 9) - 1; @@ -1050,6 +862,8 @@ void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, out: put_ldev(mdev); + + return count; } static @@ -1130,7 +944,10 @@ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector) unsigned int enr = BM_SECT_TO_EXT(sector); struct bm_extent *bm_ext; int i, sig; + int sa = 200; /* Step aside 200 times, then grab the extent and let app-IO wait. + 200 times -> 20 seconds. */ +retry: sig = wait_event_interruptible(mdev->al_wait, (bm_ext = _bme_get(mdev, enr))); if (sig) @@ -1141,16 +958,25 @@ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector) for (i = 0; i < AL_EXT_PER_BM_SECT; i++) { sig = wait_event_interruptible(mdev->al_wait, - !_is_in_al(mdev, enr * AL_EXT_PER_BM_SECT + i)); - if (sig) { + !_is_in_al(mdev, enr * AL_EXT_PER_BM_SECT + i) || + test_bit(BME_PRIORITY, &bm_ext->flags)); + + if (sig || (test_bit(BME_PRIORITY, &bm_ext->flags) && sa)) { spin_lock_irq(&mdev->al_lock); if (lc_put(mdev->resync, &bm_ext->lce) == 0) { - clear_bit(BME_NO_WRITES, &bm_ext->flags); + bm_ext->flags = 0; /* clears BME_NO_WRITES and eventually BME_PRIORITY */ mdev->resync_locked--; wake_up(&mdev->al_wait); } spin_unlock_irq(&mdev->al_lock); - return -EINTR; + if (sig) + return -EINTR; + if (schedule_timeout_interruptible(HZ/10)) + return -EINTR; + if (sa && --sa == 0) + dev_warn(DEV,"drbd_rs_begin_io() stepped aside for 20sec." + "Resync stalled?\n"); + goto retry; } } set_bit(BME_LOCKED, &bm_ext->flags); @@ -1293,8 +1119,7 @@ void drbd_rs_complete_io(struct drbd_conf *mdev, sector_t sector) } if (lc_put(mdev->resync, &bm_ext->lce) == 0) { - clear_bit(BME_LOCKED, &bm_ext->flags); - clear_bit(BME_NO_WRITES, &bm_ext->flags); + bm_ext->flags = 0; /* clear BME_LOCKED, BME_NO_WRITES and BME_PRIORITY */ mdev->resync_locked--; wake_up(&mdev->al_wait); } @@ -1385,7 +1210,7 @@ void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size) sector_t esector, nr_sectors; int wake_up = 0; - if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) { + if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) { dev_err(DEV, "drbd_rs_failed_io: sector=%llus size=%d nonsense!\n", (unsigned long long)sector, size); return; @@ -1422,7 +1247,7 @@ void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size) mdev->rs_failed += count; if (get_ldev(mdev)) { - drbd_try_clear_on_disk_bm(mdev, sector, count, FALSE); + drbd_try_clear_on_disk_bm(mdev, sector, count, false); put_ldev(mdev); } diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index fd42832..f440a02 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -28,18 +28,56 @@ #include <linux/drbd.h> #include <linux/slab.h> #include <asm/kmap_types.h> + #include "drbd_int.h" + /* OPAQUE outside this file! * interface defined in drbd_int.h * convention: * function name drbd_bm_... => used elsewhere, "public". * function name bm_... => internal to implementation, "private". + */ - * Note that since find_first_bit returns int, at the current granularity of - * the bitmap (4KB per byte), this implementation "only" supports up to - * 1<<(32+12) == 16 TB... + +/* + * LIMITATIONS: + * We want to support >= peta byte of backend storage, while for now still using + * a granularity of one bit per 4KiB of storage. + * 1 << 50 bytes backend storage (1 PiB) + * 1 << (50 - 12) bits needed + * 38 --> we need u64 to index and count bits + * 1 << (38 - 3) bitmap bytes needed + * 35 --> we still need u64 to index and count bytes + * (that's 32 GiB of bitmap for 1 PiB storage) + * 1 << (35 - 2) 32bit longs needed + * 33 --> we'd even need u64 to index and count 32bit long words. + * 1 << (35 - 3) 64bit longs needed + * 32 --> we could get away with a 32bit unsigned int to index and count + * 64bit long words, but I rather stay with unsigned long for now. + * We probably should neither count nor point to bytes or long words + * directly, but either by bitnumber, or by page index and offset. + * 1 << (35 - 12) + * 22 --> we need that much 4KiB pages of bitmap. + * 1 << (22 + 3) --> on a 64bit arch, + * we need 32 MiB to store the array of page pointers. + * + * Because I'm lazy, and because the resulting patch was too large, too ugly + * and still incomplete, on 32bit we still "only" support 16 TiB (minus some), + * (1 << 32) bits * 4k storage. + * + + * bitmap storage and IO: + * Bitmap is stored little endian on disk, and is kept little endian in + * core memory. Currently we still hold the full bitmap in core as long + * as we are "attached" to a local disk, which at 32 GiB for 1PiB storage + * seems excessive. + * + * We plan to reduce the amount of in-core bitmap pages by paging them in + * and out against their on-disk location as necessary, but need to make + * sure we don't cause too much meta data IO, and must not deadlock in + * tight memory situations. This needs some more work. */ /* @@ -55,13 +93,9 @@ struct drbd_bitmap { struct page **bm_pages; spinlock_t bm_lock; - /* WARNING unsigned long bm_*: - * 32bit number of bit offset is just enough for 512 MB bitmap. - * it will blow up if we make the bitmap bigger... - * not that it makes much sense to have a bitmap that large, - * rather change the granularity to 16k or 64k or something. - * (that implies other problems, however...) - */ + + /* see LIMITATIONS: above */ + unsigned long bm_set; /* nr of set bits; THINK maybe atomic_t? */ unsigned long bm_bits; size_t bm_words; @@ -69,29 +103,18 @@ struct drbd_bitmap { sector_t bm_dev_capacity; struct mutex bm_change; /* serializes resize operations */ - atomic_t bm_async_io; - wait_queue_head_t bm_io_wait; + wait_queue_head_t bm_io_wait; /* used to serialize IO of single pages */ - unsigned long bm_flags; + enum bm_flag bm_flags; /* debugging aid, in case we are still racy somewhere */ char *bm_why; struct task_struct *bm_task; }; -/* definition of bits in bm_flags */ -#define BM_LOCKED 0 -#define BM_MD_IO_ERROR 1 -#define BM_P_VMALLOCED 2 - static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, unsigned long e, int val, const enum km_type km); -static int bm_is_locked(struct drbd_bitmap *b) -{ - return test_bit(BM_LOCKED, &b->bm_flags); -} - #define bm_print_lock_info(m) __bm_print_lock_info(m, __func__) static void __bm_print_lock_info(struct drbd_conf *mdev, const char *func) { @@ -108,7 +131,7 @@ static void __bm_print_lock_info(struct drbd_conf *mdev, const char *func) b->bm_task == mdev->worker.task ? "worker" : "?"); } -void drbd_bm_lock(struct drbd_conf *mdev, char *why) +void drbd_bm_lock(struct drbd_conf *mdev, char *why, enum bm_flag flags) { struct drbd_bitmap *b = mdev->bitmap; int trylock_failed; @@ -131,8 +154,9 @@ void drbd_bm_lock(struct drbd_conf *mdev, char *why) b->bm_task == mdev->worker.task ? "worker" : "?"); mutex_lock(&b->bm_change); } - if (__test_and_set_bit(BM_LOCKED, &b->bm_flags)) + if (BM_LOCKED_MASK & b->bm_flags) dev_err(DEV, "FIXME bitmap already locked in bm_lock\n"); + b->bm_flags |= flags & BM_LOCKED_MASK; b->bm_why = why; b->bm_task = current; @@ -146,31 +170,137 @@ void drbd_bm_unlock(struct drbd_conf *mdev) return; } - if (!__test_and_clear_bit(BM_LOCKED, &mdev->bitmap->bm_flags)) + if (!(BM_LOCKED_MASK & mdev->bitmap->bm_flags)) dev_err(DEV, "FIXME bitmap not locked in bm_unlock\n"); + b->bm_flags &= ~BM_LOCKED_MASK; b->bm_why = NULL; b->bm_task = NULL; mutex_unlock(&b->bm_change); } -/* word offset to long pointer */ -static unsigned long *__bm_map_paddr(struct drbd_bitmap *b, unsigned long offset, const enum km_type km) +/* we store some "meta" info about our pages in page->private */ +/* at a granularity of 4k storage per bitmap bit: + * one peta byte storage: 1<<50 byte, 1<<38 * 4k storage blocks + * 1<<38 bits, + * 1<<23 4k bitmap pages. + * Use 24 bits as page index, covers 2 peta byte storage + * at a granularity of 4k per bit. + * Used to report the failed page idx on io error from the endio handlers. + */ +#define BM_PAGE_IDX_MASK ((1UL<<24)-1) +/* this page is currently read in, or written back */ +#define BM_PAGE_IO_LOCK 31 +/* if there has been an IO error for this page */ +#define BM_PAGE_IO_ERROR 30 +/* this is to be able to intelligently skip disk IO, + * set if bits have been set since last IO. */ +#define BM_PAGE_NEED_WRITEOUT 29 +/* to mark for lazy writeout once syncer cleared all clearable bits, + * we if bits have been cleared since last IO. */ +#define BM_PAGE_LAZY_WRITEOUT 28 + +/* store_page_idx uses non-atomic assignment. It is only used directly after + * allocating the page. All other bm_set_page_* and bm_clear_page_* need to + * use atomic bit manipulation, as set_out_of_sync (and therefore bitmap + * changes) may happen from various contexts, and wait_on_bit/wake_up_bit + * requires it all to be atomic as well. */ +static void bm_store_page_idx(struct page *page, unsigned long idx) +{ + BUG_ON(0 != (idx & ~BM_PAGE_IDX_MASK)); + page_private(page) |= idx; +} + +static unsigned long bm_page_to_idx(struct page *page) { - struct page *page; - unsigned long page_nr; + return page_private(page) & BM_PAGE_IDX_MASK; +} + +/* As is very unlikely that the same page is under IO from more than one + * context, we can get away with a bit per page and one wait queue per bitmap. + */ +static void bm_page_lock_io(struct drbd_conf *mdev, int page_nr) +{ + struct drbd_bitmap *b = mdev->bitmap; + void *addr = &page_private(b->bm_pages[page_nr]); + wait_event(b->bm_io_wait, !test_and_set_bit(BM_PAGE_IO_LOCK, addr)); +} + +static void bm_page_unlock_io(struct drbd_conf *mdev, int page_nr) +{ + struct drbd_bitmap *b = mdev->bitmap; + void *addr = &page_private(b->bm_pages[page_nr]); + clear_bit(BM_PAGE_IO_LOCK, addr); + smp_mb__after_clear_bit(); + wake_up(&mdev->bitmap->bm_io_wait); +} + +/* set _before_ submit_io, so it may be reset due to being changed + * while this page is in flight... will get submitted later again */ +static void bm_set_page_unchanged(struct page *page) +{ + /* use cmpxchg? */ + clear_bit(BM_PAGE_NEED_WRITEOUT, &page_private(page)); + clear_bit(BM_PAGE_LAZY_WRITEOUT, &page_private(page)); +} +static void bm_set_page_need_writeout(struct page *page) +{ + set_bit(BM_PAGE_NEED_WRITEOUT, &page_private(page)); +} + +static int bm_test_page_unchanged(struct page *page) +{ + volatile const unsigned long *addr = &page_private(page); + return (*addr & ((1UL<<BM_PAGE_NEED_WRITEOUT)|(1UL<<BM_PAGE_LAZY_WRITEOUT))) == 0; +} + +static void bm_set_page_io_err(struct page *page) +{ + set_bit(BM_PAGE_IO_ERROR, &page_private(page)); +} + +static void bm_clear_page_io_err(struct page *page) +{ + clear_bit(BM_PAGE_IO_ERROR, &page_private(page)); +} + +static void bm_set_page_lazy_writeout(struct page *page) +{ + set_bit(BM_PAGE_LAZY_WRITEOUT, &page_private(page)); +} + +static int bm_test_page_lazy_writeout(struct page *page) +{ + return test_bit(BM_PAGE_LAZY_WRITEOUT, &page_private(page)); +} + +/* on a 32bit box, this would allow for exactly (2<<38) bits. */ +static unsigned int bm_word_to_page_idx(struct drbd_bitmap *b, unsigned long long_nr) +{ /* page_nr = (word*sizeof(long)) >> PAGE_SHIFT; */ - page_nr = offset >> (PAGE_SHIFT - LN2_BPL + 3); + unsigned int page_nr = long_nr >> (PAGE_SHIFT - LN2_BPL + 3); BUG_ON(page_nr >= b->bm_number_of_pages); - page = b->bm_pages[page_nr]; + return page_nr; +} +static unsigned int bm_bit_to_page_idx(struct drbd_bitmap *b, u64 bitnr) +{ + /* page_nr = (bitnr/8) >> PAGE_SHIFT; */ + unsigned int page_nr = bitnr >> (PAGE_SHIFT + 3); + BUG_ON(page_nr >= b->bm_number_of_pages); + return page_nr; +} + +static unsigned long *__bm_map_pidx(struct drbd_bitmap *b, unsigned int idx, const enum km_type km) +{ + struct page *page = b->bm_pages[idx]; return (unsigned long *) kmap_atomic(page, km); } -static unsigned long * bm_map_paddr(struct drbd_bitmap *b, unsigned long offset) +static unsigned long *bm_map_pidx(struct drbd_bitmap *b, unsigned int idx) { - return __bm_map_paddr(b, offset, KM_IRQ1); + return __bm_map_pidx(b, idx, KM_IRQ1); } static void __bm_unmap(unsigned long *p_addr, const enum km_type km) @@ -188,7 +318,7 @@ static void bm_unmap(unsigned long *p_addr) /* word offset from start of bitmap to word number _in_page_ * modulo longs per page #define MLPP(X) ((X) % (PAGE_SIZE/sizeof(long)) - hm, well, Philipp thinks gcc might not optimze the % into & (... - 1) + hm, well, Philipp thinks gcc might not optimize the % into & (... - 1) so do it explicitly: */ #define MLPP(X) ((X) & ((PAGE_SIZE/sizeof(long))-1)) @@ -202,6 +332,7 @@ static void bm_unmap(unsigned long *p_addr) * to be able to report device specific. */ + static void bm_free_pages(struct page **pages, unsigned long number) { unsigned long i; @@ -269,6 +400,9 @@ static struct page **bm_realloc_pages(struct drbd_bitmap *b, unsigned long want) bm_vk_free(new_pages, vmalloced); return NULL; } + /* we want to know which page it is + * from the endio handlers */ + bm_store_page_idx(page, i); new_pages[i] = page; } } else { @@ -280,9 +414,9 @@ static struct page **bm_realloc_pages(struct drbd_bitmap *b, unsigned long want) } if (vmalloced) - set_bit(BM_P_VMALLOCED, &b->bm_flags); + b->bm_flags |= BM_P_VMALLOCED; else - clear_bit(BM_P_VMALLOCED, &b->bm_flags); + b->bm_flags &= ~BM_P_VMALLOCED; return new_pages; } @@ -319,7 +453,7 @@ void drbd_bm_cleanup(struct drbd_conf *mdev) { ERR_IF (!mdev->bitmap) return; bm_free_pages(mdev->bitmap->bm_pages, mdev->bitmap->bm_number_of_pages); - bm_vk_free(mdev->bitmap->bm_pages, test_bit(BM_P_VMALLOCED, &mdev->bitmap->bm_flags)); + bm_vk_free(mdev->bitmap->bm_pages, (BM_P_VMALLOCED & mdev->bitmap->bm_flags)); kfree(mdev->bitmap); mdev->bitmap = NULL; } @@ -329,22 +463,39 @@ void drbd_bm_cleanup(struct drbd_conf *mdev) * this masks out the remaining bits. * Returns the number of bits cleared. */ +#define BITS_PER_PAGE (1UL << (PAGE_SHIFT + 3)) +#define BITS_PER_PAGE_MASK (BITS_PER_PAGE - 1) +#define BITS_PER_LONG_MASK (BITS_PER_LONG - 1) static int bm_clear_surplus(struct drbd_bitmap *b) { - const unsigned long mask = (1UL << (b->bm_bits & (BITS_PER_LONG-1))) - 1; - size_t w = b->bm_bits >> LN2_BPL; - int cleared = 0; + unsigned long mask; unsigned long *p_addr, *bm; + int tmp; + int cleared = 0; - p_addr = bm_map_paddr(b, w); - bm = p_addr + MLPP(w); - if (w < b->bm_words) { + /* number of bits modulo bits per page */ + tmp = (b->bm_bits & BITS_PER_PAGE_MASK); + /* mask the used bits of the word containing the last bit */ + mask = (1UL << (tmp & BITS_PER_LONG_MASK)) -1; + /* bitmap is always stored little endian, + * on disk and in core memory alike */ + mask = cpu_to_lel(mask); + + p_addr = bm_map_pidx(b, b->bm_number_of_pages - 1); + bm = p_addr + (tmp/BITS_PER_LONG); + if (mask) { + /* If mask != 0, we are not exactly aligned, so bm now points + * to the long containing the last bit. + * If mask == 0, bm already points to the word immediately + * after the last (long word aligned) bit. */ cleared = hweight_long(*bm & ~mask); *bm &= mask; - w++; bm++; + bm++; } - if (w < b->bm_words) { + if (BITS_PER_LONG == 32 && ((bm - p_addr) & 1) == 1) { + /* on a 32bit arch, we may need to zero out + * a padding long to align with a 64bit remote */ cleared += hweight_long(*bm); *bm = 0; } @@ -354,66 +505,75 @@ static int bm_clear_surplus(struct drbd_bitmap *b) static void bm_set_surplus(struct drbd_bitmap *b) { - const unsigned long mask = (1UL << (b->bm_bits & (BITS_PER_LONG-1))) - 1; - size_t w = b->bm_bits >> LN2_BPL; + unsigned long mask; unsigned long *p_addr, *bm; - - p_addr = bm_map_paddr(b, w); - bm = p_addr + MLPP(w); - if (w < b->bm_words) { + int tmp; + + /* number of bits modulo bits per page */ + tmp = (b->bm_bits & BITS_PER_PAGE_MASK); + /* mask the used bits of the word containing the last bit */ + mask = (1UL << (tmp & BITS_PER_LONG_MASK)) -1; + /* bitmap is always stored little endian, + * on disk and in core memory alike */ + mask = cpu_to_lel(mask); + + p_addr = bm_map_pidx(b, b->bm_number_of_pages - 1); + bm = p_addr + (tmp/BITS_PER_LONG); + if (mask) { + /* If mask != 0, we are not exactly aligned, so bm now points + * to the long containing the last bit. + * If mask == 0, bm already points to the word immediately + * after the last (long word aligned) bit. */ *bm |= ~mask; - bm++; w++; + bm++; } - if (w < b->bm_words) { - *bm = ~(0UL); + if (BITS_PER_LONG == 32 && ((bm - p_addr) & 1) == 1) { + /* on a 32bit arch, we may need to zero out + * a padding long to align with a 64bit remote */ + *bm = ~0UL; } bm_unmap(p_addr); } -static unsigned long __bm_count_bits(struct drbd_bitmap *b, const int swap_endian) +/* you better not modify the bitmap while this is running, + * or its results will be stale */ +static unsigned long bm_count_bits(struct drbd_bitmap *b) { - unsigned long *p_addr, *bm, offset = 0; + unsigned long *p_addr; unsigned long bits = 0; - unsigned long i, do_now; - - while (offset < b->bm_words) { - i = do_now = min_t(size_t, b->bm_words-offset, LWPP); - p_addr = __bm_map_paddr(b, offset, KM_USER0); - bm = p_addr + MLPP(offset); - while (i--) { -#ifndef __LITTLE_ENDIAN - if (swap_endian) - *bm = lel_to_cpu(*bm); -#endif - bits += hweight_long(*bm++); - } + unsigned long mask = (1UL << (b->bm_bits & BITS_PER_LONG_MASK)) -1; + int idx, i, last_word; + + /* all but last page */ + for (idx = 0; idx < b->bm_number_of_pages - 1; idx++) { + p_addr = __bm_map_pidx(b, idx, KM_USER0); + for (i = 0; i < LWPP; i++) + bits += hweight_long(p_addr[i]); __bm_unmap(p_addr, KM_USER0); - offset += do_now; cond_resched(); } - + /* last (or only) page */ + last_word = ((b->bm_bits - 1) & BITS_PER_PAGE_MASK) >> LN2_BPL; + p_addr = __bm_map_pidx(b, idx, KM_USER0); + for (i = 0; i < last_word; i++) + bits += hweight_long(p_addr[i]); + p_addr[last_word] &= cpu_to_lel(mask); + bits += hweight_long(p_addr[last_word]); + /* 32bit arch, may have an unused padding long */ + if (BITS_PER_LONG == 32 && (last_word & 1) == 0) + p_addr[last_word+1] = 0; + __bm_unmap(p_addr, KM_USER0); return bits; } -static unsigned long bm_count_bits(struct drbd_bitmap *b) -{ - return __bm_count_bits(b, 0); -} - -static unsigned long bm_count_bits_swap_endian(struct drbd_bitmap *b) -{ - return __bm_count_bits(b, 1); -} - /* offset and len in long words.*/ static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) { unsigned long *p_addr, *bm; + unsigned int idx; size_t do_now, end; -#define BM_SECTORS_PER_BIT (BM_BLOCK_SIZE/512) - end = offset + len; if (end > b->bm_words) { @@ -423,15 +583,16 @@ static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) while (offset < end) { do_now = min_t(size_t, ALIGN(offset + 1, LWPP), end) - offset; - p_addr = bm_map_paddr(b, offset); + idx = bm_word_to_page_idx(b, offset); + p_addr = bm_map_pidx(b, idx); bm = p_addr + MLPP(offset); if (bm+do_now > p_addr + LWPP) { printk(KERN_ALERT "drbd: BUG BUG BUG! p_addr:%p bm:%p do_now:%d\n", p_addr, bm, (int)do_now); - break; /* breaks to after catch_oob_access_end() only! */ - } - memset(bm, c, do_now * sizeof(long)); + } else + memset(bm, c, do_now * sizeof(long)); bm_unmap(p_addr); + bm_set_page_need_writeout(b->bm_pages[idx]); offset += do_now; } } @@ -447,7 +608,7 @@ static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) { struct drbd_bitmap *b = mdev->bitmap; - unsigned long bits, words, owords, obits, *p_addr, *bm; + unsigned long bits, words, owords, obits; unsigned long want, have, onpages; /* number of pages */ struct page **npages, **opages = NULL; int err = 0, growing; @@ -455,7 +616,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) ERR_IF(!b) return -ENOMEM; - drbd_bm_lock(mdev, "resize"); + drbd_bm_lock(mdev, "resize", BM_LOCKED_MASK); dev_info(DEV, "drbd_bm_resize called with capacity == %llu\n", (unsigned long long)capacity); @@ -463,7 +624,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) if (capacity == b->bm_dev_capacity) goto out; - opages_vmalloced = test_bit(BM_P_VMALLOCED, &b->bm_flags); + opages_vmalloced = (BM_P_VMALLOCED & b->bm_flags); if (capacity == 0) { spin_lock_irq(&b->bm_lock); @@ -491,18 +652,23 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) words = ALIGN(bits, 64) >> LN2_BPL; if (get_ldev(mdev)) { - D_ASSERT((u64)bits <= (((u64)mdev->ldev->md.md_size_sect-MD_BM_OFFSET) << 12)); + u64 bits_on_disk = ((u64)mdev->ldev->md.md_size_sect-MD_BM_OFFSET) << 12; put_ldev(mdev); + if (bits > bits_on_disk) { + dev_info(DEV, "bits = %lu\n", bits); + dev_info(DEV, "bits_on_disk = %llu\n", bits_on_disk); + err = -ENOSPC; + goto out; + } } - /* one extra long to catch off by one errors */ - want = ALIGN((words+1)*sizeof(long), PAGE_SIZE) >> PAGE_SHIFT; + want = ALIGN(words*sizeof(long), PAGE_SIZE) >> PAGE_SHIFT; have = b->bm_number_of_pages; if (want == have) { D_ASSERT(b->bm_pages != NULL); npages = b->bm_pages; } else { - if (FAULT_ACTIVE(mdev, DRBD_FAULT_BM_ALLOC)) + if (drbd_insert_fault(mdev, DRBD_FAULT_BM_ALLOC)) npages = NULL; else npages = bm_realloc_pages(b, want); @@ -542,11 +708,6 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) bm_free_pages(opages + want, have - want); } - p_addr = bm_map_paddr(b, words); - bm = p_addr + MLPP(words); - *bm = DRBD_MAGIC; - bm_unmap(p_addr); - (void)bm_clear_surplus(b); spin_unlock_irq(&b->bm_lock); @@ -554,7 +715,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) bm_vk_free(opages, opages_vmalloced); if (!growing) b->bm_set = bm_count_bits(b); - dev_info(DEV, "resync bitmap: bits=%lu words=%lu\n", bits, words); + dev_info(DEV, "resync bitmap: bits=%lu words=%lu pages=%lu\n", bits, words, want); out: drbd_bm_unlock(mdev); @@ -624,6 +785,7 @@ void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr, *bm; unsigned long word, bits; + unsigned int idx; size_t end, do_now; end = offset + number; @@ -638,16 +800,18 @@ void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, spin_lock_irq(&b->bm_lock); while (offset < end) { do_now = min_t(size_t, ALIGN(offset+1, LWPP), end) - offset; - p_addr = bm_map_paddr(b, offset); + idx = bm_word_to_page_idx(b, offset); + p_addr = bm_map_pidx(b, idx); bm = p_addr + MLPP(offset); offset += do_now; while (do_now--) { bits = hweight_long(*bm); - word = *bm | lel_to_cpu(*buffer++); + word = *bm | *buffer++; *bm++ = word; b->bm_set += hweight_long(word) - bits; } bm_unmap(p_addr); + bm_set_page_need_writeout(b->bm_pages[idx]); } /* with 32bit <-> 64bit cross-platform connect * this is only correct for current usage, @@ -656,7 +820,6 @@ void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, */ if (end == b->bm_words) b->bm_set -= bm_clear_surplus(b); - spin_unlock_irq(&b->bm_lock); } @@ -686,11 +849,11 @@ void drbd_bm_get_lel(struct drbd_conf *mdev, size_t offset, size_t number, else { while (offset < end) { do_now = min_t(size_t, ALIGN(offset+1, LWPP), end) - offset; - p_addr = bm_map_paddr(b, offset); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, offset)); bm = p_addr + MLPP(offset); offset += do_now; while (do_now--) - *buffer++ = cpu_to_lel(*bm++); + *buffer++ = *bm++; bm_unmap(p_addr); } } @@ -724,9 +887,22 @@ void drbd_bm_clear_all(struct drbd_conf *mdev) spin_unlock_irq(&b->bm_lock); } +struct bm_aio_ctx { + struct drbd_conf *mdev; + atomic_t in_flight; + struct completion done; + unsigned flags; +#define BM_AIO_COPY_PAGES 1 + int error; +}; + +/* bv_page may be a copy, or may be the original */ static void bm_async_io_complete(struct bio *bio, int error) { - struct drbd_bitmap *b = bio->bi_private; + struct bm_aio_ctx *ctx = bio->bi_private; + struct drbd_conf *mdev = ctx->mdev; + struct drbd_bitmap *b = mdev->bitmap; + unsigned int idx = bm_page_to_idx(bio->bi_io_vec[0].bv_page); int uptodate = bio_flagged(bio, BIO_UPTODATE); @@ -737,38 +913,83 @@ static void bm_async_io_complete(struct bio *bio, int error) if (!error && !uptodate) error = -EIO; + if ((ctx->flags & BM_AIO_COPY_PAGES) == 0 && + !bm_test_page_unchanged(b->bm_pages[idx])) + dev_warn(DEV, "bitmap page idx %u changed during IO!\n", idx); + if (error) { - /* doh. what now? - * for now, set all bits, and flag MD_IO_ERROR */ - __set_bit(BM_MD_IO_ERROR, &b->bm_flags); + /* ctx error will hold the completed-last non-zero error code, + * in case error codes differ. */ + ctx->error = error; + bm_set_page_io_err(b->bm_pages[idx]); + /* Not identical to on disk version of it. + * Is BM_PAGE_IO_ERROR enough? */ + if (__ratelimit(&drbd_ratelimit_state)) + dev_err(DEV, "IO ERROR %d on bitmap page idx %u\n", + error, idx); + } else { + bm_clear_page_io_err(b->bm_pages[idx]); + dynamic_dev_dbg(DEV, "bitmap page idx %u completed\n", idx); } - if (atomic_dec_and_test(&b->bm_async_io)) - wake_up(&b->bm_io_wait); + + bm_page_unlock_io(mdev, idx); + + /* FIXME give back to page pool */ + if (ctx->flags & BM_AIO_COPY_PAGES) + put_page(bio->bi_io_vec[0].bv_page); bio_put(bio); + + if (atomic_dec_and_test(&ctx->in_flight)) + complete(&ctx->done); } -static void bm_page_io_async(struct drbd_conf *mdev, struct drbd_bitmap *b, int page_nr, int rw) __must_hold(local) +static void bm_page_io_async(struct bm_aio_ctx *ctx, int page_nr, int rw) __must_hold(local) { /* we are process context. we always get a bio */ struct bio *bio = bio_alloc(GFP_KERNEL, 1); + struct drbd_conf *mdev = ctx->mdev; + struct drbd_bitmap *b = mdev->bitmap; + struct page *page; unsigned int len; + sector_t on_disk_sector = mdev->ldev->md.md_offset + mdev->ldev->md.bm_offset; on_disk_sector += ((sector_t)page_nr) << (PAGE_SHIFT-9); /* this might happen with very small - * flexible external meta data device */ + * flexible external meta data device, + * or with PAGE_SIZE > 4k */ len = min_t(unsigned int, PAGE_SIZE, (drbd_md_last_sector(mdev->ldev) - on_disk_sector + 1)<<9); + /* serialize IO on this page */ + bm_page_lock_io(mdev, page_nr); + /* before memcpy and submit, + * so it can be redirtied any time */ + bm_set_page_unchanged(b->bm_pages[page_nr]); + + if (ctx->flags & BM_AIO_COPY_PAGES) { + /* FIXME alloc_page is good enough for now, but actually needs + * to use pre-allocated page pool */ + void *src, *dest; + page = alloc_page(__GFP_HIGHMEM|__GFP_WAIT); + dest = kmap_atomic(page, KM_USER0); + src = kmap_atomic(b->bm_pages[page_nr], KM_USER1); + memcpy(dest, src, PAGE_SIZE); + kunmap_atomic(src, KM_USER1); + kunmap_atomic(dest, KM_USER0); + bm_store_page_idx(page, page_nr); + } else + page = b->bm_pages[page_nr]; + bio->bi_bdev = mdev->ldev->md_bdev; bio->bi_sector = on_disk_sector; - bio_add_page(bio, b->bm_pages[page_nr], len, 0); - bio->bi_private = b; + bio_add_page(bio, page, len, 0); + bio->bi_private = ctx; bio->bi_end_io = bm_async_io_complete; - if (FAULT_ACTIVE(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD)) { + if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD)) { bio->bi_rw |= rw; bio_endio(bio, -EIO); } else { @@ -776,88 +997,84 @@ static void bm_page_io_async(struct drbd_conf *mdev, struct drbd_bitmap *b, int } } -# if defined(__LITTLE_ENDIAN) - /* nothing to do, on disk == in memory */ -# define bm_cpu_to_lel(x) ((void)0) -# else -static void bm_cpu_to_lel(struct drbd_bitmap *b) -{ - /* need to cpu_to_lel all the pages ... - * this may be optimized by using - * cpu_to_lel(-1) == -1 and cpu_to_lel(0) == 0; - * the following is still not optimal, but better than nothing */ - unsigned int i; - unsigned long *p_addr, *bm; - if (b->bm_set == 0) { - /* no page at all; avoid swap if all is 0 */ - i = b->bm_number_of_pages; - } else if (b->bm_set == b->bm_bits) { - /* only the last page */ - i = b->bm_number_of_pages - 1; - } else { - /* all pages */ - i = 0; - } - for (; i < b->bm_number_of_pages; i++) { - p_addr = kmap_atomic(b->bm_pages[i], KM_USER0); - for (bm = p_addr; bm < p_addr + PAGE_SIZE/sizeof(long); bm++) - *bm = cpu_to_lel(*bm); - kunmap_atomic(p_addr, KM_USER0); - } -} -# endif -/* lel_to_cpu == cpu_to_lel */ -# define bm_lel_to_cpu(x) bm_cpu_to_lel(x) - /* * bm_rw: read/write the whole bitmap from/to its on disk location. */ -static int bm_rw(struct drbd_conf *mdev, int rw) __must_hold(local) +static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_idx) __must_hold(local) { + struct bm_aio_ctx ctx = { + .mdev = mdev, + .in_flight = ATOMIC_INIT(1), + .done = COMPLETION_INITIALIZER_ONSTACK(ctx.done), + .flags = lazy_writeout_upper_idx ? BM_AIO_COPY_PAGES : 0, + }; struct drbd_bitmap *b = mdev->bitmap; - /* sector_t sector; */ - int bm_words, num_pages, i; + int num_pages, i, count = 0; unsigned long now; char ppb[10]; int err = 0; - WARN_ON(!bm_is_locked(b)); - - /* no spinlock here, the drbd_bm_lock should be enough! */ - - bm_words = drbd_bm_words(mdev); - num_pages = (bm_words*sizeof(long) + PAGE_SIZE-1) >> PAGE_SHIFT; + /* + * We are protected against bitmap disappearing/resizing by holding an + * ldev reference (caller must have called get_ldev()). + * For read/write, we are protected against changes to the bitmap by + * the bitmap lock (see drbd_bitmap_io). + * For lazy writeout, we don't care for ongoing changes to the bitmap, + * as we submit copies of pages anyways. + */ + if (!ctx.flags) + WARN_ON(!(BM_LOCKED_MASK & b->bm_flags)); - /* on disk bitmap is little endian */ - if (rw == WRITE) - bm_cpu_to_lel(b); + num_pages = b->bm_number_of_pages; now = jiffies; - atomic_set(&b->bm_async_io, num_pages); - __clear_bit(BM_MD_IO_ERROR, &b->bm_flags); /* let the layers below us try to merge these bios... */ - for (i = 0; i < num_pages; i++) - bm_page_io_async(mdev, b, i, rw); + for (i = 0; i < num_pages; i++) { + /* ignore completely unchanged pages */ + if (lazy_writeout_upper_idx && i == lazy_writeout_upper_idx) + break; + if (rw & WRITE) { + if (bm_test_page_unchanged(b->bm_pages[i])) { + dynamic_dev_dbg(DEV, "skipped bm write for idx %u\n", i); + continue; + } + /* during lazy writeout, + * ignore those pages not marked for lazy writeout. */ + if (lazy_writeout_upper_idx && + !bm_test_page_lazy_writeout(b->bm_pages[i])) { + dynamic_dev_dbg(DEV, "skipped bm lazy write for idx %u\n", i); + continue; + } + } + atomic_inc(&ctx.in_flight); + bm_page_io_async(&ctx, i, rw); + ++count; + cond_resched(); + } - drbd_blk_run_queue(bdev_get_queue(mdev->ldev->md_bdev)); - wait_event(b->bm_io_wait, atomic_read(&b->bm_async_io) == 0); + /* + * We initialize ctx.in_flight to one to make sure bm_async_io_complete + * will not complete() early, and decrement / test it here. If there + * are still some bios in flight, we need to wait for them here. + */ + if (!atomic_dec_and_test(&ctx.in_flight)) + wait_for_completion(&ctx.done); + dev_info(DEV, "bitmap %s of %u pages took %lu jiffies\n", + rw == WRITE ? "WRITE" : "READ", + count, jiffies - now); - if (test_bit(BM_MD_IO_ERROR, &b->bm_flags)) { + if (ctx.error) { dev_alert(DEV, "we had at least one MD IO ERROR during bitmap IO\n"); - drbd_chk_io_error(mdev, 1, TRUE); - err = -EIO; + drbd_chk_io_error(mdev, 1, true); + err = -EIO; /* ctx.error ? */ } now = jiffies; if (rw == WRITE) { - /* swap back endianness */ - bm_lel_to_cpu(b); - /* flush bitmap to stable storage */ drbd_md_flush(mdev); } else /* rw == READ */ { - /* just read, if necessary adjust endianness */ - b->bm_set = bm_count_bits_swap_endian(b); + b->bm_set = bm_count_bits(b); dev_info(DEV, "recounting of set bits took additional %lu jiffies\n", jiffies - now); } @@ -875,112 +1092,128 @@ static int bm_rw(struct drbd_conf *mdev, int rw) __must_hold(local) */ int drbd_bm_read(struct drbd_conf *mdev) __must_hold(local) { - return bm_rw(mdev, READ); + return bm_rw(mdev, READ, 0); } /** * drbd_bm_write() - Write the whole bitmap to its on disk location. * @mdev: DRBD device. + * + * Will only write pages that have changed since last IO. */ int drbd_bm_write(struct drbd_conf *mdev) __must_hold(local) { - return bm_rw(mdev, WRITE); + return bm_rw(mdev, WRITE, 0); } /** - * drbd_bm_write_sect: Writes a 512 (MD_SECTOR_SIZE) byte piece of the bitmap + * drbd_bm_lazy_write_out() - Write bitmap pages 0 to @upper_idx-1, if they have changed. * @mdev: DRBD device. - * @enr: Extent number in the resync lru (happens to be sector offset) - * - * The BM_EXT_SIZE is on purpose exactly the amount of the bitmap covered - * by a single sector write. Therefore enr == sector offset from the - * start of the bitmap. + * @upper_idx: 0: write all changed pages; +ve: page index to stop scanning for changed pages */ -int drbd_bm_write_sect(struct drbd_conf *mdev, unsigned long enr) __must_hold(local) +int drbd_bm_write_lazy(struct drbd_conf *mdev, unsigned upper_idx) __must_hold(local) { - sector_t on_disk_sector = enr + mdev->ldev->md.md_offset - + mdev->ldev->md.bm_offset; - int bm_words, num_words, offset; - int err = 0; + return bm_rw(mdev, WRITE, upper_idx); +} + - mutex_lock(&mdev->md_io_mutex); - bm_words = drbd_bm_words(mdev); - offset = S2W(enr); /* word offset into bitmap */ - num_words = min(S2W(1), bm_words - offset); - if (num_words < S2W(1)) - memset(page_address(mdev->md_io_page), 0, MD_SECTOR_SIZE); - drbd_bm_get_lel(mdev, offset, num_words, - page_address(mdev->md_io_page)); - if (!drbd_md_sync_page_io(mdev, mdev->ldev, on_disk_sector, WRITE)) { - int i; - err = -EIO; - dev_err(DEV, "IO ERROR writing bitmap sector %lu " - "(meta-disk sector %llus)\n", - enr, (unsigned long long)on_disk_sector); - drbd_chk_io_error(mdev, 1, TRUE); - for (i = 0; i < AL_EXT_PER_BM_SECT; i++) - drbd_bm_ALe_set_all(mdev, enr*AL_EXT_PER_BM_SECT+i); +/** + * drbd_bm_write_page: Writes a PAGE_SIZE aligned piece of bitmap + * @mdev: DRBD device. + * @idx: bitmap page index + * + * We don't want to special case on logical_block_size of the backend device, + * so we submit PAGE_SIZE aligned pieces. + * Note that on "most" systems, PAGE_SIZE is 4k. + * + * In case this becomes an issue on systems with larger PAGE_SIZE, + * we may want to change this again to write 4k aligned 4k pieces. + */ +int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local) +{ + struct bm_aio_ctx ctx = { + .mdev = mdev, + .in_flight = ATOMIC_INIT(1), + .done = COMPLETION_INITIALIZER_ONSTACK(ctx.done), + .flags = BM_AIO_COPY_PAGES, + }; + + if (bm_test_page_unchanged(mdev->bitmap->bm_pages[idx])) { + dynamic_dev_dbg(DEV, "skipped bm page write for idx %u\n", idx); + return 0; } + + bm_page_io_async(&ctx, idx, WRITE_SYNC); + wait_for_completion(&ctx.done); + + if (ctx.error) + drbd_chk_io_error(mdev, 1, true); + /* that should force detach, so the in memory bitmap will be + * gone in a moment as well. */ + mdev->bm_writ_cnt++; - mutex_unlock(&mdev->md_io_mutex); - return err; + return ctx.error; } /* NOTE * find_first_bit returns int, we return unsigned long. - * should not make much difference anyways, but ... + * For this to work on 32bit arch with bitnumbers > (1<<32), + * we'd need to return u64, and get a whole lot of other places + * fixed where we still use unsigned long. * * this returns a bit number, NOT a sector! */ -#define BPP_MASK ((1UL << (PAGE_SHIFT+3)) - 1) static unsigned long __bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, const int find_zero_bit, const enum km_type km) { struct drbd_bitmap *b = mdev->bitmap; - unsigned long i = -1UL; unsigned long *p_addr; - unsigned long bit_offset; /* bit offset of the mapped page. */ + unsigned long bit_offset; + unsigned i; + if (bm_fo > b->bm_bits) { dev_err(DEV, "bm_fo=%lu bm_bits=%lu\n", bm_fo, b->bm_bits); + bm_fo = DRBD_END_OF_BITMAP; } else { while (bm_fo < b->bm_bits) { - unsigned long offset; - bit_offset = bm_fo & ~BPP_MASK; /* bit offset of the page */ - offset = bit_offset >> LN2_BPL; /* word offset of the page */ - p_addr = __bm_map_paddr(b, offset, km); + /* bit offset of the first bit in the page */ + bit_offset = bm_fo & ~BITS_PER_PAGE_MASK; + p_addr = __bm_map_pidx(b, bm_bit_to_page_idx(b, bm_fo), km); if (find_zero_bit) - i = find_next_zero_bit(p_addr, PAGE_SIZE*8, bm_fo & BPP_MASK); + i = find_next_zero_bit_le(p_addr, + PAGE_SIZE*8, bm_fo & BITS_PER_PAGE_MASK); else - i = find_next_bit(p_addr, PAGE_SIZE*8, bm_fo & BPP_MASK); + i = find_next_bit_le(p_addr, + PAGE_SIZE*8, bm_fo & BITS_PER_PAGE_MASK); __bm_unmap(p_addr, km); if (i < PAGE_SIZE*8) { - i = bit_offset + i; - if (i >= b->bm_bits) + bm_fo = bit_offset + i; + if (bm_fo >= b->bm_bits) break; goto found; } bm_fo = bit_offset + PAGE_SIZE*8; } - i = -1UL; + bm_fo = DRBD_END_OF_BITMAP; } found: - return i; + return bm_fo; } static unsigned long bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, const int find_zero_bit) { struct drbd_bitmap *b = mdev->bitmap; - unsigned long i = -1UL; + unsigned long i = DRBD_END_OF_BITMAP; ERR_IF(!b) return i; ERR_IF(!b->bm_pages) return i; spin_lock_irq(&b->bm_lock); - if (bm_is_locked(b)) + if (BM_DONT_TEST & b->bm_flags) bm_print_lock_info(mdev); i = __bm_find_next(mdev, bm_fo, find_zero_bit, KM_IRQ1); @@ -1006,13 +1239,13 @@ unsigned long drbd_bm_find_next_zero(struct drbd_conf *mdev, unsigned long bm_fo * you must take drbd_bm_lock() first */ unsigned long _drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo) { - /* WARN_ON(!bm_is_locked(mdev)); */ + /* WARN_ON(!(BM_DONT_SET & mdev->b->bm_flags)); */ return __bm_find_next(mdev, bm_fo, 0, KM_USER1); } unsigned long _drbd_bm_find_next_zero(struct drbd_conf *mdev, unsigned long bm_fo) { - /* WARN_ON(!bm_is_locked(mdev)); */ + /* WARN_ON(!(BM_DONT_SET & mdev->b->bm_flags)); */ return __bm_find_next(mdev, bm_fo, 1, KM_USER1); } @@ -1028,8 +1261,9 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr = NULL; unsigned long bitnr; - unsigned long last_page_nr = -1UL; + unsigned int last_page_nr = -1U; int c = 0; + int changed_total = 0; if (e >= b->bm_bits) { dev_err(DEV, "ASSERT FAILED: bit_s=%lu bit_e=%lu bm_bits=%lu\n", @@ -1037,23 +1271,33 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, e = b->bm_bits ? b->bm_bits -1 : 0; } for (bitnr = s; bitnr <= e; bitnr++) { - unsigned long offset = bitnr>>LN2_BPL; - unsigned long page_nr = offset >> (PAGE_SHIFT - LN2_BPL + 3); + unsigned int page_nr = bm_bit_to_page_idx(b, bitnr); if (page_nr != last_page_nr) { if (p_addr) __bm_unmap(p_addr, km); - p_addr = __bm_map_paddr(b, offset, km); + if (c < 0) + bm_set_page_lazy_writeout(b->bm_pages[last_page_nr]); + else if (c > 0) + bm_set_page_need_writeout(b->bm_pages[last_page_nr]); + changed_total += c; + c = 0; + p_addr = __bm_map_pidx(b, page_nr, km); last_page_nr = page_nr; } if (val) - c += (0 == __test_and_set_bit(bitnr & BPP_MASK, p_addr)); + c += (0 == __test_and_set_bit_le(bitnr & BITS_PER_PAGE_MASK, p_addr)); else - c -= (0 != __test_and_clear_bit(bitnr & BPP_MASK, p_addr)); + c -= (0 != __test_and_clear_bit_le(bitnr & BITS_PER_PAGE_MASK, p_addr)); } if (p_addr) __bm_unmap(p_addr, km); - b->bm_set += c; - return c; + if (c < 0) + bm_set_page_lazy_writeout(b->bm_pages[last_page_nr]); + else if (c > 0) + bm_set_page_need_writeout(b->bm_pages[last_page_nr]); + changed_total += c; + b->bm_set += changed_total; + return changed_total; } /* returns number of bits actually changed. @@ -1071,7 +1315,7 @@ static int bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, ERR_IF(!b->bm_pages) return 0; spin_lock_irqsave(&b->bm_lock, flags); - if (bm_is_locked(b)) + if ((val ? BM_DONT_SET : BM_DONT_CLEAR) & b->bm_flags) bm_print_lock_info(mdev); c = __bm_change_bits_to(mdev, s, e, val, KM_IRQ1); @@ -1188,12 +1432,11 @@ int drbd_bm_test_bit(struct drbd_conf *mdev, const unsigned long bitnr) ERR_IF(!b->bm_pages) return 0; spin_lock_irqsave(&b->bm_lock, flags); - if (bm_is_locked(b)) + if (BM_DONT_TEST & b->bm_flags) bm_print_lock_info(mdev); if (bitnr < b->bm_bits) { - unsigned long offset = bitnr>>LN2_BPL; - p_addr = bm_map_paddr(b, offset); - i = test_bit(bitnr & BPP_MASK, p_addr) ? 1 : 0; + p_addr = bm_map_pidx(b, bm_bit_to_page_idx(b, bitnr)); + i = test_bit_le(bitnr & BITS_PER_PAGE_MASK, p_addr) ? 1 : 0; bm_unmap(p_addr); } else if (bitnr == b->bm_bits) { i = -1; @@ -1211,10 +1454,10 @@ int drbd_bm_count_bits(struct drbd_conf *mdev, const unsigned long s, const unsi { unsigned long flags; struct drbd_bitmap *b = mdev->bitmap; - unsigned long *p_addr = NULL, page_nr = -1; + unsigned long *p_addr = NULL; unsigned long bitnr; + unsigned int page_nr = -1U; int c = 0; - size_t w; /* If this is called without a bitmap, that is a bug. But just to be * robust in case we screwed up elsewhere, in that case pretend there @@ -1224,20 +1467,20 @@ int drbd_bm_count_bits(struct drbd_conf *mdev, const unsigned long s, const unsi ERR_IF(!b->bm_pages) return 1; spin_lock_irqsave(&b->bm_lock, flags); - if (bm_is_locked(b)) + if (BM_DONT_TEST & b->bm_flags) bm_print_lock_info(mdev); for (bitnr = s; bitnr <= e; bitnr++) { - w = bitnr >> LN2_BPL; - if (page_nr != w >> (PAGE_SHIFT - LN2_BPL + 3)) { - page_nr = w >> (PAGE_SHIFT - LN2_BPL + 3); + unsigned int idx = bm_bit_to_page_idx(b, bitnr); + if (page_nr != idx) { + page_nr = idx; if (p_addr) bm_unmap(p_addr); - p_addr = bm_map_paddr(b, w); + p_addr = bm_map_pidx(b, idx); } ERR_IF (bitnr >= b->bm_bits) { dev_err(DEV, "bitnr=%lu bm_bits=%lu\n", bitnr, b->bm_bits); } else { - c += (0 != test_bit(bitnr - (page_nr << (PAGE_SHIFT+3)), p_addr)); + c += (0 != test_bit_le(bitnr - (page_nr << (PAGE_SHIFT+3)), p_addr)); } } if (p_addr) @@ -1272,7 +1515,7 @@ int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr) ERR_IF(!b->bm_pages) return 0; spin_lock_irqsave(&b->bm_lock, flags); - if (bm_is_locked(b)) + if (BM_DONT_TEST & b->bm_flags) bm_print_lock_info(mdev); s = S2W(enr); @@ -1280,7 +1523,7 @@ int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr) count = 0; if (s < b->bm_words) { int n = e-s; - p_addr = bm_map_paddr(b, s); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, s)); bm = p_addr + MLPP(s); while (n--) count += hweight_long(*bm++); @@ -1292,18 +1535,20 @@ int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr) return count; } -/* set all bits covered by the AL-extent al_enr */ +/* Set all bits covered by the AL-extent al_enr. + * Returns number of bits changed. */ unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, unsigned long al_enr) { struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr, *bm; unsigned long weight; - int count, s, e, i, do_now; + unsigned long s, e; + int count, i, do_now; ERR_IF(!b) return 0; ERR_IF(!b->bm_pages) return 0; spin_lock_irq(&b->bm_lock); - if (bm_is_locked(b)) + if (BM_DONT_SET & b->bm_flags) bm_print_lock_info(mdev); weight = b->bm_set; @@ -1315,7 +1560,7 @@ unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, unsigned long al_enr) count = 0; if (s < b->bm_words) { i = do_now = e-s; - p_addr = bm_map_paddr(b, s); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, s)); bm = p_addr + MLPP(s); while (i--) { count += hweight_long(*bm); @@ -1327,7 +1572,7 @@ unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, unsigned long al_enr) if (e == b->bm_words) b->bm_set -= bm_clear_surplus(b); } else { - dev_err(DEV, "start offset (%d) too large in drbd_bm_ALe_set_all\n", s); + dev_err(DEV, "start offset (%lu) too large in drbd_bm_ALe_set_all\n", s); } weight = b->bm_set - weight; spin_unlock_irq(&b->bm_lock); diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 3803a03..ef2ceed 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -42,6 +42,7 @@ #include <linux/genhd.h> #include <net/tcp.h> #include <linux/lru_cache.h> +#include <linux/prefetch.h> #ifdef __CHECKER__ # define __protected_by(x) __attribute__((require_context(x,1,999,"rdwr"))) @@ -72,13 +73,6 @@ extern int fault_devs; extern char usermode_helper[]; -#ifndef TRUE -#define TRUE 1 -#endif -#ifndef FALSE -#define FALSE 0 -#endif - /* I don't remember why XCPU ... * This is used to wake the asender, * and to interrupt sending the sending task @@ -104,6 +98,7 @@ extern char usermode_helper[]; #define ID_SYNCER (-1ULL) #define ID_VACANT 0 #define is_syncer_block_id(id) ((id) == ID_SYNCER) +#define UUID_NEW_BM_OFFSET ((u64)0x0001000000000000ULL) struct drbd_conf; @@ -137,20 +132,19 @@ enum { DRBD_FAULT_MAX, }; -#ifdef CONFIG_DRBD_FAULT_INJECTION extern unsigned int _drbd_insert_fault(struct drbd_conf *mdev, unsigned int type); + static inline int drbd_insert_fault(struct drbd_conf *mdev, unsigned int type) { +#ifdef CONFIG_DRBD_FAULT_INJECTION return fault_rate && (enable_faults & (1<<type)) && _drbd_insert_fault(mdev, type); -} -#define FAULT_ACTIVE(_m, _t) (drbd_insert_fault((_m), (_t))) - #else -#define FAULT_ACTIVE(_m, _t) (0) + return 0; #endif +} /* integer division, round _UP_ to the next integer */ #define div_ceil(A, B) ((A)/(B) + ((A)%(B) ? 1 : 0)) @@ -212,8 +206,10 @@ enum drbd_packets { /* P_CKPT_FENCE_REQ = 0x25, * currently reserved for protocol D */ /* P_CKPT_DISABLE_REQ = 0x26, * currently reserved for protocol D */ P_DELAY_PROBE = 0x27, /* is used on BOTH sockets */ + P_OUT_OF_SYNC = 0x28, /* Mark as out of sync (Outrunning), data socket */ + P_RS_CANCEL = 0x29, /* meta: Used to cancel RS_DATA_REQUEST packet by SyncSource */ - P_MAX_CMD = 0x28, + P_MAX_CMD = 0x2A, P_MAY_IGNORE = 0x100, /* Flag to test if (cmd > P_MAY_IGNORE) ... */ P_MAX_OPT_CMD = 0x101, @@ -269,6 +265,7 @@ static inline const char *cmdname(enum drbd_packets cmd) [P_RS_IS_IN_SYNC] = "CsumRSIsInSync", [P_COMPRESSED_BITMAP] = "CBitmap", [P_DELAY_PROBE] = "DelayProbe", + [P_OUT_OF_SYNC] = "OutOfSync", [P_MAX_CMD] = NULL, }; @@ -377,7 +374,7 @@ union p_header { #define DP_HARDBARRIER 1 /* depricated */ #define DP_RW_SYNC 2 /* equals REQ_SYNC */ #define DP_MAY_SET_IN_SYNC 4 -#define DP_UNPLUG 8 /* equals REQ_UNPLUG */ +#define DP_UNPLUG 8 /* not used anymore */ #define DP_FUA 16 /* equals REQ_FUA */ #define DP_FLUSH 32 /* equals REQ_FLUSH */ #define DP_DISCARD 64 /* equals REQ_DISCARD */ @@ -512,7 +509,7 @@ struct p_sizes { u64 d_size; /* size of disk */ u64 u_size; /* user requested size */ u64 c_size; /* current exported size */ - u32 max_segment_size; /* Maximal size of a BIO */ + u32 max_bio_size; /* Maximal size of a BIO */ u16 queue_order_type; /* not yet implemented in DRBD*/ u16 dds_flags; /* use enum dds_flags here. */ } __packed; @@ -550,6 +547,13 @@ struct p_discard { u32 pad; } __packed; +struct p_block_desc { + struct p_header80 head; + u64 sector; + u32 blksize; + u32 pad; /* to multiple of 8 Byte */ +} __packed; + /* Valid values for the encoding field. * Bump proto version when changing this. */ enum drbd_bitmap_code { @@ -619,7 +623,7 @@ DCBP_set_pad_bits(struct p_compressed_bm *p, int n) /* one bitmap packet, including the p_header, * should fit within one _architecture independend_ page. * so we need to use the fixed size 4KiB page size - * most architechtures have used for a long time. + * most architectures have used for a long time. */ #define BM_PACKET_PAYLOAD_BYTES (4096 - sizeof(struct p_header80)) #define BM_PACKET_WORDS (BM_PACKET_PAYLOAD_BYTES/sizeof(long)) @@ -647,6 +651,7 @@ union p_polymorph { struct p_block_req block_req; struct p_delay_probe93 delay_probe93; struct p_rs_uuid rs_uuid; + struct p_block_desc block_desc; } __packed; /**********************************************************************/ @@ -677,13 +682,6 @@ static inline enum drbd_thread_state get_t_state(struct drbd_thread *thi) return thi->t_state; } - -/* - * Having this as the first member of a struct provides sort of "inheritance". - * "derived" structs can be "drbd_queue_work()"ed. - * The callback should know and cast back to the descendant struct. - * drbd_request and drbd_epoch_entry are descendants of drbd_work. - */ struct drbd_work; typedef int (*drbd_work_cb)(struct drbd_conf *, struct drbd_work *, int cancel); struct drbd_work { @@ -702,7 +700,7 @@ struct drbd_request { * see drbd_endio_pri(). */ struct bio *private_bio; - struct hlist_node colision; + struct hlist_node collision; sector_t sector; unsigned int size; unsigned int epoch; /* barrier_nr */ @@ -712,9 +710,6 @@ struct drbd_request { * starting a new epoch... */ - /* up to here, the struct layout is identical to drbd_epoch_entry; - * we might be able to use that to our advantage... */ - struct list_head tl_requests; /* ring list in the transfer log */ struct bio *master_bio; /* master bio pointer */ unsigned long rq_state; /* see comments above _req_mod() */ @@ -771,7 +766,7 @@ struct digest_info { struct drbd_epoch_entry { struct drbd_work w; - struct hlist_node colision; + struct hlist_node collision; struct drbd_epoch *epoch; /* for writes */ struct drbd_conf *mdev; struct page *pages; @@ -816,7 +811,7 @@ enum { /* global flag bits */ enum { - CREATE_BARRIER, /* next P_DATA is preceeded by a P_BARRIER */ + CREATE_BARRIER, /* next P_DATA is preceded by a P_BARRIER */ SIGNAL_ASENDER, /* whether asender wants to be interrupted */ SEND_PING, /* whether asender should send a ping asap */ @@ -831,7 +826,7 @@ enum { CRASHED_PRIMARY, /* This node was a crashed primary. * Gets cleared when the state.conn * goes into C_CONNECTED state. */ - WRITE_BM_AFTER_RESYNC, /* A kmalloc() during resync failed */ + NO_BARRIER_SUPP, /* underlying block device doesn't implement barriers */ CONSIDER_RESYNC, MD_NO_FUA, /* Users wants us to not use FUA/FLUSH on meta data dev */ @@ -856,10 +851,37 @@ enum { GOT_PING_ACK, /* set when we receive a ping_ack packet, misc wait gets woken */ NEW_CUR_UUID, /* Create new current UUID when thawing IO */ AL_SUSPENDED, /* Activity logging is currently suspended. */ + AHEAD_TO_SYNC_SOURCE, /* Ahead -> SyncSource queued */ }; struct drbd_bitmap; /* opaque for drbd_conf */ +/* definition of bits in bm_flags to be used in drbd_bm_lock + * and drbd_bitmap_io and friends. */ +enum bm_flag { + /* do we need to kfree, or vfree bm_pages? */ + BM_P_VMALLOCED = 0x10000, /* internal use only, will be masked out */ + + /* currently locked for bulk operation */ + BM_LOCKED_MASK = 0x7, + + /* in detail, that is: */ + BM_DONT_CLEAR = 0x1, + BM_DONT_SET = 0x2, + BM_DONT_TEST = 0x4, + + /* (test bit, count bit) allowed (common case) */ + BM_LOCKED_TEST_ALLOWED = 0x3, + + /* testing bits, as well as setting new bits allowed, but clearing bits + * would be unexpected. Used during bitmap receive. Setting new bits + * requires sending of "out-of-sync" information, though. */ + BM_LOCKED_SET_ALLOWED = 0x1, + + /* clear is not expected while bitmap is locked for bulk operation */ +}; + + /* TODO sort members for performance * MAYBE group them further */ @@ -925,6 +947,7 @@ struct drbd_md_io { struct bm_io_work { struct drbd_work w; char *why; + enum bm_flag flags; int (*io_fn)(struct drbd_conf *mdev); void (*done)(struct drbd_conf *mdev, int rv); }; @@ -963,9 +986,12 @@ struct drbd_conf { struct drbd_work resync_work, unplug_work, go_diskless, - md_sync_work; + md_sync_work, + start_resync_work; struct timer_list resync_timer; struct timer_list md_sync_timer; + struct timer_list start_resync_timer; + struct timer_list request_timer; #ifdef DRBD_DEBUG_MD_SYNC struct { unsigned int line; @@ -1000,9 +1026,9 @@ struct drbd_conf { struct hlist_head *tl_hash; unsigned int tl_hash_s; - /* blocks to sync in this run [unit BM_BLOCK_SIZE] */ + /* blocks to resync in this run [unit BM_BLOCK_SIZE] */ unsigned long rs_total; - /* number of sync IOs that failed in this run */ + /* number of resync blocks that failed in this run */ unsigned long rs_failed; /* Syncer's start time [unit jiffies] */ unsigned long rs_start; @@ -1101,7 +1127,10 @@ struct drbd_conf { int c_sync_rate; /* current resync rate after syncer throttle magic */ struct fifo_buffer rs_plan_s; /* correction values of resync planer */ int rs_in_flight; /* resync sectors in flight (to proxy, in proxy and from proxy) */ - int rs_planed; /* resync sectors already planed */ + int rs_planed; /* resync sectors already planned */ + atomic_t ap_in_flight; /* App sectors in flight (waiting for ack) */ + int peer_max_bio_size; + int local_max_bio_size; }; static inline struct drbd_conf *minor_to_mdev(unsigned int minor) @@ -1118,7 +1147,7 @@ static inline unsigned int mdev_to_minor(struct drbd_conf *mdev) return mdev->minor; } -/* returns 1 if it was successfull, +/* returns 1 if it was successful, * returns 0 if there was no data socket. * so wherever you are going to use the data.socket, e.g. do * if (!drbd_get_data_sock(mdev)) @@ -1163,14 +1192,19 @@ enum dds_flags { }; extern void drbd_init_set_defaults(struct drbd_conf *mdev); -extern int drbd_change_state(struct drbd_conf *mdev, enum chg_state_flags f, - union drbd_state mask, union drbd_state val); +extern enum drbd_state_rv drbd_change_state(struct drbd_conf *mdev, + enum chg_state_flags f, + union drbd_state mask, + union drbd_state val); extern void drbd_force_state(struct drbd_conf *, union drbd_state, union drbd_state); -extern int _drbd_request_state(struct drbd_conf *, union drbd_state, - union drbd_state, enum chg_state_flags); -extern int __drbd_set_state(struct drbd_conf *, union drbd_state, - enum chg_state_flags, struct completion *done); +extern enum drbd_state_rv _drbd_request_state(struct drbd_conf *, + union drbd_state, + union drbd_state, + enum chg_state_flags); +extern enum drbd_state_rv __drbd_set_state(struct drbd_conf *, union drbd_state, + enum chg_state_flags, + struct completion *done); extern void print_st_err(struct drbd_conf *, union drbd_state, union drbd_state, int); extern int drbd_thread_start(struct drbd_thread *thi); @@ -1186,8 +1220,6 @@ extern void drbd_free_resources(struct drbd_conf *mdev); extern void tl_release(struct drbd_conf *mdev, unsigned int barrier_nr, unsigned int set_size); extern void tl_clear(struct drbd_conf *mdev); -enum drbd_req_event; -extern void tl_restart(struct drbd_conf *mdev, enum drbd_req_event what); extern void _tl_add_barrier(struct drbd_conf *, struct drbd_tl_epoch *); extern void drbd_free_sock(struct drbd_conf *mdev); extern int drbd_send(struct drbd_conf *mdev, struct socket *sock, @@ -1195,7 +1227,7 @@ extern int drbd_send(struct drbd_conf *mdev, struct socket *sock, extern int drbd_send_protocol(struct drbd_conf *mdev); extern int drbd_send_uuids(struct drbd_conf *mdev); extern int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev); -extern int drbd_send_sync_uuid(struct drbd_conf *mdev, u64 val); +extern int drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev); extern int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags flags); extern int _drbd_send_state(struct drbd_conf *mdev); extern int drbd_send_state(struct drbd_conf *mdev); @@ -1220,11 +1252,10 @@ extern int drbd_send_ack_dp(struct drbd_conf *mdev, enum drbd_packets cmd, struct p_data *dp, int data_size); extern int drbd_send_ack_ex(struct drbd_conf *mdev, enum drbd_packets cmd, sector_t sector, int blksize, u64 block_id); +extern int drbd_send_oos(struct drbd_conf *mdev, struct drbd_request *req); extern int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, struct drbd_epoch_entry *e); extern int drbd_send_dblock(struct drbd_conf *mdev, struct drbd_request *req); -extern int _drbd_send_barrier(struct drbd_conf *mdev, - struct drbd_tl_epoch *barrier); extern int drbd_send_drequest(struct drbd_conf *mdev, int cmd, sector_t sector, int size, u64 block_id); extern int drbd_send_drequest_csum(struct drbd_conf *mdev, @@ -1235,14 +1266,13 @@ extern int drbd_send_ov_request(struct drbd_conf *mdev,sector_t sector,int size) extern int drbd_send_bitmap(struct drbd_conf *mdev); extern int _drbd_send_bitmap(struct drbd_conf *mdev); -extern int drbd_send_sr_reply(struct drbd_conf *mdev, int retcode); +extern int drbd_send_sr_reply(struct drbd_conf *mdev, enum drbd_state_rv retcode); extern void drbd_free_bc(struct drbd_backing_dev *ldev); extern void drbd_mdev_cleanup(struct drbd_conf *mdev); +void drbd_print_uuids(struct drbd_conf *mdev, const char *text); -/* drbd_meta-data.c (still in drbd_main.c) */ extern void drbd_md_sync(struct drbd_conf *mdev); extern int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev); -/* maybe define them below as inline? */ extern void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local); extern void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local); extern void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local); @@ -1261,10 +1291,12 @@ extern void drbd_md_mark_dirty_(struct drbd_conf *mdev, extern void drbd_queue_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), void (*done)(struct drbd_conf *, int), - char *why); + char *why, enum bm_flag flags); +extern int drbd_bitmap_io(struct drbd_conf *mdev, + int (*io_fn)(struct drbd_conf *), + char *why, enum bm_flag flags); extern int drbd_bmio_set_n_write(struct drbd_conf *mdev); extern int drbd_bmio_clear_n_write(struct drbd_conf *mdev); -extern int drbd_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), char *why); extern void drbd_go_diskless(struct drbd_conf *mdev); extern void drbd_ldev_destroy(struct drbd_conf *mdev); @@ -1313,6 +1345,7 @@ struct bm_extent { #define BME_NO_WRITES 0 /* bm_extent.flags: no more requests on this one! */ #define BME_LOCKED 1 /* bm_extent.flags: syncer active on this one. */ +#define BME_PRIORITY 2 /* finish resync IO on this extent ASAP! App IO waiting! */ /* drbd_bitmap.c */ /* @@ -1390,7 +1423,9 @@ struct bm_extent { * you should use 64bit OS for that much storage, anyways. */ #define DRBD_MAX_SECTORS_FLEX BM_BIT_TO_SECT(0xffff7fff) #else -#define DRBD_MAX_SECTORS_FLEX BM_BIT_TO_SECT(0x1LU << 32) +/* we allow up to 1 PiB now on 64bit architecture with "flexible" meta data */ +#define DRBD_MAX_SECTORS_FLEX (1UL << 51) +/* corresponds to (1UL << 38) bits right now. */ #endif #endif @@ -1398,7 +1433,8 @@ struct bm_extent { * With a value of 8 all IO in one 128K block make it to the same slot of the * hash table. */ #define HT_SHIFT 8 -#define DRBD_MAX_SEGMENT_SIZE (1U<<(9+HT_SHIFT)) +#define DRBD_MAX_BIO_SIZE (1U<<(9+HT_SHIFT)) +#define DRBD_MAX_BIO_SIZE_SAFE (1 << 12) /* Works always = 4k */ #define DRBD_MAX_SIZE_H80_PACKET (1 << 15) /* The old header only allows packets up to 32Kib data */ @@ -1410,16 +1446,20 @@ extern int drbd_bm_resize(struct drbd_conf *mdev, sector_t sectors, int set_new extern void drbd_bm_cleanup(struct drbd_conf *mdev); extern void drbd_bm_set_all(struct drbd_conf *mdev); extern void drbd_bm_clear_all(struct drbd_conf *mdev); +/* set/clear/test only a few bits at a time */ extern int drbd_bm_set_bits( struct drbd_conf *mdev, unsigned long s, unsigned long e); extern int drbd_bm_clear_bits( struct drbd_conf *mdev, unsigned long s, unsigned long e); -/* bm_set_bits variant for use while holding drbd_bm_lock */ +extern int drbd_bm_count_bits( + struct drbd_conf *mdev, const unsigned long s, const unsigned long e); +/* bm_set_bits variant for use while holding drbd_bm_lock, + * may process the whole bitmap in one go */ extern void _drbd_bm_set_bits(struct drbd_conf *mdev, const unsigned long s, const unsigned long e); extern int drbd_bm_test_bit(struct drbd_conf *mdev, unsigned long bitnr); extern int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr); -extern int drbd_bm_write_sect(struct drbd_conf *mdev, unsigned long enr) __must_hold(local); +extern int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local); extern int drbd_bm_read(struct drbd_conf *mdev) __must_hold(local); extern int drbd_bm_write(struct drbd_conf *mdev) __must_hold(local); extern unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, @@ -1427,6 +1467,8 @@ extern unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, extern size_t drbd_bm_words(struct drbd_conf *mdev); extern unsigned long drbd_bm_bits(struct drbd_conf *mdev); extern sector_t drbd_bm_capacity(struct drbd_conf *mdev); + +#define DRBD_END_OF_BITMAP (~(unsigned long)0) extern unsigned long drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo); /* bm_find_next variants for use while you hold drbd_bm_lock() */ extern unsigned long _drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo); @@ -1437,14 +1479,12 @@ extern int drbd_bm_rs_done(struct drbd_conf *mdev); /* for receive_bitmap */ extern void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, unsigned long *buffer); -/* for _drbd_send_bitmap and drbd_bm_write_sect */ +/* for _drbd_send_bitmap */ extern void drbd_bm_get_lel(struct drbd_conf *mdev, size_t offset, size_t number, unsigned long *buffer); -extern void drbd_bm_lock(struct drbd_conf *mdev, char *why); +extern void drbd_bm_lock(struct drbd_conf *mdev, char *why, enum bm_flag flags); extern void drbd_bm_unlock(struct drbd_conf *mdev); - -extern int drbd_bm_count_bits(struct drbd_conf *mdev, const unsigned long s, const unsigned long e); /* drbd_main.c */ extern struct kmem_cache *drbd_request_cache; @@ -1467,7 +1507,7 @@ extern void drbd_free_mdev(struct drbd_conf *mdev); extern int proc_details; /* drbd_req */ -extern int drbd_make_request_26(struct request_queue *q, struct bio *bio); +extern int drbd_make_request(struct request_queue *q, struct bio *bio); extern int drbd_read_remote(struct drbd_conf *mdev, struct drbd_request *req); extern int drbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct bio_vec *bvec); extern int is_valid_ar_handle(struct drbd_request *, sector_t); @@ -1479,11 +1519,12 @@ extern void drbd_resume_io(struct drbd_conf *mdev); extern char *ppsize(char *buf, unsigned long long size); extern sector_t drbd_new_dev_size(struct drbd_conf *, struct drbd_backing_dev *, int); enum determine_dev_size { dev_size_error = -1, unchanged = 0, shrunk = 1, grew = 2 }; -extern enum determine_dev_size drbd_determin_dev_size(struct drbd_conf *, enum dds_flags) __must_hold(local); +extern enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *, enum dds_flags) __must_hold(local); extern void resync_after_online_grow(struct drbd_conf *); -extern void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int) __must_hold(local); -extern int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, - int force); +extern void drbd_reconsider_max_bio_size(struct drbd_conf *mdev); +extern enum drbd_state_rv drbd_set_role(struct drbd_conf *mdev, + enum drbd_role new_role, + int force); extern enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev); extern void drbd_try_outdate_peer_async(struct drbd_conf *mdev); extern int drbd_khelper(struct drbd_conf *mdev, char *cmd); @@ -1499,6 +1540,7 @@ extern int drbd_resync_finished(struct drbd_conf *mdev); extern int drbd_md_sync_page_io(struct drbd_conf *mdev, struct drbd_backing_dev *bdev, sector_t sector, int rw); extern void drbd_ov_oos_found(struct drbd_conf*, sector_t, int); +extern void drbd_rs_controller_reset(struct drbd_conf *mdev); static inline void ov_oos_print(struct drbd_conf *mdev) { @@ -1522,21 +1564,23 @@ extern int w_e_end_csum_rs_req(struct drbd_conf *, struct drbd_work *, int); extern int w_e_end_ov_reply(struct drbd_conf *, struct drbd_work *, int); extern int w_e_end_ov_req(struct drbd_conf *, struct drbd_work *, int); extern int w_ov_finished(struct drbd_conf *, struct drbd_work *, int); -extern int w_resync_inactive(struct drbd_conf *, struct drbd_work *, int); +extern int w_resync_timer(struct drbd_conf *, struct drbd_work *, int); extern int w_resume_next_sg(struct drbd_conf *, struct drbd_work *, int); extern int w_send_write_hint(struct drbd_conf *, struct drbd_work *, int); -extern int w_make_resync_request(struct drbd_conf *, struct drbd_work *, int); extern int w_send_dblock(struct drbd_conf *, struct drbd_work *, int); extern int w_send_barrier(struct drbd_conf *, struct drbd_work *, int); extern int w_send_read_req(struct drbd_conf *, struct drbd_work *, int); extern int w_prev_work_done(struct drbd_conf *, struct drbd_work *, int); extern int w_e_reissue(struct drbd_conf *, struct drbd_work *, int); extern int w_restart_disk_io(struct drbd_conf *, struct drbd_work *, int); +extern int w_send_oos(struct drbd_conf *, struct drbd_work *, int); +extern int w_start_resync(struct drbd_conf *, struct drbd_work *, int); extern void resync_timer_fn(unsigned long data); +extern void start_resync_timer_fn(unsigned long data); /* drbd_receiver.c */ -extern int drbd_rs_should_slow_down(struct drbd_conf *mdev); +extern int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector); extern int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, const unsigned rw, const int fault_type); extern int drbd_release_ee(struct drbd_conf *mdev, struct list_head *list); @@ -1619,16 +1663,16 @@ extern int drbd_rs_del_all(struct drbd_conf *mdev); extern void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size); extern int drbd_al_read_log(struct drbd_conf *mdev, struct drbd_backing_dev *); +extern void drbd_advance_rs_marks(struct drbd_conf *mdev, unsigned long still_to_go); extern void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line); #define drbd_set_in_sync(mdev, sector, size) \ __drbd_set_in_sync(mdev, sector, size, __FILE__, __LINE__) -extern void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, +extern int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line); #define drbd_set_out_of_sync(mdev, sector, size) \ __drbd_set_out_of_sync(mdev, sector, size, __FILE__, __LINE__) extern void drbd_al_apply_to_bm(struct drbd_conf *mdev); -extern void drbd_al_to_on_disk_bm(struct drbd_conf *mdev); extern void drbd_al_shrink(struct drbd_conf *mdev); @@ -1747,11 +1791,11 @@ static inline void drbd_state_unlock(struct drbd_conf *mdev) wake_up(&mdev->misc_wait); } -static inline int _drbd_set_state(struct drbd_conf *mdev, - union drbd_state ns, enum chg_state_flags flags, - struct completion *done) +static inline enum drbd_state_rv +_drbd_set_state(struct drbd_conf *mdev, union drbd_state ns, + enum chg_state_flags flags, struct completion *done) { - int rv; + enum drbd_state_rv rv; read_lock(&global_state_lock); rv = __drbd_set_state(mdev, ns, flags, done); @@ -1785,6 +1829,8 @@ static inline void __drbd_chk_io_error_(struct drbd_conf *mdev, int forcedetach, if (!forcedetach) { if (__ratelimit(&drbd_ratelimit_state)) dev_err(DEV, "Local IO failed in %s.\n", where); + if (mdev->state.disk > D_INCONSISTENT) + _drbd_set_state(_NS(mdev, disk, D_INCONSISTENT), CS_HARD, NULL); break; } /* NOTE fall through to detach case if forcedetach set */ @@ -1982,17 +2028,17 @@ static inline int drbd_send_ping_ack(struct drbd_conf *mdev) static inline void drbd_thread_stop(struct drbd_thread *thi) { - _drbd_thread_stop(thi, FALSE, TRUE); + _drbd_thread_stop(thi, false, true); } static inline void drbd_thread_stop_nowait(struct drbd_thread *thi) { - _drbd_thread_stop(thi, FALSE, FALSE); + _drbd_thread_stop(thi, false, false); } static inline void drbd_thread_restart_nowait(struct drbd_thread *thi) { - _drbd_thread_stop(thi, TRUE, FALSE); + _drbd_thread_stop(thi, true, false); } /* counts how many answer packets packets we expect from our peer, @@ -2037,7 +2083,7 @@ static inline void inc_ap_pending(struct drbd_conf *mdev) /* counts how many resync-related answers we still expect from the peer * increase decrease * C_SYNC_TARGET sends P_RS_DATA_REQUEST (and expects P_RS_DATA_REPLY) - * C_SYNC_SOURCE sends P_RS_DATA_REPLY (and expects P_WRITE_ACK whith ID_SYNCER) + * C_SYNC_SOURCE sends P_RS_DATA_REPLY (and expects P_WRITE_ACK with ID_SYNCER) * (or P_NEG_ACK with ID_SYNCER) */ static inline void inc_rs_pending(struct drbd_conf *mdev) @@ -2110,6 +2156,10 @@ static inline int get_net_conf(struct drbd_conf *mdev) static inline void put_ldev(struct drbd_conf *mdev) { int i = atomic_dec_return(&mdev->local_cnt); + + /* This may be called from some endio handler, + * so we must not sleep here. */ + __release(local); D_ASSERT(i >= 0); if (i == 0) { @@ -2146,17 +2196,18 @@ extern int _get_ldev_if_state(struct drbd_conf *mdev, enum drbd_disk_state mins) static inline void drbd_get_syncer_progress(struct drbd_conf *mdev, unsigned long *bits_left, unsigned int *per_mil_done) { - /* - * this is to break it at compile time when we change that - * (we may feel 4TB maximum storage per drbd is not enough) - */ + /* this is to break it at compile time when we change that, in case we + * want to support more than (1<<32) bits on a 32bit arch. */ typecheck(unsigned long, mdev->rs_total); /* note: both rs_total and rs_left are in bits, i.e. in * units of BM_BLOCK_SIZE. * for the percentage, we don't care. */ - *bits_left = drbd_bm_total_weight(mdev) - mdev->rs_failed; + if (mdev->state.conn == C_VERIFY_S || mdev->state.conn == C_VERIFY_T) + *bits_left = mdev->ov_left; + else + *bits_left = drbd_bm_total_weight(mdev) - mdev->rs_failed; /* >> 10 to prevent overflow, * +1 to prevent division by zero */ if (*bits_left > mdev->rs_total) { @@ -2171,10 +2222,19 @@ static inline void drbd_get_syncer_progress(struct drbd_conf *mdev, *bits_left, mdev->rs_total, mdev->rs_failed); *per_mil_done = 0; } else { - /* make sure the calculation happens in long context */ - unsigned long tmp = 1000UL - - (*bits_left >> 10)*1000UL - / ((mdev->rs_total >> 10) + 1UL); + /* Make sure the division happens in long context. + * We allow up to one petabyte storage right now, + * at a granularity of 4k per bit that is 2**38 bits. + * After shift right and multiplication by 1000, + * this should still fit easily into a 32bit long, + * so we don't need a 64bit division on 32bit arch. + * Note: currently we don't support such large bitmaps on 32bit + * arch anyways, but no harm done to be prepared for it here. + */ + unsigned int shift = mdev->rs_total >= (1ULL << 32) ? 16 : 10; + unsigned long left = *bits_left >> shift; + unsigned long total = 1UL + (mdev->rs_total >> shift); + unsigned long tmp = 1000UL - left * 1000UL/total; *per_mil_done = tmp; } } @@ -2193,8 +2253,9 @@ static inline int drbd_get_max_buffers(struct drbd_conf *mdev) return mxb; } -static inline int drbd_state_is_stable(union drbd_state s) +static inline int drbd_state_is_stable(struct drbd_conf *mdev) { + union drbd_state s = mdev->state; /* DO NOT add a default clause, we want the compiler to warn us * for any newly introduced state we may have forgotten to add here */ @@ -2211,11 +2272,9 @@ static inline int drbd_state_is_stable(union drbd_state s) case C_VERIFY_T: case C_PAUSED_SYNC_S: case C_PAUSED_SYNC_T: - /* maybe stable, look at the disk state */ - break; - - /* no new io accepted during tansitional states - * like handshake or teardown */ + case C_AHEAD: + case C_BEHIND: + /* transitional states, IO allowed */ case C_DISCONNECTING: case C_UNCONNECTED: case C_TIMEOUT: @@ -2226,7 +2285,15 @@ static inline int drbd_state_is_stable(union drbd_state s) case C_WF_REPORT_PARAMS: case C_STARTING_SYNC_S: case C_STARTING_SYNC_T: + break; + + /* Allow IO in BM exchange states with new protocols */ case C_WF_BITMAP_S: + if (mdev->agreed_pro_version < 96) + return 0; + break; + + /* no new io accepted in these states */ case C_WF_BITMAP_T: case C_WF_SYNC_UUID: case C_MASK: @@ -2261,41 +2328,47 @@ static inline int is_susp(union drbd_state s) return s.susp || s.susp_nod || s.susp_fen; } -static inline int __inc_ap_bio_cond(struct drbd_conf *mdev) +static inline bool may_inc_ap_bio(struct drbd_conf *mdev) { int mxb = drbd_get_max_buffers(mdev); if (is_susp(mdev->state)) - return 0; + return false; if (test_bit(SUSPEND_IO, &mdev->flags)) - return 0; + return false; /* to avoid potential deadlock or bitmap corruption, * in various places, we only allow new application io * to start during "stable" states. */ /* no new io accepted when attaching or detaching the disk */ - if (!drbd_state_is_stable(mdev->state)) - return 0; + if (!drbd_state_is_stable(mdev)) + return false; /* since some older kernels don't have atomic_add_unless, * and we are within the spinlock anyways, we have this workaround. */ if (atomic_read(&mdev->ap_bio_cnt) > mxb) - return 0; + return false; if (test_bit(BITMAP_IO, &mdev->flags)) - return 0; - return 1; + return false; + return true; } -/* I'd like to use wait_event_lock_irq, - * but I'm not sure when it got introduced, - * and not sure when it has 3 or 4 arguments */ -static inline void inc_ap_bio(struct drbd_conf *mdev, int count) +static inline bool inc_ap_bio_cond(struct drbd_conf *mdev, int count) { - /* compare with after_state_ch, - * os.conn != C_WF_BITMAP_S && ns.conn == C_WF_BITMAP_S */ - DEFINE_WAIT(wait); + bool rv = false; + + spin_lock_irq(&mdev->req_lock); + rv = may_inc_ap_bio(mdev); + if (rv) + atomic_add(count, &mdev->ap_bio_cnt); + spin_unlock_irq(&mdev->req_lock); + return rv; +} + +static inline void inc_ap_bio(struct drbd_conf *mdev, int count) +{ /* we wait here * as long as the device is suspended * until the bitmap is no longer on the fly during connection @@ -2304,16 +2377,7 @@ static inline void inc_ap_bio(struct drbd_conf *mdev, int count) * to avoid races with the reconnect code, * we need to atomic_inc within the spinlock. */ - spin_lock_irq(&mdev->req_lock); - while (!__inc_ap_bio_cond(mdev)) { - prepare_to_wait(&mdev->misc_wait, &wait, TASK_UNINTERRUPTIBLE); - spin_unlock_irq(&mdev->req_lock); - schedule(); - finish_wait(&mdev->misc_wait, &wait); - spin_lock_irq(&mdev->req_lock); - } - atomic_add(count, &mdev->ap_bio_cnt); - spin_unlock_irq(&mdev->req_lock); + wait_event(mdev->misc_wait, inc_ap_bio_cond(mdev, count)); } static inline void dec_ap_bio(struct drbd_conf *mdev) @@ -2333,9 +2397,11 @@ static inline void dec_ap_bio(struct drbd_conf *mdev) } } -static inline void drbd_set_ed_uuid(struct drbd_conf *mdev, u64 val) +static inline int drbd_set_ed_uuid(struct drbd_conf *mdev, u64 val) { + int changed = mdev->ed_uuid != val; mdev->ed_uuid = val; + return changed; } static inline int seq_cmp(u32 a, u32 b) @@ -2382,20 +2448,6 @@ static inline int drbd_queue_order_type(struct drbd_conf *mdev) return QUEUE_ORDERED_NONE; } -static inline void drbd_blk_run_queue(struct request_queue *q) -{ - if (q && q->unplug_fn) - q->unplug_fn(q); -} - -static inline void drbd_kick_lo(struct drbd_conf *mdev) -{ - if (get_ldev(mdev)) { - drbd_blk_run_queue(bdev_get_queue(mdev->ldev->backing_bdev)); - put_ldev(mdev); - } -} - static inline void drbd_md_flush(struct drbd_conf *mdev) { int r; diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 29cd0dc..0358e55 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -85,7 +85,8 @@ MODULE_AUTHOR("Philipp Reisner <phil@linbit.com>, " MODULE_DESCRIPTION("drbd - Distributed Replicated Block Device v" REL_VERSION); MODULE_VERSION(REL_VERSION); MODULE_LICENSE("GPL"); -MODULE_PARM_DESC(minor_count, "Maximum number of drbd devices (1-255)"); +MODULE_PARM_DESC(minor_count, "Maximum number of drbd devices (" + __stringify(DRBD_MINOR_COUNT_MIN) "-" __stringify(DRBD_MINOR_COUNT_MAX) ")"); MODULE_ALIAS_BLOCKDEV_MAJOR(DRBD_MAJOR); #include <linux/moduleparam.h> @@ -115,7 +116,7 @@ module_param(fault_devs, int, 0644); #endif /* module parameter, defined */ -unsigned int minor_count = 32; +unsigned int minor_count = DRBD_MINOR_COUNT_DEF; int disable_sendpage; int allow_oos; unsigned int cn_idx = CN_IDX_DRBD; @@ -335,6 +336,7 @@ bail: drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); } + /** * _tl_restart() - Walks the transfer log, and applies an action to all requests * @mdev: DRBD device. @@ -456,7 +458,7 @@ void tl_restart(struct drbd_conf *mdev, enum drbd_req_event what) } /** - * cl_wide_st_chg() - TRUE if the state change is a cluster wide one + * cl_wide_st_chg() - true if the state change is a cluster wide one * @mdev: DRBD device. * @os: old (current) state. * @ns: new (wanted) state. @@ -473,12 +475,13 @@ static int cl_wide_st_chg(struct drbd_conf *mdev, (os.conn == C_CONNECTED && ns.conn == C_VERIFY_S); } -int drbd_change_state(struct drbd_conf *mdev, enum chg_state_flags f, - union drbd_state mask, union drbd_state val) +enum drbd_state_rv +drbd_change_state(struct drbd_conf *mdev, enum chg_state_flags f, + union drbd_state mask, union drbd_state val) { unsigned long flags; union drbd_state os, ns; - int rv; + enum drbd_state_rv rv; spin_lock_irqsave(&mdev->req_lock, flags); os = mdev->state; @@ -502,20 +505,22 @@ void drbd_force_state(struct drbd_conf *mdev, drbd_change_state(mdev, CS_HARD, mask, val); } -static int is_valid_state(struct drbd_conf *mdev, union drbd_state ns); -static int is_valid_state_transition(struct drbd_conf *, - union drbd_state, union drbd_state); +static enum drbd_state_rv is_valid_state(struct drbd_conf *, union drbd_state); +static enum drbd_state_rv is_valid_state_transition(struct drbd_conf *, + union drbd_state, + union drbd_state); static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os, union drbd_state ns, const char **warn_sync_abort); int drbd_send_state_req(struct drbd_conf *, union drbd_state, union drbd_state); -static enum drbd_state_ret_codes _req_st_cond(struct drbd_conf *mdev, - union drbd_state mask, union drbd_state val) +static enum drbd_state_rv +_req_st_cond(struct drbd_conf *mdev, union drbd_state mask, + union drbd_state val) { union drbd_state os, ns; unsigned long flags; - int rv; + enum drbd_state_rv rv; if (test_and_clear_bit(CL_ST_CHG_SUCCESS, &mdev->flags)) return SS_CW_SUCCESS; @@ -536,7 +541,7 @@ static enum drbd_state_ret_codes _req_st_cond(struct drbd_conf *mdev, if (rv == SS_SUCCESS) { rv = is_valid_state_transition(mdev, ns, os); if (rv == SS_SUCCESS) - rv = 0; /* cont waiting, otherwise fail. */ + rv = SS_UNKNOWN_ERROR; /* cont waiting, otherwise fail. */ } } spin_unlock_irqrestore(&mdev->req_lock, flags); @@ -554,14 +559,14 @@ static enum drbd_state_ret_codes _req_st_cond(struct drbd_conf *mdev, * Should not be called directly, use drbd_request_state() or * _drbd_request_state(). */ -static int drbd_req_state(struct drbd_conf *mdev, - union drbd_state mask, union drbd_state val, - enum chg_state_flags f) +static enum drbd_state_rv +drbd_req_state(struct drbd_conf *mdev, union drbd_state mask, + union drbd_state val, enum chg_state_flags f) { struct completion done; unsigned long flags; union drbd_state os, ns; - int rv; + enum drbd_state_rv rv; init_completion(&done); @@ -636,10 +641,11 @@ abort: * Cousin of drbd_request_state(), useful with the CS_WAIT_COMPLETE * flag, or when logging of failed state change requests is not desired. */ -int _drbd_request_state(struct drbd_conf *mdev, union drbd_state mask, - union drbd_state val, enum chg_state_flags f) +enum drbd_state_rv +_drbd_request_state(struct drbd_conf *mdev, union drbd_state mask, + union drbd_state val, enum chg_state_flags f) { - int rv; + enum drbd_state_rv rv; wait_event(mdev->state_wait, (rv = drbd_req_state(mdev, mask, val, f)) != SS_IN_TRANSIENT_STATE); @@ -663,8 +669,8 @@ static void print_st(struct drbd_conf *mdev, char *name, union drbd_state ns) ); } -void print_st_err(struct drbd_conf *mdev, - union drbd_state os, union drbd_state ns, int err) +void print_st_err(struct drbd_conf *mdev, union drbd_state os, + union drbd_state ns, enum drbd_state_rv err) { if (err == SS_IN_TRANSIENT_STATE) return; @@ -674,32 +680,18 @@ void print_st_err(struct drbd_conf *mdev, } -#define drbd_peer_str drbd_role_str -#define drbd_pdsk_str drbd_disk_str - -#define drbd_susp_str(A) ((A) ? "1" : "0") -#define drbd_aftr_isp_str(A) ((A) ? "1" : "0") -#define drbd_peer_isp_str(A) ((A) ? "1" : "0") -#define drbd_user_isp_str(A) ((A) ? "1" : "0") - -#define PSC(A) \ - ({ if (ns.A != os.A) { \ - pbp += sprintf(pbp, #A "( %s -> %s ) ", \ - drbd_##A##_str(os.A), \ - drbd_##A##_str(ns.A)); \ - } }) - /** * is_valid_state() - Returns an SS_ error code if ns is not valid * @mdev: DRBD device. * @ns: State to consider. */ -static int is_valid_state(struct drbd_conf *mdev, union drbd_state ns) +static enum drbd_state_rv +is_valid_state(struct drbd_conf *mdev, union drbd_state ns) { /* See drbd_state_sw_errors in drbd_strings.c */ enum drbd_fencing_p fp; - int rv = SS_SUCCESS; + enum drbd_state_rv rv = SS_SUCCESS; fp = FP_DONT_CARE; if (get_ldev(mdev)) { @@ -753,6 +745,9 @@ static int is_valid_state(struct drbd_conf *mdev, union drbd_state ns) mdev->agreed_pro_version < 88) rv = SS_NOT_SUPPORTED; + else if (ns.conn >= C_CONNECTED && ns.pdsk == D_UNKNOWN) + rv = SS_CONNECTED_OUTDATES; + return rv; } @@ -762,10 +757,11 @@ static int is_valid_state(struct drbd_conf *mdev, union drbd_state ns) * @ns: new state. * @os: old state. */ -static int is_valid_state_transition(struct drbd_conf *mdev, - union drbd_state ns, union drbd_state os) +static enum drbd_state_rv +is_valid_state_transition(struct drbd_conf *mdev, union drbd_state ns, + union drbd_state os) { - int rv = SS_SUCCESS; + enum drbd_state_rv rv = SS_SUCCESS; if ((ns.conn == C_STARTING_SYNC_T || ns.conn == C_STARTING_SYNC_S) && os.conn > C_CONNECTED) @@ -800,6 +796,10 @@ static int is_valid_state_transition(struct drbd_conf *mdev, os.conn < C_CONNECTED) rv = SS_NEED_CONNECTION; + if ((ns.conn == C_SYNC_TARGET || ns.conn == C_SYNC_SOURCE) + && os.conn < C_WF_REPORT_PARAMS) + rv = SS_NEED_CONNECTION; /* No NetworkFailure -> SyncTarget etc... */ + return rv; } @@ -817,6 +817,7 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state union drbd_state ns, const char **warn_sync_abort) { enum drbd_fencing_p fp; + enum drbd_disk_state disk_min, disk_max, pdsk_min, pdsk_max; fp = FP_DONT_CARE; if (get_ldev(mdev)) { @@ -869,56 +870,6 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state ns.conn = C_CONNECTED; } - if (ns.conn >= C_CONNECTED && - ((ns.disk == D_CONSISTENT || ns.disk == D_OUTDATED) || - (ns.disk == D_NEGOTIATING && ns.conn == C_WF_BITMAP_T))) { - switch (ns.conn) { - case C_WF_BITMAP_T: - case C_PAUSED_SYNC_T: - ns.disk = D_OUTDATED; - break; - case C_CONNECTED: - case C_WF_BITMAP_S: - case C_SYNC_SOURCE: - case C_PAUSED_SYNC_S: - ns.disk = D_UP_TO_DATE; - break; - case C_SYNC_TARGET: - ns.disk = D_INCONSISTENT; - dev_warn(DEV, "Implicitly set disk state Inconsistent!\n"); - break; - } - if (os.disk == D_OUTDATED && ns.disk == D_UP_TO_DATE) - dev_warn(DEV, "Implicitly set disk from Outdated to UpToDate\n"); - } - - if (ns.conn >= C_CONNECTED && - (ns.pdsk == D_CONSISTENT || ns.pdsk == D_OUTDATED)) { - switch (ns.conn) { - case C_CONNECTED: - case C_WF_BITMAP_T: - case C_PAUSED_SYNC_T: - case C_SYNC_TARGET: - ns.pdsk = D_UP_TO_DATE; - break; - case C_WF_BITMAP_S: - case C_PAUSED_SYNC_S: - /* remap any consistent state to D_OUTDATED, - * but disallow "upgrade" of not even consistent states. - */ - ns.pdsk = - (D_DISKLESS < os.pdsk && os.pdsk < D_OUTDATED) - ? os.pdsk : D_OUTDATED; - break; - case C_SYNC_SOURCE: - ns.pdsk = D_INCONSISTENT; - dev_warn(DEV, "Implicitly set pdsk Inconsistent!\n"); - break; - } - if (os.pdsk == D_OUTDATED && ns.pdsk == D_UP_TO_DATE) - dev_warn(DEV, "Implicitly set pdsk from Outdated to UpToDate\n"); - } - /* Connection breaks down before we finished "Negotiating" */ if (ns.conn < C_CONNECTED && ns.disk == D_NEGOTIATING && get_ldev_if_state(mdev, D_NEGOTIATING)) { @@ -933,6 +884,94 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state put_ldev(mdev); } + /* D_CONSISTENT and D_OUTDATED vanish when we get connected */ + if (ns.conn >= C_CONNECTED && ns.conn < C_AHEAD) { + if (ns.disk == D_CONSISTENT || ns.disk == D_OUTDATED) + ns.disk = D_UP_TO_DATE; + if (ns.pdsk == D_CONSISTENT || ns.pdsk == D_OUTDATED) + ns.pdsk = D_UP_TO_DATE; + } + + /* Implications of the connection stat on the disk states */ + disk_min = D_DISKLESS; + disk_max = D_UP_TO_DATE; + pdsk_min = D_INCONSISTENT; + pdsk_max = D_UNKNOWN; + switch ((enum drbd_conns)ns.conn) { + case C_WF_BITMAP_T: + case C_PAUSED_SYNC_T: + case C_STARTING_SYNC_T: + case C_WF_SYNC_UUID: + case C_BEHIND: + disk_min = D_INCONSISTENT; + disk_max = D_OUTDATED; + pdsk_min = D_UP_TO_DATE; + pdsk_max = D_UP_TO_DATE; + break; + case C_VERIFY_S: + case C_VERIFY_T: + disk_min = D_UP_TO_DATE; + disk_max = D_UP_TO_DATE; + pdsk_min = D_UP_TO_DATE; + pdsk_max = D_UP_TO_DATE; + break; + case C_CONNECTED: + disk_min = D_DISKLESS; + disk_max = D_UP_TO_DATE; + pdsk_min = D_DISKLESS; + pdsk_max = D_UP_TO_DATE; + break; + case C_WF_BITMAP_S: + case C_PAUSED_SYNC_S: + case C_STARTING_SYNC_S: + case C_AHEAD: + disk_min = D_UP_TO_DATE; + disk_max = D_UP_TO_DATE; + pdsk_min = D_INCONSISTENT; + pdsk_max = D_CONSISTENT; /* D_OUTDATED would be nice. But explicit outdate necessary*/ + break; + case C_SYNC_TARGET: + disk_min = D_INCONSISTENT; + disk_max = D_INCONSISTENT; + pdsk_min = D_UP_TO_DATE; + pdsk_max = D_UP_TO_DATE; + break; + case C_SYNC_SOURCE: + disk_min = D_UP_TO_DATE; + disk_max = D_UP_TO_DATE; + pdsk_min = D_INCONSISTENT; + pdsk_max = D_INCONSISTENT; + break; + case C_STANDALONE: + case C_DISCONNECTING: + case C_UNCONNECTED: + case C_TIMEOUT: + case C_BROKEN_PIPE: + case C_NETWORK_FAILURE: + case C_PROTOCOL_ERROR: + case C_TEAR_DOWN: + case C_WF_CONNECTION: + case C_WF_REPORT_PARAMS: + case C_MASK: + break; + } + if (ns.disk > disk_max) + ns.disk = disk_max; + + if (ns.disk < disk_min) { + dev_warn(DEV, "Implicitly set disk from %s to %s\n", + drbd_disk_str(ns.disk), drbd_disk_str(disk_min)); + ns.disk = disk_min; + } + if (ns.pdsk > pdsk_max) + ns.pdsk = pdsk_max; + + if (ns.pdsk < pdsk_min) { + dev_warn(DEV, "Implicitly set pdsk from %s to %s\n", + drbd_disk_str(ns.pdsk), drbd_disk_str(pdsk_min)); + ns.pdsk = pdsk_min; + } + if (fp == FP_STONITH && (ns.role == R_PRIMARY && ns.conn < C_CONNECTED && ns.pdsk > D_OUTDATED) && !(os.role == R_PRIMARY && os.conn < C_CONNECTED && os.pdsk > D_OUTDATED)) @@ -961,6 +1000,10 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state /* helper for __drbd_set_state */ static void set_ov_position(struct drbd_conf *mdev, enum drbd_conns cs) { + if (mdev->agreed_pro_version < 90) + mdev->ov_start_sector = 0; + mdev->rs_total = drbd_bm_bits(mdev); + mdev->ov_position = 0; if (cs == C_VERIFY_T) { /* starting online verify from an arbitrary position * does not fit well into the existing protocol. @@ -970,11 +1013,15 @@ static void set_ov_position(struct drbd_conf *mdev, enum drbd_conns cs) mdev->ov_start_sector = ~(sector_t)0; } else { unsigned long bit = BM_SECT_TO_BIT(mdev->ov_start_sector); - if (bit >= mdev->rs_total) + if (bit >= mdev->rs_total) { mdev->ov_start_sector = BM_BIT_TO_SECT(mdev->rs_total - 1); + mdev->rs_total = 1; + } else + mdev->rs_total -= bit; mdev->ov_position = mdev->ov_start_sector; } + mdev->ov_left = mdev->rs_total; } static void drbd_resume_al(struct drbd_conf *mdev) @@ -992,12 +1039,12 @@ static void drbd_resume_al(struct drbd_conf *mdev) * * Caller needs to hold req_lock, and global_state_lock. Do not call directly. */ -int __drbd_set_state(struct drbd_conf *mdev, - union drbd_state ns, enum chg_state_flags flags, - struct completion *done) +enum drbd_state_rv +__drbd_set_state(struct drbd_conf *mdev, union drbd_state ns, + enum chg_state_flags flags, struct completion *done) { union drbd_state os; - int rv = SS_SUCCESS; + enum drbd_state_rv rv = SS_SUCCESS; const char *warn_sync_abort = NULL; struct after_state_chg_work *ascw; @@ -1033,22 +1080,46 @@ int __drbd_set_state(struct drbd_conf *mdev, dev_warn(DEV, "%s aborted.\n", warn_sync_abort); { - char *pbp, pb[300]; - pbp = pb; - *pbp = 0; - PSC(role); - PSC(peer); - PSC(conn); - PSC(disk); - PSC(pdsk); - if (is_susp(ns) != is_susp(os)) - pbp += sprintf(pbp, "susp( %s -> %s ) ", - drbd_susp_str(is_susp(os)), - drbd_susp_str(is_susp(ns))); - PSC(aftr_isp); - PSC(peer_isp); - PSC(user_isp); - dev_info(DEV, "%s\n", pb); + char *pbp, pb[300]; + pbp = pb; + *pbp = 0; + if (ns.role != os.role) + pbp += sprintf(pbp, "role( %s -> %s ) ", + drbd_role_str(os.role), + drbd_role_str(ns.role)); + if (ns.peer != os.peer) + pbp += sprintf(pbp, "peer( %s -> %s ) ", + drbd_role_str(os.peer), + drbd_role_str(ns.peer)); + if (ns.conn != os.conn) + pbp += sprintf(pbp, "conn( %s -> %s ) ", + drbd_conn_str(os.conn), + drbd_conn_str(ns.conn)); + if (ns.disk != os.disk) + pbp += sprintf(pbp, "disk( %s -> %s ) ", + drbd_disk_str(os.disk), + drbd_disk_str(ns.disk)); + if (ns.pdsk != os.pdsk) + pbp += sprintf(pbp, "pdsk( %s -> %s ) ", + drbd_disk_str(os.pdsk), + drbd_disk_str(ns.pdsk)); + if (is_susp(ns) != is_susp(os)) + pbp += sprintf(pbp, "susp( %d -> %d ) ", + is_susp(os), + is_susp(ns)); + if (ns.aftr_isp != os.aftr_isp) + pbp += sprintf(pbp, "aftr_isp( %d -> %d ) ", + os.aftr_isp, + ns.aftr_isp); + if (ns.peer_isp != os.peer_isp) + pbp += sprintf(pbp, "peer_isp( %d -> %d ) ", + os.peer_isp, + ns.peer_isp); + if (ns.user_isp != os.user_isp) + pbp += sprintf(pbp, "user_isp( %d -> %d ) ", + os.user_isp, + ns.user_isp); + dev_info(DEV, "%s\n", pb); } /* solve the race between becoming unconfigured, @@ -1074,6 +1145,10 @@ int __drbd_set_state(struct drbd_conf *mdev, atomic_inc(&mdev->local_cnt); mdev->state = ns; + + if (os.disk == D_ATTACHING && ns.disk >= D_NEGOTIATING) + drbd_print_uuids(mdev, "attached to UUIDs"); + wake_up(&mdev->misc_wait); wake_up(&mdev->state_wait); @@ -1081,7 +1156,7 @@ int __drbd_set_state(struct drbd_conf *mdev, if ((os.conn == C_VERIFY_S || os.conn == C_VERIFY_T) && ns.conn < C_CONNECTED) { mdev->ov_start_sector = - BM_BIT_TO_SECT(mdev->rs_total - mdev->ov_left); + BM_BIT_TO_SECT(drbd_bm_bits(mdev) - mdev->ov_left); dev_info(DEV, "Online Verify reached sector %llu\n", (unsigned long long)mdev->ov_start_sector); } @@ -1106,14 +1181,7 @@ int __drbd_set_state(struct drbd_conf *mdev, unsigned long now = jiffies; int i; - mdev->ov_position = 0; - mdev->rs_total = drbd_bm_bits(mdev); - if (mdev->agreed_pro_version >= 90) - set_ov_position(mdev, ns.conn); - else - mdev->ov_start_sector = 0; - mdev->ov_left = mdev->rs_total - - BM_SECT_TO_BIT(mdev->ov_position); + set_ov_position(mdev, ns.conn); mdev->rs_start = now; mdev->rs_last_events = 0; mdev->rs_last_sect_ev = 0; @@ -1121,10 +1189,12 @@ int __drbd_set_state(struct drbd_conf *mdev, mdev->ov_last_oos_start = 0; for (i = 0; i < DRBD_SYNC_MARKS; i++) { - mdev->rs_mark_left[i] = mdev->rs_total; + mdev->rs_mark_left[i] = mdev->ov_left; mdev->rs_mark_time[i] = now; } + drbd_rs_controller_reset(mdev); + if (ns.conn == C_VERIFY_S) { dev_info(DEV, "Starting Online Verify from sector %llu\n", (unsigned long long)mdev->ov_position); @@ -1228,6 +1298,26 @@ static void abw_start_sync(struct drbd_conf *mdev, int rv) } } +int drbd_bitmap_io_from_worker(struct drbd_conf *mdev, + int (*io_fn)(struct drbd_conf *), + char *why, enum bm_flag flags) +{ + int rv; + + D_ASSERT(current == mdev->worker.task); + + /* open coded non-blocking drbd_suspend_io(mdev); */ + set_bit(SUSPEND_IO, &mdev->flags); + + drbd_bm_lock(mdev, why, flags); + rv = io_fn(mdev); + drbd_bm_unlock(mdev); + + drbd_resume_io(mdev); + + return rv; +} + /** * after_state_ch() - Perform after state change actions that may sleep * @mdev: DRBD device. @@ -1266,16 +1356,14 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, nsm.i = -1; if (ns.susp_nod) { - if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) { - if (ns.conn == C_CONNECTED) - what = resend, nsm.susp_nod = 0; - else /* ns.conn > C_CONNECTED */ - dev_err(DEV, "Unexpected Resynd going on!\n"); - } + if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) + what = resend; if (os.disk == D_ATTACHING && ns.disk > D_ATTACHING) - what = restart_frozen_disk_io, nsm.susp_nod = 0; + what = restart_frozen_disk_io; + if (what != nothing) + nsm.susp_nod = 0; } if (ns.susp_fen) { @@ -1306,13 +1394,30 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, spin_unlock_irq(&mdev->req_lock); } + /* Became sync source. With protocol >= 96, we still need to send out + * the sync uuid now. Need to do that before any drbd_send_state, or + * the other side may go "paused sync" before receiving the sync uuids, + * which is unexpected. */ + if ((os.conn != C_SYNC_SOURCE && os.conn != C_PAUSED_SYNC_S) && + (ns.conn == C_SYNC_SOURCE || ns.conn == C_PAUSED_SYNC_S) && + mdev->agreed_pro_version >= 96 && get_ldev(mdev)) { + drbd_gen_and_send_sync_uuid(mdev); + put_ldev(mdev); + } + /* Do not change the order of the if above and the two below... */ if (os.pdsk == D_DISKLESS && ns.pdsk > D_DISKLESS) { /* attach on the peer */ drbd_send_uuids(mdev); drbd_send_state(mdev); } - if (os.conn != C_WF_BITMAP_S && ns.conn == C_WF_BITMAP_S) - drbd_queue_bitmap_io(mdev, &drbd_send_bitmap, NULL, "send_bitmap (WFBitMapS)"); + /* No point in queuing send_bitmap if we don't have a connection + * anymore, so check also the _current_ state, not only the new state + * at the time this work was queued. */ + if (os.conn != C_WF_BITMAP_S && ns.conn == C_WF_BITMAP_S && + mdev->state.conn == C_WF_BITMAP_S) + drbd_queue_bitmap_io(mdev, &drbd_send_bitmap, NULL, + "send_bitmap (WFBitMapS)", + BM_LOCKED_TEST_ALLOWED); /* Lost contact to peer's copy of the data */ if ((os.pdsk >= D_INCONSISTENT && @@ -1343,7 +1448,23 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, /* D_DISKLESS Peer becomes secondary */ if (os.peer == R_PRIMARY && ns.peer == R_SECONDARY) - drbd_al_to_on_disk_bm(mdev); + /* We may still be Primary ourselves. + * No harm done if the bitmap still changes, + * redirtied pages will follow later. */ + drbd_bitmap_io_from_worker(mdev, &drbd_bm_write, + "demote diskless peer", BM_LOCKED_SET_ALLOWED); + put_ldev(mdev); + } + + /* Write out all changed bits on demote. + * Though, no need to da that just yet + * if there is a resync going on still */ + if (os.role == R_PRIMARY && ns.role == R_SECONDARY && + mdev->state.conn <= C_CONNECTED && get_ldev(mdev)) { + /* No changes to the bitmap expected this time, so assert that, + * even though no harm was done if it did change. */ + drbd_bitmap_io_from_worker(mdev, &drbd_bm_write, + "demote", BM_LOCKED_TEST_ALLOWED); put_ldev(mdev); } @@ -1371,15 +1492,23 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, if (os.conn == C_WF_REPORT_PARAMS && ns.conn >= C_CONNECTED) drbd_send_state(mdev); + if (os.conn != C_AHEAD && ns.conn == C_AHEAD) + drbd_send_state(mdev); + /* We are in the progress to start a full sync... */ if ((os.conn != C_STARTING_SYNC_T && ns.conn == C_STARTING_SYNC_T) || (os.conn != C_STARTING_SYNC_S && ns.conn == C_STARTING_SYNC_S)) - drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, &abw_start_sync, "set_n_write from StartingSync"); + /* no other bitmap changes expected during this phase */ + drbd_queue_bitmap_io(mdev, + &drbd_bmio_set_n_write, &abw_start_sync, + "set_n_write from StartingSync", BM_LOCKED_TEST_ALLOWED); /* We are invalidating our self... */ if (os.conn < C_CONNECTED && ns.conn < C_CONNECTED && os.disk > D_INCONSISTENT && ns.disk == D_INCONSISTENT) - drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, NULL, "set_n_write from invalidate"); + /* other bitmap operation expected during this phase */ + drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, NULL, + "set_n_write from invalidate", BM_LOCKED_MASK); /* first half of local IO error, failure to attach, * or administrative detach */ @@ -1434,13 +1563,15 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, if (drbd_send_state(mdev)) dev_warn(DEV, "Notified peer that I'm now diskless.\n"); - else - dev_err(DEV, "Sending state for being diskless failed\n"); /* corresponding get_ldev in __drbd_set_state - * this may finaly trigger drbd_ldev_destroy. */ + * this may finally trigger drbd_ldev_destroy. */ put_ldev(mdev); } + /* Notify peer that I had a local IO error, and did not detached.. */ + if (os.disk == D_UP_TO_DATE && ns.disk == D_INCONSISTENT) + drbd_send_state(mdev); + /* Disks got bigger while they were detached */ if (ns.disk > D_NEGOTIATING && ns.pdsk > D_NEGOTIATING && test_and_clear_bit(RESYNC_AFTER_NEG, &mdev->flags)) { @@ -1459,6 +1590,19 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, if (os.disk < D_UP_TO_DATE && os.conn >= C_SYNC_SOURCE && ns.conn == C_CONNECTED) drbd_send_state(mdev); + /* This triggers bitmap writeout of potentially still unwritten pages + * if the resync finished cleanly, or aborted because of peer disk + * failure, or because of connection loss. + * For resync aborted because of local disk failure, we cannot do + * any bitmap writeout anymore. + * No harm done if some bits change during this phase. + */ + if (os.conn > C_CONNECTED && ns.conn <= C_CONNECTED && get_ldev(mdev)) { + drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, + "write from resync_finished", BM_LOCKED_SET_ALLOWED); + put_ldev(mdev); + } + /* free tl_hash if we Got thawed and are C_STANDALONE */ if (ns.conn == C_STANDALONE && !is_susp(ns) && mdev->tl_hash) drbd_free_tl_hash(mdev); @@ -1559,7 +1703,7 @@ int drbd_thread_start(struct drbd_thread *thi) if (!try_module_get(THIS_MODULE)) { dev_err(DEV, "Failed to get module reference in drbd_thread_start\n"); spin_unlock_irqrestore(&thi->t_lock, flags); - return FALSE; + return false; } init_completion(&thi->stop); @@ -1576,7 +1720,7 @@ int drbd_thread_start(struct drbd_thread *thi) dev_err(DEV, "Couldn't start thread\n"); module_put(THIS_MODULE); - return FALSE; + return false; } spin_lock_irqsave(&thi->t_lock, flags); thi->task = nt; @@ -1596,7 +1740,7 @@ int drbd_thread_start(struct drbd_thread *thi) break; } - return TRUE; + return true; } @@ -1694,8 +1838,8 @@ int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, { int sent, ok; - ERR_IF(!h) return FALSE; - ERR_IF(!size) return FALSE; + ERR_IF(!h) return false; + ERR_IF(!size) return false; h->magic = BE_DRBD_MAGIC; h->command = cpu_to_be16(cmd); @@ -1704,8 +1848,8 @@ int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, sent = drbd_send(mdev, sock, h, size, msg_flags); ok = (sent == size); - if (!ok) - dev_err(DEV, "short sent %s size=%d sent=%d\n", + if (!ok && !signal_pending(current)) + dev_warn(DEV, "short sent %s size=%d sent=%d\n", cmdname(cmd), (int)size, sent); return ok; } @@ -1840,7 +1984,7 @@ int drbd_send_protocol(struct drbd_conf *mdev) else { dev_err(DEV, "--dry-run is not supported by peer"); kfree(p); - return 0; + return -1; } } p->conn_flags = cpu_to_be32(cf); @@ -1888,12 +2032,36 @@ int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev) return _drbd_send_uuids(mdev, 8); } +void drbd_print_uuids(struct drbd_conf *mdev, const char *text) +{ + if (get_ldev_if_state(mdev, D_NEGOTIATING)) { + u64 *uuid = mdev->ldev->md.uuid; + dev_info(DEV, "%s %016llX:%016llX:%016llX:%016llX\n", + text, + (unsigned long long)uuid[UI_CURRENT], + (unsigned long long)uuid[UI_BITMAP], + (unsigned long long)uuid[UI_HISTORY_START], + (unsigned long long)uuid[UI_HISTORY_END]); + put_ldev(mdev); + } else { + dev_info(DEV, "%s effective data uuid: %016llX\n", + text, + (unsigned long long)mdev->ed_uuid); + } +} -int drbd_send_sync_uuid(struct drbd_conf *mdev, u64 val) +int drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev) { struct p_rs_uuid p; + u64 uuid; - p.uuid = cpu_to_be64(val); + D_ASSERT(mdev->state.disk == D_UP_TO_DATE); + + uuid = mdev->ldev->md.uuid[UI_BITMAP] + UUID_NEW_BM_OFFSET; + drbd_uuid_set(mdev, UI_BITMAP, uuid); + drbd_print_uuids(mdev, "updated sync UUID"); + drbd_md_sync(mdev); + p.uuid = cpu_to_be64(uuid); return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_SYNC_UUID, (struct p_header80 *)&p, sizeof(p)); @@ -1903,7 +2071,7 @@ int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags fl { struct p_sizes p; sector_t d_size, u_size; - int q_order_type; + int q_order_type, max_bio_size; int ok; if (get_ldev_if_state(mdev, D_NEGOTIATING)) { @@ -1911,17 +2079,20 @@ int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags fl d_size = drbd_get_max_capacity(mdev->ldev); u_size = mdev->ldev->dc.disk_size; q_order_type = drbd_queue_order_type(mdev); + max_bio_size = queue_max_hw_sectors(mdev->ldev->backing_bdev->bd_disk->queue) << 9; + max_bio_size = min_t(int, max_bio_size, DRBD_MAX_BIO_SIZE); put_ldev(mdev); } else { d_size = 0; u_size = 0; q_order_type = QUEUE_ORDERED_NONE; + max_bio_size = DRBD_MAX_BIO_SIZE; /* ... multiple BIOs per peer_request */ } p.d_size = cpu_to_be64(d_size); p.u_size = cpu_to_be64(u_size); p.c_size = cpu_to_be64(trigger_reply ? 0 : drbd_get_capacity(mdev->this_bdev)); - p.max_segment_size = cpu_to_be32(queue_max_segment_size(mdev->rq_queue)); + p.max_bio_size = cpu_to_be32(max_bio_size); p.queue_order_type = cpu_to_be16(q_order_type); p.dds_flags = cpu_to_be16(flags); @@ -1972,7 +2143,7 @@ int drbd_send_state_req(struct drbd_conf *mdev, (struct p_header80 *)&p, sizeof(p)); } -int drbd_send_sr_reply(struct drbd_conf *mdev, int retcode) +int drbd_send_sr_reply(struct drbd_conf *mdev, enum drbd_state_rv retcode) { struct p_req_state_reply p; @@ -2076,9 +2247,15 @@ int fill_bitmap_rle_bits(struct drbd_conf *mdev, return len; } -enum { OK, FAILED, DONE } +/** + * send_bitmap_rle_or_plain + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int send_bitmap_rle_or_plain(struct drbd_conf *mdev, - struct p_header80 *h, struct bm_xfer_ctx *c) + struct p_header80 *h, struct bm_xfer_ctx *c) { struct p_compressed_bm *p = (void*)h; unsigned long num_words; @@ -2088,7 +2265,7 @@ send_bitmap_rle_or_plain(struct drbd_conf *mdev, len = fill_bitmap_rle_bits(mdev, p, c); if (len < 0) - return FAILED; + return -EIO; if (len) { DCBP_set_code(p, RLE_VLI_Bits); @@ -2118,11 +2295,14 @@ send_bitmap_rle_or_plain(struct drbd_conf *mdev, if (c->bit_offset > c->bm_bits) c->bit_offset = c->bm_bits; } - ok = ok ? ((len == 0) ? DONE : OK) : FAILED; - - if (ok == DONE) - INFO_bm_xfer_stats(mdev, "send", c); - return ok; + if (ok) { + if (len == 0) { + INFO_bm_xfer_stats(mdev, "send", c); + return 0; + } else + return 1; + } + return -EIO; } /* See the comment at receive_bitmap() */ @@ -2130,16 +2310,16 @@ int _drbd_send_bitmap(struct drbd_conf *mdev) { struct bm_xfer_ctx c; struct p_header80 *p; - int ret; + int err; - ERR_IF(!mdev->bitmap) return FALSE; + ERR_IF(!mdev->bitmap) return false; /* maybe we should use some per thread scratch page, * and allocate that during initial device creation? */ p = (struct p_header80 *) __get_free_page(GFP_NOIO); if (!p) { dev_err(DEV, "failed to allocate one page buffer in %s\n", __func__); - return FALSE; + return false; } if (get_ldev(mdev)) { @@ -2165,11 +2345,11 @@ int _drbd_send_bitmap(struct drbd_conf *mdev) }; do { - ret = send_bitmap_rle_or_plain(mdev, p, &c); - } while (ret == OK); + err = send_bitmap_rle_or_plain(mdev, p, &c); + } while (err > 0); free_page((unsigned long) p); - return (ret == DONE); + return err == 0; } int drbd_send_bitmap(struct drbd_conf *mdev) @@ -2192,7 +2372,7 @@ int drbd_send_b_ack(struct drbd_conf *mdev, u32 barrier_nr, u32 set_size) p.set_size = cpu_to_be32(set_size); if (mdev->state.conn < C_CONNECTED) - return FALSE; + return false; ok = drbd_send_cmd(mdev, USE_META_SOCKET, P_BARRIER_ACK, (struct p_header80 *)&p, sizeof(p)); return ok; @@ -2220,7 +2400,7 @@ static int _drbd_send_ack(struct drbd_conf *mdev, enum drbd_packets cmd, p.seq_num = cpu_to_be32(atomic_add_return(1, &mdev->packet_seq)); if (!mdev->meta.socket || mdev->state.conn < C_CONNECTED) - return FALSE; + return false; ok = drbd_send_cmd(mdev, USE_META_SOCKET, cmd, (struct p_header80 *)&p, sizeof(p)); return ok; @@ -2326,8 +2506,8 @@ int drbd_send_ov_request(struct drbd_conf *mdev, sector_t sector, int size) } /* called on sndtimeo - * returns FALSE if we should retry, - * TRUE if we think connection is dead + * returns false if we should retry, + * true if we think connection is dead */ static int we_should_drop_the_connection(struct drbd_conf *mdev, struct socket *sock) { @@ -2340,7 +2520,7 @@ static int we_should_drop_the_connection(struct drbd_conf *mdev, struct socket * || mdev->state.conn < C_CONNECTED; if (drop_it) - return TRUE; + return true; drop_it = !--mdev->ko_count; if (!drop_it) { @@ -2477,12 +2657,11 @@ static u32 bio_flags_to_wire(struct drbd_conf *mdev, unsigned long bi_rw) { if (mdev->agreed_pro_version >= 95) return (bi_rw & REQ_SYNC ? DP_RW_SYNC : 0) | - (bi_rw & REQ_UNPLUG ? DP_UNPLUG : 0) | (bi_rw & REQ_FUA ? DP_FUA : 0) | (bi_rw & REQ_FLUSH ? DP_FLUSH : 0) | (bi_rw & REQ_DISCARD ? DP_DISCARD : 0); else - return bi_rw & (REQ_SYNC | REQ_UNPLUG) ? DP_RW_SYNC : 0; + return bi_rw & REQ_SYNC ? DP_RW_SYNC : 0; } /* Used to send write requests @@ -2532,13 +2711,39 @@ int drbd_send_dblock(struct drbd_conf *mdev, struct drbd_request *req) if (ok && dgs) { dgb = mdev->int_dig_out; drbd_csum_bio(mdev, mdev->integrity_w_tfm, req->master_bio, dgb); - ok = drbd_send(mdev, mdev->data.socket, dgb, dgs, 0); + ok = dgs == drbd_send(mdev, mdev->data.socket, dgb, dgs, 0); } if (ok) { - if (mdev->net_conf->wire_protocol == DRBD_PROT_A) + /* For protocol A, we have to memcpy the payload into + * socket buffers, as we may complete right away + * as soon as we handed it over to tcp, at which point the data + * pages may become invalid. + * + * For data-integrity enabled, we copy it as well, so we can be + * sure that even if the bio pages may still be modified, it + * won't change the data on the wire, thus if the digest checks + * out ok after sending on this side, but does not fit on the + * receiving side, we sure have detected corruption elsewhere. + */ + if (mdev->net_conf->wire_protocol == DRBD_PROT_A || dgs) ok = _drbd_send_bio(mdev, req->master_bio); else ok = _drbd_send_zc_bio(mdev, req->master_bio); + + /* double check digest, sometimes buffers have been modified in flight. */ + if (dgs > 0 && dgs <= 64) { + /* 64 byte, 512 bit, is the largest digest size + * currently supported in kernel crypto. */ + unsigned char digest[64]; + drbd_csum_bio(mdev, mdev->integrity_w_tfm, req->master_bio, digest); + if (memcmp(mdev->int_dig_out, digest, dgs)) { + dev_warn(DEV, + "Digest mismatch, buffer modified by upper layers during write: %llus +%u\n", + (unsigned long long)req->sector, req->size); + } + } /* else if (dgs > 64) { + ... Be noisy about digest too large ... + } */ } drbd_put_data_sock(mdev); @@ -2588,7 +2793,7 @@ int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, if (ok && dgs) { dgb = mdev->int_dig_out; drbd_csum_ee(mdev, mdev->integrity_w_tfm, e, dgb); - ok = drbd_send(mdev, mdev->data.socket, dgb, dgs, 0); + ok = dgs == drbd_send(mdev, mdev->data.socket, dgb, dgs, 0); } if (ok) ok = _drbd_send_zc_ee(mdev, e); @@ -2598,6 +2803,16 @@ int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, return ok; } +int drbd_send_oos(struct drbd_conf *mdev, struct drbd_request *req) +{ + struct p_block_desc p; + + p.sector = cpu_to_be64(req->sector); + p.blksize = cpu_to_be32(req->size); + + return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_OUT_OF_SYNC, &p.head, sizeof(p)); +} + /* drbd_send distinguishes two cases: @@ -2719,35 +2934,6 @@ static int drbd_release(struct gendisk *gd, fmode_t mode) return 0; } -static void drbd_unplug_fn(struct request_queue *q) -{ - struct drbd_conf *mdev = q->queuedata; - - /* unplug FIRST */ - spin_lock_irq(q->queue_lock); - blk_remove_plug(q); - spin_unlock_irq(q->queue_lock); - - /* only if connected */ - spin_lock_irq(&mdev->req_lock); - if (mdev->state.pdsk >= D_INCONSISTENT && mdev->state.conn >= C_CONNECTED) { - D_ASSERT(mdev->state.role == R_PRIMARY); - if (test_and_clear_bit(UNPLUG_REMOTE, &mdev->flags)) { - /* add to the data.work queue, - * unless already queued. - * XXX this might be a good addition to drbd_queue_work - * anyways, to detect "double queuing" ... */ - if (list_empty(&mdev->unplug_work.list)) - drbd_queue_work(&mdev->data.work, - &mdev->unplug_work); - } - } - spin_unlock_irq(&mdev->req_lock); - - if (mdev->state.disk >= D_INCONSISTENT) - drbd_kick_lo(mdev); -} - static void drbd_set_defaults(struct drbd_conf *mdev) { /* This way we get a compile error when sync_conf grows, @@ -2800,6 +2986,7 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) atomic_set(&mdev->pp_in_use_by_net, 0); atomic_set(&mdev->rs_sect_in, 0); atomic_set(&mdev->rs_sect_ev, 0); + atomic_set(&mdev->ap_in_flight, 0); mutex_init(&mdev->md_io_mutex); mutex_init(&mdev->data.mutex); @@ -2828,19 +3015,27 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) INIT_LIST_HEAD(&mdev->unplug_work.list); INIT_LIST_HEAD(&mdev->go_diskless.list); INIT_LIST_HEAD(&mdev->md_sync_work.list); + INIT_LIST_HEAD(&mdev->start_resync_work.list); INIT_LIST_HEAD(&mdev->bm_io_work.w.list); - mdev->resync_work.cb = w_resync_inactive; + mdev->resync_work.cb = w_resync_timer; mdev->unplug_work.cb = w_send_write_hint; mdev->go_diskless.cb = w_go_diskless; mdev->md_sync_work.cb = w_md_sync; mdev->bm_io_work.w.cb = w_bitmap_io; + mdev->start_resync_work.cb = w_start_resync; init_timer(&mdev->resync_timer); init_timer(&mdev->md_sync_timer); + init_timer(&mdev->start_resync_timer); + init_timer(&mdev->request_timer); mdev->resync_timer.function = resync_timer_fn; mdev->resync_timer.data = (unsigned long) mdev; mdev->md_sync_timer.function = md_sync_timer_fn; mdev->md_sync_timer.data = (unsigned long) mdev; + mdev->start_resync_timer.function = start_resync_timer_fn; + mdev->start_resync_timer.data = (unsigned long) mdev; + mdev->request_timer.function = request_timer_fn; + mdev->request_timer.data = (unsigned long) mdev; init_waitqueue_head(&mdev->misc_wait); init_waitqueue_head(&mdev->state_wait); @@ -2856,6 +3051,8 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) mdev->agreed_pro_version = PRO_VERSION_MAX; mdev->write_ordering = WO_bdev_flush; mdev->resync_wenr = LC_FREE; + mdev->peer_max_bio_size = DRBD_MAX_BIO_SIZE_SAFE; + mdev->local_max_bio_size = DRBD_MAX_BIO_SIZE_SAFE; } void drbd_mdev_cleanup(struct drbd_conf *mdev) @@ -2911,6 +3108,8 @@ void drbd_mdev_cleanup(struct drbd_conf *mdev) D_ASSERT(list_empty(&mdev->resync_work.list)); D_ASSERT(list_empty(&mdev->unplug_work.list)); D_ASSERT(list_empty(&mdev->go_diskless.list)); + + drbd_set_defaults(mdev); } @@ -2953,7 +3152,7 @@ static void drbd_destroy_mempools(void) static int drbd_create_mempools(void) { struct page *page; - const int number = (DRBD_MAX_SEGMENT_SIZE/PAGE_SIZE) * minor_count; + const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count; int i; /* prepare our caches and mempools */ @@ -3088,7 +3287,7 @@ static void drbd_delete_device(unsigned int minor) drbd_release_ee_lists(mdev); - /* should be free'd on disconnect? */ + /* should be freed on disconnect? */ kfree(mdev->ee_hash); /* mdev->ee_hash_s = 0; @@ -3117,11 +3316,20 @@ static void drbd_cleanup(void) unregister_reboot_notifier(&drbd_notifier); + /* first remove proc, + * drbdsetup uses it's presence to detect + * whether DRBD is loaded. + * If we would get stuck in proc removal, + * but have netlink already deregistered, + * some drbdsetup commands may wait forever + * for an answer. + */ + if (drbd_proc) + remove_proc_entry("drbd", NULL); + drbd_nl_cleanup(); if (minor_table) { - if (drbd_proc) - remove_proc_entry("drbd", NULL); i = minor_count; while (i--) drbd_delete_device(i); @@ -3149,7 +3357,7 @@ static int drbd_congested(void *congested_data, int bdi_bits) char reason = '-'; int r = 0; - if (!__inc_ap_bio_cond(mdev)) { + if (!may_inc_ap_bio(mdev)) { /* DRBD has frozen IO */ r = bdi_bits; reason = 'd'; @@ -3202,7 +3410,7 @@ struct drbd_conf *drbd_new_device(unsigned int minor) goto out_no_disk; mdev->vdisk = disk; - set_disk_ro(disk, TRUE); + set_disk_ro(disk, true); disk->queue = q; disk->major = DRBD_MAJOR; @@ -3218,13 +3426,13 @@ struct drbd_conf *drbd_new_device(unsigned int minor) q->backing_dev_info.congested_fn = drbd_congested; q->backing_dev_info.congested_data = mdev; - blk_queue_make_request(q, drbd_make_request_26); - blk_queue_max_segment_size(q, DRBD_MAX_SEGMENT_SIZE); + blk_queue_make_request(q, drbd_make_request); + /* Setting the max_hw_sectors to an odd value of 8kibyte here + This triggers a max_bio_size message upon first attach or connect */ + blk_queue_max_hw_sectors(q, DRBD_MAX_BIO_SIZE_SAFE >> 8); blk_queue_bounce_limit(q, BLK_BOUNCE_ANY); blk_queue_merge_bvec(q, drbd_merge_bvec); - q->queue_lock = &mdev->req_lock; /* needed since we use */ - /* plugging on a queue, that actually has no requests! */ - q->unplug_fn = drbd_unplug_fn; + q->queue_lock = &mdev->req_lock; mdev->md_io_page = alloc_page(GFP_KERNEL); if (!mdev->md_io_page) @@ -3283,6 +3491,7 @@ void drbd_free_mdev(struct drbd_conf *mdev) put_disk(mdev->vdisk); blk_cleanup_queue(mdev->rq_queue); free_cpumask_var(mdev->cpu_mask); + drbd_free_tl_hash(mdev); kfree(mdev); } @@ -3298,7 +3507,7 @@ int __init drbd_init(void) return -EINVAL; } - if (1 > minor_count || minor_count > 255) { + if (minor_count < DRBD_MINOR_COUNT_MIN || minor_count > DRBD_MINOR_COUNT_MAX) { printk(KERN_ERR "drbd: invalid minor_count (%d)\n", minor_count); #ifdef MODULE @@ -3432,7 +3641,8 @@ struct meta_data_on_disk { /* `-- act_log->nr_elements <-- sync_conf.al_extents */ u32 bm_offset; /* offset to the bitmap, from here */ u32 bm_bytes_per_bit; /* BM_BLOCK_SIZE */ - u32 reserved_u32[4]; + u32 la_peer_max_bio_size; /* last peer max_bio_size */ + u32 reserved_u32[3]; } __packed; @@ -3473,6 +3683,7 @@ void drbd_md_sync(struct drbd_conf *mdev) buffer->device_uuid = cpu_to_be64(mdev->ldev->md.device_uuid); buffer->bm_offset = cpu_to_be32(mdev->ldev->md.bm_offset); + buffer->la_peer_max_bio_size = cpu_to_be32(mdev->peer_max_bio_size); D_ASSERT(drbd_md_ss__(mdev, mdev->ldev) == mdev->ldev->md.md_offset); sector = mdev->ldev->md.md_offset; @@ -3480,7 +3691,7 @@ void drbd_md_sync(struct drbd_conf *mdev) if (!drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) { /* this was a try anyways ... */ dev_err(DEV, "meta data update failed!\n"); - drbd_chk_io_error(mdev, 1, TRUE); + drbd_chk_io_error(mdev, 1, true); } /* Update mdev->ldev->md.la_size_sect, @@ -3496,7 +3707,7 @@ void drbd_md_sync(struct drbd_conf *mdev) * @mdev: DRBD device. * @bdev: Device from which the meta data should be read in. * - * Return 0 (NO_ERROR) on success, and an enum drbd_ret_codes in case + * Return 0 (NO_ERROR) on success, and an enum drbd_ret_code in case * something goes wrong. Currently only: ERR_IO_MD_DISK, ERR_MD_INVALID. */ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev) @@ -3511,7 +3722,7 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev) buffer = (struct meta_data_on_disk *)page_address(mdev->md_io_page); if (!drbd_md_sync_page_io(mdev, bdev, bdev->md.md_offset, READ)) { - /* NOTE: cant do normal error processing here as this is + /* NOTE: can't do normal error processing here as this is called BEFORE disk is attached */ dev_err(DEV, "Error while reading metadata.\n"); rv = ERR_IO_MD_DISK; @@ -3556,6 +3767,15 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev) mdev->sync_conf.al_extents = be32_to_cpu(buffer->al_nr_extents); bdev->md.device_uuid = be64_to_cpu(buffer->device_uuid); + spin_lock_irq(&mdev->req_lock); + if (mdev->state.conn < C_CONNECTED) { + int peer; + peer = be32_to_cpu(buffer->la_peer_max_bio_size); + peer = max_t(int, peer, DRBD_MAX_BIO_SIZE_SAFE); + mdev->peer_max_bio_size = peer; + } + spin_unlock_irq(&mdev->req_lock); + if (mdev->sync_conf.al_extents < 7) mdev->sync_conf.al_extents = 127; @@ -3566,28 +3786,6 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev) return rv; } -static void debug_drbd_uuid(struct drbd_conf *mdev, enum drbd_uuid_index index) -{ - static char *uuid_str[UI_EXTENDED_SIZE] = { - [UI_CURRENT] = "CURRENT", - [UI_BITMAP] = "BITMAP", - [UI_HISTORY_START] = "HISTORY_START", - [UI_HISTORY_END] = "HISTORY_END", - [UI_SIZE] = "SIZE", - [UI_FLAGS] = "FLAGS", - }; - - if (index >= UI_EXTENDED_SIZE) { - dev_warn(DEV, " uuid_index >= EXTENDED_SIZE\n"); - return; - } - - dynamic_dev_dbg(DEV, " uuid[%s] now %016llX\n", - uuid_str[index], - (unsigned long long)mdev->ldev->md.uuid[index]); -} - - /** * drbd_md_mark_dirty() - Mark meta data super block as dirty * @mdev: DRBD device. @@ -3617,10 +3815,8 @@ static void drbd_uuid_move_history(struct drbd_conf *mdev) __must_hold(local) { int i; - for (i = UI_HISTORY_START; i < UI_HISTORY_END; i++) { + for (i = UI_HISTORY_START; i < UI_HISTORY_END; i++) mdev->ldev->md.uuid[i+1] = mdev->ldev->md.uuid[i]; - debug_drbd_uuid(mdev, i+1); - } } void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) @@ -3635,7 +3831,6 @@ void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) } mdev->ldev->md.uuid[idx] = val; - debug_drbd_uuid(mdev, idx); drbd_md_mark_dirty(mdev); } @@ -3645,7 +3840,6 @@ void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) if (mdev->ldev->md.uuid[idx]) { drbd_uuid_move_history(mdev); mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[idx]; - debug_drbd_uuid(mdev, UI_HISTORY_START); } _drbd_uuid_set(mdev, idx, val); } @@ -3660,14 +3854,16 @@ void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local) { u64 val; + unsigned long long bm_uuid = mdev->ldev->md.uuid[UI_BITMAP]; + + if (bm_uuid) + dev_warn(DEV, "bm UUID was already set: %llX\n", bm_uuid); - dev_info(DEV, "Creating new current UUID\n"); - D_ASSERT(mdev->ldev->md.uuid[UI_BITMAP] == 0); mdev->ldev->md.uuid[UI_BITMAP] = mdev->ldev->md.uuid[UI_CURRENT]; - debug_drbd_uuid(mdev, UI_BITMAP); get_random_bytes(&val, sizeof(u64)); _drbd_uuid_set(mdev, UI_CURRENT, val); + drbd_print_uuids(mdev, "new current UUID"); /* get it to stable storage _now_ */ drbd_md_sync(mdev); } @@ -3681,16 +3877,12 @@ void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local) drbd_uuid_move_history(mdev); mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[UI_BITMAP]; mdev->ldev->md.uuid[UI_BITMAP] = 0; - debug_drbd_uuid(mdev, UI_HISTORY_START); - debug_drbd_uuid(mdev, UI_BITMAP); } else { - if (mdev->ldev->md.uuid[UI_BITMAP]) - dev_warn(DEV, "bm UUID already set"); + unsigned long long bm_uuid = mdev->ldev->md.uuid[UI_BITMAP]; + if (bm_uuid) + dev_warn(DEV, "bm UUID was already set: %llX\n", bm_uuid); - mdev->ldev->md.uuid[UI_BITMAP] = val; - mdev->ldev->md.uuid[UI_BITMAP] &= ~((u64)1); - - debug_drbd_uuid(mdev, UI_BITMAP); + mdev->ldev->md.uuid[UI_BITMAP] = val & ~((u64)1); } drbd_md_mark_dirty(mdev); } @@ -3746,15 +3938,19 @@ int drbd_bmio_clear_n_write(struct drbd_conf *mdev) static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused) { struct bm_io_work *work = container_of(w, struct bm_io_work, w); - int rv; + int rv = -EIO; D_ASSERT(atomic_read(&mdev->ap_bio_cnt) == 0); - drbd_bm_lock(mdev, work->why); - rv = work->io_fn(mdev); - drbd_bm_unlock(mdev); + if (get_ldev(mdev)) { + drbd_bm_lock(mdev, work->why, work->flags); + rv = work->io_fn(mdev); + drbd_bm_unlock(mdev); + put_ldev(mdev); + } clear_bit(BITMAP_IO, &mdev->flags); + smp_mb__after_clear_bit(); wake_up(&mdev->misc_wait); if (work->done) @@ -3762,6 +3958,7 @@ static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused) clear_bit(BITMAP_IO_QUEUED, &mdev->flags); work->why = NULL; + work->flags = 0; return 1; } @@ -3816,7 +4013,7 @@ void drbd_go_diskless(struct drbd_conf *mdev) void drbd_queue_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), void (*done)(struct drbd_conf *, int), - char *why) + char *why, enum bm_flag flags) { D_ASSERT(current == mdev->worker.task); @@ -3830,15 +4027,15 @@ void drbd_queue_bitmap_io(struct drbd_conf *mdev, mdev->bm_io_work.io_fn = io_fn; mdev->bm_io_work.done = done; mdev->bm_io_work.why = why; + mdev->bm_io_work.flags = flags; + spin_lock_irq(&mdev->req_lock); set_bit(BITMAP_IO, &mdev->flags); if (atomic_read(&mdev->ap_bio_cnt) == 0) { - if (list_empty(&mdev->bm_io_work.w.list)) { - set_bit(BITMAP_IO_QUEUED, &mdev->flags); + if (!test_and_set_bit(BITMAP_IO_QUEUED, &mdev->flags)) drbd_queue_work(&mdev->data.work, &mdev->bm_io_work.w); - } else - dev_err(DEV, "FIXME avoided double queuing bm_io_work\n"); } + spin_unlock_irq(&mdev->req_lock); } /** @@ -3850,19 +4047,22 @@ void drbd_queue_bitmap_io(struct drbd_conf *mdev, * freezes application IO while that the actual IO operations runs. This * functions MAY NOT be called from worker context. */ -int drbd_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), char *why) +int drbd_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), + char *why, enum bm_flag flags) { int rv; D_ASSERT(current != mdev->worker.task); - drbd_suspend_io(mdev); + if ((flags & BM_LOCKED_SET_ALLOWED) == 0) + drbd_suspend_io(mdev); - drbd_bm_lock(mdev, why); + drbd_bm_lock(mdev, why, flags); rv = io_fn(mdev); drbd_bm_unlock(mdev); - drbd_resume_io(mdev); + if ((flags & BM_LOCKED_SET_ALLOWED) == 0) + drbd_resume_io(mdev); return rv; } diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index 8cbfaa6..515bcd9 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -272,9 +272,28 @@ static int _try_outdate_peer_async(void *data) { struct drbd_conf *mdev = (struct drbd_conf *)data; enum drbd_disk_state nps; + union drbd_state ns; nps = drbd_try_outdate_peer(mdev); - drbd_request_state(mdev, NS(pdsk, nps)); + + /* Not using + drbd_request_state(mdev, NS(pdsk, nps)); + here, because we might were able to re-establish the connection + in the meantime. This can only partially be solved in the state's + engine is_valid_state() and is_valid_state_transition() + functions. + + nps can be D_INCONSISTENT, D_OUTDATED or D_UNKNOWN. + pdsk == D_INCONSISTENT while conn >= C_CONNECTED is valid, + therefore we have to have the pre state change check here. + */ + spin_lock_irq(&mdev->req_lock); + ns = mdev->state; + if (ns.conn < C_WF_REPORT_PARAMS) { + ns.pdsk = nps; + _drbd_set_state(mdev, ns, CS_VERBOSE, NULL); + } + spin_unlock_irq(&mdev->req_lock); return 0; } @@ -288,10 +307,11 @@ void drbd_try_outdate_peer_async(struct drbd_conf *mdev) dev_err(DEV, "out of mem, failed to invoke fence-peer helper\n"); } -int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) +enum drbd_state_rv +drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) { const int max_tries = 4; - int r = 0; + enum drbd_state_rv rv = SS_UNKNOWN_ERROR; int try = 0; int forced = 0; union drbd_state mask, val; @@ -306,17 +326,17 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) val.i = 0; val.role = new_role; while (try++ < max_tries) { - r = _drbd_request_state(mdev, mask, val, CS_WAIT_COMPLETE); + rv = _drbd_request_state(mdev, mask, val, CS_WAIT_COMPLETE); /* in case we first succeeded to outdate, * but now suddenly could establish a connection */ - if (r == SS_CW_FAILED_BY_PEER && mask.pdsk != 0) { + if (rv == SS_CW_FAILED_BY_PEER && mask.pdsk != 0) { val.pdsk = 0; mask.pdsk = 0; continue; } - if (r == SS_NO_UP_TO_DATE_DISK && force && + if (rv == SS_NO_UP_TO_DATE_DISK && force && (mdev->state.disk < D_UP_TO_DATE && mdev->state.disk >= D_INCONSISTENT)) { mask.disk = D_MASK; @@ -325,7 +345,7 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) continue; } - if (r == SS_NO_UP_TO_DATE_DISK && + if (rv == SS_NO_UP_TO_DATE_DISK && mdev->state.disk == D_CONSISTENT && mask.pdsk == 0) { D_ASSERT(mdev->state.pdsk == D_UNKNOWN); nps = drbd_try_outdate_peer(mdev); @@ -341,9 +361,9 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) continue; } - if (r == SS_NOTHING_TO_DO) + if (rv == SS_NOTHING_TO_DO) goto fail; - if (r == SS_PRIMARY_NOP && mask.pdsk == 0) { + if (rv == SS_PRIMARY_NOP && mask.pdsk == 0) { nps = drbd_try_outdate_peer(mdev); if (force && nps > D_OUTDATED) { @@ -356,25 +376,24 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) continue; } - if (r == SS_TWO_PRIMARIES) { + if (rv == SS_TWO_PRIMARIES) { /* Maybe the peer is detected as dead very soon... retry at most once more in this case. */ - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout((mdev->net_conf->ping_timeo+1)*HZ/10); + schedule_timeout_interruptible((mdev->net_conf->ping_timeo+1)*HZ/10); if (try < max_tries) try = max_tries - 1; continue; } - if (r < SS_SUCCESS) { - r = _drbd_request_state(mdev, mask, val, + if (rv < SS_SUCCESS) { + rv = _drbd_request_state(mdev, mask, val, CS_VERBOSE + CS_WAIT_COMPLETE); - if (r < SS_SUCCESS) + if (rv < SS_SUCCESS) goto fail; } break; } - if (r < SS_SUCCESS) + if (rv < SS_SUCCESS) goto fail; if (forced) @@ -384,7 +403,7 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) wait_event(mdev->misc_wait, atomic_read(&mdev->ap_pending_cnt) == 0); if (new_role == R_SECONDARY) { - set_disk_ro(mdev->vdisk, TRUE); + set_disk_ro(mdev->vdisk, true); if (get_ldev(mdev)) { mdev->ldev->md.uuid[UI_CURRENT] &= ~(u64)1; put_ldev(mdev); @@ -394,7 +413,7 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) mdev->net_conf->want_lose = 0; put_net_conf(mdev); } - set_disk_ro(mdev->vdisk, FALSE); + set_disk_ro(mdev->vdisk, false); if (get_ldev(mdev)) { if (((mdev->state.conn < C_CONNECTED || mdev->state.pdsk <= D_FAILED) @@ -406,10 +425,8 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) } } - if ((new_role == R_SECONDARY) && get_ldev(mdev)) { - drbd_al_to_on_disk_bm(mdev); - put_ldev(mdev); - } + /* writeout of activity log covered areas of the bitmap + * to stable storage done in after state change already */ if (mdev->state.conn >= C_WF_REPORT_PARAMS) { /* if this was forced, we should consider sync */ @@ -423,7 +440,7 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) kobject_uevent(&disk_to_dev(mdev->vdisk)->kobj, KOBJ_CHANGE); fail: mutex_unlock(&mdev->state_mutex); - return r; + return rv; } static struct drbd_conf *ensure_mdev(int minor, int create) @@ -528,17 +545,19 @@ static void drbd_md_set_sector_offsets(struct drbd_conf *mdev, } } +/* input size is expected to be in KB */ char *ppsize(char *buf, unsigned long long size) { - /* Needs 9 bytes at max. */ + /* Needs 9 bytes at max including trailing NUL: + * -1ULL ==> "16384 EB" */ static char units[] = { 'K', 'M', 'G', 'T', 'P', 'E' }; int base = 0; - while (size >= 10000) { + while (size >= 10000 && base < sizeof(units)-1) { /* shift + round */ size = (size >> 10) + !!(size & (1<<9)); base++; } - sprintf(buf, "%lu %cB", (long)size, units[base]); + sprintf(buf, "%u %cB", (unsigned)size, units[base]); return buf; } @@ -577,7 +596,7 @@ void drbd_resume_io(struct drbd_conf *mdev) * Returns 0 on success, negative return values indicate errors. * You should call drbd_md_sync() after calling this function. */ -enum determine_dev_size drbd_determin_dev_size(struct drbd_conf *mdev, enum dds_flags flags) __must_hold(local) +enum determine_dev_size drbd_determine_dev_size(struct drbd_conf *mdev, enum dds_flags flags) __must_hold(local) { sector_t prev_first_sect, prev_size; /* previous meta location */ sector_t la_size; @@ -642,11 +661,19 @@ enum determine_dev_size drbd_determin_dev_size(struct drbd_conf *mdev, enum dds_ || prev_size != mdev->ldev->md.md_size_sect; if (la_size_changed || md_moved) { + int err; + drbd_al_shrink(mdev); /* All extents inactive. */ dev_info(DEV, "Writing the whole bitmap, %s\n", la_size_changed && md_moved ? "size changed and md moved" : la_size_changed ? "size changed" : "md moved"); - rv = drbd_bitmap_io(mdev, &drbd_bm_write, "size changed"); /* does drbd_resume_io() ! */ + /* next line implicitly does drbd_suspend_io()+drbd_resume_io() */ + err = drbd_bitmap_io(mdev, &drbd_bm_write, + "size changed", BM_LOCKED_MASK); + if (err) { + rv = dev_size_error; + goto out; + } drbd_md_mark_dirty(mdev); } @@ -765,31 +792,78 @@ static int drbd_check_al_size(struct drbd_conf *mdev) return 0; } -void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int max_seg_s) __must_hold(local) +static void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int max_bio_size) { struct request_queue * const q = mdev->rq_queue; - struct request_queue * const b = mdev->ldev->backing_bdev->bd_disk->queue; - int max_segments = mdev->ldev->dc.max_bio_bvecs; + int max_hw_sectors = max_bio_size >> 9; + int max_segments = 0; - max_seg_s = min(queue_max_sectors(b) * queue_logical_block_size(b), max_seg_s); + if (get_ldev_if_state(mdev, D_ATTACHING)) { + struct request_queue * const b = mdev->ldev->backing_bdev->bd_disk->queue; + + max_hw_sectors = min(queue_max_hw_sectors(b), max_bio_size >> 9); + max_segments = mdev->ldev->dc.max_bio_bvecs; + put_ldev(mdev); + } - blk_queue_max_hw_sectors(q, max_seg_s >> 9); - blk_queue_max_segments(q, max_segments ? max_segments : BLK_MAX_SEGMENTS); - blk_queue_max_segment_size(q, max_seg_s); blk_queue_logical_block_size(q, 512); - blk_queue_segment_boundary(q, PAGE_SIZE-1); - blk_stack_limits(&q->limits, &b->limits, 0); + blk_queue_max_hw_sectors(q, max_hw_sectors); + /* This is the workaround for "bio would need to, but cannot, be split" */ + blk_queue_max_segments(q, max_segments ? max_segments : BLK_MAX_SEGMENTS); + blk_queue_segment_boundary(q, PAGE_CACHE_SIZE-1); + + if (get_ldev_if_state(mdev, D_ATTACHING)) { + struct request_queue * const b = mdev->ldev->backing_bdev->bd_disk->queue; - dev_info(DEV, "max_segment_size ( = BIO size ) = %u\n", queue_max_segment_size(q)); + blk_queue_stack_limits(q, b); - if (q->backing_dev_info.ra_pages != b->backing_dev_info.ra_pages) { - dev_info(DEV, "Adjusting my ra_pages to backing device's (%lu -> %lu)\n", - q->backing_dev_info.ra_pages, - b->backing_dev_info.ra_pages); - q->backing_dev_info.ra_pages = b->backing_dev_info.ra_pages; + if (q->backing_dev_info.ra_pages != b->backing_dev_info.ra_pages) { + dev_info(DEV, "Adjusting my ra_pages to backing device's (%lu -> %lu)\n", + q->backing_dev_info.ra_pages, + b->backing_dev_info.ra_pages); + q->backing_dev_info.ra_pages = b->backing_dev_info.ra_pages; + } + put_ldev(mdev); } } +void drbd_reconsider_max_bio_size(struct drbd_conf *mdev) +{ + int now, new, local, peer; + + now = queue_max_hw_sectors(mdev->rq_queue) << 9; + local = mdev->local_max_bio_size; /* Eventually last known value, from volatile memory */ + peer = mdev->peer_max_bio_size; /* Eventually last known value, from meta data */ + + if (get_ldev_if_state(mdev, D_ATTACHING)) { + local = queue_max_hw_sectors(mdev->ldev->backing_bdev->bd_disk->queue) << 9; + mdev->local_max_bio_size = local; + put_ldev(mdev); + } + + /* We may ignore peer limits if the peer is modern enough. + Because new from 8.3.8 onwards the peer can use multiple + BIOs for a single peer_request */ + if (mdev->state.conn >= C_CONNECTED) { + if (mdev->agreed_pro_version < 94) + peer = mdev->peer_max_bio_size; + else if (mdev->agreed_pro_version == 94) + peer = DRBD_MAX_SIZE_H80_PACKET; + else /* drbd 8.3.8 onwards */ + peer = DRBD_MAX_BIO_SIZE; + } + + new = min_t(int, local, peer); + + if (mdev->state.role == R_PRIMARY && new < now) + dev_err(DEV, "ASSERT FAILED new < now; (%d < %d)\n", new, now); + + if (new != now) + dev_info(DEV, "max BIO size = %u\n", new); + + drbd_setup_queue_param(mdev, new); +} + /* serialize deconfig (worker exiting, doing cleanup) * and reconfig (drbdsetup disk, drbdsetup net) * @@ -850,7 +924,7 @@ static void drbd_suspend_al(struct drbd_conf *mdev) static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { - enum drbd_ret_codes retcode; + enum drbd_ret_code retcode; enum determine_dev_size dd; sector_t max_possible_sectors; sector_t min_md_device_sectors; @@ -858,8 +932,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp struct block_device *bdev; struct lru_cache *resync_lru = NULL; union drbd_state ns, os; - unsigned int max_seg_s; - int rv; + enum drbd_state_rv rv; int cp_discovered = 0; int logical_block_size; @@ -1005,9 +1078,10 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp /* and for any other previously queued work */ drbd_flush_workqueue(mdev); - retcode = _drbd_request_state(mdev, NS(disk, D_ATTACHING), CS_VERBOSE); + rv = _drbd_request_state(mdev, NS(disk, D_ATTACHING), CS_VERBOSE); + retcode = rv; /* FIXME: Type mismatch. */ drbd_resume_io(mdev); - if (retcode < SS_SUCCESS) + if (rv < SS_SUCCESS) goto fail; if (!get_ldev_if_state(mdev, D_ATTACHING)) @@ -1109,20 +1183,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp mdev->read_cnt = 0; mdev->writ_cnt = 0; - max_seg_s = DRBD_MAX_SEGMENT_SIZE; - if (mdev->state.conn == C_CONNECTED) { - /* We are Primary, Connected, and now attach a new local - * backing store. We must not increase the user visible maximum - * bio size on this device to something the peer may not be - * able to handle. */ - if (mdev->agreed_pro_version < 94) - max_seg_s = queue_max_segment_size(mdev->rq_queue); - else if (mdev->agreed_pro_version == 94) - max_seg_s = DRBD_MAX_SIZE_H80_PACKET; - /* else: drbd 8.3.9 and later, stay with default */ - } - - drbd_setup_queue_param(mdev, max_seg_s); + drbd_reconsider_max_bio_size(mdev); /* If I am currently not R_PRIMARY, * but meta data primary indicator is set, @@ -1144,7 +1205,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp !drbd_md_test_flag(mdev->ldev, MDF_CONNECTED_IND)) set_bit(USE_DEGR_WFC_T, &mdev->flags); - dd = drbd_determin_dev_size(mdev, 0); + dd = drbd_determine_dev_size(mdev, 0); if (dd == dev_size_error) { retcode = ERR_NOMEM_BITMAP; goto force_diskless_dec; @@ -1154,12 +1215,14 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp if (drbd_md_test_flag(mdev->ldev, MDF_FULL_SYNC)) { dev_info(DEV, "Assuming that all blocks are out of sync " "(aka FullSync)\n"); - if (drbd_bitmap_io(mdev, &drbd_bmio_set_n_write, "set_n_write from attaching")) { + if (drbd_bitmap_io(mdev, &drbd_bmio_set_n_write, + "set_n_write from attaching", BM_LOCKED_MASK)) { retcode = ERR_IO_MD_DISK; goto force_diskless_dec; } } else { - if (drbd_bitmap_io(mdev, &drbd_bm_read, "read from attaching") < 0) { + if (drbd_bitmap_io(mdev, &drbd_bm_read, + "read from attaching", BM_LOCKED_MASK) < 0) { retcode = ERR_IO_MD_DISK; goto force_diskless_dec; } @@ -1167,7 +1230,11 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp if (cp_discovered) { drbd_al_apply_to_bm(mdev); - drbd_al_to_on_disk_bm(mdev); + if (drbd_bitmap_io(mdev, &drbd_bm_write, + "crashed primary apply AL", BM_LOCKED_MASK)) { + retcode = ERR_IO_MD_DISK; + goto force_diskless_dec; + } } if (_drbd_bm_total_weight(mdev) == drbd_bm_bits(mdev)) @@ -1267,11 +1334,19 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp static int drbd_nl_detach(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { + enum drbd_ret_code retcode; + int ret; drbd_suspend_io(mdev); /* so no-one is stuck in drbd_al_begin_io */ - reply->ret_code = drbd_request_state(mdev, NS(disk, D_DISKLESS)); - if (mdev->state.disk == D_DISKLESS) - wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt)); + retcode = drbd_request_state(mdev, NS(disk, D_FAILED)); + /* D_FAILED will transition to DISKLESS. */ + ret = wait_event_interruptible(mdev->misc_wait, + mdev->state.disk != D_FAILED); drbd_resume_io(mdev); + if ((int)retcode == (int)SS_IS_DISKLESS) + retcode = SS_NOTHING_TO_DO; + if (ret) + retcode = ERR_INTR; + reply->ret_code = retcode; return 0; } @@ -1279,7 +1354,7 @@ static int drbd_nl_net_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { int i, ns; - enum drbd_ret_codes retcode; + enum drbd_ret_code retcode; struct net_conf *new_conf = NULL; struct crypto_hash *tfm = NULL; struct crypto_hash *integrity_w_tfm = NULL; @@ -1324,6 +1399,8 @@ static int drbd_nl_net_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, new_conf->wire_protocol = DRBD_PROT_C; new_conf->ping_timeo = DRBD_PING_TIMEO_DEF; new_conf->rr_conflict = DRBD_RR_CONFLICT_DEF; + new_conf->on_congestion = DRBD_ON_CONGESTION_DEF; + new_conf->cong_extents = DRBD_CONG_EXTENTS_DEF; if (!net_conf_from_tags(mdev, nlp->tag_list, new_conf)) { retcode = ERR_MANDATORY_TAG; @@ -1345,6 +1422,11 @@ static int drbd_nl_net_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, } } + if (new_conf->on_congestion != OC_BLOCK && new_conf->wire_protocol != DRBD_PROT_A) { + retcode = ERR_CONG_NOT_PROTO_A; + goto fail; + } + if (mdev->state.role == R_PRIMARY && new_conf->want_lose) { retcode = ERR_DISCARD; goto fail; @@ -1525,6 +1607,21 @@ static int drbd_nl_disconnect(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl struct drbd_nl_cfg_reply *reply) { int retcode; + struct disconnect dc; + + memset(&dc, 0, sizeof(struct disconnect)); + if (!disconnect_from_tags(mdev, nlp->tag_list, &dc)) { + retcode = ERR_MANDATORY_TAG; + goto fail; + } + + if (dc.force) { + spin_lock_irq(&mdev->req_lock); + if (mdev->state.conn >= C_WF_CONNECTION) + _drbd_set_state(_NS(mdev, conn, C_DISCONNECTING), CS_HARD, NULL); + spin_unlock_irq(&mdev->req_lock); + goto done; + } retcode = _drbd_request_state(mdev, NS(conn, C_DISCONNECTING), CS_ORDERED); @@ -1622,7 +1719,7 @@ static int drbd_nl_resize(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, mdev->ldev->dc.disk_size = (sector_t)rs.resize_size; ddsf = (rs.resize_force ? DDSF_FORCED : 0) | (rs.no_resync ? DDSF_NO_RESYNC : 0); - dd = drbd_determin_dev_size(mdev, ddsf); + dd = drbd_determine_dev_size(mdev, ddsf); drbd_md_sync(mdev); put_ldev(mdev); if (dd == dev_size_error) { @@ -1842,6 +1939,10 @@ static int drbd_nl_invalidate(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl { int retcode; + /* If there is still bitmap IO pending, probably because of a previous + * resync just being finished, wait for it before requesting a new resync. */ + wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags)); + retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_T), CS_ORDERED); if (retcode < SS_SUCCESS && retcode != SS_NEED_CONNECTION) @@ -1877,6 +1978,10 @@ static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_re { int retcode; + /* If there is still bitmap IO pending, probably because of a previous + * resync just being finished, wait for it before requesting a new resync. */ + wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags)); + retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S), CS_ORDERED); if (retcode < SS_SUCCESS) { @@ -1885,9 +1990,9 @@ static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_re into a full resync. */ retcode = drbd_request_state(mdev, NS(pdsk, D_INCONSISTENT)); if (retcode >= SS_SUCCESS) { - /* open coded drbd_bitmap_io() */ if (drbd_bitmap_io(mdev, &drbd_bmio_set_susp_al, - "set_n_write from invalidate_peer")) + "set_n_write from invalidate_peer", + BM_LOCKED_SET_ALLOWED)) retcode = ERR_IO_MD_DISK; } } else @@ -1914,9 +2019,17 @@ static int drbd_nl_resume_sync(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n struct drbd_nl_cfg_reply *reply) { int retcode = NO_ERROR; + union drbd_state s; - if (drbd_request_state(mdev, NS(user_isp, 0)) == SS_NOTHING_TO_DO) - retcode = ERR_PAUSE_IS_CLEAR; + if (drbd_request_state(mdev, NS(user_isp, 0)) == SS_NOTHING_TO_DO) { + s = mdev->state; + if (s.conn == C_PAUSED_SYNC_S || s.conn == C_PAUSED_SYNC_T) { + retcode = s.aftr_isp ? ERR_PIC_AFTER_DEP : + s.peer_isp ? ERR_PIC_PEER_DEP : ERR_PAUSE_IS_CLEAR; + } else { + retcode = ERR_PAUSE_IS_CLEAR; + } + } reply->ret_code = retcode; return 0; @@ -2054,6 +2167,11 @@ static int drbd_nl_start_ov(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, reply->ret_code = ERR_MANDATORY_TAG; return 0; } + + /* If there is still bitmap IO pending, e.g. previous resync or verify + * just being finished, wait for it before requesting a new resync. */ + wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags)); + /* w_make_ov_request expects position to be aligned */ mdev->ov_start_sector = args.start_sector & ~BM_SECT_PER_BIT; reply->ret_code = drbd_request_state(mdev,NS(conn,C_VERIFY_S)); @@ -2097,7 +2215,8 @@ static int drbd_nl_new_c_uuid(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl drbd_uuid_new_current(mdev); /* New current, previous to UI_BITMAP */ if (args.clear_bm) { - err = drbd_bitmap_io(mdev, &drbd_bmio_clear_n_write, "clear_n_write from new_c_uuid"); + err = drbd_bitmap_io(mdev, &drbd_bmio_clear_n_write, + "clear_n_write from new_c_uuid", BM_LOCKED_MASK); if (err) { dev_err(DEV, "Writing bitmap failed with %d\n",err); retcode = ERR_IO_MD_DISK; @@ -2105,6 +2224,7 @@ static int drbd_nl_new_c_uuid(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl if (skip_initial_sync) { drbd_send_uuids_skip_initial_sync(mdev); _drbd_uuid_set(mdev, UI_BITMAP, 0); + drbd_print_uuids(mdev, "cleared bitmap UUID"); spin_lock_irq(&mdev->req_lock); _drbd_set_state(_NS2(mdev, disk, D_UP_TO_DATE, pdsk, D_UP_TO_DATE), CS_VERBOSE, NULL); @@ -2177,7 +2297,7 @@ static void drbd_connector_callback(struct cn_msg *req, struct netlink_skb_parms return; } - if (!cap_raised(nsp->eff_cap, CAP_SYS_ADMIN)) { + if (!cap_raised(current_cap(), CAP_SYS_ADMIN)) { retcode = ERR_PERM; goto fail; } @@ -2189,7 +2309,8 @@ static void drbd_connector_callback(struct cn_msg *req, struct netlink_skb_parms goto fail; } - if (nlp->packet_type >= P_nl_after_last_packet) { + if (nlp->packet_type >= P_nl_after_last_packet || + nlp->packet_type == P_return_code_only) { retcode = ERR_PACKET_NR; goto fail; } @@ -2205,7 +2326,7 @@ static void drbd_connector_callback(struct cn_msg *req, struct netlink_skb_parms reply_size += cm->reply_body_size; /* allocation not in the IO path, cqueue thread context */ - cn_reply = kmalloc(reply_size, GFP_KERNEL); + cn_reply = kzalloc(reply_size, GFP_KERNEL); if (!cn_reply) { retcode = ERR_NOMEM; goto fail; @@ -2213,7 +2334,7 @@ static void drbd_connector_callback(struct cn_msg *req, struct netlink_skb_parms reply = (struct drbd_nl_cfg_reply *) cn_reply->data; reply->packet_type = - cm->reply_body_size ? nlp->packet_type : P_nl_after_last_packet; + cm->reply_body_size ? nlp->packet_type : P_return_code_only; reply->minor = nlp->drbd_minor; reply->ret_code = NO_ERROR; /* Might by modified by cm->function. */ /* reply->tag_list; might be modified by cm->function. */ @@ -2376,7 +2497,7 @@ void drbd_bcast_ee(struct drbd_conf *mdev, /* receiver thread context, which is not in the writeout path (of this node), * but may be in the writeout path of the _other_ node. * GFP_NOIO to avoid potential "distributed deadlock". */ - cn_reply = kmalloc( + cn_reply = kzalloc( sizeof(struct cn_msg)+ sizeof(struct drbd_nl_cfg_reply)+ sizeof(struct dump_ee_tag_len_struct)+ @@ -2398,10 +2519,11 @@ void drbd_bcast_ee(struct drbd_conf *mdev, tl = tl_add_int(tl, T_ee_sector, &e->sector); tl = tl_add_int(tl, T_ee_block_id, &e->block_id); + /* dump the first 32k */ + len = min_t(unsigned, e->size, 32 << 10); put_unaligned(T_ee_data, tl++); - put_unaligned(e->size, tl++); + put_unaligned(len, tl++); - len = e->size; page = e->pages; page_chain_for_each(page) { void *d = kmap_atomic(page, KM_USER0); @@ -2410,6 +2532,8 @@ void drbd_bcast_ee(struct drbd_conf *mdev, kunmap_atomic(d, KM_USER0); tl = (unsigned short*)((char*)tl + l); len -= l; + if (len == 0) + break; } put_unaligned(TT_END, tl++); /* Close the tag list */ @@ -2508,6 +2632,7 @@ void drbd_nl_send_reply(struct cn_msg *req, int ret_code) (struct drbd_nl_cfg_reply *)cn_reply->data; int rr; + memset(buffer, 0, sizeof(buffer)); cn_reply->id = req->id; cn_reply->seq = req->seq; @@ -2515,6 +2640,7 @@ void drbd_nl_send_reply(struct cn_msg *req, int ret_code) cn_reply->len = sizeof(struct drbd_nl_cfg_reply); cn_reply->flags = 0; + reply->packet_type = P_return_code_only; reply->minor = ((struct drbd_nl_cfg_req *)req->data)->drbd_minor; reply->ret_code = ret_code; diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c index 7e6ac30..2959cdf 100644 --- a/drivers/block/drbd/drbd_proc.c +++ b/drivers/block/drbd/drbd_proc.c @@ -34,6 +34,7 @@ #include "drbd_int.h" static int drbd_proc_open(struct inode *inode, struct file *file); +static int drbd_proc_release(struct inode *inode, struct file *file); struct proc_dir_entry *drbd_proc; @@ -42,9 +43,22 @@ const struct file_operations drbd_proc_fops = { .open = drbd_proc_open, .read = seq_read, .llseek = seq_lseek, - .release = single_release, + .release = drbd_proc_release, }; +void seq_printf_with_thousands_grouping(struct seq_file *seq, long v) +{ + /* v is in kB/sec. We don't expect TiByte/sec yet. */ + if (unlikely(v >= 1000000)) { + /* cool: > GiByte/s */ + seq_printf(seq, "%ld,", v / 1000000); + v /= 1000000; + seq_printf(seq, "%03ld,%03ld", v/1000, v % 1000); + } else if (likely(v >= 1000)) + seq_printf(seq, "%ld,%03ld", v/1000, v % 1000); + else + seq_printf(seq, "%ld", v); +} /*lge * progress bars shamelessly adapted from driver/md/md.c @@ -71,10 +85,15 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) seq_printf(seq, "."); seq_printf(seq, "] "); - seq_printf(seq, "sync'ed:%3u.%u%% ", res / 10, res % 10); - /* if more than 1 GB display in MB */ - if (mdev->rs_total > 0x100000L) - seq_printf(seq, "(%lu/%lu)M\n\t", + if (mdev->state.conn == C_VERIFY_S || mdev->state.conn == C_VERIFY_T) + seq_printf(seq, "verified:"); + else + seq_printf(seq, "sync'ed:"); + seq_printf(seq, "%3u.%u%% ", res / 10, res % 10); + + /* if more than a few GB, display in MB */ + if (mdev->rs_total > (4UL << (30 - BM_BLOCK_SHIFT))) + seq_printf(seq, "(%lu/%lu)M", (unsigned long) Bit2KB(rs_left >> 10), (unsigned long) Bit2KB(mdev->rs_total >> 10)); else @@ -94,6 +113,7 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) /* Rolling marks. last_mark+1 may just now be modified. last_mark+2 is * at least (DRBD_SYNC_MARKS-2)*DRBD_SYNC_MARK_STEP old, and has at * least DRBD_SYNC_MARK_STEP time before it will be modified. */ + /* ------------------------ ~18s average ------------------------ */ i = (mdev->rs_last_mark + 2) % DRBD_SYNC_MARKS; dt = (jiffies - mdev->rs_mark_time[i]) / HZ; if (dt > (DRBD_SYNC_MARK_STEP * DRBD_SYNC_MARKS)) @@ -107,14 +127,24 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) seq_printf(seq, "finish: %lu:%02lu:%02lu", rt / 3600, (rt % 3600) / 60, rt % 60); - /* current speed average over (SYNC_MARKS * SYNC_MARK_STEP) jiffies */ dbdt = Bit2KB(db/dt); - if (dbdt > 1000) - seq_printf(seq, " speed: %ld,%03ld", - dbdt/1000, dbdt % 1000); - else - seq_printf(seq, " speed: %ld", dbdt); + seq_printf(seq, " speed: "); + seq_printf_with_thousands_grouping(seq, dbdt); + seq_printf(seq, " ("); + /* ------------------------- ~3s average ------------------------ */ + if (proc_details >= 1) { + /* this is what drbd_rs_should_slow_down() uses */ + i = (mdev->rs_last_mark + DRBD_SYNC_MARKS-1) % DRBD_SYNC_MARKS; + dt = (jiffies - mdev->rs_mark_time[i]) / HZ; + if (!dt) + dt++; + db = mdev->rs_mark_left[i] - rs_left; + dbdt = Bit2KB(db/dt); + seq_printf_with_thousands_grouping(seq, dbdt); + seq_printf(seq, " -- "); + } + /* --------------------- long term average ---------------------- */ /* mean speed since syncer started * we do account for PausedSync periods */ dt = (jiffies - mdev->rs_start - mdev->rs_paused) / HZ; @@ -122,20 +152,34 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) dt = 1; db = mdev->rs_total - rs_left; dbdt = Bit2KB(db/dt); - if (dbdt > 1000) - seq_printf(seq, " (%ld,%03ld)", - dbdt/1000, dbdt % 1000); - else - seq_printf(seq, " (%ld)", dbdt); + seq_printf_with_thousands_grouping(seq, dbdt); + seq_printf(seq, ")"); - if (mdev->state.conn == C_SYNC_TARGET) { - if (mdev->c_sync_rate > 1000) - seq_printf(seq, " want: %d,%03d", - mdev->c_sync_rate / 1000, mdev->c_sync_rate % 1000); - else - seq_printf(seq, " want: %d", mdev->c_sync_rate); + if (mdev->state.conn == C_SYNC_TARGET || + mdev->state.conn == C_VERIFY_S) { + seq_printf(seq, " want: "); + seq_printf_with_thousands_grouping(seq, mdev->c_sync_rate); } seq_printf(seq, " K/sec%s\n", stalled ? " (stalled)" : ""); + + if (proc_details >= 1) { + /* 64 bit: + * we convert to sectors in the display below. */ + unsigned long bm_bits = drbd_bm_bits(mdev); + unsigned long bit_pos; + if (mdev->state.conn == C_VERIFY_S || + mdev->state.conn == C_VERIFY_T) + bit_pos = bm_bits - mdev->ov_left; + else + bit_pos = mdev->bm_resync_fo; + /* Total sectors may be slightly off for oddly + * sized devices. So what. */ + seq_printf(seq, + "\t%3d%% sector pos: %llu/%llu\n", + (int)(bit_pos / (bm_bits/100+1)), + (unsigned long long)bit_pos * BM_SECT_PER_BIT, + (unsigned long long)bm_bits * BM_SECT_PER_BIT); + } } static void resync_dump_detail(struct seq_file *seq, struct lc_element *e) @@ -232,20 +276,16 @@ static int drbd_seq_show(struct seq_file *seq, void *v) mdev->epochs, write_ordering_chars[mdev->write_ordering] ); - seq_printf(seq, " oos:%lu\n", - Bit2KB(drbd_bm_total_weight(mdev))); + seq_printf(seq, " oos:%llu\n", + Bit2KB((unsigned long long) + drbd_bm_total_weight(mdev))); } if (mdev->state.conn == C_SYNC_SOURCE || - mdev->state.conn == C_SYNC_TARGET) + mdev->state.conn == C_SYNC_TARGET || + mdev->state.conn == C_VERIFY_S || + mdev->state.conn == C_VERIFY_T) drbd_syncer_progress(mdev, seq); - if (mdev->state.conn == C_VERIFY_S || mdev->state.conn == C_VERIFY_T) - seq_printf(seq, "\t%3d%% %lu/%lu\n", - (int)((mdev->rs_total-mdev->ov_left) / - (mdev->rs_total/100+1)), - mdev->rs_total - mdev->ov_left, - mdev->rs_total); - if (proc_details >= 1 && get_ldev_if_state(mdev, D_FAILED)) { lc_seq_printf_stats(seq, mdev->resync); lc_seq_printf_stats(seq, mdev->act_log); @@ -265,7 +305,15 @@ static int drbd_seq_show(struct seq_file *seq, void *v) static int drbd_proc_open(struct inode *inode, struct file *file) { - return single_open(file, drbd_seq_show, PDE(inode)->data); + if (try_module_get(THIS_MODULE)) + return single_open(file, drbd_seq_show, PDE(inode)->data); + return -ENODEV; +} + +static int drbd_proc_release(struct inode *inode, struct file *file) +{ + module_put(THIS_MODULE); + return single_release(inode, file); } /* PROC FS stuff end */ diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 24487d4..25d32c5 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -187,15 +187,6 @@ static struct page *drbd_pp_first_pages_or_try_alloc(struct drbd_conf *mdev, int return NULL; } -/* kick lower level device, if we have more than (arbitrary number) - * reference counts on it, which typically are locally submitted io - * requests. don't use unacked_cnt, so we speed up proto A and B, too. */ -static void maybe_kick_lo(struct drbd_conf *mdev) -{ - if (atomic_read(&mdev->local_cnt) >= mdev->net_conf->unplug_watermark) - drbd_kick_lo(mdev); -} - static void reclaim_net_ee(struct drbd_conf *mdev, struct list_head *to_be_freed) { struct drbd_epoch_entry *e; @@ -219,7 +210,6 @@ static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev) LIST_HEAD(reclaimed); struct drbd_epoch_entry *e, *t; - maybe_kick_lo(mdev); spin_lock_irq(&mdev->req_lock); reclaim_net_ee(mdev, &reclaimed); spin_unlock_irq(&mdev->req_lock); @@ -287,7 +277,7 @@ static void drbd_pp_free(struct drbd_conf *mdev, struct page *page, int is_net) atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use; int i; - if (drbd_pp_vacant > (DRBD_MAX_SEGMENT_SIZE/PAGE_SIZE)*minor_count) + if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE)*minor_count) i = page_chain_free(page); else { struct page *tmp; @@ -329,7 +319,7 @@ struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev, struct page *page; unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT; - if (FAULT_ACTIVE(mdev, DRBD_FAULT_AL_EE)) + if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE)) return NULL; e = mempool_alloc(drbd_ee_mempool, gfp_mask & ~__GFP_HIGHMEM); @@ -343,7 +333,7 @@ struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev, if (!page) goto fail; - INIT_HLIST_NODE(&e->colision); + INIT_HLIST_NODE(&e->collision); e->epoch = NULL; e->mdev = mdev; e->pages = page; @@ -366,7 +356,7 @@ void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, int i kfree(e->digest); drbd_pp_free(mdev, e->pages, is_net); D_ASSERT(atomic_read(&e->pending_bios) == 0); - D_ASSERT(hlist_unhashed(&e->colision)); + D_ASSERT(hlist_unhashed(&e->collision)); mempool_free(e, drbd_ee_mempool); } @@ -436,8 +426,7 @@ void _drbd_wait_ee_list_empty(struct drbd_conf *mdev, struct list_head *head) while (!list_empty(head)) { prepare_to_wait(&mdev->ee_wait, &wait, TASK_UNINTERRUPTIBLE); spin_unlock_irq(&mdev->req_lock); - drbd_kick_lo(mdev); - schedule(); + io_schedule(); finish_wait(&mdev->ee_wait, &wait); spin_lock_irq(&mdev->req_lock); } @@ -736,16 +725,16 @@ static int drbd_socket_okay(struct drbd_conf *mdev, struct socket **sock) char tb[4]; if (!*sock) - return FALSE; + return false; rr = drbd_recv_short(mdev, *sock, tb, 4, MSG_DONTWAIT | MSG_PEEK); if (rr > 0 || rr == -EAGAIN) { - return TRUE; + return true; } else { sock_release(*sock); *sock = NULL; - return FALSE; + return false; } } @@ -779,8 +768,7 @@ static int drbd_connect(struct drbd_conf *mdev) if (s || ++try >= 3) break; /* give the other side time to call bind() & listen() */ - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ / 10); + schedule_timeout_interruptible(HZ / 10); } if (s) { @@ -799,8 +787,7 @@ static int drbd_connect(struct drbd_conf *mdev) } if (sock && msock) { - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ / 10); + schedule_timeout_interruptible(mdev->net_conf->ping_timeo*HZ/10); ok = drbd_socket_okay(mdev, &sock); ok = drbd_socket_okay(mdev, &msock) && ok; if (ok) @@ -875,7 +862,7 @@ retry: msock->sk->sk_rcvtimeo = mdev->net_conf->ping_int*HZ; /* we don't want delays. - * we use TCP_CORK where apropriate, though */ + * we use TCP_CORK where appropriate, though */ drbd_tcp_nodelay(sock); drbd_tcp_nodelay(msock); @@ -912,12 +899,7 @@ retry: drbd_thread_start(&mdev->asender); - if (mdev->agreed_pro_version < 95 && get_ldev(mdev)) { - drbd_setup_queue_param(mdev, DRBD_MAX_SIZE_H80_PACKET); - put_ldev(mdev); - } - - if (!drbd_send_protocol(mdev)) + if (drbd_send_protocol(mdev) == -1) return -1; drbd_send_sync_param(mdev, &mdev->sync_conf); drbd_send_sizes(mdev, 0, 0); @@ -925,6 +907,7 @@ retry: drbd_send_state(mdev); clear_bit(USE_DEGR_WFC_T, &mdev->flags); clear_bit(RESIZE_PENDING, &mdev->flags); + mod_timer(&mdev->request_timer, jiffies + HZ); /* just start it here. */ return 1; @@ -943,8 +926,9 @@ static int drbd_recv_header(struct drbd_conf *mdev, enum drbd_packets *cmd, unsi r = drbd_recv(mdev, h, sizeof(*h)); if (unlikely(r != sizeof(*h))) { - dev_err(DEV, "short read expecting header on sock: r=%d\n", r); - return FALSE; + if (!signal_pending(current)) + dev_warn(DEV, "short read expecting header on sock: r=%d\n", r); + return false; } if (likely(h->h80.magic == BE_DRBD_MAGIC)) { @@ -958,11 +942,11 @@ static int drbd_recv_header(struct drbd_conf *mdev, enum drbd_packets *cmd, unsi be32_to_cpu(h->h80.magic), be16_to_cpu(h->h80.command), be16_to_cpu(h->h80.length)); - return FALSE; + return false; } mdev->last_received = jiffies; - return TRUE; + return true; } static void drbd_flush(struct drbd_conf *mdev) @@ -1085,6 +1069,16 @@ void drbd_bump_write_ordering(struct drbd_conf *mdev, enum write_ordering_e wo) * @mdev: DRBD device. * @e: epoch entry * @rw: flag field, see bio->bi_rw + * + * May spread the pages to multiple bios, + * depending on bio_add_page restrictions. + * + * Returns 0 if all bios have been submitted, + * -ENOMEM if we could not allocate enough bios, + * -ENOSPC (any better suggestion?) if we have not been able to bio_add_page a + * single page to an empty bio (which should never happen and likely indicates + * that the lower level IO stack is in some way broken). This has been observed + * on certain Xen deployments. */ /* TODO allocate from our own bio_set. */ int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, @@ -1097,6 +1091,7 @@ int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, unsigned ds = e->size; unsigned n_bios = 0; unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT; + int err = -ENOMEM; /* In most cases, we will only need one bio. But in case the lower * level restrictions happen to be different at this offset on this @@ -1111,8 +1106,6 @@ next_bio: /* > e->sector, unless this is the first bio */ bio->bi_sector = sector; bio->bi_bdev = mdev->ldev->backing_bdev; - /* we special case some flags in the multi-bio case, see below - * (REQ_UNPLUG) */ bio->bi_rw = rw; bio->bi_private = e; bio->bi_end_io = drbd_endio_sec; @@ -1124,8 +1117,17 @@ next_bio: page_chain_for_each(page) { unsigned len = min_t(unsigned, ds, PAGE_SIZE); if (!bio_add_page(bio, page, len, 0)) { - /* a single page must always be possible! */ - BUG_ON(bio->bi_vcnt == 0); + /* A single page must always be possible! + * But in case it fails anyways, + * we deal with it, and complain (below). */ + if (bio->bi_vcnt == 0) { + dev_err(DEV, + "bio_add_page failed for len=%u, " + "bi_vcnt=0 (bi_sector=%llu)\n", + len, (unsigned long long)bio->bi_sector); + err = -ENOSPC; + goto fail; + } goto next_bio; } ds -= len; @@ -1141,13 +1143,8 @@ next_bio: bios = bios->bi_next; bio->bi_next = NULL; - /* strip off REQ_UNPLUG unless it is the last bio */ - if (bios) - bio->bi_rw &= ~REQ_UNPLUG; - drbd_generic_make_request(mdev, fault_type, bio); } while (bios); - maybe_kick_lo(mdev); return 0; fail: @@ -1156,7 +1153,7 @@ fail: bios = bios->bi_next; bio_put(bio); } - return -ENOMEM; + return err; } static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -1167,9 +1164,6 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign inc_unacked(mdev); - if (mdev->net_conf->wire_protocol != DRBD_PROT_C) - drbd_kick_lo(mdev); - mdev->current_epoch->barrier_nr = p->barrier; rv = drbd_may_finish_epoch(mdev, mdev->current_epoch, EV_GOT_BARRIER_NR); @@ -1181,7 +1175,7 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign switch (mdev->write_ordering) { case WO_none: if (rv == FE_RECYCLED) - return TRUE; + return true; /* receiver context, in the writeout path of the other node. * avoid potential distributed deadlock */ @@ -1209,10 +1203,10 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign D_ASSERT(atomic_read(&epoch->active) == 0); D_ASSERT(epoch->flags == 0); - return TRUE; + return true; default: dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering); - return FALSE; + return false; } epoch->flags = 0; @@ -1230,7 +1224,7 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign } spin_unlock(&mdev->epoch_lock); - return TRUE; + return true; } /* used from receive_RSDataReply (recv_resync_read) @@ -1252,21 +1246,25 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ if (dgs) { rr = drbd_recv(mdev, dig_in, dgs); if (rr != dgs) { - dev_warn(DEV, "short read receiving data digest: read %d expected %d\n", - rr, dgs); + if (!signal_pending(current)) + dev_warn(DEV, + "short read receiving data digest: read %d expected %d\n", + rr, dgs); return NULL; } } data_size -= dgs; + ERR_IF(data_size == 0) return NULL; ERR_IF(data_size & 0x1ff) return NULL; - ERR_IF(data_size > DRBD_MAX_SEGMENT_SIZE) return NULL; + ERR_IF(data_size > DRBD_MAX_BIO_SIZE) return NULL; /* even though we trust out peer, * we sometimes have to double check. */ if (sector + (data_size>>9) > capacity) { - dev_err(DEV, "capacity: %llus < sector: %llus + size: %u\n", + dev_err(DEV, "request from peer beyond end of local disk: " + "capacity: %llus < sector: %llus + size: %u\n", (unsigned long long)capacity, (unsigned long long)sector, data_size); return NULL; @@ -1285,15 +1283,16 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ unsigned len = min_t(int, ds, PAGE_SIZE); data = kmap(page); rr = drbd_recv(mdev, data, len); - if (FAULT_ACTIVE(mdev, DRBD_FAULT_RECEIVE)) { + if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) { dev_err(DEV, "Fault injection: Corrupting data on receive\n"); data[0] = data[0] ^ (unsigned long)-1; } kunmap(page); if (rr != len) { drbd_free_ee(mdev, e); - dev_warn(DEV, "short read receiving data: read %d expected %d\n", - rr, len); + if (!signal_pending(current)) + dev_warn(DEV, "short read receiving data: read %d expected %d\n", + rr, len); return NULL; } ds -= rr; @@ -1302,7 +1301,8 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ if (dgs) { drbd_csum_ee(mdev, mdev->integrity_r_tfm, e, dig_vv); if (memcmp(dig_in, dig_vv, dgs)) { - dev_err(DEV, "Digest integrity check FAILED.\n"); + dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n", + (unsigned long long)sector, data_size); drbd_bcast_ee(mdev, "digest failed", dgs, dig_in, dig_vv, e); drbd_free_ee(mdev, e); @@ -1323,7 +1323,7 @@ static int drbd_drain_block(struct drbd_conf *mdev, int data_size) void *data; if (!data_size) - return TRUE; + return true; page = drbd_pp_alloc(mdev, 1, 1); @@ -1332,8 +1332,10 @@ static int drbd_drain_block(struct drbd_conf *mdev, int data_size) rr = drbd_recv(mdev, data, min_t(int, data_size, PAGE_SIZE)); if (rr != min_t(int, data_size, PAGE_SIZE)) { rv = 0; - dev_warn(DEV, "short read receiving data: read %d expected %d\n", - rr, min_t(int, data_size, PAGE_SIZE)); + if (!signal_pending(current)) + dev_warn(DEV, + "short read receiving data: read %d expected %d\n", + rr, min_t(int, data_size, PAGE_SIZE)); break; } data_size -= rr; @@ -1358,8 +1360,10 @@ static int recv_dless_read(struct drbd_conf *mdev, struct drbd_request *req, if (dgs) { rr = drbd_recv(mdev, dig_in, dgs); if (rr != dgs) { - dev_warn(DEV, "short read receiving data reply digest: read %d expected %d\n", - rr, dgs); + if (!signal_pending(current)) + dev_warn(DEV, + "short read receiving data reply digest: read %d expected %d\n", + rr, dgs); return 0; } } @@ -1380,9 +1384,10 @@ static int recv_dless_read(struct drbd_conf *mdev, struct drbd_request *req, expect); kunmap(bvec->bv_page); if (rr != expect) { - dev_warn(DEV, "short read receiving data reply: " - "read %d expected %d\n", - rr, expect); + if (!signal_pending(current)) + dev_warn(DEV, "short read receiving data reply: " + "read %d expected %d\n", + rr, expect); return 0; } data_size -= rr; @@ -1408,7 +1413,7 @@ static int e_end_resync_block(struct drbd_conf *mdev, struct drbd_work *w, int u sector_t sector = e->sector; int ok; - D_ASSERT(hlist_unhashed(&e->colision)); + D_ASSERT(hlist_unhashed(&e->collision)); if (likely((e->flags & EE_WAS_ERROR) == 0)) { drbd_set_in_sync(mdev, sector, e->size); @@ -1446,11 +1451,10 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si atomic_add(data_size >> 9, &mdev->rs_sect_ev); if (drbd_submit_ee(mdev, e, WRITE, DRBD_FAULT_RS_WR) == 0) - return TRUE; + return true; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* don't care for the reason here */ + dev_err(DEV, "submit failed, triggering re-connect\n"); spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); spin_unlock_irq(&mdev->req_lock); @@ -1458,7 +1462,7 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si drbd_free_ee(mdev, e); fail: put_ldev(mdev); - return FALSE; + return false; } static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -1475,10 +1479,10 @@ static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packets cmd, unsi spin_unlock_irq(&mdev->req_lock); if (unlikely(!req)) { dev_err(DEV, "Got a corrupt block_id/sector pair(1).\n"); - return FALSE; + return false; } - /* hlist_del(&req->colision) is done in _req_may_be_done, to avoid + /* hlist_del(&req->collision) is done in _req_may_be_done, to avoid * special casing it there for the various failure cases. * still no race with drbd_fail_pending_reads */ ok = recv_dless_read(mdev, req, sector, data_size); @@ -1549,11 +1553,11 @@ static int e_end_block(struct drbd_conf *mdev, struct drbd_work *w, int cancel) * P_WRITE_ACK / P_NEG_ACK, to get the sequence number right. */ if (mdev->net_conf->two_primaries) { spin_lock_irq(&mdev->req_lock); - D_ASSERT(!hlist_unhashed(&e->colision)); - hlist_del_init(&e->colision); + D_ASSERT(!hlist_unhashed(&e->collision)); + hlist_del_init(&e->collision); spin_unlock_irq(&mdev->req_lock); } else { - D_ASSERT(hlist_unhashed(&e->colision)); + D_ASSERT(hlist_unhashed(&e->collision)); } drbd_may_finish_epoch(mdev, e->epoch, EV_PUT + (cancel ? EV_CLEANUP : 0)); @@ -1570,8 +1574,8 @@ static int e_send_discard_ack(struct drbd_conf *mdev, struct drbd_work *w, int u ok = drbd_send_ack(mdev, P_DISCARD_ACK, e); spin_lock_irq(&mdev->req_lock); - D_ASSERT(!hlist_unhashed(&e->colision)); - hlist_del_init(&e->colision); + D_ASSERT(!hlist_unhashed(&e->collision)); + hlist_del_init(&e->collision); spin_unlock_irq(&mdev->req_lock); dec_unacked(mdev); @@ -1632,16 +1636,15 @@ static int drbd_wait_peer_seq(struct drbd_conf *mdev, const u32 packet_seq) return ret; } -static unsigned long write_flags_to_bio(struct drbd_conf *mdev, u32 dpf) +/* see also bio_flags_to_wire() + * DRBD_REQ_*, because we need to semantically map the flags to data packet + * flags and back. We may replicate to other kernel versions. */ +static unsigned long wire_flags_to_bio(struct drbd_conf *mdev, u32 dpf) { - if (mdev->agreed_pro_version >= 95) - return (dpf & DP_RW_SYNC ? REQ_SYNC : 0) | - (dpf & DP_UNPLUG ? REQ_UNPLUG : 0) | - (dpf & DP_FUA ? REQ_FUA : 0) | - (dpf & DP_FLUSH ? REQ_FUA : 0) | - (dpf & DP_DISCARD ? REQ_DISCARD : 0); - else - return dpf & DP_RW_SYNC ? (REQ_SYNC | REQ_UNPLUG) : 0; + return (dpf & DP_RW_SYNC ? REQ_SYNC : 0) | + (dpf & DP_FUA ? REQ_FUA : 0) | + (dpf & DP_FLUSH ? REQ_FLUSH : 0) | + (dpf & DP_DISCARD ? REQ_DISCARD : 0); } /* mirrored write */ @@ -1654,9 +1657,6 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned u32 dp_flags; if (!get_ldev(mdev)) { - if (__ratelimit(&drbd_ratelimit_state)) - dev_err(DEV, "Can not write mirrored data block " - "to local disk.\n"); spin_lock(&mdev->peer_seq_lock); if (mdev->peer_seq+1 == be32_to_cpu(p->seq_num)) mdev->peer_seq++; @@ -1676,23 +1676,23 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned e = read_in_block(mdev, p->block_id, sector, data_size); if (!e) { put_ldev(mdev); - return FALSE; + return false; } e->w.cb = e_end_block; + dp_flags = be32_to_cpu(p->dp_flags); + rw |= wire_flags_to_bio(mdev, dp_flags); + + if (dp_flags & DP_MAY_SET_IN_SYNC) + e->flags |= EE_MAY_SET_IN_SYNC; + spin_lock(&mdev->epoch_lock); e->epoch = mdev->current_epoch; atomic_inc(&e->epoch->epoch_size); atomic_inc(&e->epoch->active); spin_unlock(&mdev->epoch_lock); - dp_flags = be32_to_cpu(p->dp_flags); - rw |= write_flags_to_bio(mdev, dp_flags); - - if (dp_flags & DP_MAY_SET_IN_SYNC) - e->flags |= EE_MAY_SET_IN_SYNC; - /* I'm the receiver, I do hold a net_cnt reference. */ if (!mdev->net_conf->two_primaries) { spin_lock_irq(&mdev->req_lock); @@ -1750,7 +1750,7 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned spin_lock_irq(&mdev->req_lock); - hlist_add_head(&e->colision, ee_hash_slot(mdev, sector)); + hlist_add_head(&e->collision, ee_hash_slot(mdev, sector)); #define OVERLAPS overlaps(i->sector, i->size, sector, size) slot = tl_hash_slot(mdev, sector); @@ -1760,7 +1760,7 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int have_conflict = 0; prepare_to_wait(&mdev->misc_wait, &wait, TASK_INTERRUPTIBLE); - hlist_for_each_entry(i, n, slot, colision) { + hlist_for_each_entry(i, n, slot, collision) { if (OVERLAPS) { /* only ALERT on first iteration, * we may be woken up early... */ @@ -1795,11 +1795,11 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned put_ldev(mdev); wake_asender(mdev); finish_wait(&mdev->misc_wait, &wait); - return TRUE; + return true; } if (signal_pending(current)) { - hlist_del_init(&e->colision); + hlist_del_init(&e->collision); spin_unlock_irq(&mdev->req_lock); @@ -1851,25 +1851,22 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned } if (drbd_submit_ee(mdev, e, rw, DRBD_FAULT_DT_WR) == 0) - return TRUE; + return true; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* don't care for the reason here */ + dev_err(DEV, "submit failed, triggering re-connect\n"); spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); - hlist_del_init(&e->colision); + hlist_del_init(&e->collision); spin_unlock_irq(&mdev->req_lock); if (e->flags & EE_CALL_AL_COMPLETE_IO) drbd_al_complete_io(mdev, e->sector); out_interrupted: - /* yes, the epoch_size now is imbalanced. - * but we drop the connection anyways, so we don't have a chance to - * receive a barrier... atomic_inc(&mdev->epoch_size); */ + drbd_may_finish_epoch(mdev, e->epoch, EV_PUT + EV_CLEANUP); put_ldev(mdev); drbd_free_ee(mdev, e); - return FALSE; + return false; } /* We may throttle resync, if the lower device seems to be busy, @@ -1883,10 +1880,11 @@ out_interrupted: * The current sync rate used here uses only the most recent two step marks, * to have a short time average so we can react faster. */ -int drbd_rs_should_slow_down(struct drbd_conf *mdev) +int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector) { struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk; unsigned long db, dt, dbdt; + struct lc_element *tmp; int curr_events; int throttle = 0; @@ -1894,9 +1892,22 @@ int drbd_rs_should_slow_down(struct drbd_conf *mdev) if (mdev->sync_conf.c_min_rate == 0) return 0; + spin_lock_irq(&mdev->al_lock); + tmp = lc_find(mdev->resync, BM_SECT_TO_EXT(sector)); + if (tmp) { + struct bm_extent *bm_ext = lc_entry(tmp, struct bm_extent, lce); + if (test_bit(BME_PRIORITY, &bm_ext->flags)) { + spin_unlock_irq(&mdev->al_lock); + return 0; + } + /* Do not slow down if app IO is already waiting for this extent */ + } + spin_unlock_irq(&mdev->al_lock); + curr_events = (int)part_stat_read(&disk->part0, sectors[0]) + (int)part_stat_read(&disk->part0, sectors[1]) - atomic_read(&mdev->rs_sect_ev); + if (!mdev->rs_last_events || curr_events - mdev->rs_last_events > 64) { unsigned long rs_left; int i; @@ -1905,8 +1916,12 @@ int drbd_rs_should_slow_down(struct drbd_conf *mdev) /* sync speed average over the last 2*DRBD_SYNC_MARK_STEP, * approx. */ - i = (mdev->rs_last_mark + DRBD_SYNC_MARKS-2) % DRBD_SYNC_MARKS; - rs_left = drbd_bm_total_weight(mdev) - mdev->rs_failed; + i = (mdev->rs_last_mark + DRBD_SYNC_MARKS-1) % DRBD_SYNC_MARKS; + + if (mdev->state.conn == C_VERIFY_S || mdev->state.conn == C_VERIFY_T) + rs_left = mdev->ov_left; + else + rs_left = drbd_bm_total_weight(mdev) - mdev->rs_failed; dt = ((long)jiffies - (long)mdev->rs_mark_time[i]) / HZ; if (!dt) @@ -1934,15 +1949,15 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un sector = be64_to_cpu(p->sector); size = be32_to_cpu(p->blksize); - if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) { + if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) { dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__, (unsigned long long)sector, size); - return FALSE; + return false; } if (sector + (size>>9) > capacity) { dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__, (unsigned long long)sector, size); - return FALSE; + return false; } if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) { @@ -1979,7 +1994,7 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un e = drbd_alloc_ee(mdev, p->block_id, sector, size, GFP_NOIO); if (!e) { put_ldev(mdev); - return FALSE; + return false; } switch (cmd) { @@ -1992,6 +2007,8 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un case P_RS_DATA_REQUEST: e->w.cb = w_e_end_rsdata_req; fault_type = DRBD_FAULT_RS_RD; + /* used in the sector offset progress display */ + mdev->bm_resync_fo = BM_SECT_TO_BIT(sector); break; case P_OV_REPLY: @@ -2013,7 +2030,11 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un if (cmd == P_CSUM_RS_REQUEST) { D_ASSERT(mdev->agreed_pro_version >= 89); e->w.cb = w_e_end_csum_rs_req; + /* used in the sector offset progress display */ + mdev->bm_resync_fo = BM_SECT_TO_BIT(sector); } else if (cmd == P_OV_REPLY) { + /* track progress, we may need to throttle */ + atomic_add(size >> 9, &mdev->rs_sect_in); e->w.cb = w_e_end_ov_reply; dec_rs_pending(mdev); /* drbd_rs_begin_io done when we sent this request, @@ -2025,9 +2046,16 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un case P_OV_REQUEST: if (mdev->ov_start_sector == ~(sector_t)0 && mdev->agreed_pro_version >= 90) { + unsigned long now = jiffies; + int i; mdev->ov_start_sector = sector; mdev->ov_position = sector; - mdev->ov_left = mdev->rs_total - BM_SECT_TO_BIT(sector); + mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector); + mdev->rs_total = mdev->ov_left; + for (i = 0; i < DRBD_SYNC_MARKS; i++) { + mdev->rs_mark_left[i] = mdev->ov_left; + mdev->rs_mark_time[i] = now; + } dev_info(DEV, "Online Verify start sector: %llu\n", (unsigned long long)sector); } @@ -2064,9 +2092,9 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un * we would also throttle its application reads. * In that case, throttling is done on the SyncTarget only. */ - if (mdev->state.peer != R_PRIMARY && drbd_rs_should_slow_down(mdev)) - msleep(100); - if (drbd_rs_begin_io(mdev, e->sector)) + if (mdev->state.peer != R_PRIMARY && drbd_rs_should_slow_down(mdev, sector)) + schedule_timeout_uninterruptible(HZ/10); + if (drbd_rs_begin_io(mdev, sector)) goto out_free_e; submit_for_resync: @@ -2079,11 +2107,10 @@ submit: spin_unlock_irq(&mdev->req_lock); if (drbd_submit_ee(mdev, e, READ, fault_type) == 0) - return TRUE; + return true; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* don't care for the reason here */ + dev_err(DEV, "submit failed, triggering re-connect\n"); spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); spin_unlock_irq(&mdev->req_lock); @@ -2092,7 +2119,7 @@ submit: out_free_e: put_ldev(mdev); drbd_free_ee(mdev, e); - return FALSE; + return false; } static int drbd_asb_recover_0p(struct drbd_conf *mdev) __must_hold(local) @@ -2169,10 +2196,7 @@ static int drbd_asb_recover_0p(struct drbd_conf *mdev) __must_hold(local) static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local) { - int self, peer, hg, rv = -100; - - self = mdev->ldev->md.uuid[UI_BITMAP] & 1; - peer = mdev->p_uuid[UI_BITMAP] & 1; + int hg, rv = -100; switch (mdev->net_conf->after_sb_1p) { case ASB_DISCARD_YOUNGER_PRI: @@ -2199,12 +2223,14 @@ static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local) case ASB_CALL_HELPER: hg = drbd_asb_recover_0p(mdev); if (hg == -1 && mdev->state.role == R_PRIMARY) { - self = drbd_set_role(mdev, R_SECONDARY, 0); + enum drbd_state_rv rv2; + + drbd_set_role(mdev, R_SECONDARY, 0); /* drbd_change_state() does not sleep while in SS_IN_TRANSIENT_STATE, * we might be here in C_WF_REPORT_PARAMS which is transient. * we do not need to wait for the after state change work either. */ - self = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); - if (self != SS_SUCCESS) { + rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); + if (rv2 != SS_SUCCESS) { drbd_khelper(mdev, "pri-lost-after-sb"); } else { dev_warn(DEV, "Successfully gave up primary role.\n"); @@ -2219,10 +2245,7 @@ static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local) static int drbd_asb_recover_2p(struct drbd_conf *mdev) __must_hold(local) { - int self, peer, hg, rv = -100; - - self = mdev->ldev->md.uuid[UI_BITMAP] & 1; - peer = mdev->p_uuid[UI_BITMAP] & 1; + int hg, rv = -100; switch (mdev->net_conf->after_sb_2p) { case ASB_DISCARD_YOUNGER_PRI: @@ -2242,11 +2265,13 @@ static int drbd_asb_recover_2p(struct drbd_conf *mdev) __must_hold(local) case ASB_CALL_HELPER: hg = drbd_asb_recover_0p(mdev); if (hg == -1) { + enum drbd_state_rv rv2; + /* drbd_change_state() does not sleep while in SS_IN_TRANSIENT_STATE, * we might be here in C_WF_REPORT_PARAMS which is transient. * we do not need to wait for the after state change work either. */ - self = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); - if (self != SS_SUCCESS) { + rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); + if (rv2 != SS_SUCCESS) { drbd_khelper(mdev, "pri-lost-after-sb"); } else { dev_warn(DEV, "Successfully gave up primary role.\n"); @@ -2285,6 +2310,8 @@ static void drbd_uuid_dump(struct drbd_conf *mdev, char *text, u64 *uuid, -2 C_SYNC_TARGET set BitMap -100 after split brain, disconnect -1000 unrelated data +-1091 requires proto 91 +-1096 requires proto 96 */ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(local) { @@ -2314,7 +2341,7 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l if (mdev->p_uuid[UI_BITMAP] == (u64)0 && mdev->ldev->md.uuid[UI_BITMAP] != (u64)0) { if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; if ((mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) && (mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1))) { @@ -2335,7 +2362,7 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l if (mdev->ldev->md.uuid[UI_BITMAP] == (u64)0 && mdev->p_uuid[UI_BITMAP] != (u64)0) { if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; if ((mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_BITMAP] & ~((u64)1)) && (mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1))) { @@ -2380,17 +2407,22 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l *rule_nr = 51; peer = mdev->p_uuid[UI_HISTORY_START] & ~((u64)1); if (self == peer) { - self = mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1); - peer = mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1); - if (self == peer) { + if (mdev->agreed_pro_version < 96 ? + (mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == + (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1)) : + peer + UUID_NEW_BM_OFFSET == (mdev->p_uuid[UI_BITMAP] & ~((u64)1))) { /* The last P_SYNC_UUID did not get though. Undo the last start of resync as sync source modifications of the peer's UUIDs. */ if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START]; mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1]; + + dev_info(DEV, "Did not got last syncUUID packet, corrected:\n"); + drbd_uuid_dump(mdev, "peer", mdev->p_uuid, mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]); + return -1; } } @@ -2412,20 +2444,20 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l *rule_nr = 71; self = mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1); if (self == peer) { - self = mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1); - peer = mdev->p_uuid[UI_HISTORY_START] & ~((u64)1); - if (self == peer) { + if (mdev->agreed_pro_version < 96 ? + (mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1)) == + (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) : + self + UUID_NEW_BM_OFFSET == (mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1))) { /* The last P_SYNC_UUID did not get though. Undo the last start of resync as sync source modifications of our UUIDs. */ if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; _drbd_uuid_set(mdev, UI_BITMAP, mdev->ldev->md.uuid[UI_HISTORY_START]); _drbd_uuid_set(mdev, UI_HISTORY_START, mdev->ldev->md.uuid[UI_HISTORY_START + 1]); - dev_info(DEV, "Undid last start of resync:\n"); - + dev_info(DEV, "Last syncUUID did not get through, corrected:\n"); drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid, mdev->state.disk >= D_NEGOTIATING ? drbd_bm_total_weight(mdev) : 0, 0); @@ -2488,8 +2520,8 @@ static enum drbd_conns drbd_sync_handshake(struct drbd_conf *mdev, enum drbd_rol dev_alert(DEV, "Unrelated data, aborting!\n"); return C_MASK; } - if (hg == -1001) { - dev_alert(DEV, "To resolve this both sides have to support at least protocol\n"); + if (hg < -1000) { + dev_alert(DEV, "To resolve this both sides have to support at least protocol %d\n", -hg - 1000); return C_MASK; } @@ -2588,7 +2620,8 @@ static enum drbd_conns drbd_sync_handshake(struct drbd_conf *mdev, enum drbd_rol if (abs(hg) >= 2) { dev_info(DEV, "Writing the whole bitmap, full sync required after drbd_sync_handshake.\n"); - if (drbd_bitmap_io(mdev, &drbd_bmio_set_n_write, "set_n_write from sync_handshake")) + if (drbd_bitmap_io(mdev, &drbd_bmio_set_n_write, "set_n_write from sync_handshake", + BM_LOCKED_SET_ALLOWED)) return C_MASK; } @@ -2682,7 +2715,7 @@ static int receive_protocol(struct drbd_conf *mdev, enum drbd_packets cmd, unsig unsigned char *my_alg = mdev->net_conf->integrity_alg; if (drbd_recv(mdev, p_integrity_alg, data_size) != data_size) - return FALSE; + return false; p_integrity_alg[SHARED_SECRET_MAX-1] = 0; if (strcmp(p_integrity_alg, my_alg)) { @@ -2693,11 +2726,11 @@ static int receive_protocol(struct drbd_conf *mdev, enum drbd_packets cmd, unsig my_alg[0] ? my_alg : (unsigned char *)"<not-used>"); } - return TRUE; + return true; disconnect: drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } /* helper function @@ -2729,7 +2762,7 @@ struct crypto_hash *drbd_crypto_alloc_digest_safe(const struct drbd_conf *mdev, static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int packet_size) { - int ok = TRUE; + int ok = true; struct p_rs_param_95 *p = &mdev->data.rbuf.rs_param_95; unsigned int header_size, data_size, exp_max_sz; struct crypto_hash *verify_tfm = NULL; @@ -2747,7 +2780,7 @@ static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsi if (packet_size > exp_max_sz) { dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n", packet_size, exp_max_sz); - return FALSE; + return false; } if (apv <= 88) { @@ -2767,7 +2800,7 @@ static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsi memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX); if (drbd_recv(mdev, &p->head.payload, header_size) != header_size) - return FALSE; + return false; mdev->sync_conf.rate = be32_to_cpu(p->rate); @@ -2777,11 +2810,11 @@ static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsi dev_err(DEV, "verify-alg too long, " "peer wants %u, accepting only %u byte\n", data_size, SHARED_SECRET_MAX); - return FALSE; + return false; } if (drbd_recv(mdev, p->verify_alg, data_size) != data_size) - return FALSE; + return false; /* we expect NUL terminated string */ /* but just in case someone tries to be evil */ @@ -2875,13 +2908,7 @@ disconnect: /* but free the verify_tfm again, if csums_tfm did not work out */ crypto_free_hash(verify_tfm); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; -} - -static void drbd_setup_order_type(struct drbd_conf *mdev, int peer) -{ - /* sorry, we currently have no working implementation - * of distributed TCQ */ + return false; } /* warn if the arguments differ by more than 12.5% */ @@ -2901,7 +2928,6 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned { struct p_sizes *p = &mdev->data.rbuf.sizes; enum determine_dev_size dd = unchanged; - unsigned int max_seg_s; sector_t p_size, p_usize, my_usize; int ldsc = 0; /* local disk size changed */ enum dds_flags ddsf; @@ -2912,7 +2938,7 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (p_size == 0 && mdev->state.disk == D_DISKLESS) { dev_err(DEV, "some backing storage is needed\n"); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } /* just store the peer's disk size for now. @@ -2949,41 +2975,32 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); mdev->ldev->dc.disk_size = my_usize; put_ldev(mdev); - return FALSE; + return false; } put_ldev(mdev); } -#undef min_not_zero ddsf = be16_to_cpu(p->dds_flags); if (get_ldev(mdev)) { - dd = drbd_determin_dev_size(mdev, ddsf); + dd = drbd_determine_dev_size(mdev, ddsf); put_ldev(mdev); if (dd == dev_size_error) - return FALSE; + return false; drbd_md_sync(mdev); } else { /* I am diskless, need to accept the peer's size. */ drbd_set_my_capacity(mdev, p_size); } + mdev->peer_max_bio_size = be32_to_cpu(p->max_bio_size); + drbd_reconsider_max_bio_size(mdev); + if (get_ldev(mdev)) { if (mdev->ldev->known_size != drbd_get_capacity(mdev->ldev->backing_bdev)) { mdev->ldev->known_size = drbd_get_capacity(mdev->ldev->backing_bdev); ldsc = 1; } - if (mdev->agreed_pro_version < 94) - max_seg_s = be32_to_cpu(p->max_segment_size); - else if (mdev->agreed_pro_version == 94) - max_seg_s = DRBD_MAX_SIZE_H80_PACKET; - else /* drbd 8.3.8 onwards */ - max_seg_s = DRBD_MAX_SEGMENT_SIZE; - - if (max_seg_s != queue_max_segment_size(mdev->rq_queue)) - drbd_setup_queue_param(mdev, max_seg_s); - - drbd_setup_order_type(mdev, be16_to_cpu(p->queue_order_type)); put_ldev(mdev); } @@ -3007,14 +3024,14 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned } } - return TRUE; + return true; } static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { struct p_uuids *p = &mdev->data.rbuf.uuids; u64 *p_uuid; - int i; + int i, updated_uuids = 0; p_uuid = kmalloc(sizeof(u64)*UI_EXTENDED_SIZE, GFP_NOIO); @@ -3031,7 +3048,7 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned dev_err(DEV, "Can only connect to data with current UUID=%016llX\n", (unsigned long long)mdev->ed_uuid); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } if (get_ldev(mdev)) { @@ -3043,19 +3060,21 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (skip_initial_sync) { dev_info(DEV, "Accepted new current UUID, preparing to skip initial sync\n"); drbd_bitmap_io(mdev, &drbd_bmio_clear_n_write, - "clear_n_write from receive_uuids"); + "clear_n_write from receive_uuids", + BM_LOCKED_TEST_ALLOWED); _drbd_uuid_set(mdev, UI_CURRENT, p_uuid[UI_CURRENT]); _drbd_uuid_set(mdev, UI_BITMAP, 0); _drbd_set_state(_NS2(mdev, disk, D_UP_TO_DATE, pdsk, D_UP_TO_DATE), CS_VERBOSE, NULL); drbd_md_sync(mdev); + updated_uuids = 1; } put_ldev(mdev); } else if (mdev->state.disk < D_INCONSISTENT && mdev->state.role == R_PRIMARY) { /* I am a diskless primary, the peer just created a new current UUID for me. */ - drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); + updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); } /* Before we test for the disk state, we should wait until an eventually @@ -3064,9 +3083,12 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned new disk state... */ wait_event(mdev->misc_wait, !test_bit(CLUSTER_ST_CHANGE, &mdev->flags)); if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT) - drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); + updated_uuids |= drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); - return TRUE; + if (updated_uuids) + drbd_print_uuids(mdev, "receiver updated UUIDs to"); + + return true; } /** @@ -3103,7 +3125,7 @@ static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsi { struct p_req_state *p = &mdev->data.rbuf.req_state; union drbd_state mask, val; - int rv; + enum drbd_state_rv rv; mask.i = be32_to_cpu(p->mask); val.i = be32_to_cpu(p->val); @@ -3111,7 +3133,7 @@ static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsi if (test_bit(DISCARD_CONCURRENT, &mdev->flags) && test_bit(CLUSTER_ST_CHANGE, &mdev->flags)) { drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG); - return TRUE; + return true; } mask = convert_state(mask); @@ -3122,7 +3144,7 @@ static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsi drbd_send_sr_reply(mdev, rv); drbd_md_sync(mdev); - return TRUE; + return true; } static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -3167,7 +3189,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned peer_state.conn == C_CONNECTED) { if (drbd_bm_total_weight(mdev) <= mdev->rs_failed) drbd_resync_finished(mdev); - return TRUE; + return true; } } @@ -3183,6 +3205,9 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (ns.conn == C_WF_REPORT_PARAMS) ns.conn = C_CONNECTED; + if (peer_state.conn == C_AHEAD) + ns.conn = C_BEHIND; + if (mdev->p_uuid && peer_state.disk >= D_NEGOTIATING && get_ldev_if_state(mdev, D_NEGOTIATING)) { int cr; /* consider resync */ @@ -3217,10 +3242,10 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned real_peer_disk = D_DISKLESS; } else { if (test_and_clear_bit(CONN_DRY_RUN, &mdev->flags)) - return FALSE; + return false; D_ASSERT(os.conn == C_WF_REPORT_PARAMS); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } } } @@ -3245,7 +3270,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_uuid_new_current(mdev); clear_bit(NEW_CUR_UUID, &mdev->flags); drbd_force_state(mdev, NS2(conn, C_PROTOCOL_ERROR, susp, 0)); - return FALSE; + return false; } rv = _drbd_set_state(mdev, ns, cs_flags, NULL); ns = mdev->state; @@ -3253,7 +3278,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (rv < SS_SUCCESS) { drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } if (os.conn > C_WF_REPORT_PARAMS) { @@ -3271,7 +3296,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_md_sync(mdev); /* update connected indicator, la_size, ... */ - return TRUE; + return true; } static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -3280,6 +3305,7 @@ static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsi wait_event(mdev->misc_wait, mdev->state.conn == C_WF_SYNC_UUID || + mdev->state.conn == C_BEHIND || mdev->state.conn < C_CONNECTED || mdev->state.disk < D_NEGOTIATING); @@ -3291,32 +3317,42 @@ static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsi _drbd_uuid_set(mdev, UI_CURRENT, be64_to_cpu(p->uuid)); _drbd_uuid_set(mdev, UI_BITMAP, 0UL); + drbd_print_uuids(mdev, "updated sync uuid"); drbd_start_resync(mdev, C_SYNC_TARGET); put_ldev(mdev); } else dev_err(DEV, "Ignoring SyncUUID packet!\n"); - return TRUE; + return true; } -enum receive_bitmap_ret { OK, DONE, FAILED }; - -static enum receive_bitmap_ret +/** + * receive_bitmap_plain + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int receive_bitmap_plain(struct drbd_conf *mdev, unsigned int data_size, unsigned long *buffer, struct bm_xfer_ctx *c) { unsigned num_words = min_t(size_t, BM_PACKET_WORDS, c->bm_words - c->word_offset); unsigned want = num_words * sizeof(long); + int err; if (want != data_size) { dev_err(DEV, "%s:want (%u) != data_size (%u)\n", __func__, want, data_size); - return FAILED; + return -EIO; } if (want == 0) - return DONE; - if (drbd_recv(mdev, buffer, want) != want) - return FAILED; + return 0; + err = drbd_recv(mdev, buffer, want); + if (err != want) { + if (err >= 0) + err = -EIO; + return err; + } drbd_bm_merge_lel(mdev, c->word_offset, num_words, buffer); @@ -3325,10 +3361,16 @@ receive_bitmap_plain(struct drbd_conf *mdev, unsigned int data_size, if (c->bit_offset > c->bm_bits) c->bit_offset = c->bm_bits; - return OK; + return 1; } -static enum receive_bitmap_ret +/** + * recv_bm_rle_bits + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int recv_bm_rle_bits(struct drbd_conf *mdev, struct p_compressed_bm *p, struct bm_xfer_ctx *c) @@ -3348,18 +3390,18 @@ recv_bm_rle_bits(struct drbd_conf *mdev, bits = bitstream_get_bits(&bs, &look_ahead, 64); if (bits < 0) - return FAILED; + return -EIO; for (have = bits; have > 0; s += rl, toggle = !toggle) { bits = vli_decode_bits(&rl, look_ahead); if (bits <= 0) - return FAILED; + return -EIO; if (toggle) { e = s + rl -1; if (e >= c->bm_bits) { dev_err(DEV, "bitmap overflow (e:%lu) while decoding bm RLE packet\n", e); - return FAILED; + return -EIO; } _drbd_bm_set_bits(mdev, s, e); } @@ -3369,14 +3411,14 @@ recv_bm_rle_bits(struct drbd_conf *mdev, have, bits, look_ahead, (unsigned int)(bs.cur.b - p->code), (unsigned int)bs.buf_len); - return FAILED; + return -EIO; } look_ahead >>= bits; have -= bits; bits = bitstream_get_bits(&bs, &tmp, 64 - have); if (bits < 0) - return FAILED; + return -EIO; look_ahead |= tmp << have; have += bits; } @@ -3384,10 +3426,16 @@ recv_bm_rle_bits(struct drbd_conf *mdev, c->bit_offset = s; bm_xfer_ctx_bit_to_word_offset(c); - return (s == c->bm_bits) ? DONE : OK; + return (s != c->bm_bits); } -static enum receive_bitmap_ret +/** + * decode_bitmap_c + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int decode_bitmap_c(struct drbd_conf *mdev, struct p_compressed_bm *p, struct bm_xfer_ctx *c) @@ -3401,7 +3449,7 @@ decode_bitmap_c(struct drbd_conf *mdev, dev_err(DEV, "receive_bitmap_c: unknown encoding %u\n", p->encoding); drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); - return FAILED; + return -EIO; } void INFO_bm_xfer_stats(struct drbd_conf *mdev, @@ -3450,13 +3498,13 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne { struct bm_xfer_ctx c; void *buffer; - enum receive_bitmap_ret ret; - int ok = FALSE; + int err; + int ok = false; struct p_header80 *h = &mdev->data.rbuf.header.h80; - wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); - - drbd_bm_lock(mdev, "receive bitmap"); + drbd_bm_lock(mdev, "receive bitmap", BM_LOCKED_SET_ALLOWED); + /* you are supposed to send additional out-of-sync information + * if you actually set bits during this phase */ /* maybe we should use some per thread scratch page, * and allocate that during initial device creation? */ @@ -3471,9 +3519,9 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne .bm_words = drbd_bm_words(mdev), }; - do { + for(;;) { if (cmd == P_BITMAP) { - ret = receive_bitmap_plain(mdev, data_size, buffer, &c); + err = receive_bitmap_plain(mdev, data_size, buffer, &c); } else if (cmd == P_COMPRESSED_BITMAP) { /* MAYBE: sanity check that we speak proto >= 90, * and the feature is enabled! */ @@ -3490,9 +3538,9 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne goto out; if (data_size <= (sizeof(*p) - sizeof(p->head))) { dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", data_size); - return FAILED; + goto out; } - ret = decode_bitmap_c(mdev, p, &c); + err = decode_bitmap_c(mdev, p, &c); } else { dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", cmd); goto out; @@ -3501,24 +3549,26 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne c.packets[cmd == P_BITMAP]++; c.bytes[cmd == P_BITMAP] += sizeof(struct p_header80) + data_size; - if (ret != OK) + if (err <= 0) { + if (err < 0) + goto out; break; - + } if (!drbd_recv_header(mdev, &cmd, &data_size)) goto out; - } while (ret == OK); - if (ret == FAILED) - goto out; + } INFO_bm_xfer_stats(mdev, "receive", &c); if (mdev->state.conn == C_WF_BITMAP_T) { + enum drbd_state_rv rv; + ok = !drbd_send_bitmap(mdev); if (!ok) goto out; /* Omit CS_ORDERED with this state transition to avoid deadlocks. */ - ok = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE); - D_ASSERT(ok == SS_SUCCESS); + rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE); + D_ASSERT(rv == SS_SUCCESS); } else if (mdev->state.conn != C_WF_BITMAP_S) { /* admin may have requested C_DISCONNECTING, * other threads may have noticed network errors */ @@ -3526,7 +3576,7 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne drbd_conn_str(mdev->state.conn)); } - ok = TRUE; + ok = true; out: drbd_bm_unlock(mdev); if (ok && mdev->state.conn == C_WF_BITMAP_S) @@ -3556,14 +3606,30 @@ static int receive_skip(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned static int receive_UnplugRemote(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - if (mdev->state.disk >= D_INCONSISTENT) - drbd_kick_lo(mdev); - /* Make sure we've acked all the TCP data associated * with the data requests being unplugged */ drbd_tcp_quickack(mdev->data.socket); - return TRUE; + return true; +} + +static int receive_out_of_sync(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) +{ + struct p_block_desc *p = &mdev->data.rbuf.block_desc; + + switch (mdev->state.conn) { + case C_WF_SYNC_UUID: + case C_WF_BITMAP_T: + case C_BEHIND: + break; + default: + dev_err(DEV, "ASSERT FAILED cstate = %s, expected: WFSyncUUID|WFBitMapT|Behind\n", + drbd_conn_str(mdev->state.conn)); + } + + drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize)); + + return true; } typedef int (*drbd_cmd_handler_f)(struct drbd_conf *, enum drbd_packets cmd, unsigned int to_receive); @@ -3596,6 +3662,7 @@ static struct data_cmd drbd_cmd_handler[] = { [P_OV_REPLY] = { 1, sizeof(struct p_block_req), receive_DataRequest }, [P_CSUM_RS_REQUEST] = { 1, sizeof(struct p_block_req), receive_DataRequest }, [P_DELAY_PROBE] = { 0, sizeof(struct p_delay_probe93), receive_skip }, + [P_OUT_OF_SYNC] = { 0, sizeof(struct p_block_desc), receive_out_of_sync }, /* anything missing from this table is in * the asender_tbl, see get_asender_cmd */ [P_MAX_CMD] = { 0, 0, NULL }, @@ -3635,7 +3702,8 @@ static void drbdd(struct drbd_conf *mdev) if (shs) { rv = drbd_recv(mdev, &header->h80.payload, shs); if (unlikely(rv != shs)) { - dev_err(DEV, "short read while reading sub header: rv=%d\n", rv); + if (!signal_pending(current)) + dev_warn(DEV, "short read while reading sub header: rv=%d\n", rv); goto err_out; } } @@ -3707,9 +3775,6 @@ static void drbd_disconnect(struct drbd_conf *mdev) if (mdev->state.conn == C_STANDALONE) return; - if (mdev->state.conn >= C_WF_CONNECTION) - dev_err(DEV, "ASSERT FAILED cstate = %s, expected < WFConnection\n", - drbd_conn_str(mdev->state.conn)); /* asender does not clean up anything. it must not interfere, either */ drbd_thread_stop(&mdev->asender); @@ -3738,6 +3803,8 @@ static void drbd_disconnect(struct drbd_conf *mdev) atomic_set(&mdev->rs_pending_cnt, 0); wake_up(&mdev->misc_wait); + del_timer(&mdev->request_timer); + /* make sure syncer is stopped and w_resume_next_sg queued */ del_timer_sync(&mdev->resync_timer); resync_timer_fn((unsigned long)mdev); @@ -3783,13 +3850,6 @@ static void drbd_disconnect(struct drbd_conf *mdev) if (os.conn == C_DISCONNECTING) { wait_event(mdev->net_cnt_wait, atomic_read(&mdev->net_cnt) == 0); - if (!is_susp(mdev->state)) { - /* we must not free the tl_hash - * while application io is still on the fly */ - wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); - drbd_free_tl_hash(mdev); - } - crypto_free_hash(mdev->cram_hmac_tfm); mdev->cram_hmac_tfm = NULL; @@ -3798,6 +3858,10 @@ static void drbd_disconnect(struct drbd_conf *mdev) drbd_request_state(mdev, NS(conn, C_STANDALONE)); } + /* serialize with bitmap writeout triggered by the state change, + * if any. */ + wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags)); + /* tcp_close and release of sendpage pages can be deferred. I don't * want to use SO_LINGER, because apparently it can be deferred for * more than 20 seconds (longest time I checked). @@ -3898,7 +3962,8 @@ static int drbd_do_handshake(struct drbd_conf *mdev) rv = drbd_recv(mdev, &p->head.payload, expect); if (rv != expect) { - dev_err(DEV, "short read receiving handshake packet: l=%u\n", rv); + if (!signal_pending(current)) + dev_warn(DEV, "short read receiving handshake packet: l=%u\n", rv); return 0; } @@ -4000,7 +4065,8 @@ static int drbd_do_auth(struct drbd_conf *mdev) rv = drbd_recv(mdev, peers_ch, length); if (rv != length) { - dev_err(DEV, "short read AuthChallenge: l=%u\n", rv); + if (!signal_pending(current)) + dev_warn(DEV, "short read AuthChallenge: l=%u\n", rv); rv = 0; goto fail; } @@ -4047,7 +4113,8 @@ static int drbd_do_auth(struct drbd_conf *mdev) rv = drbd_recv(mdev, response , resp_size); if (rv != resp_size) { - dev_err(DEV, "short read receiving AuthResponse: l=%u\n", rv); + if (!signal_pending(current)) + dev_warn(DEV, "short read receiving AuthResponse: l=%u\n", rv); rv = 0; goto fail; } @@ -4099,8 +4166,7 @@ int drbdd_init(struct drbd_thread *thi) h = drbd_connect(mdev); if (h == 0) { drbd_disconnect(mdev); - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ); + schedule_timeout_interruptible(HZ); } if (h == -1) { dev_warn(DEV, "Discarding network configuration.\n"); @@ -4138,7 +4204,7 @@ static int got_RqSReply(struct drbd_conf *mdev, struct p_header80 *h) } wake_up(&mdev->state_wait); - return TRUE; + return true; } static int got_Ping(struct drbd_conf *mdev, struct p_header80 *h) @@ -4154,7 +4220,7 @@ static int got_PingAck(struct drbd_conf *mdev, struct p_header80 *h) if (!test_and_set_bit(GOT_PING_ACK, &mdev->flags)) wake_up(&mdev->misc_wait); - return TRUE; + return true; } static int got_IsInSync(struct drbd_conf *mdev, struct p_header80 *h) @@ -4177,7 +4243,7 @@ static int got_IsInSync(struct drbd_conf *mdev, struct p_header80 *h) dec_rs_pending(mdev); atomic_add(blksize >> 9, &mdev->rs_sect_in); - return TRUE; + return true; } /* when we receive the ACK for a write request, @@ -4189,7 +4255,7 @@ static struct drbd_request *_ack_id_to_req(struct drbd_conf *mdev, struct hlist_node *n; struct drbd_request *req; - hlist_for_each_entry(req, n, slot, colision) { + hlist_for_each_entry(req, n, slot, collision) { if ((unsigned long)req == (unsigned long)id) { if (req->sector != sector) { dev_err(DEV, "_ack_id_to_req: found req %p but it has " @@ -4201,8 +4267,6 @@ static struct drbd_request *_ack_id_to_req(struct drbd_conf *mdev, return req; } } - dev_err(DEV, "_ack_id_to_req: failed to find req %p, sector %llus in list\n", - (void *)(unsigned long)id, (unsigned long long)sector); return NULL; } @@ -4220,15 +4284,17 @@ static int validate_req_change_req_state(struct drbd_conf *mdev, req = validator(mdev, id, sector); if (unlikely(!req)) { spin_unlock_irq(&mdev->req_lock); - dev_err(DEV, "%s: got a corrupt block_id/sector pair\n", func); - return FALSE; + + dev_err(DEV, "%s: failed to find req %p, sector %llus\n", func, + (void *)(unsigned long)id, (unsigned long long)sector); + return false; } __req_mod(req, what, &m); spin_unlock_irq(&mdev->req_lock); if (m.bio) complete_master_bio(mdev, &m); - return TRUE; + return true; } static int got_BlockAck(struct drbd_conf *mdev, struct p_header80 *h) @@ -4243,7 +4309,7 @@ static int got_BlockAck(struct drbd_conf *mdev, struct p_header80 *h) if (is_syncer_block_id(p->block_id)) { drbd_set_in_sync(mdev, sector, blksize); dec_rs_pending(mdev); - return TRUE; + return true; } switch (be16_to_cpu(h->command)) { case P_RS_WRITE_ACK: @@ -4264,7 +4330,7 @@ static int got_BlockAck(struct drbd_conf *mdev, struct p_header80 *h) break; default: D_ASSERT(0); - return FALSE; + return false; } return validate_req_change_req_state(mdev, p->block_id, sector, @@ -4275,20 +4341,44 @@ static int got_NegAck(struct drbd_conf *mdev, struct p_header80 *h) { struct p_block_ack *p = (struct p_block_ack *)h; sector_t sector = be64_to_cpu(p->sector); - - if (__ratelimit(&drbd_ratelimit_state)) - dev_warn(DEV, "Got NegAck packet. Peer is in troubles?\n"); + int size = be32_to_cpu(p->blksize); + struct drbd_request *req; + struct bio_and_error m; update_peer_seq(mdev, be32_to_cpu(p->seq_num)); if (is_syncer_block_id(p->block_id)) { - int size = be32_to_cpu(p->blksize); dec_rs_pending(mdev); drbd_rs_failed_io(mdev, sector, size); - return TRUE; + return true; } - return validate_req_change_req_state(mdev, p->block_id, sector, - _ack_id_to_req, __func__ , neg_acked); + + spin_lock_irq(&mdev->req_lock); + req = _ack_id_to_req(mdev, p->block_id, sector); + if (!req) { + spin_unlock_irq(&mdev->req_lock); + if (mdev->net_conf->wire_protocol == DRBD_PROT_A || + mdev->net_conf->wire_protocol == DRBD_PROT_B) { + /* Protocol A has no P_WRITE_ACKs, but has P_NEG_ACKs. + The master bio might already be completed, therefore the + request is no longer in the collision hash. + => Do not try to validate block_id as request. */ + /* In Protocol B we might already have got a P_RECV_ACK + but then get a P_NEG_ACK after wards. */ + drbd_set_out_of_sync(mdev, sector, size); + return true; + } else { + dev_err(DEV, "%s: failed to find req %p, sector %llus\n", __func__, + (void *)(unsigned long)p->block_id, (unsigned long long)sector); + return false; + } + } + __req_mod(req, neg_acked, &m); + spin_unlock_irq(&mdev->req_lock); + + if (m.bio) + complete_master_bio(mdev, &m); + return true; } static int got_NegDReply(struct drbd_conf *mdev, struct p_header80 *h) @@ -4319,11 +4409,20 @@ static int got_NegRSDReply(struct drbd_conf *mdev, struct p_header80 *h) if (get_ldev_if_state(mdev, D_FAILED)) { drbd_rs_complete_io(mdev, sector); - drbd_rs_failed_io(mdev, sector, size); + switch (be16_to_cpu(h->command)) { + case P_NEG_RS_DREPLY: + drbd_rs_failed_io(mdev, sector, size); + case P_RS_CANCEL: + break; + default: + D_ASSERT(0); + put_ldev(mdev); + return false; + } put_ldev(mdev); } - return TRUE; + return true; } static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) @@ -4332,7 +4431,14 @@ static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) tl_release(mdev, p->barrier, be32_to_cpu(p->set_size)); - return TRUE; + if (mdev->state.conn == C_AHEAD && + atomic_read(&mdev->ap_in_flight) == 0 && + !test_and_set_bit(AHEAD_TO_SYNC_SOURCE, &mdev->current_epoch->flags)) { + mdev->start_resync_timer.expires = jiffies + HZ; + add_timer(&mdev->start_resync_timer); + } + + return true; } static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) @@ -4353,12 +4459,18 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) ov_oos_print(mdev); if (!get_ldev(mdev)) - return TRUE; + return true; drbd_rs_complete_io(mdev, sector); dec_rs_pending(mdev); - if (--mdev->ov_left == 0) { + --mdev->ov_left; + + /* let's advance progress step marks only for every other megabyte */ + if ((mdev->ov_left & 0x200) == 0x200) + drbd_advance_rs_marks(mdev, mdev->ov_left); + + if (mdev->ov_left == 0) { w = kmalloc(sizeof(*w), GFP_NOIO); if (w) { w->cb = w_ov_finished; @@ -4370,12 +4482,12 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) } } put_ldev(mdev); - return TRUE; + return true; } static int got_skip(struct drbd_conf *mdev, struct p_header80 *h) { - return TRUE; + return true; } struct asender_cmd { @@ -4403,6 +4515,7 @@ static struct asender_cmd *get_asender_cmd(int cmd) [P_STATE_CHG_REPLY] = { sizeof(struct p_req_state_reply), got_RqSReply }, [P_RS_IS_IN_SYNC] = { sizeof(struct p_block_ack), got_IsInSync }, [P_DELAY_PROBE] = { sizeof(struct p_delay_probe93), got_skip }, + [P_RS_CANCEL] = { sizeof(struct p_block_ack), got_NegRSDReply}, [P_MAX_CMD] = { 0, NULL }, }; if (cmd > P_MAX_CMD || asender_tbl[cmd].process == NULL) @@ -4421,6 +4534,7 @@ int drbd_asender(struct drbd_thread *thi) int received = 0; int expect = sizeof(struct p_header80); int empty; + int ping_timeout_active = 0; sprintf(current->comm, "drbd%d_asender", mdev_to_minor(mdev)); @@ -4433,6 +4547,7 @@ int drbd_asender(struct drbd_thread *thi) ERR_IF(!drbd_send_ping(mdev)) goto reconnect; mdev->meta.socket->sk->sk_rcvtimeo = mdev->net_conf->ping_timeo*HZ/10; + ping_timeout_active = 1; } /* conditionally cork; @@ -4487,8 +4602,7 @@ int drbd_asender(struct drbd_thread *thi) dev_err(DEV, "meta connection shut down by peer.\n"); goto reconnect; } else if (rv == -EAGAIN) { - if (mdev->meta.socket->sk->sk_rcvtimeo == - mdev->net_conf->ping_timeo*HZ/10) { + if (ping_timeout_active) { dev_err(DEV, "PingAck did not arrive in time.\n"); goto reconnect; } @@ -4527,6 +4641,11 @@ int drbd_asender(struct drbd_thread *thi) if (!cmd->process(mdev, h)) goto reconnect; + /* the idle_timeout (ping-int) + * has been restored in got_PingAck() */ + if (cmd == get_asender_cmd(P_PING_ACK)) + ping_timeout_active = 0; + buf = h; received = 0; expect = sizeof(struct p_header80); diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index 11a75d3..3424d67 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -140,9 +140,14 @@ static void _about_to_complete_local_write(struct drbd_conf *mdev, struct hlist_node *n; struct hlist_head *slot; - /* before we can signal completion to the upper layers, - * we may need to close the current epoch */ + /* Before we can signal completion to the upper layers, + * we may need to close the current epoch. + * We can skip this, if this request has not even been sent, because we + * did not have a fully established connection yet/anymore, during + * bitmap exchange, or while we are C_AHEAD due to congestion policy. + */ if (mdev->state.conn >= C_CONNECTED && + (s & RQ_NET_SENT) != 0 && req->epoch == mdev->newest_tle->br_number) queue_barrier(mdev); @@ -158,7 +163,7 @@ static void _about_to_complete_local_write(struct drbd_conf *mdev, * they must have been failed on the spot */ #define OVERLAPS overlaps(sector, size, i->sector, i->size) slot = tl_hash_slot(mdev, sector); - hlist_for_each_entry(i, n, slot, colision) { + hlist_for_each_entry(i, n, slot, collision) { if (OVERLAPS) { dev_alert(DEV, "LOGIC BUG: completed: %p %llus +%u; " "other: %p %llus +%u\n", @@ -182,7 +187,7 @@ static void _about_to_complete_local_write(struct drbd_conf *mdev, #undef OVERLAPS #define OVERLAPS overlaps(sector, size, e->sector, e->size) slot = ee_hash_slot(mdev, req->sector); - hlist_for_each_entry(e, n, slot, colision) { + hlist_for_each_entry(e, n, slot, collision) { if (OVERLAPS) { wake_up(&mdev->misc_wait); break; @@ -255,8 +260,8 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m) /* remove the request from the conflict detection * respective block_id verification hash */ - if (!hlist_unhashed(&req->colision)) - hlist_del(&req->colision); + if (!hlist_unhashed(&req->collision)) + hlist_del(&req->collision); else D_ASSERT((s & (RQ_NET_MASK & ~RQ_NET_DONE)) == 0); @@ -324,7 +329,7 @@ static int _req_conflicts(struct drbd_request *req) struct hlist_node *n; struct hlist_head *slot; - D_ASSERT(hlist_unhashed(&req->colision)); + D_ASSERT(hlist_unhashed(&req->collision)); if (!get_net_conf(mdev)) return 0; @@ -336,7 +341,7 @@ static int _req_conflicts(struct drbd_request *req) #define OVERLAPS overlaps(i->sector, i->size, sector, size) slot = tl_hash_slot(mdev, sector); - hlist_for_each_entry(i, n, slot, colision) { + hlist_for_each_entry(i, n, slot, collision) { if (OVERLAPS) { dev_alert(DEV, "%s[%u] Concurrent local write detected! " "[DISCARD L] new: %llus +%u; " @@ -354,7 +359,7 @@ static int _req_conflicts(struct drbd_request *req) #undef OVERLAPS #define OVERLAPS overlaps(e->sector, e->size, sector, size) slot = ee_hash_slot(mdev, sector); - hlist_for_each_entry(e, n, slot, colision) { + hlist_for_each_entry(e, n, slot, collision) { if (OVERLAPS) { dev_alert(DEV, "%s[%u] Concurrent remote write detected!" " [DISCARD L] new: %llus +%u; " @@ -440,7 +445,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state |= RQ_LOCAL_COMPLETED; req->rq_state &= ~RQ_LOCAL_PENDING; - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); _req_may_be_done_not_susp(req, m); put_ldev(mdev); break; @@ -461,7 +466,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, D_ASSERT(!(req->rq_state & RQ_NET_MASK)); - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); put_ldev(mdev); /* no point in retrying if there is no good remote data, @@ -486,7 +491,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, /* so we can verify the handle in the answer packet * corresponding hlist_del is in _req_may_be_done() */ - hlist_add_head(&req->colision, ar_hash_slot(mdev, req->sector)); + hlist_add_head(&req->collision, ar_hash_slot(mdev, req->sector)); set_bit(UNPLUG_REMOTE, &mdev->flags); @@ -502,7 +507,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, /* assert something? */ /* from drbd_make_request_common only */ - hlist_add_head(&req->colision, tl_hash_slot(mdev, req->sector)); + hlist_add_head(&req->collision, tl_hash_slot(mdev, req->sector)); /* corresponding hlist_del is in _req_may_be_done() */ /* NOTE @@ -545,6 +550,14 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, break; + case queue_for_send_oos: + req->rq_state |= RQ_NET_QUEUED; + req->w.cb = w_send_oos; + drbd_queue_work(&mdev->data.work, &req->w); + break; + + case oos_handed_to_network: + /* actually the same */ case send_canceled: /* treat it the same */ case send_failed: @@ -558,6 +571,9 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, case handed_over_to_network: /* assert something? */ + if (bio_data_dir(req->master_bio) == WRITE) + atomic_add(req->size>>9, &mdev->ap_in_flight); + if (bio_data_dir(req->master_bio) == WRITE && mdev->net_conf->wire_protocol == DRBD_PROT_A) { /* this is what is dangerous about protocol A: @@ -591,6 +607,9 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, dec_ap_pending(mdev); req->rq_state &= ~(RQ_NET_OK|RQ_NET_PENDING); req->rq_state |= RQ_NET_DONE; + if (req->rq_state & RQ_NET_SENT && req->rq_state & RQ_WRITE) + atomic_sub(req->size>>9, &mdev->ap_in_flight); + /* if it is still queued, we may not complete it here. * it will be canceled soon. */ if (!(req->rq_state & RQ_NET_QUEUED)) @@ -628,14 +647,17 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state |= RQ_NET_OK; D_ASSERT(req->rq_state & RQ_NET_PENDING); dec_ap_pending(mdev); + atomic_sub(req->size>>9, &mdev->ap_in_flight); req->rq_state &= ~RQ_NET_PENDING; _req_may_be_done_not_susp(req, m); break; case neg_acked: /* assert something? */ - if (req->rq_state & RQ_NET_PENDING) + if (req->rq_state & RQ_NET_PENDING) { dec_ap_pending(mdev); + atomic_sub(req->size>>9, &mdev->ap_in_flight); + } req->rq_state &= ~(RQ_NET_OK|RQ_NET_PENDING); req->rq_state |= RQ_NET_DONE; @@ -690,8 +712,11 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, dev_err(DEV, "FIXME (barrier_acked but pending)\n"); list_move(&req->tl_requests, &mdev->out_of_sequence_requests); } - D_ASSERT(req->rq_state & RQ_NET_SENT); - req->rq_state |= RQ_NET_DONE; + if ((req->rq_state & RQ_NET_MASK) != 0) { + req->rq_state |= RQ_NET_DONE; + if (mdev->net_conf->wire_protocol == DRBD_PROT_A) + atomic_sub(req->size>>9, &mdev->ap_in_flight); + } _req_may_be_done(req, m); /* Allowed while state.susp */ break; @@ -738,14 +763,14 @@ static int drbd_may_do_local_read(struct drbd_conf *mdev, sector_t sector, int s return 0 == drbd_bm_count_bits(mdev, sbnr, ebnr); } -static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) +static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio, unsigned long start_time) { const int rw = bio_rw(bio); const int size = bio->bi_size; const sector_t sector = bio->bi_sector; struct drbd_tl_epoch *b = NULL; struct drbd_request *req; - int local, remote; + int local, remote, send_oos = 0; int err = -EIO; int ret = 0; @@ -759,6 +784,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) bio_endio(bio, -ENOMEM); return 0; } + req->start_time = start_time; local = get_ldev(mdev); if (!local) { @@ -808,9 +834,9 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) drbd_al_begin_io(mdev, sector); } - remote = remote && (mdev->state.pdsk == D_UP_TO_DATE || - (mdev->state.pdsk == D_INCONSISTENT && - mdev->state.conn >= C_CONNECTED)); + remote = remote && drbd_should_do_remote(mdev->state); + send_oos = rw == WRITE && drbd_should_send_oos(mdev->state); + D_ASSERT(!(remote && send_oos)); if (!(local || remote) && !is_susp(mdev->state)) { if (__ratelimit(&drbd_ratelimit_state)) @@ -824,7 +850,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) * but there is a race between testing the bit and pointer outside the * spinlock, and grabbing the spinlock. * if we lost that race, we retry. */ - if (rw == WRITE && remote && + if (rw == WRITE && (remote || send_oos) && mdev->unused_spare_tle == NULL && test_bit(CREATE_BARRIER, &mdev->flags)) { allocate_barrier: @@ -842,18 +868,19 @@ allocate_barrier: if (is_susp(mdev->state)) { /* If we got suspended, use the retry mechanism of generic_make_request() to restart processing of this - bio. In the next call to drbd_make_request_26 + bio. In the next call to drbd_make_request we sleep in inc_ap_bio() */ ret = 1; spin_unlock_irq(&mdev->req_lock); goto fail_free_complete; } - if (remote) { - remote = (mdev->state.pdsk == D_UP_TO_DATE || - (mdev->state.pdsk == D_INCONSISTENT && - mdev->state.conn >= C_CONNECTED)); - if (!remote) + if (remote || send_oos) { + remote = drbd_should_do_remote(mdev->state); + send_oos = rw == WRITE && drbd_should_send_oos(mdev->state); + D_ASSERT(!(remote && send_oos)); + + if (!(remote || send_oos)) dev_warn(DEV, "lost connection while grabbing the req_lock!\n"); if (!(local || remote)) { dev_err(DEV, "IO ERROR: neither local nor remote disk\n"); @@ -866,7 +893,7 @@ allocate_barrier: mdev->unused_spare_tle = b; b = NULL; } - if (rw == WRITE && remote && + if (rw == WRITE && (remote || send_oos) && mdev->unused_spare_tle == NULL && test_bit(CREATE_BARRIER, &mdev->flags)) { /* someone closed the current epoch @@ -889,7 +916,7 @@ allocate_barrier: * barrier packet. To get the write ordering right, we only have to * make sure that, if this is a write request and it triggered a * barrier packet, this request is queued within the same spinlock. */ - if (remote && mdev->unused_spare_tle && + if ((remote || send_oos) && mdev->unused_spare_tle && test_and_clear_bit(CREATE_BARRIER, &mdev->flags)) { _tl_add_barrier(mdev, mdev->unused_spare_tle); mdev->unused_spare_tle = NULL; @@ -937,6 +964,34 @@ allocate_barrier: ? queue_for_net_write : queue_for_net_read); } + if (send_oos && drbd_set_out_of_sync(mdev, sector, size)) + _req_mod(req, queue_for_send_oos); + + if (remote && + mdev->net_conf->on_congestion != OC_BLOCK && mdev->agreed_pro_version >= 96) { + int congested = 0; + + if (mdev->net_conf->cong_fill && + atomic_read(&mdev->ap_in_flight) >= mdev->net_conf->cong_fill) { + dev_info(DEV, "Congestion-fill threshold reached\n"); + congested = 1; + } + + if (mdev->act_log->used >= mdev->net_conf->cong_extents) { + dev_info(DEV, "Congestion-extents threshold reached\n"); + congested = 1; + } + + if (congested) { + queue_barrier(mdev); /* last barrier, after mirrored writes */ + + if (mdev->net_conf->on_congestion == OC_PULL_AHEAD) + _drbd_set_state(_NS(mdev, conn, C_AHEAD), 0, NULL); + else /*mdev->net_conf->on_congestion == OC_DISCONNECT */ + _drbd_set_state(_NS(mdev, conn, C_DISCONNECTING), 0, NULL); + } + } + spin_unlock_irq(&mdev->req_lock); kfree(b); /* if someone else has beaten us to it... */ @@ -949,9 +1004,9 @@ allocate_barrier: * stable storage, and this is a WRITE, we may not even submit * this bio. */ if (get_ldev(mdev)) { - if (FAULT_ACTIVE(mdev, rw == WRITE ? DRBD_FAULT_DT_WR - : rw == READ ? DRBD_FAULT_DT_RD - : DRBD_FAULT_DT_RA)) + if (drbd_insert_fault(mdev, rw == WRITE ? DRBD_FAULT_DT_WR + : rw == READ ? DRBD_FAULT_DT_RD + : DRBD_FAULT_DT_RA)) bio_endio(req->private_bio, -EIO); else generic_make_request(req->private_bio); @@ -960,10 +1015,6 @@ allocate_barrier: bio_endio(req->private_bio, -EIO); } - /* we need to plug ALWAYS since we possibly need to kick lo_dev. - * we plug after submit, so we won't miss an unplug event */ - drbd_plug_device(mdev); - return 0; fail_conflicting: @@ -982,7 +1033,7 @@ fail_conflicting: err = 0; fail_free_complete: - if (rw == WRITE && local) + if (req->rq_state & RQ_IN_ACT_LOG) drbd_al_complete_io(mdev, sector); fail_and_free_req: if (local) { @@ -1022,16 +1073,19 @@ static int drbd_fail_request_early(struct drbd_conf *mdev, int is_write) return 0; } -int drbd_make_request_26(struct request_queue *q, struct bio *bio) +int drbd_make_request(struct request_queue *q, struct bio *bio) { unsigned int s_enr, e_enr; struct drbd_conf *mdev = (struct drbd_conf *) q->queuedata; + unsigned long start_time; if (drbd_fail_request_early(mdev, bio_data_dir(bio) & WRITE)) { bio_endio(bio, -EPERM); return 0; } + start_time = jiffies; + /* * what we "blindly" assume: */ @@ -1046,12 +1100,12 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) if (likely(s_enr == e_enr)) { inc_ap_bio(mdev, 1); - return drbd_make_request_common(mdev, bio); + return drbd_make_request_common(mdev, bio, start_time); } /* can this bio be split generically? * Maybe add our own split-arbitrary-bios function. */ - if (bio->bi_vcnt != 1 || bio->bi_idx != 0 || bio->bi_size > DRBD_MAX_SEGMENT_SIZE) { + if (bio->bi_vcnt != 1 || bio->bi_idx != 0 || bio->bi_size > DRBD_MAX_BIO_SIZE) { /* rather error out here than BUG in bio_split */ dev_err(DEV, "bio would need to, but cannot, be split: " "(vcnt=%u,idx=%u,size=%u,sector=%llu)\n", @@ -1073,11 +1127,7 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) const int sps = 1 << HT_SHIFT; /* sectors per slot */ const int mask = sps - 1; const sector_t first_sectors = sps - (sect & mask); - bp = bio_split(bio, -#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) - bio_split_pool, -#endif - first_sectors); + bp = bio_split(bio, first_sectors); /* we need to get a "reference count" (ap_bio_cnt) * to avoid races with the disconnect/reconnect/suspend code. @@ -1088,10 +1138,10 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) D_ASSERT(e_enr == s_enr + 1); - while (drbd_make_request_common(mdev, &bp->bio1)) + while (drbd_make_request_common(mdev, &bp->bio1, start_time)) inc_ap_bio(mdev, 1); - while (drbd_make_request_common(mdev, &bp->bio2)) + while (drbd_make_request_common(mdev, &bp->bio2, start_time)) inc_ap_bio(mdev, 1); dec_ap_bio(mdev); @@ -1102,7 +1152,7 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) } /* This is called by bio_add_page(). With this function we reduce - * the number of BIOs that span over multiple DRBD_MAX_SEGMENT_SIZEs + * the number of BIOs that span over multiple DRBD_MAX_BIO_SIZEs * units (was AL_EXTENTs). * * we do the calculation within the lower 32bit of the byte offsets, @@ -1112,7 +1162,7 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) * As long as the BIO is empty we have to allow at least one bvec, * regardless of size and offset. so the resulting bio may still * cross extent boundaries. those are dealt with (bio_split) in - * drbd_make_request_26. + * drbd_make_request. */ int drbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct bio_vec *bvec) { @@ -1122,8 +1172,8 @@ int drbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct unsigned int bio_size = bvm->bi_size; int limit, backing_limit; - limit = DRBD_MAX_SEGMENT_SIZE - - ((bio_offset & (DRBD_MAX_SEGMENT_SIZE-1)) + bio_size); + limit = DRBD_MAX_BIO_SIZE + - ((bio_offset & (DRBD_MAX_BIO_SIZE-1)) + bio_size); if (limit < 0) limit = 0; if (bio_size == 0) { @@ -1140,3 +1190,42 @@ int drbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct } return limit; } + +void request_timer_fn(unsigned long data) +{ + struct drbd_conf *mdev = (struct drbd_conf *) data; + struct drbd_request *req; /* oldest request */ + struct list_head *le; + unsigned long et = 0; /* effective timeout = ko_count * timeout */ + + if (get_net_conf(mdev)) { + et = mdev->net_conf->timeout*HZ/10 * mdev->net_conf->ko_count; + put_net_conf(mdev); + } + if (!et || mdev->state.conn < C_WF_REPORT_PARAMS) + return; /* Recurring timer stopped */ + + spin_lock_irq(&mdev->req_lock); + le = &mdev->oldest_tle->requests; + if (list_empty(le)) { + spin_unlock_irq(&mdev->req_lock); + mod_timer(&mdev->request_timer, jiffies + et); + return; + } + + le = le->prev; + req = list_entry(le, struct drbd_request, tl_requests); + if (time_is_before_eq_jiffies(req->start_time + et)) { + if (req->rq_state & RQ_NET_PENDING) { + dev_warn(DEV, "Remote failed to finish a request within ko-count * timeout\n"); + _drbd_set_state(_NS(mdev, conn, C_TIMEOUT), CS_VERBOSE, NULL); + } else { + dev_warn(DEV, "Local backing block device frozen?\n"); + mod_timer(&mdev->request_timer, jiffies + et); + } + } else { + mod_timer(&mdev->request_timer, req->start_time + et); + } + + spin_unlock_irq(&mdev->req_lock); +} diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h index ab2bd09..68a234a 100644 --- a/drivers/block/drbd/drbd_req.h +++ b/drivers/block/drbd/drbd_req.h @@ -82,14 +82,16 @@ enum drbd_req_event { to_be_submitted, /* XXX yes, now I am inconsistent... - * these two are not "events" but "actions" + * these are not "events" but "actions" * oh, well... */ queue_for_net_write, queue_for_net_read, + queue_for_send_oos, send_canceled, send_failed, handed_over_to_network, + oos_handed_to_network, connection_lost_while_pending, read_retry_remote_canceled, recv_acked_by_peer, @@ -254,7 +256,7 @@ static inline struct drbd_request *_ar_id_to_req(struct drbd_conf *mdev, struct hlist_node *n; struct drbd_request *req; - hlist_for_each_entry(req, n, slot, colision) { + hlist_for_each_entry(req, n, slot, collision) { if ((unsigned long)req == (unsigned long)id) { D_ASSERT(req->sector == sector); return req; @@ -289,8 +291,7 @@ static inline struct drbd_request *drbd_req_new(struct drbd_conf *mdev, req->epoch = 0; req->sector = bio_src->bi_sector; req->size = bio_src->bi_size; - req->start_time = jiffies; - INIT_HLIST_NODE(&req->colision); + INIT_HLIST_NODE(&req->collision); INIT_LIST_HEAD(&req->tl_requests); INIT_LIST_HEAD(&req->w.list); } @@ -321,6 +322,8 @@ extern int __req_mod(struct drbd_request *req, enum drbd_req_event what, struct bio_and_error *m); extern void complete_master_bio(struct drbd_conf *mdev, struct bio_and_error *m); +extern void request_timer_fn(unsigned long data); +extern void tl_restart(struct drbd_conf *mdev, enum drbd_req_event what); /* use this if you don't want to deal with calling complete_master_bio() * outside the spinlock, e.g. when walking some list on cleanup. */ @@ -338,23 +341,43 @@ static inline int _req_mod(struct drbd_request *req, enum drbd_req_event what) return rv; } -/* completion of master bio is outside of spinlock. - * If you need it irqsave, do it your self! - * Which means: don't use from bio endio callback. */ +/* completion of master bio is outside of our spinlock. + * We still may or may not be inside some irqs disabled section + * of the lower level driver completion callback, so we need to + * spin_lock_irqsave here. */ static inline int req_mod(struct drbd_request *req, enum drbd_req_event what) { + unsigned long flags; struct drbd_conf *mdev = req->mdev; struct bio_and_error m; int rv; - spin_lock_irq(&mdev->req_lock); + spin_lock_irqsave(&mdev->req_lock, flags); rv = __req_mod(req, what, &m); - spin_unlock_irq(&mdev->req_lock); + spin_unlock_irqrestore(&mdev->req_lock, flags); if (m.bio) complete_master_bio(mdev, &m); return rv; } + +static inline bool drbd_should_do_remote(union drbd_state s) +{ + return s.pdsk == D_UP_TO_DATE || + (s.pdsk >= D_INCONSISTENT && + s.conn >= C_WF_BITMAP_T && + s.conn < C_AHEAD); + /* Before proto 96 that was >= CONNECTED instead of >= C_WF_BITMAP_T. + That is equivalent since before 96 IO was frozen in the C_WF_BITMAP* + states. */ +} +static inline bool drbd_should_send_oos(union drbd_state s) +{ + return s.conn == C_AHEAD || s.conn == C_WF_BITMAP_S; + /* pdsk = D_INCONSISTENT as a consequence. Protocol 96 check not necessary + since we enter state C_AHEAD only if proto >= 96 */ +} + #endif diff --git a/drivers/block/drbd/drbd_strings.c b/drivers/block/drbd/drbd_strings.c index 85179e1..c44a2a6 100644 --- a/drivers/block/drbd/drbd_strings.c +++ b/drivers/block/drbd/drbd_strings.c @@ -48,6 +48,8 @@ static const char *drbd_conn_s_names[] = { [C_PAUSED_SYNC_T] = "PausedSyncT", [C_VERIFY_S] = "VerifyS", [C_VERIFY_T] = "VerifyT", + [C_AHEAD] = "Ahead", + [C_BEHIND] = "Behind", }; static const char *drbd_role_s_names[] = { @@ -92,7 +94,7 @@ static const char *drbd_state_sw_errors[] = { const char *drbd_conn_str(enum drbd_conns s) { /* enums are unsigned... */ - return s > C_PAUSED_SYNC_T ? "TOO_LARGE" : drbd_conn_s_names[s]; + return s > C_BEHIND ? "TOO_LARGE" : drbd_conn_s_names[s]; } const char *drbd_role_str(enum drbd_role s) @@ -105,7 +107,7 @@ const char *drbd_disk_str(enum drbd_disk_state s) return s > D_UP_TO_DATE ? "TOO_LARGE" : drbd_disk_s_names[s]; } -const char *drbd_set_st_err_str(enum drbd_state_ret_codes err) +const char *drbd_set_st_err_str(enum drbd_state_rv err) { return err <= SS_AFTER_LAST_ERROR ? "TOO_SMALL" : err > SS_TWO_PRIMARIES ? "TOO_LARGE" diff --git a/drivers/block/drbd/drbd_vli.h b/drivers/block/drbd/drbd_vli.h index fc82400..8cb1532 100644 --- a/drivers/block/drbd/drbd_vli.h +++ b/drivers/block/drbd/drbd_vli.h @@ -32,7 +32,7 @@ * the bitmap transfer time can take much too long, * if transmitted in plain text. * - * We try to reduce the transfered bitmap information + * We try to reduce the transferred bitmap information * by encoding runlengths of bit polarity. * * We never actually need to encode a "zero" (runlengths are positive). diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 34f224b..4d76b06 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -39,18 +39,17 @@ #include "drbd_req.h" static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel); +static int w_make_resync_request(struct drbd_conf *mdev, + struct drbd_work *w, int cancel); -/* defined here: - drbd_md_io_complete - drbd_endio_sec - drbd_endio_pri - - * more endio handlers: - atodb_endio in drbd_actlog.c - drbd_bm_async_io_complete in drbd_bitmap.c - +/* endio handlers: + * drbd_md_io_complete (defined here) + * drbd_endio_pri (defined here) + * drbd_endio_sec (defined here) + * bm_async_io_complete (defined in drbd_bitmap.c) + * * For all these callbacks, note the following: * The callbacks will be called in irq context by the IDE drivers, * and in Softirqs/Tasklets/BH context by the SCSI drivers. @@ -94,7 +93,7 @@ void drbd_endio_read_sec_final(struct drbd_epoch_entry *e) __releases(local) if (list_empty(&mdev->read_ee)) wake_up(&mdev->ee_wait); if (test_bit(__EE_WAS_ERROR, &e->flags)) - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); spin_unlock_irqrestore(&mdev->req_lock, flags); drbd_queue_work(&mdev->data.work, &e->w); @@ -127,7 +126,7 @@ static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(lo list_del(&e->w.list); /* has been on active_ee or sync_ee */ list_add_tail(&e->w.list, &mdev->done_ee); - /* No hlist_del_init(&e->colision) here, we did not send the Ack yet, + /* No hlist_del_init(&e->collision) here, we did not send the Ack yet, * neither did we wake possibly waiting conflicting requests. * done from "drbd_process_done_ee" within the appropriate w.cb * (e_end_block/e_end_resync_block) or from _drbd_clear_done_ee */ @@ -137,7 +136,7 @@ static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(lo : list_empty(&mdev->active_ee); if (test_bit(__EE_WAS_ERROR, &e->flags)) - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); spin_unlock_irqrestore(&mdev->req_lock, flags); if (is_syncer_req) @@ -163,14 +162,15 @@ void drbd_endio_sec(struct bio *bio, int error) int uptodate = bio_flagged(bio, BIO_UPTODATE); int is_write = bio_data_dir(bio) == WRITE; - if (error) + if (error && __ratelimit(&drbd_ratelimit_state)) dev_warn(DEV, "%s: error=%d s=%llus\n", is_write ? "write" : "read", error, (unsigned long long)e->sector); if (!error && !uptodate) { - dev_warn(DEV, "%s: setting error to -EIO s=%llus\n", - is_write ? "write" : "read", - (unsigned long long)e->sector); + if (__ratelimit(&drbd_ratelimit_state)) + dev_warn(DEV, "%s: setting error to -EIO s=%llus\n", + is_write ? "write" : "read", + (unsigned long long)e->sector); /* strange behavior of some lower level drivers... * fail the request by clearing the uptodate flag, * but do not return any error?! */ @@ -250,13 +250,6 @@ int w_read_retry_remote(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return w_send_read_req(mdev, w, 0); } -int w_resync_inactive(struct drbd_conf *mdev, struct drbd_work *w, int cancel) -{ - ERR_IF(cancel) return 1; - dev_err(DEV, "resync inactive, but callback triggered??\n"); - return 1; /* Simply ignore this! */ -} - void drbd_csum_ee(struct drbd_conf *mdev, struct crypto_hash *tfm, struct drbd_epoch_entry *e, void *digest) { struct hash_desc desc; @@ -304,42 +297,48 @@ void drbd_csum_bio(struct drbd_conf *mdev, struct crypto_hash *tfm, struct bio * crypto_hash_final(&desc, digest); } -static int w_e_send_csum(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +/* TODO merge common code with w_e_end_ov_req */ +int w_e_send_csum(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w); int digest_size; void *digest; - int ok; + int ok = 1; D_ASSERT(e->block_id == DRBD_MAGIC + 0xbeef); - if (unlikely(cancel)) { - drbd_free_ee(mdev, e); - return 1; - } + if (unlikely(cancel)) + goto out; - if (likely((e->flags & EE_WAS_ERROR) == 0)) { - digest_size = crypto_hash_digestsize(mdev->csums_tfm); - digest = kmalloc(digest_size, GFP_NOIO); - if (digest) { - drbd_csum_ee(mdev, mdev->csums_tfm, e, digest); + if (likely((e->flags & EE_WAS_ERROR) != 0)) + goto out; - inc_rs_pending(mdev); - ok = drbd_send_drequest_csum(mdev, - e->sector, - e->size, - digest, - digest_size, - P_CSUM_RS_REQUEST); - kfree(digest); - } else { - dev_err(DEV, "kmalloc() of digest failed.\n"); - ok = 0; - } - } else - ok = 1; + digest_size = crypto_hash_digestsize(mdev->csums_tfm); + digest = kmalloc(digest_size, GFP_NOIO); + if (digest) { + sector_t sector = e->sector; + unsigned int size = e->size; + drbd_csum_ee(mdev, mdev->csums_tfm, e, digest); + /* Free e and pages before send. + * In case we block on congestion, we could otherwise run into + * some distributed deadlock, if the other side blocks on + * congestion as well, because our receiver blocks in + * drbd_pp_alloc due to pp_in_use > max_buffers. */ + drbd_free_ee(mdev, e); + e = NULL; + inc_rs_pending(mdev); + ok = drbd_send_drequest_csum(mdev, sector, size, + digest, digest_size, + P_CSUM_RS_REQUEST); + kfree(digest); + } else { + dev_err(DEV, "kmalloc() of digest failed.\n"); + ok = 0; + } - drbd_free_ee(mdev, e); +out: + if (e) + drbd_free_ee(mdev, e); if (unlikely(!ok)) dev_err(DEV, "drbd_send_drequest(..., csum) failed\n"); @@ -355,7 +354,7 @@ static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size) if (!get_ldev(mdev)) return -EIO; - if (drbd_rs_should_slow_down(mdev)) + if (drbd_rs_should_slow_down(mdev, sector)) goto defer; /* GFP_TRY, because if there is no memory available right now, this may @@ -373,9 +372,10 @@ static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size) if (drbd_submit_ee(mdev, e, READ, DRBD_FAULT_RS_RD) == 0) return 0; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* If it failed because of ENOMEM, retry should help. If it failed + * because bio_add_page failed (probably broken lower level driver), + * retry may or may not help. + * If it does not, you may need to force disconnect. */ spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); spin_unlock_irq(&mdev->req_lock); @@ -386,26 +386,25 @@ defer: return -EAGAIN; } -void resync_timer_fn(unsigned long data) +int w_resync_timer(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { - struct drbd_conf *mdev = (struct drbd_conf *) data; - int queue; - - queue = 1; switch (mdev->state.conn) { case C_VERIFY_S: - mdev->resync_work.cb = w_make_ov_request; + w_make_ov_request(mdev, w, cancel); break; case C_SYNC_TARGET: - mdev->resync_work.cb = w_make_resync_request; + w_make_resync_request(mdev, w, cancel); break; - default: - queue = 0; - mdev->resync_work.cb = w_resync_inactive; } - /* harmless race: list_empty outside data.work.q_lock */ - if (list_empty(&mdev->resync_work.list) && queue) + return 1; +} + +void resync_timer_fn(unsigned long data) +{ + struct drbd_conf *mdev = (struct drbd_conf *) data; + + if (list_empty(&mdev->resync_work.list)) drbd_queue_work(&mdev->data.work, &mdev->resync_work); } @@ -438,7 +437,7 @@ static void fifo_add_val(struct fifo_buffer *fb, int value) fb->values[i] += value; } -int drbd_rs_controller(struct drbd_conf *mdev) +static int drbd_rs_controller(struct drbd_conf *mdev) { unsigned int sect_in; /* Number of sectors that came in since the last turn */ unsigned int want; /* The number of sectors we want in the proxy */ @@ -492,29 +491,36 @@ int drbd_rs_controller(struct drbd_conf *mdev) return req_sect; } -int w_make_resync_request(struct drbd_conf *mdev, - struct drbd_work *w, int cancel) +static int drbd_rs_number_requests(struct drbd_conf *mdev) +{ + int number; + if (mdev->rs_plan_s.size) { /* mdev->sync_conf.c_plan_ahead */ + number = drbd_rs_controller(mdev) >> (BM_BLOCK_SHIFT - 9); + mdev->c_sync_rate = number * HZ * (BM_BLOCK_SIZE / 1024) / SLEEP_TIME; + } else { + mdev->c_sync_rate = mdev->sync_conf.rate; + number = SLEEP_TIME * mdev->c_sync_rate / ((BM_BLOCK_SIZE / 1024) * HZ); + } + + /* ignore the amount of pending requests, the resync controller should + * throttle down to incoming reply rate soon enough anyways. */ + return number; +} + +static int w_make_resync_request(struct drbd_conf *mdev, + struct drbd_work *w, int cancel) { unsigned long bit; sector_t sector; const sector_t capacity = drbd_get_capacity(mdev->this_bdev); - int max_segment_size; - int number, rollback_i, size, pe, mx; + int max_bio_size; + int number, rollback_i, size; int align, queued, sndbuf; int i = 0; if (unlikely(cancel)) return 1; - if (unlikely(mdev->state.conn < C_CONNECTED)) { - dev_err(DEV, "Confused in w_make_resync_request()! cstate < Connected"); - return 0; - } - - if (mdev->state.conn != C_SYNC_TARGET) - dev_err(DEV, "%s in w_make_resync_request\n", - drbd_conn_str(mdev->state.conn)); - if (mdev->rs_total == 0) { /* empty resync? */ drbd_resync_finished(mdev); @@ -527,49 +533,19 @@ int w_make_resync_request(struct drbd_conf *mdev, to continue resync with a broken disk makes no sense at all */ dev_err(DEV, "Disk broke down during resync!\n"); - mdev->resync_work.cb = w_resync_inactive; return 1; } /* starting with drbd 8.3.8, we can handle multi-bio EEs, * if it should be necessary */ - max_segment_size = - mdev->agreed_pro_version < 94 ? queue_max_segment_size(mdev->rq_queue) : - mdev->agreed_pro_version < 95 ? DRBD_MAX_SIZE_H80_PACKET : DRBD_MAX_SEGMENT_SIZE; - - if (mdev->rs_plan_s.size) { /* mdev->sync_conf.c_plan_ahead */ - number = drbd_rs_controller(mdev) >> (BM_BLOCK_SHIFT - 9); - mdev->c_sync_rate = number * HZ * (BM_BLOCK_SIZE / 1024) / SLEEP_TIME; - } else { - mdev->c_sync_rate = mdev->sync_conf.rate; - number = SLEEP_TIME * mdev->c_sync_rate / ((BM_BLOCK_SIZE / 1024) * HZ); - } + max_bio_size = + mdev->agreed_pro_version < 94 ? queue_max_hw_sectors(mdev->rq_queue) << 9 : + mdev->agreed_pro_version < 95 ? DRBD_MAX_SIZE_H80_PACKET : DRBD_MAX_BIO_SIZE; - /* Throttle resync on lower level disk activity, which may also be - * caused by application IO on Primary/SyncTarget. - * Keep this after the call to drbd_rs_controller, as that assumes - * to be called as precisely as possible every SLEEP_TIME, - * and would be confused otherwise. */ - if (drbd_rs_should_slow_down(mdev)) + number = drbd_rs_number_requests(mdev); + if (number == 0) goto requeue; - mutex_lock(&mdev->data.mutex); - if (mdev->data.socket) - mx = mdev->data.socket->sk->sk_rcvbuf / sizeof(struct p_block_req); - else - mx = 1; - mutex_unlock(&mdev->data.mutex); - - /* For resync rates >160MB/sec, allow more pending RS requests */ - if (number > mx) - mx = number; - - /* Limit the number of pending RS requests to no more than the peer's receive buffer */ - pe = atomic_read(&mdev->rs_pending_cnt); - if ((pe + number) > mx) { - number = mx - pe; - } - for (i = 0; i < number; i++) { /* Stop generating RS requests, when half of the send buffer is filled */ mutex_lock(&mdev->data.mutex); @@ -588,16 +564,16 @@ next_sector: size = BM_BLOCK_SIZE; bit = drbd_bm_find_next(mdev, mdev->bm_resync_fo); - if (bit == -1UL) { + if (bit == DRBD_END_OF_BITMAP) { mdev->bm_resync_fo = drbd_bm_bits(mdev); - mdev->resync_work.cb = w_resync_inactive; put_ldev(mdev); return 1; } sector = BM_BIT_TO_SECT(bit); - if (drbd_try_rs_begin_io(mdev, sector)) { + if (drbd_rs_should_slow_down(mdev, sector) || + drbd_try_rs_begin_io(mdev, sector)) { mdev->bm_resync_fo = bit; goto requeue; } @@ -608,7 +584,7 @@ next_sector: goto next_sector; } -#if DRBD_MAX_SEGMENT_SIZE > BM_BLOCK_SIZE +#if DRBD_MAX_BIO_SIZE > BM_BLOCK_SIZE /* try to find some adjacent bits. * we stop if we have already the maximum req size. * @@ -618,7 +594,7 @@ next_sector: align = 1; rollback_i = i; for (;;) { - if (size + BM_BLOCK_SIZE > max_segment_size) + if (size + BM_BLOCK_SIZE > max_bio_size) break; /* Be always aligned */ @@ -685,7 +661,6 @@ next_sector: * resync data block, and the last bit is cleared. * until then resync "work" is "inactive" ... */ - mdev->resync_work.cb = w_resync_inactive; put_ldev(mdev); return 1; } @@ -706,27 +681,18 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca if (unlikely(cancel)) return 1; - if (unlikely(mdev->state.conn < C_CONNECTED)) { - dev_err(DEV, "Confused in w_make_ov_request()! cstate < Connected"); - return 0; - } - - number = SLEEP_TIME*mdev->sync_conf.rate / ((BM_BLOCK_SIZE/1024)*HZ); - if (atomic_read(&mdev->rs_pending_cnt) > number) - goto requeue; - - number -= atomic_read(&mdev->rs_pending_cnt); + number = drbd_rs_number_requests(mdev); sector = mdev->ov_position; for (i = 0; i < number; i++) { if (sector >= capacity) { - mdev->resync_work.cb = w_resync_inactive; return 1; } size = BM_BLOCK_SIZE; - if (drbd_try_rs_begin_io(mdev, sector)) { + if (drbd_rs_should_slow_down(mdev, sector) || + drbd_try_rs_begin_io(mdev, sector)) { mdev->ov_position = sector; goto requeue; } @@ -744,11 +710,33 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca mdev->ov_position = sector; requeue: + mdev->rs_in_flight += (i << (BM_BLOCK_SHIFT - 9)); mod_timer(&mdev->resync_timer, jiffies + SLEEP_TIME); return 1; } +void start_resync_timer_fn(unsigned long data) +{ + struct drbd_conf *mdev = (struct drbd_conf *) data; + + drbd_queue_work(&mdev->data.work, &mdev->start_resync_work); +} + +int w_start_resync(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +{ + if (atomic_read(&mdev->unacked_cnt) || atomic_read(&mdev->rs_pending_cnt)) { + dev_warn(DEV, "w_start_resync later...\n"); + mdev->start_resync_timer.expires = jiffies + HZ/10; + add_timer(&mdev->start_resync_timer); + return 1; + } + + drbd_start_resync(mdev, C_SYNC_SOURCE); + clear_bit(AHEAD_TO_SYNC_SOURCE, &mdev->current_epoch->flags); + return 1; +} + int w_ov_finished(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { kfree(w); @@ -782,6 +770,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) union drbd_state os, ns; struct drbd_work *w; char *khelper_cmd = NULL; + int verify_done = 0; /* Remove all elements from the resync LRU. Since future actions * might set bits in the (main) bitmap, then the entries in the @@ -792,9 +781,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) * queue (or even the read operations for those packets * is not finished by now). Retry in 100ms. */ - drbd_kick_lo(mdev); - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ / 10); + schedule_timeout_interruptible(HZ / 10); w = kmalloc(sizeof(struct drbd_work), GFP_ATOMIC); if (w) { w->cb = w_resync_finished; @@ -819,6 +806,8 @@ int drbd_resync_finished(struct drbd_conf *mdev) spin_lock_irq(&mdev->req_lock); os = mdev->state; + verify_done = (os.conn == C_VERIFY_S || os.conn == C_VERIFY_T); + /* This protects us against multiple calls (that can happen in the presence of application IO), and against connectivity loss just before we arrive here. */ if (os.conn <= C_CONNECTED) @@ -828,8 +817,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) ns.conn = C_CONNECTED; dev_info(DEV, "%s done (total %lu sec; paused %lu sec; %lu K/sec)\n", - (os.conn == C_VERIFY_S || os.conn == C_VERIFY_T) ? - "Online verify " : "Resync", + verify_done ? "Online verify " : "Resync", dt + mdev->rs_paused, mdev->rs_paused, dbdt); n_oos = drbd_bm_total_weight(mdev); @@ -852,7 +840,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) const int ratio = (t == 0) ? 0 : (t < 100000) ? ((s*100)/t) : (s/(t/100)); - dev_info(DEV, "%u %% had equal check sums, eliminated: %luK; " + dev_info(DEV, "%u %% had equal checksums, eliminated: %luK; " "transferred %luK total %luK\n", ratio, Bit2KB(mdev->rs_same_csum), @@ -887,14 +875,18 @@ int drbd_resync_finished(struct drbd_conf *mdev) } } - drbd_uuid_set_bm(mdev, 0UL); - - if (mdev->p_uuid) { - /* Now the two UUID sets are equal, update what we - * know of the peer. */ - int i; - for (i = UI_CURRENT ; i <= UI_HISTORY_END ; i++) - mdev->p_uuid[i] = mdev->ldev->md.uuid[i]; + if (!(os.conn == C_VERIFY_S || os.conn == C_VERIFY_T)) { + /* for verify runs, we don't update uuids here, + * so there would be nothing to report. */ + drbd_uuid_set_bm(mdev, 0UL); + drbd_print_uuids(mdev, "updated UUIDs"); + if (mdev->p_uuid) { + /* Now the two UUID sets are equal, update what we + * know of the peer. */ + int i; + for (i = UI_CURRENT ; i <= UI_HISTORY_END ; i++) + mdev->p_uuid[i] = mdev->ldev->md.uuid[i]; + } } } @@ -906,15 +898,11 @@ out: mdev->rs_total = 0; mdev->rs_failed = 0; mdev->rs_paused = 0; - mdev->ov_start_sector = 0; + if (verify_done) + mdev->ov_start_sector = 0; drbd_md_sync(mdev); - if (test_and_clear_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags)) { - dev_info(DEV, "Writing the whole bitmap\n"); - drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished"); - } - if (khelper_cmd) drbd_khelper(mdev, khelper_cmd); @@ -995,7 +983,9 @@ int w_e_end_rsdata_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) put_ldev(mdev); } - if (likely((e->flags & EE_WAS_ERROR) == 0)) { + if (mdev->state.conn == C_AHEAD) { + ok = drbd_send_ack(mdev, P_RS_CANCEL, e); + } else if (likely((e->flags & EE_WAS_ERROR) == 0)) { if (likely(mdev->state.pdsk >= D_INCONSISTENT)) { inc_rs_pending(mdev); ok = drbd_send_block(mdev, P_RS_DATA_REPLY, e); @@ -1087,9 +1077,12 @@ int w_e_end_csum_rs_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return ok; } +/* TODO merge common code with w_e_send_csum */ int w_e_end_ov_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w); + sector_t sector = e->sector; + unsigned int size = e->size; int digest_size; void *digest; int ok = 1; @@ -1097,27 +1090,37 @@ int w_e_end_ov_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) if (unlikely(cancel)) goto out; - if (unlikely((e->flags & EE_WAS_ERROR) != 0)) - goto out; - digest_size = crypto_hash_digestsize(mdev->verify_tfm); - /* FIXME if this allocation fails, online verify will not terminate! */ digest = kmalloc(digest_size, GFP_NOIO); - if (digest) { - drbd_csum_ee(mdev, mdev->verify_tfm, e, digest); - inc_rs_pending(mdev); - ok = drbd_send_drequest_csum(mdev, e->sector, e->size, - digest, digest_size, P_OV_REPLY); - if (!ok) - dec_rs_pending(mdev); - kfree(digest); + if (!digest) { + ok = 0; /* terminate the connection in case the allocation failed */ + goto out; } -out: + if (likely(!(e->flags & EE_WAS_ERROR))) + drbd_csum_ee(mdev, mdev->verify_tfm, e, digest); + else + memset(digest, 0, digest_size); + + /* Free e and pages before send. + * In case we block on congestion, we could otherwise run into + * some distributed deadlock, if the other side blocks on + * congestion as well, because our receiver blocks in + * drbd_pp_alloc due to pp_in_use > max_buffers. */ drbd_free_ee(mdev, e); + e = NULL; + inc_rs_pending(mdev); + ok = drbd_send_drequest_csum(mdev, sector, size, + digest, digest_size, + P_OV_REPLY); + if (!ok) + dec_rs_pending(mdev); + kfree(digest); +out: + if (e) + drbd_free_ee(mdev, e); dec_unacked(mdev); - return ok; } @@ -1130,15 +1133,16 @@ void drbd_ov_oos_found(struct drbd_conf *mdev, sector_t sector, int size) mdev->ov_last_oos_size = size>>9; } drbd_set_out_of_sync(mdev, sector, size); - set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags); } int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { struct drbd_epoch_entry *e = container_of(w, struct drbd_epoch_entry, w); struct digest_info *di; - int digest_size; void *digest; + sector_t sector = e->sector; + unsigned int size = e->size; + int digest_size; int ok, eq = 0; if (unlikely(cancel)) { @@ -1166,24 +1170,31 @@ int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) eq = !memcmp(digest, di->digest, digest_size); kfree(digest); } - } else { - ok = drbd_send_ack(mdev, P_NEG_RS_DREPLY, e); - if (__ratelimit(&drbd_ratelimit_state)) - dev_err(DEV, "Sending NegDReply. I guess it gets messy.\n"); } - dec_unacked(mdev); + /* Free e and pages before send. + * In case we block on congestion, we could otherwise run into + * some distributed deadlock, if the other side blocks on + * congestion as well, because our receiver blocks in + * drbd_pp_alloc due to pp_in_use > max_buffers. */ + drbd_free_ee(mdev, e); if (!eq) - drbd_ov_oos_found(mdev, e->sector, e->size); + drbd_ov_oos_found(mdev, sector, size); else ov_oos_print(mdev); - ok = drbd_send_ack_ex(mdev, P_OV_RESULT, e->sector, e->size, + ok = drbd_send_ack_ex(mdev, P_OV_RESULT, sector, size, eq ? ID_IN_SYNC : ID_OUT_OF_SYNC); - drbd_free_ee(mdev, e); + dec_unacked(mdev); + + --mdev->ov_left; - if (--mdev->ov_left == 0) { + /* let's advance progress step marks only for every other megabyte */ + if ((mdev->ov_left & 0x200) == 0x200) + drbd_advance_rs_marks(mdev, mdev->ov_left); + + if (mdev->ov_left == 0) { ov_oos_print(mdev); drbd_resync_finished(mdev); } @@ -1236,6 +1247,22 @@ int w_send_write_hint(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return drbd_send_short_cmd(mdev, P_UNPLUG_REMOTE); } +int w_send_oos(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +{ + struct drbd_request *req = container_of(w, struct drbd_request, w); + int ok; + + if (unlikely(cancel)) { + req_mod(req, send_canceled); + return 1; + } + + ok = drbd_send_oos(mdev, req); + req_mod(req, oos_handed_to_network); + + return ok; +} + /** * w_send_dblock() - Worker callback to send a P_DATA packet in order to mirror a write request * @mdev: DRBD device. @@ -1431,6 +1458,17 @@ int drbd_alter_sa(struct drbd_conf *mdev, int na) return retcode; } +void drbd_rs_controller_reset(struct drbd_conf *mdev) +{ + atomic_set(&mdev->rs_sect_in, 0); + atomic_set(&mdev->rs_sect_ev, 0); + mdev->rs_in_flight = 0; + mdev->rs_planed = 0; + spin_lock(&mdev->peer_seq_lock); + fifo_set(&mdev->rs_plan_s, 0); + spin_unlock(&mdev->peer_seq_lock); +} + /** * drbd_start_resync() - Start the resync process * @mdev: DRBD device. @@ -1444,13 +1482,18 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) union drbd_state ns; int r; - if (mdev->state.conn >= C_SYNC_SOURCE) { + if (mdev->state.conn >= C_SYNC_SOURCE && mdev->state.conn < C_AHEAD) { dev_err(DEV, "Resync already running!\n"); return; } - /* In case a previous resync run was aborted by an IO error/detach on the peer. */ - drbd_rs_cancel_all(mdev); + if (mdev->state.conn < C_AHEAD) { + /* In case a previous resync run was aborted by an IO error/detach on the peer. */ + drbd_rs_cancel_all(mdev); + /* This should be done when we abort the resync. We definitely do not + want to have this for connections going back and forth between + Ahead/Behind and SyncSource/SyncTarget */ + } if (side == C_SYNC_TARGET) { /* Since application IO was locked out during C_WF_BITMAP_T and @@ -1464,6 +1507,20 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); return; } + } else /* C_SYNC_SOURCE */ { + r = drbd_khelper(mdev, "before-resync-source"); + r = (r >> 8) & 0xff; + if (r > 0) { + if (r == 3) { + dev_info(DEV, "before-resync-source handler returned %d, " + "ignoring. Old userland tools?", r); + } else { + dev_info(DEV, "before-resync-source handler returned %d, " + "dropping connection.\n", r); + drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); + return; + } + } } drbd_state_lock(mdev); @@ -1473,18 +1530,6 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) return; } - if (side == C_SYNC_TARGET) { - mdev->bm_resync_fo = 0; - } else /* side == C_SYNC_SOURCE */ { - u64 uuid; - - get_random_bytes(&uuid, sizeof(u64)); - drbd_uuid_set(mdev, UI_BITMAP, uuid); - drbd_send_sync_uuid(mdev, uuid); - - D_ASSERT(mdev->state.disk == D_UP_TO_DATE); - } - write_lock_irq(&global_state_lock); ns = mdev->state; @@ -1522,13 +1567,24 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) _drbd_pause_after(mdev); } write_unlock_irq(&global_state_lock); - put_ldev(mdev); if (r == SS_SUCCESS) { dev_info(DEV, "Began resync as %s (will sync %lu KB [%lu bits set]).\n", drbd_conn_str(ns.conn), (unsigned long) mdev->rs_total << (BM_BLOCK_SHIFT-10), (unsigned long) mdev->rs_total); + if (side == C_SYNC_TARGET) + mdev->bm_resync_fo = 0; + + /* Since protocol 96, we must serialize drbd_gen_and_send_sync_uuid + * with w_send_oos, or the sync target will get confused as to + * how much bits to resync. We cannot do that always, because for an + * empty resync and protocol < 95, we need to do it here, as we call + * drbd_resync_finished from here in that case. + * We drbd_gen_and_send_sync_uuid here for protocol < 96, + * and from after_state_ch otherwise. */ + if (side == C_SYNC_SOURCE && mdev->agreed_pro_version < 96) + drbd_gen_and_send_sync_uuid(mdev); if (mdev->agreed_pro_version < 95 && mdev->rs_total == 0) { /* This still has a race (about when exactly the peers @@ -1548,13 +1604,7 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) drbd_resync_finished(mdev); } - atomic_set(&mdev->rs_sect_in, 0); - atomic_set(&mdev->rs_sect_ev, 0); - mdev->rs_in_flight = 0; - mdev->rs_planed = 0; - spin_lock(&mdev->peer_seq_lock); - fifo_set(&mdev->rs_plan_s, 0); - spin_unlock(&mdev->peer_seq_lock); + drbd_rs_controller_reset(mdev); /* ns.conn may already be != mdev->state.conn, * we may have been paused in between, or become paused until * the timer triggers. @@ -1564,6 +1614,7 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) drbd_md_sync(mdev); } + put_ldev(mdev); drbd_state_unlock(mdev); } diff --git a/drivers/block/drbd/drbd_wrappers.h b/drivers/block/drbd/drbd_wrappers.h index defdb50..151f1a3 100644 --- a/drivers/block/drbd/drbd_wrappers.h +++ b/drivers/block/drbd/drbd_wrappers.h @@ -39,30 +39,12 @@ static inline void drbd_generic_make_request(struct drbd_conf *mdev, return; } - if (FAULT_ACTIVE(mdev, fault_type)) + if (drbd_insert_fault(mdev, fault_type)) bio_endio(bio, -EIO); else generic_make_request(bio); } -static inline void drbd_plug_device(struct drbd_conf *mdev) -{ - struct request_queue *q; - q = bdev_get_queue(mdev->this_bdev); - - spin_lock_irq(q->queue_lock); - -/* XXX the check on !blk_queue_plugged is redundant, - * implicitly checked in blk_plug_device */ - - if (!blk_queue_plugged(q)) { - blk_plug_device(q); - del_timer(&q->unplug_timer); - /* unplugging should not happen automatically... */ - } - spin_unlock_irq(q->queue_lock); -} - static inline int drbd_crypto_is_hash(struct crypto_tfm *tfm) { return (crypto_tfm_alg_type(tfm) & CRYPTO_ALG_TYPE_HASH_MASK) diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 20aea9b..98de8f4 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -3771,13 +3771,14 @@ out2: /* * Check if the disk has been changed or if a change has been faked. */ -static int check_floppy_change(struct gendisk *disk) +static unsigned int floppy_check_events(struct gendisk *disk, + unsigned int clearing) { int drive = (long)disk->private_data; if (test_bit(FD_DISK_CHANGED_BIT, &UDRS->flags) || test_bit(FD_VERIFY_BIT, &UDRS->flags)) - return 1; + return DISK_EVENT_MEDIA_CHANGE; if (time_after(jiffies, UDRS->last_checked + UDP->checkfreq)) { lock_fdc(drive, false); @@ -3789,7 +3790,7 @@ static int check_floppy_change(struct gendisk *disk) test_bit(FD_VERIFY_BIT, &UDRS->flags) || test_bit(drive, &fake_change) || drive_no_geom(drive)) - return 1; + return DISK_EVENT_MEDIA_CHANGE; return 0; } @@ -3838,7 +3839,6 @@ static int __floppy_read_block_0(struct block_device *bdev) bio.bi_end_io = floppy_rb0_complete; submit_bio(READ, &bio); - generic_unplug_device(bdev_get_queue(bdev)); process_fd_request(); wait_for_completion(&complete); @@ -3899,7 +3899,7 @@ static const struct block_device_operations floppy_fops = { .release = floppy_release, .ioctl = fd_ioctl, .getgeo = fd_getgeo, - .media_changed = check_floppy_change, + .check_events = floppy_check_events, .revalidate_disk = floppy_revalidate, }; diff --git a/drivers/block/hd.c b/drivers/block/hd.c index 30ec6b3..007c630 100644 --- a/drivers/block/hd.c +++ b/drivers/block/hd.c @@ -733,7 +733,7 @@ static int __init hd_init(void) * the BIOS or CMOS. This doesn't work all that well, * since this assumes that this is a primary or secondary * drive, and if we're using this legacy driver, it's - * probably an auxilliary controller added to recover + * probably an auxiliary controller added to recover * legacy data off an ST-506 drive. Either way, it's * definitely safest to have the user explicitly specify * the information. diff --git a/drivers/block/loop.c b/drivers/block/loop.c index dbf31ec..76c8da7 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -540,17 +540,6 @@ out: return 0; } -/* - * kick off io on the underlying address space - */ -static void loop_unplug(struct request_queue *q) -{ - struct loop_device *lo = q->queuedata; - - queue_flag_clear_unlocked(QUEUE_FLAG_PLUGGED, q); - blk_run_address_space(lo->lo_backing_file->f_mapping); -} - struct switch_request { struct file *file; struct completion wait; @@ -917,7 +906,6 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, */ blk_queue_make_request(lo->lo_queue, loop_make_request); lo->lo_queue->queuedata = lo; - lo->lo_queue->unplug_fn = loop_unplug; if (!(lo_flags & LO_FLAGS_READ_ONLY) && file->f_op->fsync) blk_queue_flush(lo->lo_queue, REQ_FLUSH); @@ -1019,7 +1007,6 @@ static int loop_clr_fd(struct loop_device *lo, struct block_device *bdev) kthread_stop(lo->lo_thread); - lo->lo_queue->unplug_fn = NULL; lo->lo_backing_file = NULL; loop_release_xfer(lo); @@ -1553,9 +1540,9 @@ static const struct block_device_operations lo_fops = { * And now the modules code and kernel interface. */ static int max_loop; -module_param(max_loop, int, 0); +module_param(max_loop, int, S_IRUGO); MODULE_PARM_DESC(max_loop, "Maximum number of loop devices"); -module_param(max_part, int, 0); +module_param(max_part, int, S_IRUGO); MODULE_PARM_DESC(max_part, "Maximum number of partitions per loop device"); MODULE_LICENSE("GPL"); MODULE_ALIAS_BLOCKDEV_MAJOR(LOOP_MAJOR); @@ -1636,9 +1623,6 @@ out: static void loop_free(struct loop_device *lo) { - if (!lo->lo_queue->queue_lock) - lo->lo_queue->queue_lock = &lo->lo_queue->__queue_lock; - blk_cleanup_queue(lo->lo_queue); put_disk(lo->lo_disk); list_del(&lo->lo_list); @@ -1674,7 +1658,7 @@ static struct kobject *loop_probe(dev_t dev, int *part, void *data) struct kobject *kobj; mutex_lock(&loop_devices_mutex); - lo = loop_init_one(dev & MINORMASK); + lo = loop_init_one(MINOR(dev) >> part_shift); kobj = lo ? get_disk(lo->lo_disk) : ERR_PTR(-ENOMEM); mutex_unlock(&loop_devices_mutex); @@ -1704,18 +1688,32 @@ static int __init loop_init(void) */ part_shift = 0; - if (max_part > 0) + if (max_part > 0) { part_shift = fls(max_part); + /* + * Adjust max_part according to part_shift as it is exported + * to user space so that user can decide correct minor number + * if [s]he want to create more devices. + * + * Note that -1 is required because partition 0 is reserved + * for the whole disk. + */ + max_part = (1UL << part_shift) - 1; + } + + if ((1UL << part_shift) > DISK_MAX_PARTS) + return -EINVAL; + if (max_loop > 1UL << (MINORBITS - part_shift)) return -EINVAL; if (max_loop) { nr = max_loop; - range = max_loop; + range = max_loop << part_shift; } else { nr = 8; - range = 1UL << (MINORBITS - part_shift); + range = 1UL << MINORBITS; } if (register_blkdev(LOOP_MAJOR, "loop")) @@ -1754,7 +1752,7 @@ static void __exit loop_exit(void) unsigned long range; struct loop_device *lo, *next; - range = max_loop ? max_loop : 1UL << (MINORBITS - part_shift); + range = max_loop ? max_loop << part_shift : 1UL << MINORBITS; list_for_each_entry_safe(lo, next, &loop_devices, lo_list) loop_del_one(lo); diff --git a/drivers/block/paride/pcd.c b/drivers/block/paride/pcd.c index 62cec6a..a0aabd9 100644 --- a/drivers/block/paride/pcd.c +++ b/drivers/block/paride/pcd.c @@ -172,7 +172,8 @@ module_param_array(drive3, int, NULL, 0); static int pcd_open(struct cdrom_device_info *cdi, int purpose); static void pcd_release(struct cdrom_device_info *cdi); static int pcd_drive_status(struct cdrom_device_info *cdi, int slot_nr); -static int pcd_media_changed(struct cdrom_device_info *cdi, int slot_nr); +static unsigned int pcd_check_events(struct cdrom_device_info *cdi, + unsigned int clearing, int slot_nr); static int pcd_tray_move(struct cdrom_device_info *cdi, int position); static int pcd_lock_door(struct cdrom_device_info *cdi, int lock); static int pcd_drive_reset(struct cdrom_device_info *cdi); @@ -257,10 +258,11 @@ static int pcd_block_ioctl(struct block_device *bdev, fmode_t mode, return ret; } -static int pcd_block_media_changed(struct gendisk *disk) +static unsigned int pcd_block_check_events(struct gendisk *disk, + unsigned int clearing) { struct pcd_unit *cd = disk->private_data; - return cdrom_media_changed(&cd->info); + return cdrom_check_events(&cd->info, clearing); } static const struct block_device_operations pcd_bdops = { @@ -268,14 +270,14 @@ static const struct block_device_operations pcd_bdops = { .open = pcd_block_open, .release = pcd_block_release, .ioctl = pcd_block_ioctl, - .media_changed = pcd_block_media_changed, + .check_events = pcd_block_check_events, }; static struct cdrom_device_ops pcd_dops = { .open = pcd_open, .release = pcd_release, .drive_status = pcd_drive_status, - .media_changed = pcd_media_changed, + .check_events = pcd_check_events, .tray_move = pcd_tray_move, .lock_door = pcd_lock_door, .get_mcn = pcd_get_mcn, @@ -318,6 +320,8 @@ static void pcd_init_units(void) disk->first_minor = unit; strcpy(disk->disk_name, cd->name); /* umm... */ disk->fops = &pcd_bdops; + disk->flags = GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE; + disk->events = DISK_EVENT_MEDIA_CHANGE; } } @@ -502,13 +506,14 @@ static int pcd_packet(struct cdrom_device_info *cdi, struct packet_command *cgc) #define DBMSG(msg) ((verbose>1)?(msg):NULL) -static int pcd_media_changed(struct cdrom_device_info *cdi, int slot_nr) +static unsigned int pcd_check_events(struct cdrom_device_info *cdi, + unsigned int clearing, int slot_nr) { struct pcd_unit *cd = cdi->handle; int res = cd->changed; if (res) cd->changed = 0; - return res; + return res ? DISK_EVENT_MEDIA_CHANGE : 0; } static int pcd_lock_door(struct cdrom_device_info *cdi, int lock) diff --git a/drivers/block/paride/pd.c b/drivers/block/paride/pd.c index c0ee155..869e767 100644 --- a/drivers/block/paride/pd.c +++ b/drivers/block/paride/pd.c @@ -794,7 +794,7 @@ static int pd_release(struct gendisk *p, fmode_t mode) return 0; } -static int pd_check_media(struct gendisk *p) +static unsigned int pd_check_events(struct gendisk *p, unsigned int clearing) { struct pd_unit *disk = p->private_data; int r; @@ -803,7 +803,7 @@ static int pd_check_media(struct gendisk *p) pd_special_command(disk, pd_media_check); r = disk->changed; disk->changed = 0; - return r; + return r ? DISK_EVENT_MEDIA_CHANGE : 0; } static int pd_revalidate(struct gendisk *p) @@ -822,7 +822,7 @@ static const struct block_device_operations pd_fops = { .release = pd_release, .ioctl = pd_ioctl, .getgeo = pd_getgeo, - .media_changed = pd_check_media, + .check_events = pd_check_events, .revalidate_disk= pd_revalidate }; diff --git a/drivers/block/paride/pf.c b/drivers/block/paride/pf.c index 635f25d..f21b520 100644 --- a/drivers/block/paride/pf.c +++ b/drivers/block/paride/pf.c @@ -243,7 +243,8 @@ static struct pf_unit units[PF_UNITS]; static int pf_identify(struct pf_unit *pf); static void pf_lock(struct pf_unit *pf, int func); static void pf_eject(struct pf_unit *pf); -static int pf_check_media(struct gendisk *disk); +static unsigned int pf_check_events(struct gendisk *disk, + unsigned int clearing); static char pf_scratch[512]; /* scratch block buffer */ @@ -270,7 +271,7 @@ static const struct block_device_operations pf_fops = { .release = pf_release, .ioctl = pf_ioctl, .getgeo = pf_getgeo, - .media_changed = pf_check_media, + .check_events = pf_check_events, }; static void __init pf_init_units(void) @@ -377,9 +378,9 @@ static int pf_release(struct gendisk *disk, fmode_t mode) } -static int pf_check_media(struct gendisk *disk) +static unsigned int pf_check_events(struct gendisk *disk, unsigned int clearing) { - return 1; + return DISK_EVENT_MEDIA_CHANGE; } static inline int status_reg(struct pf_unit *pf) diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c index 77d70ee..07a382e 100644 --- a/drivers/block/pktcdvd.c +++ b/drivers/block/pktcdvd.c @@ -1606,8 +1606,6 @@ static int kcdrwd(void *foobar) min_sleep_time = pkt->sleep_time; } - generic_unplug_device(bdev_get_queue(pd->bdev)); - VPRINTK("kcdrwd: sleeping\n"); residue = schedule_timeout(min_sleep_time); VPRINTK("kcdrwd: wake up\n"); @@ -2796,7 +2794,8 @@ static int pkt_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, return ret; } -static int pkt_media_changed(struct gendisk *disk) +static unsigned int pkt_check_events(struct gendisk *disk, + unsigned int clearing) { struct pktcdvd_device *pd = disk->private_data; struct gendisk *attached_disk; @@ -2806,9 +2805,9 @@ static int pkt_media_changed(struct gendisk *disk) if (!pd->bdev) return 0; attached_disk = pd->bdev->bd_disk; - if (!attached_disk) + if (!attached_disk || !attached_disk->fops->check_events) return 0; - return attached_disk->fops->media_changed(attached_disk); + return attached_disk->fops->check_events(attached_disk, clearing); } static const struct block_device_operations pktcdvd_ops = { @@ -2816,7 +2815,7 @@ static const struct block_device_operations pktcdvd_ops = { .open = pkt_open, .release = pkt_close, .ioctl = pkt_ioctl, - .media_changed = pkt_media_changed, + .check_events = pkt_check_events, }; static char *pktcdvd_devnode(struct gendisk *gd, mode_t *mode) @@ -2889,6 +2888,10 @@ static int pkt_setup_dev(dev_t dev, dev_t* pkt_dev) if (ret) goto out_new_dev; + /* inherit events of the host device */ + disk->events = pd->bdev->bd_disk->events; + disk->async_events = pd->bdev->bd_disk->async_events; + add_disk(disk); pkt_sysfs_dev_new(pd); diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index e1e38b1..1278098 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -31,6 +31,7 @@ #include <linux/ceph/osd_client.h> #include <linux/ceph/mon_client.h> #include <linux/ceph/decode.h> +#include <linux/parser.h> #include <linux/kernel.h> #include <linux/device.h> @@ -54,6 +55,8 @@ #define DEV_NAME_LEN 32 +#define RBD_NOTIFY_TIMEOUT_DEFAULT 10 + /* * block device image metadata (in-memory version) */ @@ -71,6 +74,12 @@ struct rbd_image_header { char *snap_names; u64 *snap_sizes; + + u64 obj_version; +}; + +struct rbd_options { + int notify_timeout; }; /* @@ -78,10 +87,13 @@ struct rbd_image_header { */ struct rbd_client { struct ceph_client *client; + struct rbd_options *rbd_opts; struct kref kref; struct list_head node; }; +struct rbd_req_coll; + /* * a single io request */ @@ -90,6 +102,24 @@ struct rbd_request { struct bio *bio; /* cloned bio */ struct page **pages; /* list of used pages */ u64 len; + int coll_index; + struct rbd_req_coll *coll; +}; + +struct rbd_req_status { + int done; + int rc; + u64 bytes; +}; + +/* + * a collection of requests + */ +struct rbd_req_coll { + int total; + int num_done; + struct kref kref; + struct rbd_req_status status[0]; }; struct rbd_snap { @@ -124,6 +154,9 @@ struct rbd_device { char pool_name[RBD_MAX_POOL_NAME_LEN]; int poolid; + struct ceph_osd_event *watch_event; + struct ceph_osd_request *watch_request; + char snap_name[RBD_MAX_SNAP_NAME_LEN]; u32 cur_snap; /* index+1 of current snapshot within snap context 0 - for the head */ @@ -177,6 +210,8 @@ static void rbd_put_dev(struct rbd_device *rbd_dev) put_device(&rbd_dev->dev); } +static int __rbd_update_snaps(struct rbd_device *rbd_dev); + static int rbd_open(struct block_device *bdev, fmode_t mode) { struct gendisk *disk = bdev->bd_disk; @@ -211,7 +246,8 @@ static const struct block_device_operations rbd_bd_ops = { * Initialize an rbd client instance. * We own *opt. */ -static struct rbd_client *rbd_client_create(struct ceph_options *opt) +static struct rbd_client *rbd_client_create(struct ceph_options *opt, + struct rbd_options *rbd_opts) { struct rbd_client *rbdc; int ret = -ENOMEM; @@ -233,6 +269,8 @@ static struct rbd_client *rbd_client_create(struct ceph_options *opt) if (ret < 0) goto out_err; + rbdc->rbd_opts = rbd_opts; + spin_lock(&node_lock); list_add_tail(&rbdc->node, &rbd_client_list); spin_unlock(&node_lock); @@ -267,6 +305,59 @@ static struct rbd_client *__rbd_client_find(struct ceph_options *opt) } /* + * mount options + */ +enum { + Opt_notify_timeout, + Opt_last_int, + /* int args above */ + Opt_last_string, + /* string args above */ +}; + +static match_table_t rbdopt_tokens = { + {Opt_notify_timeout, "notify_timeout=%d"}, + /* int args above */ + /* string args above */ + {-1, NULL} +}; + +static int parse_rbd_opts_token(char *c, void *private) +{ + struct rbd_options *rbdopt = private; + substring_t argstr[MAX_OPT_ARGS]; + int token, intval, ret; + + token = match_token((char *)c, rbdopt_tokens, argstr); + if (token < 0) + return -EINVAL; + + if (token < Opt_last_int) { + ret = match_int(&argstr[0], &intval); + if (ret < 0) { + pr_err("bad mount option arg (not int) " + "at '%s'\n", c); + return ret; + } + dout("got int token %d val %d\n", token, intval); + } else if (token > Opt_last_int && token < Opt_last_string) { + dout("got string token %d val %s\n", token, + argstr[0].from); + } else { + dout("got token %d\n", token); + } + + switch (token) { + case Opt_notify_timeout: + rbdopt->notify_timeout = intval; + break; + default: + BUG_ON(token); + } + return 0; +} + +/* * Get a ceph client with specific addr and configuration, if one does * not exist create it. */ @@ -276,11 +367,18 @@ static int rbd_get_client(struct rbd_device *rbd_dev, const char *mon_addr, struct rbd_client *rbdc; struct ceph_options *opt; int ret; + struct rbd_options *rbd_opts; + + rbd_opts = kzalloc(sizeof(*rbd_opts), GFP_KERNEL); + if (!rbd_opts) + return -ENOMEM; + + rbd_opts->notify_timeout = RBD_NOTIFY_TIMEOUT_DEFAULT; ret = ceph_parse_options(&opt, options, mon_addr, - mon_addr + strlen(mon_addr), NULL, NULL); + mon_addr + strlen(mon_addr), parse_rbd_opts_token, rbd_opts); if (ret < 0) - return ret; + goto done_err; spin_lock(&node_lock); rbdc = __rbd_client_find(opt); @@ -296,13 +394,18 @@ static int rbd_get_client(struct rbd_device *rbd_dev, const char *mon_addr, } spin_unlock(&node_lock); - rbdc = rbd_client_create(opt); - if (IS_ERR(rbdc)) - return PTR_ERR(rbdc); + rbdc = rbd_client_create(opt, rbd_opts); + if (IS_ERR(rbdc)) { + ret = PTR_ERR(rbdc); + goto done_err; + } rbd_dev->rbd_client = rbdc; rbd_dev->client = rbdc->client; return 0; +done_err: + kfree(rbd_opts); + return ret; } /* @@ -318,6 +421,7 @@ static void rbd_client_release(struct kref *kref) spin_unlock(&node_lock); ceph_destroy_client(rbdc->client); + kfree(rbdc->rbd_opts); kfree(rbdc); } @@ -332,6 +436,17 @@ static void rbd_put_client(struct rbd_device *rbd_dev) rbd_dev->client = NULL; } +/* + * Destroy requests collection + */ +static void rbd_coll_release(struct kref *kref) +{ + struct rbd_req_coll *coll = + container_of(kref, struct rbd_req_coll, kref); + + dout("rbd_coll_release %p\n", coll); + kfree(coll); +} /* * Create a new header structure, translate header format from the on-disk @@ -506,6 +621,14 @@ static u64 rbd_get_segment(struct rbd_image_header *header, return len; } +static int rbd_get_num_segments(struct rbd_image_header *header, + u64 ofs, u64 len) +{ + u64 start_seg = ofs >> header->obj_order; + u64 end_seg = (ofs + len - 1) >> header->obj_order; + return end_seg - start_seg + 1; +} + /* * bio helpers */ @@ -651,6 +774,50 @@ static void rbd_destroy_ops(struct ceph_osd_req_op *ops) kfree(ops); } +static void rbd_coll_end_req_index(struct request *rq, + struct rbd_req_coll *coll, + int index, + int ret, u64 len) +{ + struct request_queue *q; + int min, max, i; + + dout("rbd_coll_end_req_index %p index %d ret %d len %lld\n", + coll, index, ret, len); + + if (!rq) + return; + + if (!coll) { + blk_end_request(rq, ret, len); + return; + } + + q = rq->q; + + spin_lock_irq(q->queue_lock); + coll->status[index].done = 1; + coll->status[index].rc = ret; + coll->status[index].bytes = len; + max = min = coll->num_done; + while (max < coll->total && coll->status[max].done) + max++; + + for (i = min; i<max; i++) { + __blk_end_request(rq, coll->status[i].rc, + coll->status[i].bytes); + coll->num_done++; + kref_put(&coll->kref, rbd_coll_release); + } + spin_unlock_irq(q->queue_lock); +} + +static void rbd_coll_end_req(struct rbd_request *req, + int ret, u64 len) +{ + rbd_coll_end_req_index(req->rq, req->coll, req->coll_index, ret, len); +} + /* * Send ceph osd request */ @@ -665,8 +832,12 @@ static int rbd_do_request(struct request *rq, int flags, struct ceph_osd_req_op *ops, int num_reply, + struct rbd_req_coll *coll, + int coll_index, void (*rbd_cb)(struct ceph_osd_request *req, - struct ceph_msg *msg)) + struct ceph_msg *msg), + struct ceph_osd_request **linger_req, + u64 *ver) { struct ceph_osd_request *req; struct ceph_file_layout *layout; @@ -677,12 +848,20 @@ static int rbd_do_request(struct request *rq, struct ceph_osd_request_head *reqhead; struct rbd_image_header *header = &dev->header; - ret = -ENOMEM; req_data = kzalloc(sizeof(*req_data), GFP_NOIO); - if (!req_data) - goto done; + if (!req_data) { + if (coll) + rbd_coll_end_req_index(rq, coll, coll_index, + -ENOMEM, len); + return -ENOMEM; + } - dout("rbd_do_request len=%lld ofs=%lld\n", len, ofs); + if (coll) { + req_data->coll = coll; + req_data->coll_index = coll_index; + } + + dout("rbd_do_request obj=%s ofs=%lld len=%lld\n", obj, len, ofs); down_read(&header->snap_rwsem); @@ -691,9 +870,9 @@ static int rbd_do_request(struct request *rq, ops, false, GFP_NOIO, pages, bio); - if (IS_ERR(req)) { + if (!req) { up_read(&header->snap_rwsem); - ret = PTR_ERR(req); + ret = -ENOMEM; goto done_pages; } @@ -729,12 +908,21 @@ static int rbd_do_request(struct request *rq, req->r_oid, req->r_oid_len); up_read(&header->snap_rwsem); + if (linger_req) { + ceph_osdc_set_request_linger(&dev->client->osdc, req); + *linger_req = req; + } + ret = ceph_osdc_start_request(&dev->client->osdc, req, false); if (ret < 0) goto done_err; if (!rbd_cb) { ret = ceph_osdc_wait_request(&dev->client->osdc, req); + if (ver) + *ver = le64_to_cpu(req->r_reassert_version.version); + dout("reassert_ver=%lld\n", + le64_to_cpu(req->r_reassert_version.version)); ceph_osdc_put_request(req); } return ret; @@ -743,10 +931,8 @@ done_err: bio_chain_put(req_data->bio); ceph_osdc_put_request(req); done_pages: + rbd_coll_end_req(req_data, ret, len); kfree(req_data); -done: - if (rq) - blk_end_request(rq, ret, len); return ret; } @@ -780,7 +966,7 @@ static void rbd_req_cb(struct ceph_osd_request *req, struct ceph_msg *msg) bytes = req_data->len; } - blk_end_request(req_data->rq, rc, bytes); + rbd_coll_end_req(req_data, rc, bytes); if (req_data->bio) bio_chain_put(req_data->bio); @@ -789,6 +975,11 @@ static void rbd_req_cb(struct ceph_osd_request *req, struct ceph_msg *msg) kfree(req_data); } +static void rbd_simple_req_cb(struct ceph_osd_request *req, struct ceph_msg *msg) +{ + ceph_osdc_put_request(req); +} + /* * Do a synchronous ceph osd operation */ @@ -801,7 +992,9 @@ static int rbd_req_sync_op(struct rbd_device *dev, int num_reply, const char *obj, u64 ofs, u64 len, - char *buf) + char *buf, + struct ceph_osd_request **linger_req, + u64 *ver) { int ret; struct page **pages; @@ -833,7 +1026,9 @@ static int rbd_req_sync_op(struct rbd_device *dev, flags, ops, 2, - NULL); + NULL, 0, + NULL, + linger_req, ver); if (ret < 0) goto done_ops; @@ -857,7 +1052,9 @@ static int rbd_do_op(struct request *rq, u64 snapid, int opcode, int flags, int num_reply, u64 ofs, u64 len, - struct bio *bio) + struct bio *bio, + struct rbd_req_coll *coll, + int coll_index) { char *seg_name; u64 seg_ofs; @@ -893,7 +1090,10 @@ static int rbd_do_op(struct request *rq, flags, ops, num_reply, - rbd_req_cb); + coll, coll_index, + rbd_req_cb, 0, NULL); + + rbd_destroy_ops(ops); done: kfree(seg_name); return ret; @@ -906,13 +1106,15 @@ static int rbd_req_write(struct request *rq, struct rbd_device *rbd_dev, struct ceph_snap_context *snapc, u64 ofs, u64 len, - struct bio *bio) + struct bio *bio, + struct rbd_req_coll *coll, + int coll_index) { return rbd_do_op(rq, rbd_dev, snapc, CEPH_NOSNAP, CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, 2, - ofs, len, bio); + ofs, len, bio, coll, coll_index); } /* @@ -922,14 +1124,16 @@ static int rbd_req_read(struct request *rq, struct rbd_device *rbd_dev, u64 snapid, u64 ofs, u64 len, - struct bio *bio) + struct bio *bio, + struct rbd_req_coll *coll, + int coll_index) { return rbd_do_op(rq, rbd_dev, NULL, (snapid ? snapid : CEPH_NOSNAP), CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ, 2, - ofs, len, bio); + ofs, len, bio, coll, coll_index); } /* @@ -940,18 +1144,182 @@ static int rbd_req_sync_read(struct rbd_device *dev, u64 snapid, const char *obj, u64 ofs, u64 len, - char *buf) + char *buf, + u64 *ver) { return rbd_req_sync_op(dev, NULL, (snapid ? snapid : CEPH_NOSNAP), CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ, NULL, - 1, obj, ofs, len, buf); + 1, obj, ofs, len, buf, NULL, ver); } /* - * Request sync osd read + * Request sync osd watch + */ +static int rbd_req_sync_notify_ack(struct rbd_device *dev, + u64 ver, + u64 notify_id, + const char *obj) +{ + struct ceph_osd_req_op *ops; + struct page **pages = NULL; + int ret; + + ret = rbd_create_rw_ops(&ops, 1, CEPH_OSD_OP_NOTIFY_ACK, 0); + if (ret < 0) + return ret; + + ops[0].watch.ver = cpu_to_le64(dev->header.obj_version); + ops[0].watch.cookie = notify_id; + ops[0].watch.flag = 0; + + ret = rbd_do_request(NULL, dev, NULL, CEPH_NOSNAP, + obj, 0, 0, NULL, + pages, 0, + CEPH_OSD_FLAG_READ, + ops, + 1, + NULL, 0, + rbd_simple_req_cb, 0, NULL); + + rbd_destroy_ops(ops); + return ret; +} + +static void rbd_watch_cb(u64 ver, u64 notify_id, u8 opcode, void *data) +{ + struct rbd_device *dev = (struct rbd_device *)data; + int rc; + + if (!dev) + return; + + dout("rbd_watch_cb %s notify_id=%lld opcode=%d\n", dev->obj_md_name, + notify_id, (int)opcode); + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + rc = __rbd_update_snaps(dev); + mutex_unlock(&ctl_mutex); + if (rc) + pr_warning(DRV_NAME "%d got notification but failed to update" + " snaps: %d\n", dev->major, rc); + + rbd_req_sync_notify_ack(dev, ver, notify_id, dev->obj_md_name); +} + +/* + * Request sync osd watch + */ +static int rbd_req_sync_watch(struct rbd_device *dev, + const char *obj, + u64 ver) +{ + struct ceph_osd_req_op *ops; + struct ceph_osd_client *osdc = &dev->client->osdc; + + int ret = rbd_create_rw_ops(&ops, 1, CEPH_OSD_OP_WATCH, 0); + if (ret < 0) + return ret; + + ret = ceph_osdc_create_event(osdc, rbd_watch_cb, 0, + (void *)dev, &dev->watch_event); + if (ret < 0) + goto fail; + + ops[0].watch.ver = cpu_to_le64(ver); + ops[0].watch.cookie = cpu_to_le64(dev->watch_event->cookie); + ops[0].watch.flag = 1; + + ret = rbd_req_sync_op(dev, NULL, + CEPH_NOSNAP, + 0, + CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, + ops, + 1, obj, 0, 0, NULL, + &dev->watch_request, NULL); + + if (ret < 0) + goto fail_event; + + rbd_destroy_ops(ops); + return 0; + +fail_event: + ceph_osdc_cancel_event(dev->watch_event); + dev->watch_event = NULL; +fail: + rbd_destroy_ops(ops); + return ret; +} + +struct rbd_notify_info { + struct rbd_device *dev; +}; + +static void rbd_notify_cb(u64 ver, u64 notify_id, u8 opcode, void *data) +{ + struct rbd_device *dev = (struct rbd_device *)data; + if (!dev) + return; + + dout("rbd_notify_cb %s notify_id=%lld opcode=%d\n", dev->obj_md_name, + notify_id, (int)opcode); +} + +/* + * Request sync osd notify + */ +static int rbd_req_sync_notify(struct rbd_device *dev, + const char *obj) +{ + struct ceph_osd_req_op *ops; + struct ceph_osd_client *osdc = &dev->client->osdc; + struct ceph_osd_event *event; + struct rbd_notify_info info; + int payload_len = sizeof(u32) + sizeof(u32); + int ret; + + ret = rbd_create_rw_ops(&ops, 1, CEPH_OSD_OP_NOTIFY, payload_len); + if (ret < 0) + return ret; + + info.dev = dev; + + ret = ceph_osdc_create_event(osdc, rbd_notify_cb, 1, + (void *)&info, &event); + if (ret < 0) + goto fail; + + ops[0].watch.ver = 1; + ops[0].watch.flag = 1; + ops[0].watch.cookie = event->cookie; + ops[0].watch.prot_ver = RADOS_NOTIFY_VER; + ops[0].watch.timeout = 12; + + ret = rbd_req_sync_op(dev, NULL, + CEPH_NOSNAP, + 0, + CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, + ops, + 1, obj, 0, 0, NULL, NULL, NULL); + if (ret < 0) + goto fail_event; + + ret = ceph_osdc_wait_event(event, CEPH_OSD_TIMEOUT_DEFAULT); + dout("ceph_osdc_wait_event returned %d\n", ret); + rbd_destroy_ops(ops); + return 0; + +fail_event: + ceph_osdc_cancel_event(event); +fail: + rbd_destroy_ops(ops); + return ret; +} + +/* + * Request sync osd rollback */ static int rbd_req_sync_rollback_obj(struct rbd_device *dev, u64 snapid, @@ -969,13 +1337,10 @@ static int rbd_req_sync_rollback_obj(struct rbd_device *dev, 0, CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, ops, - 1, obj, 0, 0, NULL); + 1, obj, 0, 0, NULL, NULL, NULL); rbd_destroy_ops(ops); - if (ret < 0) - return ret; - return ret; } @@ -987,7 +1352,8 @@ static int rbd_req_sync_exec(struct rbd_device *dev, const char *cls, const char *method, const char *data, - int len) + int len, + u64 *ver) { struct ceph_osd_req_op *ops; int cls_len = strlen(cls); @@ -1010,7 +1376,7 @@ static int rbd_req_sync_exec(struct rbd_device *dev, 0, CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, ops, - 1, obj, 0, 0, NULL); + 1, obj, 0, 0, NULL, NULL, ver); rbd_destroy_ops(ops); @@ -1018,6 +1384,20 @@ static int rbd_req_sync_exec(struct rbd_device *dev, return ret; } +static struct rbd_req_coll *rbd_alloc_coll(int num_reqs) +{ + struct rbd_req_coll *coll = + kzalloc(sizeof(struct rbd_req_coll) + + sizeof(struct rbd_req_status) * num_reqs, + GFP_ATOMIC); + + if (!coll) + return NULL; + coll->total = num_reqs; + kref_init(&coll->kref); + return coll; +} + /* * block device queue callback */ @@ -1035,6 +1415,8 @@ static void rbd_rq_fn(struct request_queue *q) bool do_write; int size, op_size = 0; u64 ofs; + int num_segs, cur_seg = 0; + struct rbd_req_coll *coll; /* peek at request from block layer */ if (!rq) @@ -1065,6 +1447,14 @@ static void rbd_rq_fn(struct request_queue *q) do_write ? "write" : "read", size, blk_rq_pos(rq) * 512ULL); + num_segs = rbd_get_num_segments(&rbd_dev->header, ofs, size); + coll = rbd_alloc_coll(num_segs); + if (!coll) { + spin_lock_irq(q->queue_lock); + __blk_end_request_all(rq, -ENOMEM); + goto next; + } + do { /* a bio clone to be passed down to OSD req */ dout("rq->bio->bi_vcnt=%d\n", rq->bio->bi_vcnt); @@ -1072,35 +1462,41 @@ static void rbd_rq_fn(struct request_queue *q) rbd_dev->header.block_name, ofs, size, NULL, NULL); + kref_get(&coll->kref); bio = bio_chain_clone(&rq_bio, &next_bio, &bp, op_size, GFP_ATOMIC); if (!bio) { - spin_lock_irq(q->queue_lock); - __blk_end_request_all(rq, -ENOMEM); - goto next; + rbd_coll_end_req_index(rq, coll, cur_seg, + -ENOMEM, op_size); + goto next_seg; } + /* init OSD command: write or read */ if (do_write) rbd_req_write(rq, rbd_dev, rbd_dev->header.snapc, ofs, - op_size, bio); + op_size, bio, + coll, cur_seg); else rbd_req_read(rq, rbd_dev, cur_snap_id(rbd_dev), ofs, - op_size, bio); + op_size, bio, + coll, cur_seg); +next_seg: size -= op_size; ofs += op_size; + cur_seg++; rq_bio = next_bio; } while (size > 0); + kref_put(&coll->kref, rbd_coll_release); if (bp) bio_pair_release(bp); - spin_lock_irq(q->queue_lock); next: rq = blk_fetch_request(q); @@ -1156,6 +1552,7 @@ static int rbd_read_header(struct rbd_device *rbd_dev, struct rbd_image_header_ondisk *dh; int snap_count = 0; u64 snap_names_len = 0; + u64 ver; while (1) { int len = sizeof(*dh) + @@ -1171,7 +1568,7 @@ static int rbd_read_header(struct rbd_device *rbd_dev, NULL, CEPH_NOSNAP, rbd_dev->obj_md_name, 0, len, - (char *)dh); + (char *)dh, &ver); if (rc < 0) goto out_dh; @@ -1188,6 +1585,7 @@ static int rbd_read_header(struct rbd_device *rbd_dev, } break; } + header->obj_version = ver; out_dh: kfree(dh); @@ -1204,7 +1602,8 @@ static int rbd_header_add_snap(struct rbd_device *dev, int name_len = strlen(snap_name); u64 new_snapid; int ret; - void *data, *data_start, *data_end; + void *data, *p, *e; + u64 ver; /* we should create a snapshot only if we're pointing at the head */ if (dev->cur_snap) @@ -1220,16 +1619,16 @@ static int rbd_header_add_snap(struct rbd_device *dev, if (!data) return -ENOMEM; - data_start = data; - data_end = data + name_len + 16; + p = data; + e = data + name_len + 16; - ceph_encode_string_safe(&data, data_end, snap_name, name_len, bad); - ceph_encode_64_safe(&data, data_end, new_snapid, bad); + ceph_encode_string_safe(&p, e, snap_name, name_len, bad); + ceph_encode_64_safe(&p, e, new_snapid, bad); ret = rbd_req_sync_exec(dev, dev->obj_md_name, "rbd", "snap_add", - data_start, data - data_start); + data, p - data, &ver); - kfree(data_start); + kfree(data); if (ret < 0) return ret; @@ -1259,14 +1658,23 @@ static int __rbd_update_snaps(struct rbd_device *rbd_dev) int ret; struct rbd_image_header h; u64 snap_seq; + int follow_seq = 0; ret = rbd_read_header(rbd_dev, &h); if (ret < 0) return ret; + /* resized? */ + set_capacity(rbd_dev->disk, h.image_size / 512ULL); + down_write(&rbd_dev->header.snap_rwsem); snap_seq = rbd_dev->header.snapc->seq; + if (rbd_dev->header.total_snaps && + rbd_dev->header.snapc->snaps[0] == snap_seq) + /* pointing at the head, will need to follow that + if head moves */ + follow_seq = 1; kfree(rbd_dev->header.snapc); kfree(rbd_dev->header.snap_names); @@ -1277,7 +1685,10 @@ static int __rbd_update_snaps(struct rbd_device *rbd_dev) rbd_dev->header.snap_names = h.snap_names; rbd_dev->header.snap_names_len = h.snap_names_len; rbd_dev->header.snap_sizes = h.snap_sizes; - rbd_dev->header.snapc->seq = snap_seq; + if (follow_seq) + rbd_dev->header.snapc->seq = rbd_dev->header.snapc->snaps[0]; + else + rbd_dev->header.snapc->seq = snap_seq; ret = __rbd_init_snaps_header(rbd_dev); @@ -1313,7 +1724,8 @@ static int rbd_init_disk(struct rbd_device *rbd_dev) if (!disk) goto out; - sprintf(disk->disk_name, DRV_NAME "%d", rbd_dev->id); + snprintf(disk->disk_name, sizeof(disk->disk_name), DRV_NAME "%d", + rbd_dev->id); disk->major = rbd_dev->major; disk->first_minor = 0; disk->fops = &rbd_bd_ops; @@ -1699,7 +2111,28 @@ static void rbd_bus_del_dev(struct rbd_device *rbd_dev) device_unregister(&rbd_dev->dev); } -static ssize_t rbd_add(struct bus_type *bus, const char *buf, size_t count) +static int rbd_init_watch_dev(struct rbd_device *rbd_dev) +{ + int ret, rc; + + do { + ret = rbd_req_sync_watch(rbd_dev, rbd_dev->obj_md_name, + rbd_dev->header.obj_version); + if (ret == -ERANGE) { + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + rc = __rbd_update_snaps(rbd_dev); + mutex_unlock(&ctl_mutex); + if (rc < 0) + return rc; + } + } while (ret == -ERANGE); + + return ret; +} + +static ssize_t rbd_add(struct bus_type *bus, + const char *buf, + size_t count) { struct ceph_osd_client *osdc; struct rbd_device *rbd_dev; @@ -1797,6 +2230,10 @@ static ssize_t rbd_add(struct bus_type *bus, const char *buf, size_t count) if (rc) goto err_out_bus; + rc = rbd_init_watch_dev(rbd_dev); + if (rc) + goto err_out_bus; + return count; err_out_bus: @@ -1849,6 +2286,12 @@ static void rbd_dev_release(struct device *dev) struct rbd_device *rbd_dev = container_of(dev, struct rbd_device, dev); + if (rbd_dev->watch_request) + ceph_osdc_unregister_linger_request(&rbd_dev->client->osdc, + rbd_dev->watch_request); + if (rbd_dev->watch_event) + ceph_osdc_cancel_event(rbd_dev->watch_event); + rbd_put_client(rbd_dev); /* clean up and free blkdev */ @@ -1914,14 +2357,24 @@ static ssize_t rbd_snap_add(struct device *dev, ret = rbd_header_add_snap(rbd_dev, name, GFP_KERNEL); if (ret < 0) - goto done_unlock; + goto err_unlock; ret = __rbd_update_snaps(rbd_dev); if (ret < 0) - goto done_unlock; + goto err_unlock; + + /* shouldn't hold ctl_mutex when notifying.. notify might + trigger a watch callback that would need to get that mutex */ + mutex_unlock(&ctl_mutex); + + /* make a best effort, don't error if failed */ + rbd_req_sync_notify(rbd_dev, rbd_dev->obj_md_name); ret = count; -done_unlock: + kfree(name); + return ret; + +err_unlock: mutex_unlock(&ctl_mutex); kfree(name); return ret; diff --git a/drivers/block/smart1,2.h b/drivers/block/smart1,2.h index a0b403a..e5565fb 100644 --- a/drivers/block/smart1,2.h +++ b/drivers/block/smart1,2.h @@ -95,7 +95,7 @@ static unsigned long smart4_completed(ctlr_info_t *h) /* * This hardware returns interrupt pending at a different place and * it does not tell us if the fifo is empty, we will have check - * that by getting a 0 back from the comamnd_completed call. + * that by getting a 0 back from the command_completed call. */ static unsigned long smart4_intr_pending(ctlr_info_t *h) { diff --git a/drivers/block/swim.c b/drivers/block/swim.c index 75333d0..fd5adcd 100644 --- a/drivers/block/swim.c +++ b/drivers/block/swim.c @@ -741,11 +741,12 @@ static int floppy_getgeo(struct block_device *bdev, struct hd_geometry *geo) return 0; } -static int floppy_check_change(struct gendisk *disk) +static unsigned int floppy_check_events(struct gendisk *disk, + unsigned int clearing) { struct floppy_state *fs = disk->private_data; - return fs->ejected; + return fs->ejected ? DISK_EVENT_MEDIA_CHANGE : 0; } static int floppy_revalidate(struct gendisk *disk) @@ -772,7 +773,7 @@ static const struct block_device_operations floppy_fops = { .release = floppy_release, .ioctl = floppy_ioctl, .getgeo = floppy_getgeo, - .media_changed = floppy_check_change, + .check_events = floppy_check_events, .revalidate_disk = floppy_revalidate, }; diff --git a/drivers/block/swim3.c b/drivers/block/swim3.c index bf3a5b8..773bfa7 100644 --- a/drivers/block/swim3.c +++ b/drivers/block/swim3.c @@ -250,7 +250,8 @@ static int floppy_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, unsigned long param); static int floppy_open(struct block_device *bdev, fmode_t mode); static int floppy_release(struct gendisk *disk, fmode_t mode); -static int floppy_check_change(struct gendisk *disk); +static unsigned int floppy_check_events(struct gendisk *disk, + unsigned int clearing); static int floppy_revalidate(struct gendisk *disk); static bool swim3_end_request(int err, unsigned int nr_bytes) @@ -975,10 +976,11 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) return 0; } -static int floppy_check_change(struct gendisk *disk) +static unsigned int floppy_check_events(struct gendisk *disk, + unsigned int clearing) { struct floppy_state *fs = disk->private_data; - return fs->ejected; + return fs->ejected ? DISK_EVENT_MEDIA_CHANGE : 0; } static int floppy_revalidate(struct gendisk *disk) @@ -1025,7 +1027,7 @@ static const struct block_device_operations floppy_fops = { .open = floppy_unlocked_open, .release = floppy_release, .ioctl = floppy_ioctl, - .media_changed = floppy_check_change, + .check_events = floppy_check_events, .revalidate_disk= floppy_revalidate, }; diff --git a/drivers/block/ub.c b/drivers/block/ub.c index 9ae3bb7..0e376d4 100644 --- a/drivers/block/ub.c +++ b/drivers/block/ub.c @@ -1788,7 +1788,8 @@ static int ub_bd_revalidate(struct gendisk *disk) * * The return code is bool! */ -static int ub_bd_media_changed(struct gendisk *disk) +static unsigned int ub_bd_check_events(struct gendisk *disk, + unsigned int clearing) { struct ub_lun *lun = disk->private_data; @@ -1806,10 +1807,10 @@ static int ub_bd_media_changed(struct gendisk *disk) */ if (ub_sync_tur(lun->udev, lun) != 0) { lun->changed = 1; - return 1; + return DISK_EVENT_MEDIA_CHANGE; } - return lun->changed; + return lun->changed ? DISK_EVENT_MEDIA_CHANGE : 0; } static const struct block_device_operations ub_bd_fops = { @@ -1817,7 +1818,7 @@ static const struct block_device_operations ub_bd_fops = { .open = ub_bd_unlocked_open, .release = ub_bd_release, .ioctl = ub_bd_ioctl, - .media_changed = ub_bd_media_changed, + .check_events = ub_bd_check_events, .revalidate_disk = ub_bd_revalidate, }; diff --git a/drivers/block/umem.c b/drivers/block/umem.c index 8be5715..031ca72 100644 --- a/drivers/block/umem.c +++ b/drivers/block/umem.c @@ -241,8 +241,7 @@ static void dump_dmastat(struct cardinfo *card, unsigned int dmastat) * * Whenever IO on the active page completes, the Ready page is activated * and the ex-Active page is clean out and made Ready. - * Otherwise the Ready page is only activated when it becomes full, or - * when mm_unplug_device is called via the unplug_io_fn. + * Otherwise the Ready page is only activated when it becomes full. * * If a request arrives while both pages a full, it is queued, and b_rdev is * overloaded to record whether it was a read or a write. @@ -333,17 +332,6 @@ static inline void reset_page(struct mm_page *page) page->biotail = &page->bio; } -static void mm_unplug_device(struct request_queue *q) -{ - struct cardinfo *card = q->queuedata; - unsigned long flags; - - spin_lock_irqsave(&card->lock, flags); - if (blk_remove_plug(q)) - activate(card); - spin_unlock_irqrestore(&card->lock, flags); -} - /* * If there is room on Ready page, take * one bh off list and add it. @@ -535,7 +523,6 @@ static int mm_make_request(struct request_queue *q, struct bio *bio) *card->biotail = bio; bio->bi_next = NULL; card->biotail = &bio->bi_next; - blk_plug_device(q); spin_unlock_irq(&card->lock); return 0; @@ -779,20 +766,10 @@ static int mm_getgeo(struct block_device *bdev, struct hd_geometry *geo) return 0; } -/* - * Future support for removable devices - */ -static int mm_check_change(struct gendisk *disk) -{ -/* struct cardinfo *dev = disk->private_data; */ - return 0; -} - static const struct block_device_operations mm_fops = { .owner = THIS_MODULE, .getgeo = mm_getgeo, .revalidate_disk = mm_revalidate, - .media_changed = mm_check_change, }; static int __devinit mm_pci_probe(struct pci_dev *dev, @@ -907,7 +884,6 @@ static int __devinit mm_pci_probe(struct pci_dev *dev, blk_queue_make_request(card->queue, mm_make_request); card->queue->queue_lock = &card->lock; card->queue->queuedata = card; - card->queue->unplug_fn = mm_unplug_device; tasklet_init(&card->tasklet, process_page, (unsigned long)card); diff --git a/drivers/block/viodasd.c b/drivers/block/viodasd.c index e2ff697..9a5b2a2 100644 --- a/drivers/block/viodasd.c +++ b/drivers/block/viodasd.c @@ -94,7 +94,7 @@ static const struct vio_error_entry viodasd_err_table[] = { { 0x0204, EIO, "Use Error" }, { 0x0205, EIO, "Release Error" }, { 0x0206, EINVAL, "Invalid Disk" }, - { 0x0207, EBUSY, "Cant Lock" }, + { 0x0207, EBUSY, "Can't Lock" }, { 0x0208, EIO, "Already Locked" }, { 0x0209, EIO, "Already Unlocked" }, { 0x020A, EIO, "Invalid Arg" }, diff --git a/drivers/block/xen-blkback/Makefile b/drivers/block/xen-blkback/Makefile new file mode 100644 index 0000000..e491c1b --- /dev/null +++ b/drivers/block/xen-blkback/Makefile @@ -0,0 +1,3 @@ +obj-$(CONFIG_XEN_BLKDEV_BACKEND) := xen-blkback.o + +xen-blkback-y := blkback.o xenbus.o diff --git a/drivers/block/xen-blkback/blkback.c b/drivers/block/xen-blkback/blkback.c new file mode 100644 index 0000000..c73910c --- /dev/null +++ b/drivers/block/xen-blkback/blkback.c @@ -0,0 +1,824 @@ +/****************************************************************************** + * + * Back-end of the driver for virtual block devices. This portion of the + * driver exports a 'unified' block-device interface that can be accessed + * by any operating system that implements a compatible front end. A + * reference front-end implementation can be found in: + * drivers/block/xen-blkfront.c + * + * Copyright (c) 2003-2004, Keir Fraser & Steve Hand + * Copyright (c) 2005, Christopher Clark + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version 2 + * as published by the Free Software Foundation; or, when distributed + * separately from the Linux kernel or incorporated into other + * software packages, subject to the following license: + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this source file (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, copy, modify, + * merge, publish, distribute, sublicense, and/or sell copies of the Software, + * and to permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ + +#include <linux/spinlock.h> +#include <linux/kthread.h> +#include <linux/list.h> +#include <linux/delay.h> +#include <linux/freezer.h> + +#include <xen/events.h> +#include <xen/page.h> +#include <asm/xen/hypervisor.h> +#include <asm/xen/hypercall.h> +#include "common.h" + +/* + * These are rather arbitrary. They are fairly large because adjacent requests + * pulled from a communication ring are quite likely to end up being part of + * the same scatter/gather request at the disc. + * + * ** TRY INCREASING 'xen_blkif_reqs' IF WRITE SPEEDS SEEM TOO LOW ** + * + * This will increase the chances of being able to write whole tracks. + * 64 should be enough to keep us competitive with Linux. + */ +static int xen_blkif_reqs = 64; +module_param_named(reqs, xen_blkif_reqs, int, 0); +MODULE_PARM_DESC(reqs, "Number of blkback requests to allocate"); + +/* Run-time switchable: /sys/module/blkback/parameters/ */ +static unsigned int log_stats; +module_param(log_stats, int, 0644); + +/* + * Each outstanding request that we've passed to the lower device layers has a + * 'pending_req' allocated to it. Each buffer_head that completes decrements + * the pendcnt towards zero. When it hits zero, the specified domain has a + * response queued for it, with the saved 'id' passed back. + */ +struct pending_req { + struct xen_blkif *blkif; + u64 id; + int nr_pages; + atomic_t pendcnt; + unsigned short operation; + int status; + struct list_head free_list; +}; + +#define BLKBACK_INVALID_HANDLE (~0) + +struct xen_blkbk { + struct pending_req *pending_reqs; + /* List of all 'pending_req' available */ + struct list_head pending_free; + /* And its spinlock. */ + spinlock_t pending_free_lock; + wait_queue_head_t pending_free_wq; + /* The list of all pages that are available. */ + struct page **pending_pages; + /* And the grant handles that are available. */ + grant_handle_t *pending_grant_handles; +}; + +static struct xen_blkbk *blkbk; + +/* + * Little helpful macro to figure out the index and virtual address of the + * pending_pages[..]. For each 'pending_req' we have have up to + * BLKIF_MAX_SEGMENTS_PER_REQUEST (11) pages. The seg would be from 0 through + * 10 and would index in the pending_pages[..]. + */ +static inline int vaddr_pagenr(struct pending_req *req, int seg) +{ + return (req - blkbk->pending_reqs) * + BLKIF_MAX_SEGMENTS_PER_REQUEST + seg; +} + +#define pending_page(req, seg) pending_pages[vaddr_pagenr(req, seg)] + +static inline unsigned long vaddr(struct pending_req *req, int seg) +{ + unsigned long pfn = page_to_pfn(blkbk->pending_page(req, seg)); + return (unsigned long)pfn_to_kaddr(pfn); +} + +#define pending_handle(_req, _seg) \ + (blkbk->pending_grant_handles[vaddr_pagenr(_req, _seg)]) + + +static int do_block_io_op(struct xen_blkif *blkif); +static int dispatch_rw_block_io(struct xen_blkif *blkif, + struct blkif_request *req, + struct pending_req *pending_req); +static void make_response(struct xen_blkif *blkif, u64 id, + unsigned short op, int st); + +/* + * Retrieve from the 'pending_reqs' a free pending_req structure to be used. + */ +static struct pending_req *alloc_req(void) +{ + struct pending_req *req = NULL; + unsigned long flags; + + spin_lock_irqsave(&blkbk->pending_free_lock, flags); + if (!list_empty(&blkbk->pending_free)) { + req = list_entry(blkbk->pending_free.next, struct pending_req, + free_list); + list_del(&req->free_list); + } + spin_unlock_irqrestore(&blkbk->pending_free_lock, flags); + return req; +} + +/* + * Return the 'pending_req' structure back to the freepool. We also + * wake up the thread if it was waiting for a free page. + */ +static void free_req(struct pending_req *req) +{ + unsigned long flags; + int was_empty; + + spin_lock_irqsave(&blkbk->pending_free_lock, flags); + was_empty = list_empty(&blkbk->pending_free); + list_add(&req->free_list, &blkbk->pending_free); + spin_unlock_irqrestore(&blkbk->pending_free_lock, flags); + if (was_empty) + wake_up(&blkbk->pending_free_wq); +} + +/* + * Routines for managing virtual block devices (vbds). + */ +static int xen_vbd_translate(struct phys_req *req, struct xen_blkif *blkif, + int operation) +{ + struct xen_vbd *vbd = &blkif->vbd; + int rc = -EACCES; + + if ((operation != READ) && vbd->readonly) + goto out; + + if (likely(req->nr_sects)) { + blkif_sector_t end = req->sector_number + req->nr_sects; + + if (unlikely(end < req->sector_number)) + goto out; + if (unlikely(end > vbd_sz(vbd))) + goto out; + } + + req->dev = vbd->pdevice; + req->bdev = vbd->bdev; + rc = 0; + + out: + return rc; +} + +static void xen_vbd_resize(struct xen_blkif *blkif) +{ + struct xen_vbd *vbd = &blkif->vbd; + struct xenbus_transaction xbt; + int err; + struct xenbus_device *dev = xen_blkbk_xenbus(blkif->be); + unsigned long long new_size = vbd_sz(vbd); + + pr_info(DRV_PFX "VBD Resize: Domid: %d, Device: (%d, %d)\n", + blkif->domid, MAJOR(vbd->pdevice), MINOR(vbd->pdevice)); + pr_info(DRV_PFX "VBD Resize: new size %llu\n", new_size); + vbd->size = new_size; +again: + err = xenbus_transaction_start(&xbt); + if (err) { + pr_warn(DRV_PFX "Error starting transaction"); + return; + } + err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu", + (unsigned long long)vbd_sz(vbd)); + if (err) { + pr_warn(DRV_PFX "Error writing new size"); + goto abort; + } + /* + * Write the current state; we will use this to synchronize + * the front-end. If the current state is "connected" the + * front-end will get the new size information online. + */ + err = xenbus_printf(xbt, dev->nodename, "state", "%d", dev->state); + if (err) { + pr_warn(DRV_PFX "Error writing the state"); + goto abort; + } + + err = xenbus_transaction_end(xbt, 0); + if (err == -EAGAIN) + goto again; + if (err) + pr_warn(DRV_PFX "Error ending transaction"); + return; +abort: + xenbus_transaction_end(xbt, 1); +} + +/* + * Notification from the guest OS. + */ +static void blkif_notify_work(struct xen_blkif *blkif) +{ + blkif->waiting_reqs = 1; + wake_up(&blkif->wq); +} + +irqreturn_t xen_blkif_be_int(int irq, void *dev_id) +{ + blkif_notify_work(dev_id); + return IRQ_HANDLED; +} + +/* + * SCHEDULER FUNCTIONS + */ + +static void print_stats(struct xen_blkif *blkif) +{ + pr_info("xen-blkback (%s): oo %3d | rd %4d | wr %4d | f %4d\n", + current->comm, blkif->st_oo_req, + blkif->st_rd_req, blkif->st_wr_req, blkif->st_f_req); + blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000); + blkif->st_rd_req = 0; + blkif->st_wr_req = 0; + blkif->st_oo_req = 0; +} + +int xen_blkif_schedule(void *arg) +{ + struct xen_blkif *blkif = arg; + struct xen_vbd *vbd = &blkif->vbd; + + xen_blkif_get(blkif); + + while (!kthread_should_stop()) { + if (try_to_freeze()) + continue; + if (unlikely(vbd->size != vbd_sz(vbd))) + xen_vbd_resize(blkif); + + wait_event_interruptible( + blkif->wq, + blkif->waiting_reqs || kthread_should_stop()); + wait_event_interruptible( + blkbk->pending_free_wq, + !list_empty(&blkbk->pending_free) || + kthread_should_stop()); + + blkif->waiting_reqs = 0; + smp_mb(); /* clear flag *before* checking for work */ + + if (do_block_io_op(blkif)) + blkif->waiting_reqs = 1; + + if (log_stats && time_after(jiffies, blkif->st_print)) + print_stats(blkif); + } + + if (log_stats) + print_stats(blkif); + + blkif->xenblkd = NULL; + xen_blkif_put(blkif); + + return 0; +} + +struct seg_buf { + unsigned long buf; + unsigned int nsec; +}; +/* + * Unmap the grant references, and also remove the M2P over-rides + * used in the 'pending_req'. + */ +static void xen_blkbk_unmap(struct pending_req *req) +{ + struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST]; + unsigned int i, invcount = 0; + grant_handle_t handle; + int ret; + + for (i = 0; i < req->nr_pages; i++) { + handle = pending_handle(req, i); + if (handle == BLKBACK_INVALID_HANDLE) + continue; + gnttab_set_unmap_op(&unmap[invcount], vaddr(req, i), + GNTMAP_host_map, handle); + pending_handle(req, i) = BLKBACK_INVALID_HANDLE; + invcount++; + } + + ret = HYPERVISOR_grant_table_op( + GNTTABOP_unmap_grant_ref, unmap, invcount); + BUG_ON(ret); + /* + * Note, we use invcount, so nr->pages, so we can't index + * using vaddr(req, i). + */ + for (i = 0; i < invcount; i++) { + ret = m2p_remove_override( + virt_to_page(unmap[i].host_addr), false); + if (ret) { + pr_alert(DRV_PFX "Failed to remove M2P override for %lx\n", + (unsigned long)unmap[i].host_addr); + continue; + } + } +} + +static int xen_blkbk_map(struct blkif_request *req, + struct pending_req *pending_req, + struct seg_buf seg[]) +{ + struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST]; + int i; + int nseg = req->nr_segments; + int ret = 0; + + /* + * Fill out preq.nr_sects with proper amount of sectors, and setup + * assign map[..] with the PFN of the page in our domain with the + * corresponding grant reference for each page. + */ + for (i = 0; i < nseg; i++) { + uint32_t flags; + + flags = GNTMAP_host_map; + if (pending_req->operation != BLKIF_OP_READ) + flags |= GNTMAP_readonly; + gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags, + req->u.rw.seg[i].gref, + pending_req->blkif->domid); + } + + ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nseg); + BUG_ON(ret); + + /* + * Now swizzle the MFN in our domain with the MFN from the other domain + * so that when we access vaddr(pending_req,i) it has the contents of + * the page from the other domain. + */ + for (i = 0; i < nseg; i++) { + if (unlikely(map[i].status != 0)) { + pr_debug(DRV_PFX "invalid buffer -- could not remap it\n"); + map[i].handle = BLKBACK_INVALID_HANDLE; + ret |= 1; + } + + pending_handle(pending_req, i) = map[i].handle; + + if (ret) + continue; + + ret = m2p_add_override(PFN_DOWN(map[i].dev_bus_addr), + blkbk->pending_page(pending_req, i), false); + if (ret) { + pr_alert(DRV_PFX "Failed to install M2P override for %lx (ret: %d)\n", + (unsigned long)map[i].dev_bus_addr, ret); + /* We could switch over to GNTTABOP_copy */ + continue; + } + + seg[i].buf = map[i].dev_bus_addr | + (req->u.rw.seg[i].first_sect << 9); + } + return ret; +} + +/* + * Completion callback on the bio's. Called as bh->b_end_io() + */ + +static void __end_block_io_op(struct pending_req *pending_req, int error) +{ + /* An error fails the entire request. */ + if ((pending_req->operation == BLKIF_OP_FLUSH_DISKCACHE) && + (error == -EOPNOTSUPP)) { + pr_debug(DRV_PFX "flush diskcache op failed, not supported\n"); + xen_blkbk_flush_diskcache(XBT_NIL, pending_req->blkif->be, 0); + pending_req->status = BLKIF_RSP_EOPNOTSUPP; + } else if (error) { + pr_debug(DRV_PFX "Buffer not up-to-date at end of operation," + " error=%d\n", error); + pending_req->status = BLKIF_RSP_ERROR; + } + + /* + * If all of the bio's have completed it is time to unmap + * the grant references associated with 'request' and provide + * the proper response on the ring. + */ + if (atomic_dec_and_test(&pending_req->pendcnt)) { + xen_blkbk_unmap(pending_req); + make_response(pending_req->blkif, pending_req->id, + pending_req->operation, pending_req->status); + xen_blkif_put(pending_req->blkif); + free_req(pending_req); + } +} + +/* + * bio callback. + */ +static void end_block_io_op(struct bio *bio, int error) +{ + __end_block_io_op(bio->bi_private, error); + bio_put(bio); +} + + + +/* + * Function to copy the from the ring buffer the 'struct blkif_request' + * (which has the sectors we want, number of them, grant references, etc), + * and transmute it to the block API to hand it over to the proper block disk. + */ +static int do_block_io_op(struct xen_blkif *blkif) +{ + union blkif_back_rings *blk_rings = &blkif->blk_rings; + struct blkif_request req; + struct pending_req *pending_req; + RING_IDX rc, rp; + int more_to_do = 0; + + rc = blk_rings->common.req_cons; + rp = blk_rings->common.sring->req_prod; + rmb(); /* Ensure we see queued requests up to 'rp'. */ + + while (rc != rp) { + + if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc)) + break; + + if (kthread_should_stop()) { + more_to_do = 1; + break; + } + + pending_req = alloc_req(); + if (NULL == pending_req) { + blkif->st_oo_req++; + more_to_do = 1; + break; + } + + switch (blkif->blk_protocol) { + case BLKIF_PROTOCOL_NATIVE: + memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc), sizeof(req)); + break; + case BLKIF_PROTOCOL_X86_32: + blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc)); + break; + case BLKIF_PROTOCOL_X86_64: + blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc)); + break; + default: + BUG(); + } + blk_rings->common.req_cons = ++rc; /* before make_response() */ + + /* Apply all sanity checks to /private copy/ of request. */ + barrier(); + + if (dispatch_rw_block_io(blkif, &req, pending_req)) + break; + + /* Yield point for this unbounded loop. */ + cond_resched(); + } + + return more_to_do; +} + +/* + * Transmutation of the 'struct blkif_request' to a proper 'struct bio' + * and call the 'submit_bio' to pass it to the underlying storage. + */ +static int dispatch_rw_block_io(struct xen_blkif *blkif, + struct blkif_request *req, + struct pending_req *pending_req) +{ + struct phys_req preq; + struct seg_buf seg[BLKIF_MAX_SEGMENTS_PER_REQUEST]; + unsigned int nseg; + struct bio *bio = NULL; + struct bio *biolist[BLKIF_MAX_SEGMENTS_PER_REQUEST]; + int i, nbio = 0; + int operation; + struct blk_plug plug; + + switch (req->operation) { + case BLKIF_OP_READ: + blkif->st_rd_req++; + operation = READ; + break; + case BLKIF_OP_WRITE: + blkif->st_wr_req++; + operation = WRITE_ODIRECT; + break; + case BLKIF_OP_FLUSH_DISKCACHE: + blkif->st_f_req++; + operation = WRITE_FLUSH; + break; + case BLKIF_OP_WRITE_BARRIER: + default: + operation = 0; /* make gcc happy */ + goto fail_response; + break; + } + + /* Check that the number of segments is sane. */ + nseg = req->nr_segments; + if (unlikely(nseg == 0 && operation != WRITE_FLUSH) || + unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) { + pr_debug(DRV_PFX "Bad number of segments in request (%d)\n", + nseg); + /* Haven't submitted any bio's yet. */ + goto fail_response; + } + + preq.dev = req->handle; + preq.sector_number = req->u.rw.sector_number; + preq.nr_sects = 0; + + pending_req->blkif = blkif; + pending_req->id = req->id; + pending_req->operation = req->operation; + pending_req->status = BLKIF_RSP_OKAY; + pending_req->nr_pages = nseg; + + for (i = 0; i < nseg; i++) { + seg[i].nsec = req->u.rw.seg[i].last_sect - + req->u.rw.seg[i].first_sect + 1; + if ((req->u.rw.seg[i].last_sect >= (PAGE_SIZE >> 9)) || + (req->u.rw.seg[i].last_sect < req->u.rw.seg[i].first_sect)) + goto fail_response; + preq.nr_sects += seg[i].nsec; + + } + + if (xen_vbd_translate(&preq, blkif, operation) != 0) { + pr_debug(DRV_PFX "access denied: %s of [%llu,%llu] on dev=%04x\n", + operation == READ ? "read" : "write", + preq.sector_number, + preq.sector_number + preq.nr_sects, preq.dev); + goto fail_response; + } + + /* + * This check _MUST_ be done after xen_vbd_translate as the preq.bdev + * is set there. + */ + for (i = 0; i < nseg; i++) { + if (((int)preq.sector_number|(int)seg[i].nsec) & + ((bdev_logical_block_size(preq.bdev) >> 9) - 1)) { + pr_debug(DRV_PFX "Misaligned I/O request from domain %d", + blkif->domid); + goto fail_response; + } + } + + /* + * If we have failed at this point, we need to undo the M2P override, + * set gnttab_set_unmap_op on all of the grant references and perform + * the hypercall to unmap the grants - that is all done in + * xen_blkbk_unmap. + */ + if (xen_blkbk_map(req, pending_req, seg)) + goto fail_flush; + + /* This corresponding xen_blkif_put is done in __end_block_io_op */ + xen_blkif_get(blkif); + + for (i = 0; i < nseg; i++) { + while ((bio == NULL) || + (bio_add_page(bio, + blkbk->pending_page(pending_req, i), + seg[i].nsec << 9, + seg[i].buf & ~PAGE_MASK) == 0)) { + + bio = bio_alloc(GFP_KERNEL, nseg-i); + if (unlikely(bio == NULL)) + goto fail_put_bio; + + biolist[nbio++] = bio; + bio->bi_bdev = preq.bdev; + bio->bi_private = pending_req; + bio->bi_end_io = end_block_io_op; + bio->bi_sector = preq.sector_number; + } + + preq.sector_number += seg[i].nsec; + } + + /* This will be hit if the operation was a flush. */ + if (!bio) { + BUG_ON(operation != WRITE_FLUSH); + + bio = bio_alloc(GFP_KERNEL, 0); + if (unlikely(bio == NULL)) + goto fail_put_bio; + + biolist[nbio++] = bio; + bio->bi_bdev = preq.bdev; + bio->bi_private = pending_req; + bio->bi_end_io = end_block_io_op; + } + + /* + * We set it one so that the last submit_bio does not have to call + * atomic_inc. + */ + atomic_set(&pending_req->pendcnt, nbio); + + /* Get a reference count for the disk queue and start sending I/O */ + blk_start_plug(&plug); + + for (i = 0; i < nbio; i++) + submit_bio(operation, biolist[i]); + + /* Let the I/Os go.. */ + blk_finish_plug(&plug); + + if (operation == READ) + blkif->st_rd_sect += preq.nr_sects; + else if (operation == WRITE || operation == WRITE_FLUSH) + blkif->st_wr_sect += preq.nr_sects; + + return 0; + + fail_flush: + xen_blkbk_unmap(pending_req); + fail_response: + /* Haven't submitted any bio's yet. */ + make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR); + free_req(pending_req); + msleep(1); /* back off a bit */ + return -EIO; + + fail_put_bio: + for (i = 0; i < nbio; i++) + bio_put(biolist[i]); + __end_block_io_op(pending_req, -EINVAL); + msleep(1); /* back off a bit */ + return -EIO; +} + + + +/* + * Put a response on the ring on how the operation fared. + */ +static void make_response(struct xen_blkif *blkif, u64 id, + unsigned short op, int st) +{ + struct blkif_response resp; + unsigned long flags; + union blkif_back_rings *blk_rings = &blkif->blk_rings; + int more_to_do = 0; + int notify; + + resp.id = id; + resp.operation = op; + resp.status = st; + + spin_lock_irqsave(&blkif->blk_ring_lock, flags); + /* Place on the response ring for the relevant domain. */ + switch (blkif->blk_protocol) { + case BLKIF_PROTOCOL_NATIVE: + memcpy(RING_GET_RESPONSE(&blk_rings->native, blk_rings->native.rsp_prod_pvt), + &resp, sizeof(resp)); + break; + case BLKIF_PROTOCOL_X86_32: + memcpy(RING_GET_RESPONSE(&blk_rings->x86_32, blk_rings->x86_32.rsp_prod_pvt), + &resp, sizeof(resp)); + break; + case BLKIF_PROTOCOL_X86_64: + memcpy(RING_GET_RESPONSE(&blk_rings->x86_64, blk_rings->x86_64.rsp_prod_pvt), + &resp, sizeof(resp)); + break; + default: + BUG(); + } + blk_rings->common.rsp_prod_pvt++; + RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify); + if (blk_rings->common.rsp_prod_pvt == blk_rings->common.req_cons) { + /* + * Tail check for pending requests. Allows frontend to avoid + * notifications if requests are already in flight (lower + * overheads and promotes batching). + */ + RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do); + + } else if (RING_HAS_UNCONSUMED_REQUESTS(&blk_rings->common)) { + more_to_do = 1; + } + + spin_unlock_irqrestore(&blkif->blk_ring_lock, flags); + + if (more_to_do) + blkif_notify_work(blkif); + if (notify) + notify_remote_via_irq(blkif->irq); +} + +static int __init xen_blkif_init(void) +{ + int i, mmap_pages; + int rc = 0; + + if (!xen_pv_domain()) + return -ENODEV; + + blkbk = kzalloc(sizeof(struct xen_blkbk), GFP_KERNEL); + if (!blkbk) { + pr_alert(DRV_PFX "%s: out of memory!\n", __func__); + return -ENOMEM; + } + + mmap_pages = xen_blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST; + + blkbk->pending_reqs = kmalloc(sizeof(blkbk->pending_reqs[0]) * + xen_blkif_reqs, GFP_KERNEL); + blkbk->pending_grant_handles = kzalloc(sizeof(blkbk->pending_grant_handles[0]) * + mmap_pages, GFP_KERNEL); + blkbk->pending_pages = kzalloc(sizeof(blkbk->pending_pages[0]) * + mmap_pages, GFP_KERNEL); + + if (!blkbk->pending_reqs || !blkbk->pending_grant_handles || + !blkbk->pending_pages) { + rc = -ENOMEM; + goto out_of_memory; + } + + for (i = 0; i < mmap_pages; i++) { + blkbk->pending_grant_handles[i] = BLKBACK_INVALID_HANDLE; + blkbk->pending_pages[i] = alloc_page(GFP_KERNEL); + if (blkbk->pending_pages[i] == NULL) { + rc = -ENOMEM; + goto out_of_memory; + } + } + rc = xen_blkif_interface_init(); + if (rc) + goto failed_init; + + memset(blkbk->pending_reqs, 0, sizeof(blkbk->pending_reqs)); + + INIT_LIST_HEAD(&blkbk->pending_free); + spin_lock_init(&blkbk->pending_free_lock); + init_waitqueue_head(&blkbk->pending_free_wq); + + for (i = 0; i < xen_blkif_reqs; i++) + list_add_tail(&blkbk->pending_reqs[i].free_list, + &blkbk->pending_free); + + rc = xen_blkif_xenbus_init(); + if (rc) + goto failed_init; + + return 0; + + out_of_memory: + pr_alert(DRV_PFX "%s: out of memory\n", __func__); + failed_init: + kfree(blkbk->pending_reqs); + kfree(blkbk->pending_grant_handles); + for (i = 0; i < mmap_pages; i++) { + if (blkbk->pending_pages[i]) + __free_page(blkbk->pending_pages[i]); + } + kfree(blkbk->pending_pages); + kfree(blkbk); + blkbk = NULL; + return rc; +} + +module_init(xen_blkif_init); + +MODULE_LICENSE("Dual BSD/GPL"); diff --git a/drivers/block/xen-blkback/common.h b/drivers/block/xen-blkback/common.h new file mode 100644 index 0000000..9e40b28 --- /dev/null +++ b/drivers/block/xen-blkback/common.h @@ -0,0 +1,233 @@ +/* + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version 2 + * as published by the Free Software Foundation; or, when distributed + * separately from the Linux kernel or incorporated into other + * software packages, subject to the following license: + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this source file (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, copy, modify, + * merge, publish, distribute, sublicense, and/or sell copies of the Software, + * and to permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ + +#ifndef __XEN_BLKIF__BACKEND__COMMON_H__ +#define __XEN_BLKIF__BACKEND__COMMON_H__ + +#include <linux/version.h> +#include <linux/module.h> +#include <linux/interrupt.h> +#include <linux/slab.h> +#include <linux/blkdev.h> +#include <linux/vmalloc.h> +#include <linux/wait.h> +#include <linux/io.h> +#include <asm/setup.h> +#include <asm/pgalloc.h> +#include <asm/hypervisor.h> +#include <xen/grant_table.h> +#include <xen/xenbus.h> +#include <xen/interface/io/ring.h> +#include <xen/interface/io/blkif.h> +#include <xen/interface/io/protocols.h> + +#define DRV_PFX "xen-blkback:" +#define DPRINTK(fmt, args...) \ + pr_debug(DRV_PFX "(%s:%d) " fmt ".\n", \ + __func__, __LINE__, ##args) + + +/* Not a real protocol. Used to generate ring structs which contain + * the elements common to all protocols only. This way we get a + * compiler-checkable way to use common struct elements, so we can + * avoid using switch(protocol) in a number of places. */ +struct blkif_common_request { + char dummy; +}; +struct blkif_common_response { + char dummy; +}; + +/* i386 protocol version */ +#pragma pack(push, 4) +struct blkif_x86_32_request { + uint8_t operation; /* BLKIF_OP_??? */ + uint8_t nr_segments; /* number of segments */ + blkif_vdev_t handle; /* only for read/write requests */ + uint64_t id; /* private guest value, echoed in resp */ + blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ + struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST]; +}; +struct blkif_x86_32_response { + uint64_t id; /* copied from request */ + uint8_t operation; /* copied from request */ + int16_t status; /* BLKIF_RSP_??? */ +}; +#pragma pack(pop) + +/* x86_64 protocol version */ +struct blkif_x86_64_request { + uint8_t operation; /* BLKIF_OP_??? */ + uint8_t nr_segments; /* number of segments */ + blkif_vdev_t handle; /* only for read/write requests */ + uint64_t __attribute__((__aligned__(8))) id; + blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ + struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST]; +}; +struct blkif_x86_64_response { + uint64_t __attribute__((__aligned__(8))) id; + uint8_t operation; /* copied from request */ + int16_t status; /* BLKIF_RSP_??? */ +}; + +DEFINE_RING_TYPES(blkif_common, struct blkif_common_request, + struct blkif_common_response); +DEFINE_RING_TYPES(blkif_x86_32, struct blkif_x86_32_request, + struct blkif_x86_32_response); +DEFINE_RING_TYPES(blkif_x86_64, struct blkif_x86_64_request, + struct blkif_x86_64_response); + +union blkif_back_rings { + struct blkif_back_ring native; + struct blkif_common_back_ring common; + struct blkif_x86_32_back_ring x86_32; + struct blkif_x86_64_back_ring x86_64; +}; + +enum blkif_protocol { + BLKIF_PROTOCOL_NATIVE = 1, + BLKIF_PROTOCOL_X86_32 = 2, + BLKIF_PROTOCOL_X86_64 = 3, +}; + +struct xen_vbd { + /* What the domain refers to this vbd as. */ + blkif_vdev_t handle; + /* Non-zero -> read-only */ + unsigned char readonly; + /* VDISK_xxx */ + unsigned char type; + /* phys device that this vbd maps to. */ + u32 pdevice; + struct block_device *bdev; + /* Cached size parameter. */ + sector_t size; + bool flush_support; +}; + +struct backend_info; + +struct xen_blkif { + /* Unique identifier for this interface. */ + domid_t domid; + unsigned int handle; + /* Physical parameters of the comms window. */ + unsigned int irq; + /* Comms information. */ + enum blkif_protocol blk_protocol; + union blkif_back_rings blk_rings; + struct vm_struct *blk_ring_area; + /* The VBD attached to this interface. */ + struct xen_vbd vbd; + /* Back pointer to the backend_info. */ + struct backend_info *be; + /* Private fields. */ + spinlock_t blk_ring_lock; + atomic_t refcnt; + + wait_queue_head_t wq; + /* One thread per one blkif. */ + struct task_struct *xenblkd; + unsigned int waiting_reqs; + + /* statistics */ + unsigned long st_print; + int st_rd_req; + int st_wr_req; + int st_oo_req; + int st_f_req; + int st_rd_sect; + int st_wr_sect; + + wait_queue_head_t waiting_to_free; + + grant_handle_t shmem_handle; + grant_ref_t shmem_ref; +}; + + +#define vbd_sz(_v) ((_v)->bdev->bd_part ? \ + (_v)->bdev->bd_part->nr_sects : \ + get_capacity((_v)->bdev->bd_disk)) + +#define xen_blkif_get(_b) (atomic_inc(&(_b)->refcnt)) +#define xen_blkif_put(_b) \ + do { \ + if (atomic_dec_and_test(&(_b)->refcnt)) \ + wake_up(&(_b)->waiting_to_free);\ + } while (0) + +struct phys_req { + unsigned short dev; + unsigned short nr_sects; + struct block_device *bdev; + blkif_sector_t sector_number; +}; +int xen_blkif_interface_init(void); + +int xen_blkif_xenbus_init(void); + +irqreturn_t xen_blkif_be_int(int irq, void *dev_id); +int xen_blkif_schedule(void *arg); + +int xen_blkbk_flush_diskcache(struct xenbus_transaction xbt, + struct backend_info *be, int state); + +struct xenbus_device *xen_blkbk_xenbus(struct backend_info *be); + +static inline void blkif_get_x86_32_req(struct blkif_request *dst, + struct blkif_x86_32_request *src) +{ + int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST; + dst->operation = src->operation; + dst->nr_segments = src->nr_segments; + dst->handle = src->handle; + dst->id = src->id; + dst->u.rw.sector_number = src->sector_number; + barrier(); + if (n > dst->nr_segments) + n = dst->nr_segments; + for (i = 0; i < n; i++) + dst->u.rw.seg[i] = src->seg[i]; +} + +static inline void blkif_get_x86_64_req(struct blkif_request *dst, + struct blkif_x86_64_request *src) +{ + int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST; + dst->operation = src->operation; + dst->nr_segments = src->nr_segments; + dst->handle = src->handle; + dst->id = src->id; + dst->u.rw.sector_number = src->sector_number; + barrier(); + if (n > dst->nr_segments) + n = dst->nr_segments; + for (i = 0; i < n; i++) + dst->u.rw.seg[i] = src->seg[i]; +} + +#endif /* __XEN_BLKIF__BACKEND__COMMON_H__ */ diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c new file mode 100644 index 0000000..3457082 --- /dev/null +++ b/drivers/block/xen-blkback/xenbus.c @@ -0,0 +1,768 @@ +/* Xenbus code for blkif backend + Copyright (C) 2005 Rusty Russell <rusty@rustcorp.com.au> + Copyright (C) 2005 XenSource Ltd + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + +*/ + +#include <stdarg.h> +#include <linux/module.h> +#include <linux/kthread.h> +#include <xen/events.h> +#include <xen/grant_table.h> +#include "common.h" + +struct backend_info { + struct xenbus_device *dev; + struct xen_blkif *blkif; + struct xenbus_watch backend_watch; + unsigned major; + unsigned minor; + char *mode; +}; + +static struct kmem_cache *xen_blkif_cachep; +static void connect(struct backend_info *); +static int connect_ring(struct backend_info *); +static void backend_changed(struct xenbus_watch *, const char **, + unsigned int); + +struct xenbus_device *xen_blkbk_xenbus(struct backend_info *be) +{ + return be->dev; +} + +static int blkback_name(struct xen_blkif *blkif, char *buf) +{ + char *devpath, *devname; + struct xenbus_device *dev = blkif->be->dev; + + devpath = xenbus_read(XBT_NIL, dev->nodename, "dev", NULL); + if (IS_ERR(devpath)) + return PTR_ERR(devpath); + + devname = strstr(devpath, "/dev/"); + if (devname != NULL) + devname += strlen("/dev/"); + else + devname = devpath; + + snprintf(buf, TASK_COMM_LEN, "blkback.%d.%s", blkif->domid, devname); + kfree(devpath); + + return 0; +} + +static void xen_update_blkif_status(struct xen_blkif *blkif) +{ + int err; + char name[TASK_COMM_LEN]; + + /* Not ready to connect? */ + if (!blkif->irq || !blkif->vbd.bdev) + return; + + /* Already connected? */ + if (blkif->be->dev->state == XenbusStateConnected) + return; + + /* Attempt to connect: exit if we fail to. */ + connect(blkif->be); + if (blkif->be->dev->state != XenbusStateConnected) + return; + + err = blkback_name(blkif, name); + if (err) { + xenbus_dev_error(blkif->be->dev, err, "get blkback dev name"); + return; + } + + err = filemap_write_and_wait(blkif->vbd.bdev->bd_inode->i_mapping); + if (err) { + xenbus_dev_error(blkif->be->dev, err, "block flush"); + return; + } + invalidate_inode_pages2(blkif->vbd.bdev->bd_inode->i_mapping); + + blkif->xenblkd = kthread_run(xen_blkif_schedule, blkif, name); + if (IS_ERR(blkif->xenblkd)) { + err = PTR_ERR(blkif->xenblkd); + blkif->xenblkd = NULL; + xenbus_dev_error(blkif->be->dev, err, "start xenblkd"); + } +} + +static struct xen_blkif *xen_blkif_alloc(domid_t domid) +{ + struct xen_blkif *blkif; + + blkif = kmem_cache_alloc(xen_blkif_cachep, GFP_KERNEL); + if (!blkif) + return ERR_PTR(-ENOMEM); + + memset(blkif, 0, sizeof(*blkif)); + blkif->domid = domid; + spin_lock_init(&blkif->blk_ring_lock); + atomic_set(&blkif->refcnt, 1); + init_waitqueue_head(&blkif->wq); + blkif->st_print = jiffies; + init_waitqueue_head(&blkif->waiting_to_free); + + return blkif; +} + +static int map_frontend_page(struct xen_blkif *blkif, unsigned long shared_page) +{ + struct gnttab_map_grant_ref op; + + gnttab_set_map_op(&op, (unsigned long)blkif->blk_ring_area->addr, + GNTMAP_host_map, shared_page, blkif->domid); + + if (HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1)) + BUG(); + + if (op.status) { + DPRINTK("Grant table operation failure !\n"); + return op.status; + } + + blkif->shmem_ref = shared_page; + blkif->shmem_handle = op.handle; + + return 0; +} + +static void unmap_frontend_page(struct xen_blkif *blkif) +{ + struct gnttab_unmap_grant_ref op; + + gnttab_set_unmap_op(&op, (unsigned long)blkif->blk_ring_area->addr, + GNTMAP_host_map, blkif->shmem_handle); + + if (HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1)) + BUG(); +} + +static int xen_blkif_map(struct xen_blkif *blkif, unsigned long shared_page, + unsigned int evtchn) +{ + int err; + + /* Already connected through? */ + if (blkif->irq) + return 0; + + blkif->blk_ring_area = alloc_vm_area(PAGE_SIZE); + if (!blkif->blk_ring_area) + return -ENOMEM; + + err = map_frontend_page(blkif, shared_page); + if (err) { + free_vm_area(blkif->blk_ring_area); + return err; + } + + switch (blkif->blk_protocol) { + case BLKIF_PROTOCOL_NATIVE: + { + struct blkif_sring *sring; + sring = (struct blkif_sring *)blkif->blk_ring_area->addr; + BACK_RING_INIT(&blkif->blk_rings.native, sring, PAGE_SIZE); + break; + } + case BLKIF_PROTOCOL_X86_32: + { + struct blkif_x86_32_sring *sring_x86_32; + sring_x86_32 = (struct blkif_x86_32_sring *)blkif->blk_ring_area->addr; + BACK_RING_INIT(&blkif->blk_rings.x86_32, sring_x86_32, PAGE_SIZE); + break; + } + case BLKIF_PROTOCOL_X86_64: + { + struct blkif_x86_64_sring *sring_x86_64; + sring_x86_64 = (struct blkif_x86_64_sring *)blkif->blk_ring_area->addr; + BACK_RING_INIT(&blkif->blk_rings.x86_64, sring_x86_64, PAGE_SIZE); + break; + } + default: + BUG(); + } + + err = bind_interdomain_evtchn_to_irqhandler(blkif->domid, evtchn, + xen_blkif_be_int, 0, + "blkif-backend", blkif); + if (err < 0) { + unmap_frontend_page(blkif); + free_vm_area(blkif->blk_ring_area); + blkif->blk_rings.common.sring = NULL; + return err; + } + blkif->irq = err; + + return 0; +} + +static void xen_blkif_disconnect(struct xen_blkif *blkif) +{ + if (blkif->xenblkd) { + kthread_stop(blkif->xenblkd); + blkif->xenblkd = NULL; + } + + atomic_dec(&blkif->refcnt); + wait_event(blkif->waiting_to_free, atomic_read(&blkif->refcnt) == 0); + atomic_inc(&blkif->refcnt); + + if (blkif->irq) { + unbind_from_irqhandler(blkif->irq, blkif); + blkif->irq = 0; + } + + if (blkif->blk_rings.common.sring) { + unmap_frontend_page(blkif); + free_vm_area(blkif->blk_ring_area); + blkif->blk_rings.common.sring = NULL; + } +} + +void xen_blkif_free(struct xen_blkif *blkif) +{ + if (!atomic_dec_and_test(&blkif->refcnt)) + BUG(); + kmem_cache_free(xen_blkif_cachep, blkif); +} + +int __init xen_blkif_interface_init(void) +{ + xen_blkif_cachep = kmem_cache_create("blkif_cache", + sizeof(struct xen_blkif), + 0, 0, NULL); + if (!xen_blkif_cachep) + return -ENOMEM; + + return 0; +} + +/* + * sysfs interface for VBD I/O requests + */ + +#define VBD_SHOW(name, format, args...) \ + static ssize_t show_##name(struct device *_dev, \ + struct device_attribute *attr, \ + char *buf) \ + { \ + struct xenbus_device *dev = to_xenbus_device(_dev); \ + struct backend_info *be = dev_get_drvdata(&dev->dev); \ + \ + return sprintf(buf, format, ##args); \ + } \ + static DEVICE_ATTR(name, S_IRUGO, show_##name, NULL) + +VBD_SHOW(oo_req, "%d\n", be->blkif->st_oo_req); +VBD_SHOW(rd_req, "%d\n", be->blkif->st_rd_req); +VBD_SHOW(wr_req, "%d\n", be->blkif->st_wr_req); +VBD_SHOW(f_req, "%d\n", be->blkif->st_f_req); +VBD_SHOW(rd_sect, "%d\n", be->blkif->st_rd_sect); +VBD_SHOW(wr_sect, "%d\n", be->blkif->st_wr_sect); + +static struct attribute *xen_vbdstat_attrs[] = { + &dev_attr_oo_req.attr, + &dev_attr_rd_req.attr, + &dev_attr_wr_req.attr, + &dev_attr_f_req.attr, + &dev_attr_rd_sect.attr, + &dev_attr_wr_sect.attr, + NULL +}; + +static struct attribute_group xen_vbdstat_group = { + .name = "statistics", + .attrs = xen_vbdstat_attrs, +}; + +VBD_SHOW(physical_device, "%x:%x\n", be->major, be->minor); +VBD_SHOW(mode, "%s\n", be->mode); + +int xenvbd_sysfs_addif(struct xenbus_device *dev) +{ + int error; + + error = device_create_file(&dev->dev, &dev_attr_physical_device); + if (error) + goto fail1; + + error = device_create_file(&dev->dev, &dev_attr_mode); + if (error) + goto fail2; + + error = sysfs_create_group(&dev->dev.kobj, &xen_vbdstat_group); + if (error) + goto fail3; + + return 0; + +fail3: sysfs_remove_group(&dev->dev.kobj, &xen_vbdstat_group); +fail2: device_remove_file(&dev->dev, &dev_attr_mode); +fail1: device_remove_file(&dev->dev, &dev_attr_physical_device); + return error; +} + +void xenvbd_sysfs_delif(struct xenbus_device *dev) +{ + sysfs_remove_group(&dev->dev.kobj, &xen_vbdstat_group); + device_remove_file(&dev->dev, &dev_attr_mode); + device_remove_file(&dev->dev, &dev_attr_physical_device); +} + + +static void xen_vbd_free(struct xen_vbd *vbd) +{ + if (vbd->bdev) + blkdev_put(vbd->bdev, vbd->readonly ? FMODE_READ : FMODE_WRITE); + vbd->bdev = NULL; +} + +static int xen_vbd_create(struct xen_blkif *blkif, blkif_vdev_t handle, + unsigned major, unsigned minor, int readonly, + int cdrom) +{ + struct xen_vbd *vbd; + struct block_device *bdev; + struct request_queue *q; + + vbd = &blkif->vbd; + vbd->handle = handle; + vbd->readonly = readonly; + vbd->type = 0; + + vbd->pdevice = MKDEV(major, minor); + + bdev = blkdev_get_by_dev(vbd->pdevice, vbd->readonly ? + FMODE_READ : FMODE_WRITE, NULL); + + if (IS_ERR(bdev)) { + DPRINTK("xen_vbd_create: device %08x could not be opened.\n", + vbd->pdevice); + return -ENOENT; + } + + vbd->bdev = bdev; + vbd->size = vbd_sz(vbd); + + if (vbd->bdev->bd_disk == NULL) { + DPRINTK("xen_vbd_create: device %08x doesn't exist.\n", + vbd->pdevice); + xen_vbd_free(vbd); + return -ENOENT; + } + + if (vbd->bdev->bd_disk->flags & GENHD_FL_CD || cdrom) + vbd->type |= VDISK_CDROM; + if (vbd->bdev->bd_disk->flags & GENHD_FL_REMOVABLE) + vbd->type |= VDISK_REMOVABLE; + + q = bdev_get_queue(bdev); + if (q && q->flush_flags) + vbd->flush_support = true; + + DPRINTK("Successful creation of handle=%04x (dom=%u)\n", + handle, blkif->domid); + return 0; +} +static int xen_blkbk_remove(struct xenbus_device *dev) +{ + struct backend_info *be = dev_get_drvdata(&dev->dev); + + DPRINTK(""); + + if (be->major || be->minor) + xenvbd_sysfs_delif(dev); + + if (be->backend_watch.node) { + unregister_xenbus_watch(&be->backend_watch); + kfree(be->backend_watch.node); + be->backend_watch.node = NULL; + } + + if (be->blkif) { + xen_blkif_disconnect(be->blkif); + xen_vbd_free(&be->blkif->vbd); + xen_blkif_free(be->blkif); + be->blkif = NULL; + } + + kfree(be); + dev_set_drvdata(&dev->dev, NULL); + return 0; +} + +int xen_blkbk_flush_diskcache(struct xenbus_transaction xbt, + struct backend_info *be, int state) +{ + struct xenbus_device *dev = be->dev; + int err; + + err = xenbus_printf(xbt, dev->nodename, "feature-flush-cache", + "%d", state); + if (err) + xenbus_dev_fatal(dev, err, "writing feature-flush-cache"); + + return err; +} + +/* + * Entry point to this code when a new device is created. Allocate the basic + * structures, and watch the store waiting for the hotplug scripts to tell us + * the device's physical major and minor numbers. Switch to InitWait. + */ +static int xen_blkbk_probe(struct xenbus_device *dev, + const struct xenbus_device_id *id) +{ + int err; + struct backend_info *be = kzalloc(sizeof(struct backend_info), + GFP_KERNEL); + if (!be) { + xenbus_dev_fatal(dev, -ENOMEM, + "allocating backend structure"); + return -ENOMEM; + } + be->dev = dev; + dev_set_drvdata(&dev->dev, be); + + be->blkif = xen_blkif_alloc(dev->otherend_id); + if (IS_ERR(be->blkif)) { + err = PTR_ERR(be->blkif); + be->blkif = NULL; + xenbus_dev_fatal(dev, err, "creating block interface"); + goto fail; + } + + /* setup back pointer */ + be->blkif->be = be; + + err = xenbus_watch_pathfmt(dev, &be->backend_watch, backend_changed, + "%s/%s", dev->nodename, "physical-device"); + if (err) + goto fail; + + err = xenbus_switch_state(dev, XenbusStateInitWait); + if (err) + goto fail; + + return 0; + +fail: + DPRINTK("failed"); + xen_blkbk_remove(dev); + return err; +} + + +/* + * Callback received when the hotplug scripts have placed the physical-device + * node. Read it and the mode node, and create a vbd. If the frontend is + * ready, connect. + */ +static void backend_changed(struct xenbus_watch *watch, + const char **vec, unsigned int len) +{ + int err; + unsigned major; + unsigned minor; + struct backend_info *be + = container_of(watch, struct backend_info, backend_watch); + struct xenbus_device *dev = be->dev; + int cdrom = 0; + char *device_type; + + DPRINTK(""); + + err = xenbus_scanf(XBT_NIL, dev->nodename, "physical-device", "%x:%x", + &major, &minor); + if (XENBUS_EXIST_ERR(err)) { + /* + * Since this watch will fire once immediately after it is + * registered, we expect this. Ignore it, and wait for the + * hotplug scripts. + */ + return; + } + if (err != 2) { + xenbus_dev_fatal(dev, err, "reading physical-device"); + return; + } + + if ((be->major || be->minor) && + ((be->major != major) || (be->minor != minor))) { + pr_warn(DRV_PFX "changing physical device (from %x:%x to %x:%x) not supported.\n", + be->major, be->minor, major, minor); + return; + } + + be->mode = xenbus_read(XBT_NIL, dev->nodename, "mode", NULL); + if (IS_ERR(be->mode)) { + err = PTR_ERR(be->mode); + be->mode = NULL; + xenbus_dev_fatal(dev, err, "reading mode"); + return; + } + + device_type = xenbus_read(XBT_NIL, dev->otherend, "device-type", NULL); + if (!IS_ERR(device_type)) { + cdrom = strcmp(device_type, "cdrom") == 0; + kfree(device_type); + } + + if (be->major == 0 && be->minor == 0) { + /* Front end dir is a number, which is used as the handle. */ + + char *p = strrchr(dev->otherend, '/') + 1; + long handle; + err = strict_strtoul(p, 0, &handle); + if (err) + return; + + be->major = major; + be->minor = minor; + + err = xen_vbd_create(be->blkif, handle, major, minor, + (NULL == strchr(be->mode, 'w')), cdrom); + if (err) { + be->major = 0; + be->minor = 0; + xenbus_dev_fatal(dev, err, "creating vbd structure"); + return; + } + + err = xenvbd_sysfs_addif(dev); + if (err) { + xen_vbd_free(&be->blkif->vbd); + be->major = 0; + be->minor = 0; + xenbus_dev_fatal(dev, err, "creating sysfs entries"); + return; + } + + /* We're potentially connected now */ + xen_update_blkif_status(be->blkif); + } +} + + +/* + * Callback received when the frontend's state changes. + */ +static void frontend_changed(struct xenbus_device *dev, + enum xenbus_state frontend_state) +{ + struct backend_info *be = dev_get_drvdata(&dev->dev); + int err; + + DPRINTK("%s", xenbus_strstate(frontend_state)); + + switch (frontend_state) { + case XenbusStateInitialising: + if (dev->state == XenbusStateClosed) { + pr_info(DRV_PFX "%s: prepare for reconnect\n", + dev->nodename); + xenbus_switch_state(dev, XenbusStateInitWait); + } + break; + + case XenbusStateInitialised: + case XenbusStateConnected: + /* + * Ensure we connect even when two watches fire in + * close successsion and we miss the intermediate value + * of frontend_state. + */ + if (dev->state == XenbusStateConnected) + break; + + /* + * Enforce precondition before potential leak point. + * blkif_disconnect() is idempotent. + */ + xen_blkif_disconnect(be->blkif); + + err = connect_ring(be); + if (err) + break; + xen_update_blkif_status(be->blkif); + break; + + case XenbusStateClosing: + xen_blkif_disconnect(be->blkif); + xenbus_switch_state(dev, XenbusStateClosing); + break; + + case XenbusStateClosed: + xenbus_switch_state(dev, XenbusStateClosed); + if (xenbus_dev_is_online(dev)) + break; + /* fall through if not online */ + case XenbusStateUnknown: + /* implies blkif_disconnect() via blkback_remove() */ + device_unregister(&dev->dev); + break; + + default: + xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend", + frontend_state); + break; + } +} + + +/* ** Connection ** */ + + +/* + * Write the physical details regarding the block device to the store, and + * switch to Connected state. + */ +static void connect(struct backend_info *be) +{ + struct xenbus_transaction xbt; + int err; + struct xenbus_device *dev = be->dev; + + DPRINTK("%s", dev->otherend); + + /* Supply the information about the device the frontend needs */ +again: + err = xenbus_transaction_start(&xbt); + if (err) { + xenbus_dev_fatal(dev, err, "starting transaction"); + return; + } + + err = xen_blkbk_flush_diskcache(xbt, be, be->blkif->vbd.flush_support); + if (err) + goto abort; + + err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu", + (unsigned long long)vbd_sz(&be->blkif->vbd)); + if (err) { + xenbus_dev_fatal(dev, err, "writing %s/sectors", + dev->nodename); + goto abort; + } + + /* FIXME: use a typename instead */ + err = xenbus_printf(xbt, dev->nodename, "info", "%u", + be->blkif->vbd.type | + (be->blkif->vbd.readonly ? VDISK_READONLY : 0)); + if (err) { + xenbus_dev_fatal(dev, err, "writing %s/info", + dev->nodename); + goto abort; + } + err = xenbus_printf(xbt, dev->nodename, "sector-size", "%lu", + (unsigned long) + bdev_logical_block_size(be->blkif->vbd.bdev)); + if (err) { + xenbus_dev_fatal(dev, err, "writing %s/sector-size", + dev->nodename); + goto abort; + } + + err = xenbus_transaction_end(xbt, 0); + if (err == -EAGAIN) + goto again; + if (err) + xenbus_dev_fatal(dev, err, "ending transaction"); + + err = xenbus_switch_state(dev, XenbusStateConnected); + if (err) + xenbus_dev_fatal(dev, err, "switching to Connected state", + dev->nodename); + + return; + abort: + xenbus_transaction_end(xbt, 1); +} + + +static int connect_ring(struct backend_info *be) +{ + struct xenbus_device *dev = be->dev; + unsigned long ring_ref; + unsigned int evtchn; + char protocol[64] = ""; + int err; + + DPRINTK("%s", dev->otherend); + + err = xenbus_gather(XBT_NIL, dev->otherend, "ring-ref", "%lu", + &ring_ref, "event-channel", "%u", &evtchn, NULL); + if (err) { + xenbus_dev_fatal(dev, err, + "reading %s/ring-ref and event-channel", + dev->otherend); + return err; + } + + be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE; + err = xenbus_gather(XBT_NIL, dev->otherend, "protocol", + "%63s", protocol, NULL); + if (err) + strcpy(protocol, "unspecified, assuming native"); + else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_NATIVE)) + be->blkif->blk_protocol = BLKIF_PROTOCOL_NATIVE; + else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_32)) + be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_32; + else if (0 == strcmp(protocol, XEN_IO_PROTO_ABI_X86_64)) + be->blkif->blk_protocol = BLKIF_PROTOCOL_X86_64; + else { + xenbus_dev_fatal(dev, err, "unknown fe protocol %s", protocol); + return -1; + } + pr_info(DRV_PFX "ring-ref %ld, event-channel %d, protocol %d (%s)\n", + ring_ref, evtchn, be->blkif->blk_protocol, protocol); + + /* Map the shared frame, irq etc. */ + err = xen_blkif_map(be->blkif, ring_ref, evtchn); + if (err) { + xenbus_dev_fatal(dev, err, "mapping ring-ref %lu port %u", + ring_ref, evtchn); + return err; + } + + return 0; +} + + +/* ** Driver Registration ** */ + + +static const struct xenbus_device_id xen_blkbk_ids[] = { + { "vbd" }, + { "" } +}; + + +static struct xenbus_driver xen_blkbk = { + .name = "vbd", + .owner = THIS_MODULE, + .ids = xen_blkbk_ids, + .probe = xen_blkbk_probe, + .remove = xen_blkbk_remove, + .otherend_changed = frontend_changed +}; + + +int xen_blkif_xenbus_init(void) +{ + return xenbus_register_backend(&xen_blkbk); +} diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index d7aa39e..b536a9c 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -97,6 +97,7 @@ struct blkfront_info struct blk_shadow shadow[BLK_RING_SIZE]; unsigned long shadow_free; unsigned int feature_flush; + unsigned int flush_op; int is_ready; }; @@ -120,6 +121,10 @@ static DEFINE_SPINLOCK(minor_lock); #define EXTENDED (1<<EXT_SHIFT) #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED)) #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED)) +#define EMULATED_HD_DISK_MINOR_OFFSET (0) +#define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256) +#define EMULATED_SD_DISK_MINOR_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET + (4 * 16)) +#define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_HD_DISK_NAME_OFFSET + 4) #define DEV_NAME "xvd" /* name in /dev */ @@ -246,8 +251,7 @@ static int blkif_ioctl(struct block_device *bdev, fmode_t mode, /* * Generate a Xen blkfront IO request from a blk layer request. Reads - * and writes are handled as expected. Since we lack a loose flush - * request, we map flushes into a full ordered barrier. + * and writes are handled as expected. * * @req: a request struct */ @@ -281,7 +285,7 @@ static int blkif_queue_request(struct request *req) info->shadow[id].request = req; ring_req->id = id; - ring_req->sector_number = (blkif_sector_t)blk_rq_pos(req); + ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req); ring_req->handle = info->handle; ring_req->operation = rq_data_dir(req) ? @@ -289,14 +293,13 @@ static int blkif_queue_request(struct request *req) if (req->cmd_flags & (REQ_FLUSH | REQ_FUA)) { /* - * Ideally we could just do an unordered - * flush-to-disk, but all we have is a full write - * barrier at the moment. However, a barrier write is + * Ideally we can do an unordered flush-to-disk. In case the + * backend onlysupports barriers, use that. A barrier request * a superset of FUA, so we can implement it the same * way. (It's also a FLUSH+FUA, since it is * guaranteed ordered WRT previous writes.) */ - ring_req->operation = BLKIF_OP_WRITE_BARRIER; + ring_req->operation = info->flush_op; } ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg); @@ -317,7 +320,7 @@ static int blkif_queue_request(struct request *req) rq_data_dir(req) ); info->shadow[id].frame[i] = mfn_to_pfn(buffer_mfn); - ring_req->seg[i] = + ring_req->u.rw.seg[i] = (struct blkif_request_segment) { .gref = ref, .first_sect = fsect, @@ -429,11 +432,73 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size) static void xlvbd_flush(struct blkfront_info *info) { blk_queue_flush(info->rq, info->feature_flush); - printk(KERN_INFO "blkfront: %s: barriers %s\n", + printk(KERN_INFO "blkfront: %s: %s: %s\n", info->gd->disk_name, + info->flush_op == BLKIF_OP_WRITE_BARRIER ? + "barrier" : (info->flush_op == BLKIF_OP_FLUSH_DISKCACHE ? + "flush diskcache" : "barrier or flush"), info->feature_flush ? "enabled" : "disabled"); } +static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset) +{ + int major; + major = BLKIF_MAJOR(vdevice); + *minor = BLKIF_MINOR(vdevice); + switch (major) { + case XEN_IDE0_MAJOR: + *offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET; + *minor = ((*minor / 64) * PARTS_PER_DISK) + + EMULATED_HD_DISK_MINOR_OFFSET; + break; + case XEN_IDE1_MAJOR: + *offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET; + *minor = (((*minor / 64) + 2) * PARTS_PER_DISK) + + EMULATED_HD_DISK_MINOR_OFFSET; + break; + case XEN_SCSI_DISK0_MAJOR: + *offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET; + *minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET; + break; + case XEN_SCSI_DISK1_MAJOR: + case XEN_SCSI_DISK2_MAJOR: + case XEN_SCSI_DISK3_MAJOR: + case XEN_SCSI_DISK4_MAJOR: + case XEN_SCSI_DISK5_MAJOR: + case XEN_SCSI_DISK6_MAJOR: + case XEN_SCSI_DISK7_MAJOR: + *offset = (*minor / PARTS_PER_DISK) + + ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) + + EMULATED_SD_DISK_NAME_OFFSET; + *minor = *minor + + ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) + + EMULATED_SD_DISK_MINOR_OFFSET; + break; + case XEN_SCSI_DISK8_MAJOR: + case XEN_SCSI_DISK9_MAJOR: + case XEN_SCSI_DISK10_MAJOR: + case XEN_SCSI_DISK11_MAJOR: + case XEN_SCSI_DISK12_MAJOR: + case XEN_SCSI_DISK13_MAJOR: + case XEN_SCSI_DISK14_MAJOR: + case XEN_SCSI_DISK15_MAJOR: + *offset = (*minor / PARTS_PER_DISK) + + ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) + + EMULATED_SD_DISK_NAME_OFFSET; + *minor = *minor + + ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) + + EMULATED_SD_DISK_MINOR_OFFSET; + break; + case XENVBD_MAJOR: + *offset = *minor / PARTS_PER_DISK; + break; + default: + printk(KERN_WARNING "blkfront: your disk configuration is " + "incorrect, please use an xvd device instead\n"); + return -ENODEV; + } + return 0; +} static int xlvbd_alloc_gendisk(blkif_sector_t capacity, struct blkfront_info *info, @@ -441,7 +506,7 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity, { struct gendisk *gd; int nr_minors = 1; - int err = -ENODEV; + int err; unsigned int offset; int minor; int nr_parts; @@ -456,12 +521,20 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity, } if (!VDEV_IS_EXTENDED(info->vdevice)) { - minor = BLKIF_MINOR(info->vdevice); - nr_parts = PARTS_PER_DISK; + err = xen_translate_vdev(info->vdevice, &minor, &offset); + if (err) + return err; + nr_parts = PARTS_PER_DISK; } else { minor = BLKIF_MINOR_EXT(info->vdevice); nr_parts = PARTS_PER_EXT_DISK; + offset = minor / nr_parts; + if (xen_hvm_domain() && offset <= EMULATED_HD_DISK_NAME_OFFSET + 4) + printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with " + "emulated IDE disks,\n\t choose an xvd device name" + "from xvde on\n", info->vdevice); } + err = -ENODEV; if ((minor % nr_parts) == 0) nr_minors = nr_parts; @@ -475,8 +548,6 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity, if (gd == NULL) goto release; - offset = minor / nr_parts; - if (nr_minors > 1) { if (offset < 26) sprintf(gd->disk_name, "%s%c", DEV_NAME, 'a' + offset); @@ -615,7 +686,7 @@ static void blkif_completion(struct blk_shadow *s) { int i; for (i = 0; i < s->req.nr_segments; i++) - gnttab_end_foreign_access(s->req.seg[i].gref, 0, 0UL); + gnttab_end_foreign_access(s->req.u.rw.seg[i].gref, 0, 0UL); } static irqreturn_t blkif_interrupt(int irq, void *dev_id) @@ -651,15 +722,20 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO; switch (bret->operation) { + case BLKIF_OP_FLUSH_DISKCACHE: case BLKIF_OP_WRITE_BARRIER: if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) { - printk(KERN_WARNING "blkfront: %s: write barrier op failed\n", + printk(KERN_WARNING "blkfront: %s: write %s op failed\n", + info->flush_op == BLKIF_OP_WRITE_BARRIER ? + "barrier" : "flush disk cache", info->gd->disk_name); error = -EOPNOTSUPP; } if (unlikely(bret->status == BLKIF_RSP_ERROR && info->shadow[id].req.nr_segments == 0)) { - printk(KERN_WARNING "blkfront: %s: empty write barrier op failed\n", + printk(KERN_WARNING "blkfront: %s: empty write %s op failed\n", + info->flush_op == BLKIF_OP_WRITE_BARRIER ? + "barrier" : "flush disk cache", info->gd->disk_name); error = -EOPNOTSUPP; } @@ -667,6 +743,7 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) if (error == -EOPNOTSUPP) error = 0; info->feature_flush = 0; + info->flush_op = 0; xlvbd_flush(info); } /* fall through */ @@ -932,7 +1009,7 @@ static int blkif_recover(struct blkfront_info *info) /* Rewrite any grant references invalidated by susp/resume. */ for (j = 0; j < req->nr_segments; j++) gnttab_grant_foreign_access_ref( - req->seg[j].gref, + req->u.rw.seg[j].gref, info->xbdev->otherend_id, pfn_to_mfn(info->shadow[req->id].frame[j]), rq_data_dir(info->shadow[req->id].request)); @@ -1031,7 +1108,7 @@ static void blkfront_connect(struct blkfront_info *info) unsigned long sector_size; unsigned int binfo; int err; - int barrier; + int barrier, flush; switch (info->connected) { case BLKIF_STATE_CONNECTED: @@ -1071,8 +1148,11 @@ static void blkfront_connect(struct blkfront_info *info) return; } + info->feature_flush = 0; + info->flush_op = 0; + err = xenbus_gather(XBT_NIL, info->xbdev->otherend, - "feature-barrier", "%lu", &barrier, + "feature-barrier", "%d", &barrier, NULL); /* @@ -1082,11 +1162,23 @@ static void blkfront_connect(struct blkfront_info *info) * * If there are barriers, then we use flush. */ - info->feature_flush = 0; - - if (!err && barrier) + if (!err && barrier) { info->feature_flush = REQ_FLUSH | REQ_FUA; + info->flush_op = BLKIF_OP_WRITE_BARRIER; + } + /* + * And if there is "feature-flush-cache" use that above + * barriers. + */ + err = xenbus_gather(XBT_NIL, info->xbdev->otherend, + "feature-flush-cache", "%d", &flush, + NULL); + if (!err && flush) { + info->feature_flush = REQ_FLUSH; + info->flush_op = BLKIF_OP_FLUSH_DISKCACHE; + } + err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size); if (err) { xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s", diff --git a/drivers/block/xsysace.c b/drivers/block/xsysace.c index 829161e..6c7fd7d 100644 --- a/drivers/block/xsysace.c +++ b/drivers/block/xsysace.c @@ -621,7 +621,7 @@ static void ace_fsm_dostate(struct ace_device *ace) ace_dump_mem(ace->cf_id, 512); /* Debug: Dump out disk ID */ if (ace->data_result) { - /* Error occured, disable the disk */ + /* Error occurred, disable the disk */ ace->media_change = 1; set_capacity(ace->gd, 0); dev_err(ace->dev, "error fetching CF id (%i)\n", @@ -801,7 +801,7 @@ static int ace_interrupt_checkstate(struct ace_device *ace) u32 sreg = ace_in32(ace, ACE_STATUS); u16 creg = ace_in(ace, ACE_CTRL); - /* Check for error occurance */ + /* Check for error occurrence */ if ((sreg & (ACE_STATUS_CFGERROR | ACE_STATUS_CFCERROR)) && (creg & ACE_CTRL_ERRORIRQ)) { dev_err(ace->dev, "transfer failure\n"); @@ -867,12 +867,12 @@ static void ace_request(struct request_queue * q) } } -static int ace_media_changed(struct gendisk *gd) +static unsigned int ace_check_events(struct gendisk *gd, unsigned int clearing) { struct ace_device *ace = gd->private_data; - dev_dbg(ace->dev, "ace_media_changed(): %i\n", ace->media_change); + dev_dbg(ace->dev, "ace_check_events(): %i\n", ace->media_change); - return ace->media_change; + return ace->media_change ? DISK_EVENT_MEDIA_CHANGE : 0; } static int ace_revalidate_disk(struct gendisk *gd) @@ -953,7 +953,7 @@ static const struct block_device_operations ace_fops = { .owner = THIS_MODULE, .open = ace_open, .release = ace_release, - .media_changed = ace_media_changed, + .check_events = ace_check_events, .revalidate_disk = ace_revalidate_disk, .getgeo = ace_getgeo, }; @@ -1168,7 +1168,7 @@ static int __devinit ace_probe(struct platform_device *dev) irq = dev->resource[i].start; } - /* Call the bus-independant setup code */ + /* Call the bus-independent setup code */ return ace_alloc(&dev->dev, id, physaddr, irq, bus_width); } @@ -1195,16 +1195,13 @@ static struct platform_driver ace_platform_driver = { */ #if defined(CONFIG_OF) -static int __devinit -ace_of_probe(struct platform_device *op, const struct of_device_id *match) +static int __devinit ace_of_probe(struct platform_device *op) { struct resource res; resource_size_t physaddr; const u32 *id; int irq, bus_width, rc; - dev_dbg(&op->dev, "ace_of_probe(%p, %p)\n", op, match); - /* device id */ id = of_get_property(op->dev.of_node, "port-number", NULL); @@ -1224,7 +1221,7 @@ ace_of_probe(struct platform_device *op, const struct of_device_id *match) if (of_find_property(op->dev.of_node, "8-bit", NULL)) bus_width = ACE_BUS_WIDTH_8; - /* Call the bus-independant setup code */ + /* Call the bus-independent setup code */ return ace_alloc(&op->dev, id ? be32_to_cpup(id) : 0, physaddr, irq, bus_width); } @@ -1245,7 +1242,7 @@ static const struct of_device_id ace_of_match[] __devinitconst = { }; MODULE_DEVICE_TABLE(of, ace_of_match); -static struct of_platform_driver ace_of_driver = { +static struct platform_driver ace_of_driver = { .probe = ace_of_probe, .remove = __devexit_p(ace_of_remove), .driver = { @@ -1259,12 +1256,12 @@ static struct of_platform_driver ace_of_driver = { static inline int __init ace_of_register(void) { pr_debug("xsysace: registering OF binding\n"); - return of_register_platform_driver(&ace_of_driver); + return platform_driver_register(&ace_of_driver); } static inline void __exit ace_of_unregister(void) { - of_unregister_platform_driver(&ace_of_driver); + platform_driver_unregister(&ace_of_driver); } #else /* CONFIG_OF */ /* CONFIG_OF not enabled; do nothing helpers */ |