diff options
Diffstat (limited to 'drivers/block')
46 files changed, 5070 insertions, 2154 deletions
diff --git a/drivers/block/DAC960.c b/drivers/block/DAC960.c index 4e2c367..1f286ab 100644 --- a/drivers/block/DAC960.c +++ b/drivers/block/DAC960.c @@ -36,7 +36,7 @@ #include <linux/ioport.h> #include <linux/mm.h> #include <linux/slab.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/proc_fs.h> #include <linux/seq_file.h> #include <linux/reboot.h> @@ -54,6 +54,7 @@ #define DAC960_GAM_MINOR 252 +static DEFINE_MUTEX(DAC960_mutex); static DAC960_Controller_T *DAC960_Controllers[DAC960_MaxControllers]; static int DAC960_ControllerCount; static struct proc_dir_entry *DAC960_ProcDirectoryEntry; @@ -81,7 +82,7 @@ static int DAC960_open(struct block_device *bdev, fmode_t mode) int drive_nr = (long)disk->private_data; int ret = -ENXIO; - lock_kernel(); + mutex_lock(&DAC960_mutex); if (p->FirmwareType == DAC960_V1_Controller) { if (p->V1.LogicalDriveInformation[drive_nr]. LogicalDriveState == DAC960_V1_LogicalDrive_Offline) @@ -99,7 +100,7 @@ static int DAC960_open(struct block_device *bdev, fmode_t mode) goto out; ret = 0; out: - unlock_kernel(); + mutex_unlock(&DAC960_mutex); return ret; } @@ -6625,7 +6626,7 @@ static long DAC960_gam_ioctl(struct file *file, unsigned int Request, long ErrorCode = 0; if (!capable(CAP_SYS_ADMIN)) return -EACCES; - lock_kernel(); + mutex_lock(&DAC960_mutex); switch (Request) { case DAC960_IOCTL_GET_CONTROLLER_COUNT: @@ -7056,13 +7057,14 @@ static long DAC960_gam_ioctl(struct file *file, unsigned int Request, default: ErrorCode = -ENOTTY; } - unlock_kernel(); + mutex_unlock(&DAC960_mutex); return ErrorCode; } static const struct file_operations DAC960_gam_fops = { .owner = THIS_MODULE, - .unlocked_ioctl = DAC960_gam_ioctl + .unlocked_ioctl = DAC960_gam_ioctl, + .llseek = noop_llseek, }; static struct miscdevice DAC960_gam_dev = { diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig index de27768..83c32cb 100644 --- a/drivers/block/Kconfig +++ b/drivers/block/Kconfig @@ -464,6 +464,7 @@ config XEN_BLKDEV_FRONTEND tristate "Xen virtual block device support" depends on XEN default y + select XEN_XENBUS_FRONTEND help This driver implements the front-end of the Xen virtual block device driver. It communicates with a back-end driver @@ -488,4 +489,21 @@ config BLK_DEV_HD If unsure, say N. +config BLK_DEV_RBD + tristate "Rados block device (RBD)" + depends on INET && EXPERIMENTAL && BLOCK + select CEPH_LIB + select LIBCRC32C + select CRYPTO_AES + select CRYPTO + default n + help + Say Y here if you want include the Rados block device, which stripes + a block device over objects stored in the Ceph distributed object + store. + + More information at http://ceph.newdream.net/. + + If unsure, say N. + endif # BLK_DEV diff --git a/drivers/block/Makefile b/drivers/block/Makefile index aff5ac9..d7f463d 100644 --- a/drivers/block/Makefile +++ b/drivers/block/Makefile @@ -37,5 +37,6 @@ obj-$(CONFIG_BLK_DEV_HD) += hd.o obj-$(CONFIG_XEN_BLKDEV_FRONTEND) += xen-blkfront.o obj-$(CONFIG_BLK_DEV_DRBD) += drbd/ +obj-$(CONFIG_BLK_DEV_RBD) += rbd.o swim_mod-objs := swim.o swim_asm.o diff --git a/drivers/block/amiflop.c b/drivers/block/amiflop.c index 76f114f..7888501 100644 --- a/drivers/block/amiflop.c +++ b/drivers/block/amiflop.c @@ -60,7 +60,7 @@ #include <linux/hdreg.h> #include <linux/delay.h> #include <linux/init.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/amifdreg.h> #include <linux/amifd.h> #include <linux/buffer_head.h> @@ -109,13 +109,12 @@ #define FD_HD_3 0x55555555 /* high-density 3.5" (1760K) drive */ #define FD_DD_5 0xaaaaaaaa /* double-density 5.25" (440K) drive */ +static DEFINE_MUTEX(amiflop_mutex); static unsigned long int fd_def_df0 = FD_DD_3; /* default for df0 if it doesn't identify */ module_param(fd_def_df0, ulong, 0); MODULE_LICENSE("GPL"); -static struct request_queue *floppy_queue; - /* * Macros */ @@ -164,6 +163,7 @@ static volatile int selected = -1; /* currently selected drive */ static int writepending; static int writefromint; static char *raw_buf; +static int fdc_queue; static DEFINE_SPINLOCK(amiflop_lock); @@ -1334,6 +1334,42 @@ static int get_track(int drive, int track) return -1; } +/* + * Round-robin between our available drives, doing one request from each + */ +static struct request *set_next_request(void) +{ + struct request_queue *q; + int cnt = FD_MAX_UNITS; + struct request *rq = NULL; + + /* Find next queue we can dispatch from */ + fdc_queue = fdc_queue + 1; + if (fdc_queue == FD_MAX_UNITS) + fdc_queue = 0; + + for(cnt = FD_MAX_UNITS; cnt > 0; cnt--) { + + if (unit[fdc_queue].type->code == FD_NODRIVE) { + if (++fdc_queue == FD_MAX_UNITS) + fdc_queue = 0; + continue; + } + + q = unit[fdc_queue].gendisk->queue; + if (q) { + rq = blk_fetch_request(q); + if (rq) + break; + } + + if (++fdc_queue == FD_MAX_UNITS) + fdc_queue = 0; + } + + return rq; +} + static void redo_fd_request(void) { struct request *rq; @@ -1345,7 +1381,7 @@ static void redo_fd_request(void) int err; next_req: - rq = blk_fetch_request(floppy_queue); + rq = set_next_request(); if (!rq) { /* Nothing left to do */ return; @@ -1506,9 +1542,9 @@ static int fd_ioctl(struct block_device *bdev, fmode_t mode, { int ret; - lock_kernel(); + mutex_lock(&amiflop_mutex); ret = fd_locked_ioctl(bdev, mode, cmd, param); - unlock_kernel(); + mutex_unlock(&amiflop_mutex); return ret; } @@ -1555,11 +1591,11 @@ static int floppy_open(struct block_device *bdev, fmode_t mode) int old_dev; unsigned long flags; - lock_kernel(); + mutex_lock(&amiflop_mutex); old_dev = fd_device[drive]; if (fd_ref[drive] && old_dev != system) { - unlock_kernel(); + mutex_unlock(&amiflop_mutex); return -EBUSY; } @@ -1575,7 +1611,7 @@ static int floppy_open(struct block_device *bdev, fmode_t mode) rel_fdc(); if (wrprot) { - unlock_kernel(); + mutex_unlock(&amiflop_mutex); return -EROFS; } } @@ -1594,7 +1630,7 @@ static int floppy_open(struct block_device *bdev, fmode_t mode) printk(KERN_INFO "fd%d: accessing %s-disk with %s-layout\n",drive, unit[drive].type->name, data_types[system].name); - unlock_kernel(); + mutex_unlock(&amiflop_mutex); return 0; } @@ -1603,7 +1639,7 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) struct amiga_floppy_struct *p = disk->private_data; int drive = p - unit; - lock_kernel(); + mutex_lock(&amiflop_mutex); if (unit[drive].dirty == 1) { del_timer (flush_track_timer + drive); non_int_flush_track (drive); @@ -1617,7 +1653,7 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) /* the mod_use counter is handled this way */ floppy_off (drive | 0x40000000); #endif - unlock_kernel(); + mutex_unlock(&amiflop_mutex); return 0; } @@ -1682,6 +1718,13 @@ static int __init fd_probe_drives(void) continue; } unit[drive].gendisk = disk; + + disk->queue = blk_init_queue(do_fd_request, &amiflop_lock); + if (!disk->queue) { + unit[drive].type->code = FD_NODRIVE; + continue; + } + drives++; if ((unit[drive].trackbuf = kmalloc(FLOPPY_MAX_SECTORS * 512, GFP_KERNEL)) == NULL) { printk("no mem for "); @@ -1695,7 +1738,6 @@ static int __init fd_probe_drives(void) disk->fops = &floppy_fops; sprintf(disk->disk_name, "fd%d", drive); disk->private_data = &unit[drive]; - disk->queue = floppy_queue; set_capacity(disk, 880*2); add_disk(disk); } @@ -1743,11 +1785,6 @@ static int __init amiga_floppy_probe(struct platform_device *pdev) goto out_irq2; } - ret = -ENOMEM; - floppy_queue = blk_init_queue(do_fd_request, &amiflop_lock); - if (!floppy_queue) - goto out_queue; - ret = -ENODEV; if (fd_probe_drives() < 1) /* No usable drives */ goto out_probe; @@ -1791,8 +1828,6 @@ static int __init amiga_floppy_probe(struct platform_device *pdev) return 0; out_probe: - blk_cleanup_queue(floppy_queue); -out_queue: free_irq(IRQ_AMIGA_CIAA_TB, NULL); out_irq2: free_irq(IRQ_AMIGA_DSKBLK, NULL); @@ -1810,9 +1845,12 @@ static int __exit amiga_floppy_remove(struct platform_device *pdev) for( i = 0; i < FD_MAX_UNITS; i++) { if (unit[i].type->code != FD_NODRIVE) { + struct request_queue *q = unit[i].gendisk->queue; del_gendisk(unit[i].gendisk); put_disk(unit[i].gendisk); kfree(unit[i].trackbuf); + if (q) + blk_cleanup_queue(q); } } blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); @@ -1820,7 +1858,6 @@ static int __exit amiga_floppy_remove(struct platform_device *pdev) free_irq(IRQ_AMIGA_DSKBLK, NULL); custom.dmacon = DMAF_DISK; /* disable DMA */ amiga_chip_free(raw_buf); - blk_cleanup_queue(floppy_queue); unregister_blkdev(FLOPPY_MAJOR, "fd"); } #endif diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c index a946929..528f631 100644 --- a/drivers/block/aoe/aoeblk.c +++ b/drivers/block/aoe/aoeblk.c @@ -4,17 +4,20 @@ * block device routines */ +#include <linux/kernel.h> #include <linux/hdreg.h> #include <linux/blkdev.h> #include <linux/backing-dev.h> #include <linux/fs.h> #include <linux/ioctl.h> #include <linux/slab.h> +#include <linux/ratelimit.h> #include <linux/genhd.h> #include <linux/netdevice.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include "aoe.h" +static DEFINE_MUTEX(aoeblk_mutex); static struct kmem_cache *buf_pool_cache; static ssize_t aoedisk_show_state(struct device *dev, @@ -125,16 +128,16 @@ aoeblk_open(struct block_device *bdev, fmode_t mode) struct aoedev *d = bdev->bd_disk->private_data; ulong flags; - lock_kernel(); + mutex_lock(&aoeblk_mutex); spin_lock_irqsave(&d->lock, flags); if (d->flags & DEVFL_UP) { d->nopen++; spin_unlock_irqrestore(&d->lock, flags); - unlock_kernel(); + mutex_unlock(&aoeblk_mutex); return 0; } spin_unlock_irqrestore(&d->lock, flags); - unlock_kernel(); + mutex_unlock(&aoeblk_mutex); return -ENODEV; } @@ -177,9 +180,6 @@ aoeblk_make_request(struct request_queue *q, struct bio *bio) BUG(); bio_endio(bio, -ENXIO); return 0; - } else if (bio->bi_rw & REQ_HARDBARRIER) { - bio_endio(bio, -EOPNOTSUPP); - return 0; } else if (bio->bi_io_vec == NULL) { printk(KERN_ERR "aoe: bi_io_vec is NULL\n"); BUG(); @@ -206,7 +206,7 @@ aoeblk_make_request(struct request_queue *q, struct bio *bio) spin_lock_irqsave(&d->lock, flags); if ((d->flags & DEVFL_UP) == 0) { - printk(KERN_INFO "aoe: device %ld.%d is not up\n", + pr_info_ratelimited("aoe: device %ld.%d is not up\n", d->aoemajor, d->aoeminor); spin_unlock_irqrestore(&d->lock, flags); mempool_free(buf, d->bufpool); diff --git a/drivers/block/aoe/aoechr.c b/drivers/block/aoe/aoechr.c index 4a1b9e74..146296c 100644 --- a/drivers/block/aoe/aoechr.c +++ b/drivers/block/aoe/aoechr.c @@ -9,7 +9,7 @@ #include <linux/completion.h> #include <linux/delay.h> #include <linux/slab.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/skbuff.h> #include "aoe.h" @@ -37,6 +37,7 @@ struct ErrMsg { char *msg; }; +static DEFINE_MUTEX(aoechr_mutex); static struct ErrMsg emsgs[NMSG]; static int emsgs_head_idx, emsgs_tail_idx; static struct completion emsgs_comp; @@ -183,16 +184,16 @@ aoechr_open(struct inode *inode, struct file *filp) { int n, i; - lock_kernel(); + mutex_lock(&aoechr_mutex); n = iminor(inode); filp->private_data = (void *) (unsigned long) n; for (i = 0; i < ARRAY_SIZE(chardevs); ++i) if (chardevs[i].minor == n) { - unlock_kernel(); + mutex_unlock(&aoechr_mutex); return 0; } - unlock_kernel(); + mutex_unlock(&aoechr_mutex); return -EINVAL; } @@ -265,6 +266,7 @@ static const struct file_operations aoe_fops = { .open = aoechr_open, .release = aoechr_rel, .owner = THIS_MODULE, + .llseek = noop_llseek, }; static char *aoe_devnode(struct device *dev, mode_t *mode) diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c index 5674bd0..de0435e 100644 --- a/drivers/block/aoe/aoecmd.c +++ b/drivers/block/aoe/aoecmd.c @@ -297,8 +297,8 @@ aoecmd_cfg_pkts(ushort aoemajor, unsigned char aoeminor, struct sk_buff_head *qu struct sk_buff *skb; struct net_device *ifp; - read_lock(&dev_base_lock); - for_each_netdev(&init_net, ifp) { + rcu_read_lock(); + for_each_netdev_rcu(&init_net, ifp) { dev_hold(ifp); if (!is_aoe_netif(ifp)) goto cont; @@ -325,7 +325,7 @@ aoecmd_cfg_pkts(ushort aoemajor, unsigned char aoeminor, struct sk_buff_head *qu cont: dev_put(ifp); } - read_unlock(&dev_base_lock); + rcu_read_unlock(); } static void diff --git a/drivers/block/aoe/aoedev.c b/drivers/block/aoe/aoedev.c index 0849280..6b5110a 100644 --- a/drivers/block/aoe/aoedev.c +++ b/drivers/block/aoe/aoedev.c @@ -102,6 +102,7 @@ aoedev_freedev(struct aoedev *d) { struct aoetgt **t, **e; + cancel_work_sync(&d->work); if (d->gd) { aoedisk_rm_sysfs(d); del_gendisk(d->gd); @@ -135,7 +136,6 @@ aoedev_flush(const char __user *str, size_t cnt) all = !strncmp(buf, "all", 3); } - flush_scheduled_work(); spin_lock_irqsave(&devlist_lock, flags); dd = &devlist; while ((d = *dd)) { @@ -257,8 +257,6 @@ aoedev_exit(void) struct aoedev *d; ulong flags; - flush_scheduled_work(); - while ((d = devlist)) { devlist = d->next; diff --git a/drivers/block/ataflop.c b/drivers/block/ataflop.c index aceb964..605a67e 100644 --- a/drivers/block/ataflop.c +++ b/drivers/block/ataflop.c @@ -67,7 +67,7 @@ #include <linux/delay.h> #include <linux/init.h> #include <linux/blkdev.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <asm/atafd.h> #include <asm/atafdreg.h> @@ -79,8 +79,9 @@ #undef DEBUG -static struct request_queue *floppy_queue; +static DEFINE_MUTEX(ataflop_mutex); static struct request *fd_request; +static int fdc_queue; /* Disk types: DD, HD, ED */ static struct atari_disk_type { @@ -1391,6 +1392,29 @@ static void setup_req_params( int drive ) ReqTrack, ReqSector, (unsigned long)ReqData )); } +/* + * Round-robin between our available drives, doing one request from each + */ +static struct request *set_next_request(void) +{ + struct request_queue *q; + int old_pos = fdc_queue; + struct request *rq = NULL; + + do { + q = unit[fdc_queue].disk->queue; + if (++fdc_queue == FD_MAX_UNITS) + fdc_queue = 0; + if (q) { + rq = blk_fetch_request(q); + if (rq) + break; + } + } while (fdc_queue != old_pos); + + return rq; +} + static void redo_fd_request(void) { @@ -1405,7 +1429,7 @@ static void redo_fd_request(void) repeat: if (!fd_request) { - fd_request = blk_fetch_request(floppy_queue); + fd_request = set_next_request(); if (!fd_request) goto the_end; } @@ -1671,9 +1695,9 @@ static int fd_ioctl(struct block_device *bdev, fmode_t mode, { int ret; - lock_kernel(); + mutex_lock(&ataflop_mutex); ret = fd_locked_ioctl(bdev, mode, cmd, arg); - unlock_kernel(); + mutex_unlock(&ataflop_mutex); return ret; } @@ -1854,9 +1878,9 @@ static int floppy_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&ataflop_mutex); ret = floppy_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&ataflop_mutex); return ret; } @@ -1864,14 +1888,14 @@ static int floppy_unlocked_open(struct block_device *bdev, fmode_t mode) static int floppy_release(struct gendisk *disk, fmode_t mode) { struct atari_floppy_struct *p = disk->private_data; - lock_kernel(); + mutex_lock(&ataflop_mutex); if (p->ref < 0) p->ref = 0; else if (!p->ref--) { printk(KERN_ERR "floppy_release with fd_ref == 0"); p->ref = 0; } - unlock_kernel(); + mutex_unlock(&ataflop_mutex); return 0; } @@ -1932,10 +1956,6 @@ static int __init atari_floppy_init (void) PhysTrackBuffer = virt_to_phys(TrackBuffer); BufferDrive = BufferSide = BufferTrack = -1; - floppy_queue = blk_init_queue(do_fd_request, &ataflop_lock); - if (!floppy_queue) - goto Enomem; - for (i = 0; i < FD_MAX_UNITS; i++) { unit[i].track = -1; unit[i].flags = 0; @@ -1944,7 +1964,10 @@ static int __init atari_floppy_init (void) sprintf(unit[i].disk->disk_name, "fd%d", i); unit[i].disk->fops = &floppy_fops; unit[i].disk->private_data = &unit[i]; - unit[i].disk->queue = floppy_queue; + unit[i].disk->queue = blk_init_queue(do_fd_request, + &ataflop_lock); + if (!unit[i].disk->queue) + goto Enomem; set_capacity(unit[i].disk, MAX_DISK_SIZE * 2); add_disk(unit[i].disk); } @@ -1959,10 +1982,14 @@ static int __init atari_floppy_init (void) return 0; Enomem: - while (i--) + while (i--) { + struct request_queue *q = unit[i].disk->queue; + put_disk(unit[i].disk); - if (floppy_queue) - blk_cleanup_queue(floppy_queue); + if (q) + blk_cleanup_queue(q); + } + unregister_blkdev(FLOPPY_MAJOR, "fd"); return -ENOMEM; } @@ -2011,12 +2038,14 @@ static void __exit atari_floppy_exit(void) int i; blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); for (i = 0; i < FD_MAX_UNITS; i++) { + struct request_queue *q = unit[i].disk->queue; + del_gendisk(unit[i].disk); put_disk(unit[i].disk); + blk_cleanup_queue(q); } unregister_blkdev(FLOPPY_MAJOR, "fd"); - blk_cleanup_queue(floppy_queue); del_timer_sync(&fd_timer); atari_stram_free( DMABuffer ); } diff --git a/drivers/block/brd.c b/drivers/block/brd.c index 1c7f637..b7f51e4 100644 --- a/drivers/block/brd.c +++ b/drivers/block/brd.c @@ -15,7 +15,7 @@ #include <linux/blkdev.h> #include <linux/bio.h> #include <linux/highmem.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/radix-tree.h> #include <linux/buffer_head.h> /* invalidate_bh_lrus() */ #include <linux/slab.h> @@ -55,6 +55,7 @@ struct brd_device { /* * Look up and return a brd's page for a given sector. */ +static DEFINE_MUTEX(brd_mutex); static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) { pgoff_t idx; @@ -402,7 +403,7 @@ static int brd_ioctl(struct block_device *bdev, fmode_t mode, * ram device BLKFLSBUF has special semantics, we want to actually * release and destroy the ramdisk data. */ - lock_kernel(); + mutex_lock(&brd_mutex); mutex_lock(&bdev->bd_mutex); error = -EBUSY; if (bdev->bd_openers <= 1) { @@ -419,7 +420,7 @@ static int brd_ioctl(struct block_device *bdev, fmode_t mode, error = 0; } mutex_unlock(&bdev->bd_mutex); - unlock_kernel(); + mutex_unlock(&brd_mutex); return error; } @@ -482,7 +483,6 @@ static struct brd_device *brd_alloc(int i) if (!brd->brd_queue) goto out_free_dev; blk_queue_make_request(brd->brd_queue, brd_make_request); - blk_queue_ordered(brd->brd_queue, QUEUE_ORDERED_TAG); blk_queue_max_hw_sectors(brd->brd_queue, 1024); blk_queue_bounce_limit(brd->brd_queue, BLK_BOUNCE_ANY); diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c index a54f7f4..516d5bb 100644 --- a/drivers/block/cciss.c +++ b/drivers/block/cciss.c @@ -26,7 +26,6 @@ #include <linux/pci.h> #include <linux/kernel.h> #include <linux/slab.h> -#include <linux/smp_lock.h> #include <linux/delay.h> #include <linux/major.h> #include <linux/fs.h> @@ -66,11 +65,8 @@ MODULE_SUPPORTED_DEVICE("HP Smart Array Controllers"); MODULE_VERSION("3.6.26"); MODULE_LICENSE("GPL"); -static int cciss_allow_hpsa; -module_param(cciss_allow_hpsa, int, S_IRUGO|S_IWUSR); -MODULE_PARM_DESC(cciss_allow_hpsa, - "Prevent cciss driver from accessing hardware known to be " - " supported by the hpsa driver"); +static DEFINE_MUTEX(cciss_mutex); +static struct proc_dir_entry *proc_cciss; #include "cciss_cmd.h" #include "cciss.h" @@ -98,18 +94,6 @@ static const struct pci_device_id cciss_pci_device_id[] = { {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSD, 0x103C, 0x3215}, {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x3237}, {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x323D}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3241}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3243}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3245}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3247}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3249}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x324A}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x324B}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3250}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3251}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3252}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3253}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSE, 0x103C, 0x3254}, {0,} }; @@ -130,6 +114,8 @@ static struct board_type products[] = { {0x409D0E11, "Smart Array 6400 EM", &SA5_access}, {0x40910E11, "Smart Array 6i", &SA5_access}, {0x3225103C, "Smart Array P600", &SA5_access}, + {0x3223103C, "Smart Array P800", &SA5_access}, + {0x3234103C, "Smart Array P400", &SA5_access}, {0x3235103C, "Smart Array P400i", &SA5_access}, {0x3211103C, "Smart Array E200i", &SA5_access}, {0x3212103C, "Smart Array E200", &SA5_access}, @@ -137,23 +123,9 @@ static struct board_type products[] = { {0x3214103C, "Smart Array E200i", &SA5_access}, {0x3215103C, "Smart Array E200i", &SA5_access}, {0x3237103C, "Smart Array E500", &SA5_access}, -/* controllers below this line are also supported by the hpsa driver. */ -#define HPSA_BOUNDARY 0x3223103C {0x3223103C, "Smart Array P800", &SA5_access}, {0x3234103C, "Smart Array P400", &SA5_access}, {0x323D103C, "Smart Array P700m", &SA5_access}, - {0x3241103C, "Smart Array P212", &SA5_access}, - {0x3243103C, "Smart Array P410", &SA5_access}, - {0x3245103C, "Smart Array P410i", &SA5_access}, - {0x3247103C, "Smart Array P411", &SA5_access}, - {0x3249103C, "Smart Array P812", &SA5_access}, - {0x324A103C, "Smart Array P712m", &SA5_access}, - {0x324B103C, "Smart Array P711m", &SA5_access}, - {0x3250103C, "Smart Array", &SA5_access}, - {0x3251103C, "Smart Array", &SA5_access}, - {0x3252103C, "Smart Array", &SA5_access}, - {0x3253103C, "Smart Array", &SA5_access}, - {0x3254103C, "Smart Array", &SA5_access}, }; /* How long to wait (in milliseconds) for board to go into simple mode */ @@ -392,8 +364,6 @@ static const char *raid_label[] = { "0", "4", "1(1+0)", "5", "5+1", "ADG", #define ENG_GIG_FACTOR (ENG_GIG/512) #define ENGAGE_SCSI "engage scsi" -static struct proc_dir_entry *proc_cciss; - static void cciss_seq_show_header(struct seq_file *seq) { ctlr_info_t *h = seq->private; @@ -1059,9 +1029,9 @@ static int cciss_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&cciss_mutex); ret = cciss_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&cciss_mutex); return ret; } @@ -1074,13 +1044,13 @@ static int cciss_release(struct gendisk *disk, fmode_t mode) ctlr_info_t *h; drive_info_struct *drv; - lock_kernel(); + mutex_lock(&cciss_mutex); h = get_host(disk); drv = get_drv(disk); dev_dbg(&h->pdev->dev, "cciss_release %s\n", disk->disk_name); drv->usage_count--; h->usage_count--; - unlock_kernel(); + mutex_unlock(&cciss_mutex); return 0; } @@ -1088,9 +1058,9 @@ static int do_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd, unsigned long arg) { int ret; - lock_kernel(); + mutex_lock(&cciss_mutex); ret = cciss_ioctl(bdev, mode, cmd, arg); - unlock_kernel(); + mutex_unlock(&cciss_mutex); return ret; } @@ -1182,6 +1152,7 @@ static int cciss_ioctl32_big_passthru(struct block_device *bdev, fmode_t mode, int err; u32 cp; + memset(&arg64, 0, sizeof(arg64)); err = 0; err |= copy_from_user(&arg64.LUN_info, &arg32->LUN_info, @@ -1232,470 +1203,452 @@ static void check_ioctl_unit_attention(ctlr_info_t *h, CommandList_struct *c) c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION) (void)check_for_unit_attention(h, c); } -/* - * ioctl - */ -static int cciss_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg) + +static int cciss_getpciinfo(ctlr_info_t *h, void __user *argp) { - struct gendisk *disk = bdev->bd_disk; - ctlr_info_t *h = get_host(disk); - drive_info_struct *drv = get_drv(disk); - void __user *argp = (void __user *)arg; + cciss_pci_info_struct pciinfo; - dev_dbg(&h->pdev->dev, "cciss_ioctl: Called with cmd=%x %lx\n", - cmd, arg); - switch (cmd) { - case CCISS_GETPCIINFO: - { - cciss_pci_info_struct pciinfo; - - if (!arg) - return -EINVAL; - pciinfo.domain = pci_domain_nr(h->pdev->bus); - pciinfo.bus = h->pdev->bus->number; - pciinfo.dev_fn = h->pdev->devfn; - pciinfo.board_id = h->board_id; - if (copy_to_user - (argp, &pciinfo, sizeof(cciss_pci_info_struct))) - return -EFAULT; - return 0; - } - case CCISS_GETINTINFO: - { - cciss_coalint_struct intinfo; - if (!arg) - return -EINVAL; - intinfo.delay = - readl(&h->cfgtable->HostWrite.CoalIntDelay); - intinfo.count = - readl(&h->cfgtable->HostWrite.CoalIntCount); - if (copy_to_user - (argp, &intinfo, sizeof(cciss_coalint_struct))) - return -EFAULT; - return 0; - } - case CCISS_SETINTINFO: - { - cciss_coalint_struct intinfo; - unsigned long flags; - int i; - - if (!arg) - return -EINVAL; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - if (copy_from_user - (&intinfo, argp, sizeof(cciss_coalint_struct))) - return -EFAULT; - if ((intinfo.delay == 0) && (intinfo.count == 0)) - return -EINVAL; - spin_lock_irqsave(&h->lock, flags); - /* Update the field, and then ring the doorbell */ - writel(intinfo.delay, - &(h->cfgtable->HostWrite.CoalIntDelay)); - writel(intinfo.count, - &(h->cfgtable->HostWrite.CoalIntCount)); - writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); - - for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) { - if (!(readl(h->vaddr + SA5_DOORBELL) - & CFGTBL_ChangeReq)) - break; - /* delay and try again */ - udelay(1000); - } - spin_unlock_irqrestore(&h->lock, flags); - if (i >= MAX_IOCTL_CONFIG_WAIT) - return -EAGAIN; - return 0; - } - case CCISS_GETNODENAME: - { - NodeName_type NodeName; - int i; - - if (!arg) - return -EINVAL; - for (i = 0; i < 16; i++) - NodeName[i] = - readb(&h->cfgtable->ServerName[i]); - if (copy_to_user(argp, NodeName, sizeof(NodeName_type))) - return -EFAULT; - return 0; - } - case CCISS_SETNODENAME: - { - NodeName_type NodeName; - unsigned long flags; - int i; + if (!argp) + return -EINVAL; + pciinfo.domain = pci_domain_nr(h->pdev->bus); + pciinfo.bus = h->pdev->bus->number; + pciinfo.dev_fn = h->pdev->devfn; + pciinfo.board_id = h->board_id; + if (copy_to_user(argp, &pciinfo, sizeof(cciss_pci_info_struct))) + return -EFAULT; + return 0; +} - if (!arg) - return -EINVAL; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; +static int cciss_getintinfo(ctlr_info_t *h, void __user *argp) +{ + cciss_coalint_struct intinfo; - if (copy_from_user - (NodeName, argp, sizeof(NodeName_type))) - return -EFAULT; + if (!argp) + return -EINVAL; + intinfo.delay = readl(&h->cfgtable->HostWrite.CoalIntDelay); + intinfo.count = readl(&h->cfgtable->HostWrite.CoalIntCount); + if (copy_to_user + (argp, &intinfo, sizeof(cciss_coalint_struct))) + return -EFAULT; + return 0; +} - spin_lock_irqsave(&h->lock, flags); +static int cciss_setintinfo(ctlr_info_t *h, void __user *argp) +{ + cciss_coalint_struct intinfo; + unsigned long flags; + int i; - /* Update the field, and then ring the doorbell */ - for (i = 0; i < 16; i++) - writeb(NodeName[i], - &h->cfgtable->ServerName[i]); + if (!argp) + return -EINVAL; + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + if (copy_from_user(&intinfo, argp, sizeof(intinfo))) + return -EFAULT; + if ((intinfo.delay == 0) && (intinfo.count == 0)) + return -EINVAL; + spin_lock_irqsave(&h->lock, flags); + /* Update the field, and then ring the doorbell */ + writel(intinfo.delay, &(h->cfgtable->HostWrite.CoalIntDelay)); + writel(intinfo.count, &(h->cfgtable->HostWrite.CoalIntCount)); + writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); - writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); + for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) { + if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) + break; + udelay(1000); /* delay and try again */ + } + spin_unlock_irqrestore(&h->lock, flags); + if (i >= MAX_IOCTL_CONFIG_WAIT) + return -EAGAIN; + return 0; +} - for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) { - if (!(readl(h->vaddr + SA5_DOORBELL) - & CFGTBL_ChangeReq)) - break; - /* delay and try again */ - udelay(1000); - } - spin_unlock_irqrestore(&h->lock, flags); - if (i >= MAX_IOCTL_CONFIG_WAIT) - return -EAGAIN; - return 0; - } +static int cciss_getnodename(ctlr_info_t *h, void __user *argp) +{ + NodeName_type NodeName; + int i; - case CCISS_GETHEARTBEAT: - { - Heartbeat_type heartbeat; - - if (!arg) - return -EINVAL; - heartbeat = readl(&h->cfgtable->HeartBeat); - if (copy_to_user - (argp, &heartbeat, sizeof(Heartbeat_type))) - return -EFAULT; - return 0; - } - case CCISS_GETBUSTYPES: - { - BusTypes_type BusTypes; - - if (!arg) - return -EINVAL; - BusTypes = readl(&h->cfgtable->BusTypes); - if (copy_to_user - (argp, &BusTypes, sizeof(BusTypes_type))) - return -EFAULT; - return 0; - } - case CCISS_GETFIRMVER: - { - FirmwareVer_type firmware; + if (!argp) + return -EINVAL; + for (i = 0; i < 16; i++) + NodeName[i] = readb(&h->cfgtable->ServerName[i]); + if (copy_to_user(argp, NodeName, sizeof(NodeName_type))) + return -EFAULT; + return 0; +} - if (!arg) - return -EINVAL; - memcpy(firmware, h->firm_ver, 4); +static int cciss_setnodename(ctlr_info_t *h, void __user *argp) +{ + NodeName_type NodeName; + unsigned long flags; + int i; - if (copy_to_user - (argp, firmware, sizeof(FirmwareVer_type))) - return -EFAULT; - return 0; - } - case CCISS_GETDRIVVER: - { - DriverVer_type DriverVer = DRIVER_VERSION; + if (!argp) + return -EINVAL; + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + if (copy_from_user(NodeName, argp, sizeof(NodeName_type))) + return -EFAULT; + spin_lock_irqsave(&h->lock, flags); + /* Update the field, and then ring the doorbell */ + for (i = 0; i < 16; i++) + writeb(NodeName[i], &h->cfgtable->ServerName[i]); + writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); + for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) { + if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) + break; + udelay(1000); /* delay and try again */ + } + spin_unlock_irqrestore(&h->lock, flags); + if (i >= MAX_IOCTL_CONFIG_WAIT) + return -EAGAIN; + return 0; +} - if (!arg) - return -EINVAL; +static int cciss_getheartbeat(ctlr_info_t *h, void __user *argp) +{ + Heartbeat_type heartbeat; - if (copy_to_user - (argp, &DriverVer, sizeof(DriverVer_type))) - return -EFAULT; - return 0; - } + if (!argp) + return -EINVAL; + heartbeat = readl(&h->cfgtable->HeartBeat); + if (copy_to_user(argp, &heartbeat, sizeof(Heartbeat_type))) + return -EFAULT; + return 0; +} - case CCISS_DEREGDISK: - case CCISS_REGNEWD: - case CCISS_REVALIDVOLS: - return rebuild_lun_table(h, 0, 1); +static int cciss_getbustypes(ctlr_info_t *h, void __user *argp) +{ + BusTypes_type BusTypes; - case CCISS_GETLUNINFO:{ - LogvolInfo_struct luninfo; + if (!argp) + return -EINVAL; + BusTypes = readl(&h->cfgtable->BusTypes); + if (copy_to_user(argp, &BusTypes, sizeof(BusTypes_type))) + return -EFAULT; + return 0; +} - memcpy(&luninfo.LunID, drv->LunID, - sizeof(luninfo.LunID)); - luninfo.num_opens = drv->usage_count; - luninfo.num_parts = 0; - if (copy_to_user(argp, &luninfo, - sizeof(LogvolInfo_struct))) - return -EFAULT; - return 0; +static int cciss_getfirmver(ctlr_info_t *h, void __user *argp) +{ + FirmwareVer_type firmware; + + if (!argp) + return -EINVAL; + memcpy(firmware, h->firm_ver, 4); + + if (copy_to_user + (argp, firmware, sizeof(FirmwareVer_type))) + return -EFAULT; + return 0; +} + +static int cciss_getdrivver(ctlr_info_t *h, void __user *argp) +{ + DriverVer_type DriverVer = DRIVER_VERSION; + + if (!argp) + return -EINVAL; + if (copy_to_user(argp, &DriverVer, sizeof(DriverVer_type))) + return -EFAULT; + return 0; +} + +static int cciss_getluninfo(ctlr_info_t *h, + struct gendisk *disk, void __user *argp) +{ + LogvolInfo_struct luninfo; + drive_info_struct *drv = get_drv(disk); + + if (!argp) + return -EINVAL; + memcpy(&luninfo.LunID, drv->LunID, sizeof(luninfo.LunID)); + luninfo.num_opens = drv->usage_count; + luninfo.num_parts = 0; + if (copy_to_user(argp, &luninfo, sizeof(LogvolInfo_struct))) + return -EFAULT; + return 0; +} + +static int cciss_passthru(ctlr_info_t *h, void __user *argp) +{ + IOCTL_Command_struct iocommand; + CommandList_struct *c; + char *buff = NULL; + u64bit temp64; + DECLARE_COMPLETION_ONSTACK(wait); + + if (!argp) + return -EINVAL; + + if (!capable(CAP_SYS_RAWIO)) + return -EPERM; + + if (copy_from_user + (&iocommand, argp, sizeof(IOCTL_Command_struct))) + return -EFAULT; + if ((iocommand.buf_size < 1) && + (iocommand.Request.Type.Direction != XFER_NONE)) { + return -EINVAL; + } + if (iocommand.buf_size > 0) { + buff = kmalloc(iocommand.buf_size, GFP_KERNEL); + if (buff == NULL) + return -EFAULT; + } + if (iocommand.Request.Type.Direction == XFER_WRITE) { + /* Copy the data into the buffer we created */ + if (copy_from_user(buff, iocommand.buf, iocommand.buf_size)) { + kfree(buff); + return -EFAULT; } - case CCISS_PASSTHRU: - { - IOCTL_Command_struct iocommand; - CommandList_struct *c; - char *buff = NULL; - u64bit temp64; - DECLARE_COMPLETION_ONSTACK(wait); - - if (!arg) - return -EINVAL; - - if (!capable(CAP_SYS_RAWIO)) - return -EPERM; - - if (copy_from_user - (&iocommand, argp, sizeof(IOCTL_Command_struct))) - return -EFAULT; - if ((iocommand.buf_size < 1) && - (iocommand.Request.Type.Direction != XFER_NONE)) { - return -EINVAL; - } -#if 0 /* 'buf_size' member is 16-bits, and always smaller than kmalloc limit */ - /* Check kmalloc limits */ - if (iocommand.buf_size > 128000) - return -EINVAL; -#endif - if (iocommand.buf_size > 0) { - buff = kmalloc(iocommand.buf_size, GFP_KERNEL); - if (buff == NULL) - return -EFAULT; - } - if (iocommand.Request.Type.Direction == XFER_WRITE) { - /* Copy the data into the buffer we created */ - if (copy_from_user - (buff, iocommand.buf, iocommand.buf_size)) { - kfree(buff); - return -EFAULT; - } - } else { - memset(buff, 0, iocommand.buf_size); - } - c = cmd_special_alloc(h); - if (!c) { - kfree(buff); - return -ENOMEM; - } - /* Fill in the command type */ - c->cmd_type = CMD_IOCTL_PEND; - /* Fill in Command Header */ - c->Header.ReplyQueue = 0; /* unused in simple mode */ - if (iocommand.buf_size > 0) /* buffer to fill */ - { - c->Header.SGList = 1; - c->Header.SGTotal = 1; - } else /* no buffers to fill */ - { - c->Header.SGList = 0; - c->Header.SGTotal = 0; - } - c->Header.LUN = iocommand.LUN_info; - /* use the kernel address the cmd block for tag */ - c->Header.Tag.lower = c->busaddr; - - /* Fill in Request block */ - c->Request = iocommand.Request; - - /* Fill in the scatter gather information */ - if (iocommand.buf_size > 0) { - temp64.val = pci_map_single(h->pdev, buff, - iocommand.buf_size, - PCI_DMA_BIDIRECTIONAL); - c->SG[0].Addr.lower = temp64.val32.lower; - c->SG[0].Addr.upper = temp64.val32.upper; - c->SG[0].Len = iocommand.buf_size; - c->SG[0].Ext = 0; /* we are not chaining */ - } - c->waiting = &wait; + } else { + memset(buff, 0, iocommand.buf_size); + } + c = cmd_special_alloc(h); + if (!c) { + kfree(buff); + return -ENOMEM; + } + /* Fill in the command type */ + c->cmd_type = CMD_IOCTL_PEND; + /* Fill in Command Header */ + c->Header.ReplyQueue = 0; /* unused in simple mode */ + if (iocommand.buf_size > 0) { /* buffer to fill */ + c->Header.SGList = 1; + c->Header.SGTotal = 1; + } else { /* no buffers to fill */ + c->Header.SGList = 0; + c->Header.SGTotal = 0; + } + c->Header.LUN = iocommand.LUN_info; + /* use the kernel address the cmd block for tag */ + c->Header.Tag.lower = c->busaddr; - enqueue_cmd_and_start_io(h, c); - wait_for_completion(&wait); + /* Fill in Request block */ + c->Request = iocommand.Request; - /* unlock the buffers from DMA */ - temp64.val32.lower = c->SG[0].Addr.lower; - temp64.val32.upper = c->SG[0].Addr.upper; - pci_unmap_single(h->pdev, (dma_addr_t) temp64.val, - iocommand.buf_size, - PCI_DMA_BIDIRECTIONAL); + /* Fill in the scatter gather information */ + if (iocommand.buf_size > 0) { + temp64.val = pci_map_single(h->pdev, buff, + iocommand.buf_size, PCI_DMA_BIDIRECTIONAL); + c->SG[0].Addr.lower = temp64.val32.lower; + c->SG[0].Addr.upper = temp64.val32.upper; + c->SG[0].Len = iocommand.buf_size; + c->SG[0].Ext = 0; /* we are not chaining */ + } + c->waiting = &wait; - check_ioctl_unit_attention(h, c); + enqueue_cmd_and_start_io(h, c); + wait_for_completion(&wait); - /* Copy the error information out */ - iocommand.error_info = *(c->err_info); - if (copy_to_user - (argp, &iocommand, sizeof(IOCTL_Command_struct))) { - kfree(buff); - cmd_special_free(h, c); - return -EFAULT; - } + /* unlock the buffers from DMA */ + temp64.val32.lower = c->SG[0].Addr.lower; + temp64.val32.upper = c->SG[0].Addr.upper; + pci_unmap_single(h->pdev, (dma_addr_t) temp64.val, iocommand.buf_size, + PCI_DMA_BIDIRECTIONAL); + check_ioctl_unit_attention(h, c); + + /* Copy the error information out */ + iocommand.error_info = *(c->err_info); + if (copy_to_user(argp, &iocommand, sizeof(IOCTL_Command_struct))) { + kfree(buff); + cmd_special_free(h, c); + return -EFAULT; + } - if (iocommand.Request.Type.Direction == XFER_READ) { - /* Copy the data out of the buffer we created */ - if (copy_to_user - (iocommand.buf, buff, iocommand.buf_size)) { - kfree(buff); - cmd_special_free(h, c); - return -EFAULT; - } - } + if (iocommand.Request.Type.Direction == XFER_READ) { + /* Copy the data out of the buffer we created */ + if (copy_to_user(iocommand.buf, buff, iocommand.buf_size)) { kfree(buff); cmd_special_free(h, c); - return 0; + return -EFAULT; } - case CCISS_BIG_PASSTHRU:{ - BIG_IOCTL_Command_struct *ioc; - CommandList_struct *c; - unsigned char **buff = NULL; - int *buff_size = NULL; - u64bit temp64; - BYTE sg_used = 0; - int status = 0; - int i; - DECLARE_COMPLETION_ONSTACK(wait); - __u32 left; - __u32 sz; - BYTE __user *data_ptr; - - if (!arg) - return -EINVAL; - if (!capable(CAP_SYS_RAWIO)) - return -EPERM; - ioc = (BIG_IOCTL_Command_struct *) - kmalloc(sizeof(*ioc), GFP_KERNEL); - if (!ioc) { - status = -ENOMEM; - goto cleanup1; - } - if (copy_from_user(ioc, argp, sizeof(*ioc))) { + } + kfree(buff); + cmd_special_free(h, c); + return 0; +} + +static int cciss_bigpassthru(ctlr_info_t *h, void __user *argp) +{ + BIG_IOCTL_Command_struct *ioc; + CommandList_struct *c; + unsigned char **buff = NULL; + int *buff_size = NULL; + u64bit temp64; + BYTE sg_used = 0; + int status = 0; + int i; + DECLARE_COMPLETION_ONSTACK(wait); + __u32 left; + __u32 sz; + BYTE __user *data_ptr; + + if (!argp) + return -EINVAL; + if (!capable(CAP_SYS_RAWIO)) + return -EPERM; + ioc = (BIG_IOCTL_Command_struct *) + kmalloc(sizeof(*ioc), GFP_KERNEL); + if (!ioc) { + status = -ENOMEM; + goto cleanup1; + } + if (copy_from_user(ioc, argp, sizeof(*ioc))) { + status = -EFAULT; + goto cleanup1; + } + if ((ioc->buf_size < 1) && + (ioc->Request.Type.Direction != XFER_NONE)) { + status = -EINVAL; + goto cleanup1; + } + /* Check kmalloc limits using all SGs */ + if (ioc->malloc_size > MAX_KMALLOC_SIZE) { + status = -EINVAL; + goto cleanup1; + } + if (ioc->buf_size > ioc->malloc_size * MAXSGENTRIES) { + status = -EINVAL; + goto cleanup1; + } + buff = kzalloc(MAXSGENTRIES * sizeof(char *), GFP_KERNEL); + if (!buff) { + status = -ENOMEM; + goto cleanup1; + } + buff_size = kmalloc(MAXSGENTRIES * sizeof(int), GFP_KERNEL); + if (!buff_size) { + status = -ENOMEM; + goto cleanup1; + } + left = ioc->buf_size; + data_ptr = ioc->buf; + while (left) { + sz = (left > ioc->malloc_size) ? ioc->malloc_size : left; + buff_size[sg_used] = sz; + buff[sg_used] = kmalloc(sz, GFP_KERNEL); + if (buff[sg_used] == NULL) { + status = -ENOMEM; + goto cleanup1; + } + if (ioc->Request.Type.Direction == XFER_WRITE) { + if (copy_from_user(buff[sg_used], data_ptr, sz)) { status = -EFAULT; goto cleanup1; } - if ((ioc->buf_size < 1) && - (ioc->Request.Type.Direction != XFER_NONE)) { - status = -EINVAL; - goto cleanup1; - } - /* Check kmalloc limits using all SGs */ - if (ioc->malloc_size > MAX_KMALLOC_SIZE) { - status = -EINVAL; - goto cleanup1; - } - if (ioc->buf_size > ioc->malloc_size * MAXSGENTRIES) { - status = -EINVAL; - goto cleanup1; - } - buff = - kzalloc(MAXSGENTRIES * sizeof(char *), GFP_KERNEL); - if (!buff) { - status = -ENOMEM; - goto cleanup1; - } - buff_size = kmalloc(MAXSGENTRIES * sizeof(int), - GFP_KERNEL); - if (!buff_size) { - status = -ENOMEM; - goto cleanup1; - } - left = ioc->buf_size; - data_ptr = ioc->buf; - while (left) { - sz = (left > - ioc->malloc_size) ? ioc-> - malloc_size : left; - buff_size[sg_used] = sz; - buff[sg_used] = kmalloc(sz, GFP_KERNEL); - if (buff[sg_used] == NULL) { - status = -ENOMEM; - goto cleanup1; - } - if (ioc->Request.Type.Direction == XFER_WRITE) { - if (copy_from_user - (buff[sg_used], data_ptr, sz)) { - status = -EFAULT; - goto cleanup1; - } - } else { - memset(buff[sg_used], 0, sz); - } - left -= sz; - data_ptr += sz; - sg_used++; - } - c = cmd_special_alloc(h); - if (!c) { - status = -ENOMEM; - goto cleanup1; - } - c->cmd_type = CMD_IOCTL_PEND; - c->Header.ReplyQueue = 0; + } else { + memset(buff[sg_used], 0, sz); + } + left -= sz; + data_ptr += sz; + sg_used++; + } + c = cmd_special_alloc(h); + if (!c) { + status = -ENOMEM; + goto cleanup1; + } + c->cmd_type = CMD_IOCTL_PEND; + c->Header.ReplyQueue = 0; + c->Header.SGList = sg_used; + c->Header.SGTotal = sg_used; + c->Header.LUN = ioc->LUN_info; + c->Header.Tag.lower = c->busaddr; - if (ioc->buf_size > 0) { - c->Header.SGList = sg_used; - c->Header.SGTotal = sg_used; - } else { - c->Header.SGList = 0; - c->Header.SGTotal = 0; - } - c->Header.LUN = ioc->LUN_info; - c->Header.Tag.lower = c->busaddr; - - c->Request = ioc->Request; - if (ioc->buf_size > 0) { - for (i = 0; i < sg_used; i++) { - temp64.val = - pci_map_single(h->pdev, buff[i], - buff_size[i], - PCI_DMA_BIDIRECTIONAL); - c->SG[i].Addr.lower = - temp64.val32.lower; - c->SG[i].Addr.upper = - temp64.val32.upper; - c->SG[i].Len = buff_size[i]; - c->SG[i].Ext = 0; /* we are not chaining */ - } - } - c->waiting = &wait; - enqueue_cmd_and_start_io(h, c); - wait_for_completion(&wait); - /* unlock the buffers from DMA */ - for (i = 0; i < sg_used; i++) { - temp64.val32.lower = c->SG[i].Addr.lower; - temp64.val32.upper = c->SG[i].Addr.upper; - pci_unmap_single(h->pdev, - (dma_addr_t) temp64.val, buff_size[i], - PCI_DMA_BIDIRECTIONAL); - } - check_ioctl_unit_attention(h, c); - /* Copy the error information out */ - ioc->error_info = *(c->err_info); - if (copy_to_user(argp, ioc, sizeof(*ioc))) { + c->Request = ioc->Request; + for (i = 0; i < sg_used; i++) { + temp64.val = pci_map_single(h->pdev, buff[i], buff_size[i], + PCI_DMA_BIDIRECTIONAL); + c->SG[i].Addr.lower = temp64.val32.lower; + c->SG[i].Addr.upper = temp64.val32.upper; + c->SG[i].Len = buff_size[i]; + c->SG[i].Ext = 0; /* we are not chaining */ + } + c->waiting = &wait; + enqueue_cmd_and_start_io(h, c); + wait_for_completion(&wait); + /* unlock the buffers from DMA */ + for (i = 0; i < sg_used; i++) { + temp64.val32.lower = c->SG[i].Addr.lower; + temp64.val32.upper = c->SG[i].Addr.upper; + pci_unmap_single(h->pdev, + (dma_addr_t) temp64.val, buff_size[i], + PCI_DMA_BIDIRECTIONAL); + } + check_ioctl_unit_attention(h, c); + /* Copy the error information out */ + ioc->error_info = *(c->err_info); + if (copy_to_user(argp, ioc, sizeof(*ioc))) { + cmd_special_free(h, c); + status = -EFAULT; + goto cleanup1; + } + if (ioc->Request.Type.Direction == XFER_READ) { + /* Copy the data out of the buffer we created */ + BYTE __user *ptr = ioc->buf; + for (i = 0; i < sg_used; i++) { + if (copy_to_user(ptr, buff[i], buff_size[i])) { cmd_special_free(h, c); status = -EFAULT; goto cleanup1; } - if (ioc->Request.Type.Direction == XFER_READ) { - /* Copy the data out of the buffer we created */ - BYTE __user *ptr = ioc->buf; - for (i = 0; i < sg_used; i++) { - if (copy_to_user - (ptr, buff[i], buff_size[i])) { - cmd_special_free(h, c); - status = -EFAULT; - goto cleanup1; - } - ptr += buff_size[i]; - } - } - cmd_special_free(h, c); - status = 0; - cleanup1: - if (buff) { - for (i = 0; i < sg_used; i++) - kfree(buff[i]); - kfree(buff); - } - kfree(buff_size); - kfree(ioc); - return status; + ptr += buff_size[i]; } + } + cmd_special_free(h, c); + status = 0; +cleanup1: + if (buff) { + for (i = 0; i < sg_used; i++) + kfree(buff[i]); + kfree(buff); + } + kfree(buff_size); + kfree(ioc); + return status; +} + +static int cciss_ioctl(struct block_device *bdev, fmode_t mode, + unsigned int cmd, unsigned long arg) +{ + struct gendisk *disk = bdev->bd_disk; + ctlr_info_t *h = get_host(disk); + void __user *argp = (void __user *)arg; + + dev_dbg(&h->pdev->dev, "cciss_ioctl: Called with cmd=%x %lx\n", + cmd, arg); + switch (cmd) { + case CCISS_GETPCIINFO: + return cciss_getpciinfo(h, argp); + case CCISS_GETINTINFO: + return cciss_getintinfo(h, argp); + case CCISS_SETINTINFO: + return cciss_setintinfo(h, argp); + case CCISS_GETNODENAME: + return cciss_getnodename(h, argp); + case CCISS_SETNODENAME: + return cciss_setnodename(h, argp); + case CCISS_GETHEARTBEAT: + return cciss_getheartbeat(h, argp); + case CCISS_GETBUSTYPES: + return cciss_getbustypes(h, argp); + case CCISS_GETFIRMVER: + return cciss_getfirmver(h, argp); + case CCISS_GETDRIVVER: + return cciss_getdrivver(h, argp); + case CCISS_DEREGDISK: + case CCISS_REGNEWD: + case CCISS_REVALIDVOLS: + return rebuild_lun_table(h, 0, 1); + case CCISS_GETLUNINFO: + return cciss_getluninfo(h, disk, argp); + case CCISS_PASSTHRU: + return cciss_passthru(h, argp); + case CCISS_BIG_PASSTHRU: + return cciss_bigpassthru(h, argp); /* scsi_cmd_ioctl handles these, below, though some are not */ /* very meaningful for cciss. SG_IO is the main one people want. */ @@ -2881,6 +2834,8 @@ static int cciss_revalidate(struct gendisk *disk) InquiryData_struct *inq_buff = NULL; for (logvol = 0; logvol < CISS_MAX_LUN; logvol++) { + if (!h->drv[logvol]) + continue; if (memcmp(h->drv[logvol]->LunID, drv->LunID, sizeof(drv->LunID)) == 0) { FOUND = 1; @@ -3800,7 +3755,7 @@ static void __devinit cciss_wait_for_mode_change_ack(ctlr_info_t *h) for (i = 0; i < MAX_CONFIG_WAIT; i++) { if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) break; - msleep(10); + usleep_range(10000, 20000); } } @@ -3984,13 +3939,9 @@ static int __devinit cciss_lookup_board_id(struct pci_dev *pdev, u32 *board_id) *board_id = ((subsystem_device_id << 16) & 0xffff0000) | subsystem_vendor_id; - for (i = 0; i < ARRAY_SIZE(products); i++) { - /* Stand aside for hpsa driver on request */ - if (cciss_allow_hpsa && products[i].board_id == HPSA_BOUNDARY) - return -ENODEV; + for (i = 0; i < ARRAY_SIZE(products); i++) if (*board_id == products[i].board_id) return i; - } dev_warn(&pdev->dev, "unrecognized board ID: 0x%08x, ignoring.\n", *board_id); return -ENODEV; @@ -4021,18 +3972,31 @@ static int __devinit cciss_pci_find_memory_BAR(struct pci_dev *pdev, return -ENODEV; } -static int __devinit cciss_wait_for_board_ready(ctlr_info_t *h) +static int __devinit cciss_wait_for_board_state(struct pci_dev *pdev, + void __iomem *vaddr, int wait_for_ready) +#define BOARD_READY 1 +#define BOARD_NOT_READY 0 { - int i; + int i, iterations; u32 scratchpad; - for (i = 0; i < CCISS_BOARD_READY_ITERATIONS; i++) { - scratchpad = readl(h->vaddr + SA5_SCRATCHPAD_OFFSET); - if (scratchpad == CCISS_FIRMWARE_READY) - return 0; + if (wait_for_ready) + iterations = CCISS_BOARD_READY_ITERATIONS; + else + iterations = CCISS_BOARD_NOT_READY_ITERATIONS; + + for (i = 0; i < iterations; i++) { + scratchpad = readl(vaddr + SA5_SCRATCHPAD_OFFSET); + if (wait_for_ready) { + if (scratchpad == CCISS_FIRMWARE_READY) + return 0; + } else { + if (scratchpad != CCISS_FIRMWARE_READY) + return 0; + } msleep(CCISS_BOARD_READY_POLL_INTERVAL_MSECS); } - dev_warn(&h->pdev->dev, "board not ready, timed out.\n"); + dev_warn(&pdev->dev, "board not ready, timed out.\n"); return -ENODEV; } @@ -4081,6 +4045,11 @@ static int __devinit cciss_find_cfgtables(ctlr_info_t *h) static void __devinit cciss_get_max_perf_mode_cmds(struct ctlr_info *h) { h->max_commands = readl(&(h->cfgtable->MaxPerformantModeCommands)); + + /* Limit commands in memory limited kdump scenario. */ + if (reset_devices && h->max_commands > 32) + h->max_commands = 32; + if (h->max_commands < 16) { dev_warn(&h->pdev->dev, "Controller reports " "max supported commands of %d, an obvious lie. " @@ -4198,7 +4167,7 @@ static int __devinit cciss_pci_init(ctlr_info_t *h) err = -ENOMEM; goto err_out_free_res; } - err = cciss_wait_for_board_ready(h); + err = cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_READY); if (err) goto err_out_free_res; err = cciss_find_cfgtables(h); @@ -4363,36 +4332,6 @@ static __devinit int cciss_message(struct pci_dev *pdev, unsigned char opcode, u #define cciss_soft_reset_controller(p) cciss_message(p, 1, 0) #define cciss_noop(p) cciss_message(p, 3, 0) -static __devinit int cciss_reset_msi(struct pci_dev *pdev) -{ -/* the #defines are stolen from drivers/pci/msi.h. */ -#define msi_control_reg(base) (base + PCI_MSI_FLAGS) -#define PCI_MSIX_FLAGS_ENABLE (1 << 15) - - int pos; - u16 control = 0; - - pos = pci_find_capability(pdev, PCI_CAP_ID_MSI); - if (pos) { - pci_read_config_word(pdev, msi_control_reg(pos), &control); - if (control & PCI_MSI_FLAGS_ENABLE) { - dev_info(&pdev->dev, "resetting MSI\n"); - pci_write_config_word(pdev, msi_control_reg(pos), control & ~PCI_MSI_FLAGS_ENABLE); - } - } - - pos = pci_find_capability(pdev, PCI_CAP_ID_MSIX); - if (pos) { - pci_read_config_word(pdev, msi_control_reg(pos), &control); - if (control & PCI_MSIX_FLAGS_ENABLE) { - dev_info(&pdev->dev, "resetting MSI-X\n"); - pci_write_config_word(pdev, msi_control_reg(pos), control & ~PCI_MSIX_FLAGS_ENABLE); - } - } - - return 0; -} - static int cciss_controller_hard_reset(struct pci_dev *pdev, void * __iomem vaddr, bool use_doorbell) { @@ -4447,17 +4386,17 @@ static int cciss_controller_hard_reset(struct pci_dev *pdev, * states or using the doorbell register. */ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) { - u16 saved_config_space[32]; u64 cfg_offset; u32 cfg_base_addr; u64 cfg_base_addr_index; void __iomem *vaddr; unsigned long paddr; u32 misc_fw_support, active_transport; - int rc, i; + int rc; CfgTable_struct __iomem *cfgtable; bool use_doorbell; u32 board_id; + u16 command_register; /* For controllers as old a the p600, this is very nearly * the same thing as @@ -4467,14 +4406,6 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) * pci_set_power_state(pci_dev, PCI_D0); * pci_restore_state(pci_dev); * - * but we can't use these nice canned kernel routines on - * kexec, because they also check the MSI/MSI-X state in PCI - * configuration space and do the wrong thing when it is - * set/cleared. Also, the pci_save/restore_state functions - * violate the ordering requirements for restoring the - * configuration space from the CCISS document (see the - * comment below). So we roll our own .... - * * For controllers newer than the P600, the pci power state * method of resetting doesn't work so we have another way * using the doorbell register. @@ -4493,8 +4424,13 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) return -ENODEV; } - for (i = 0; i < 32; i++) - pci_read_config_word(pdev, 2*i, &saved_config_space[i]); + /* 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() + * won't turn the board on before the rest of config space is ready. + */ + pci_disable_device(pdev); + pci_save_state(pdev); /* find the first memory BAR, so we can find the cfg table */ rc = cciss_pci_find_memory_BAR(pdev, &paddr); @@ -4529,26 +4465,32 @@ static __devinit int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) rc = cciss_controller_hard_reset(pdev, vaddr, use_doorbell); if (rc) goto unmap_cfgtable; - - /* Restore the PCI configuration space. The Open CISS - * Specification says, "Restore the PCI Configuration - * Registers, offsets 00h through 60h. It is important to - * restore the command register, 16-bits at offset 04h, - * last. Do not restore the configuration status register, - * 16-bits at offset 06h." Note that the offset is 2*i. - */ - for (i = 0; i < 32; i++) { - if (i == 2 || i == 3) - continue; - pci_write_config_word(pdev, 2*i, saved_config_space[i]); + pci_restore_state(pdev); + rc = pci_enable_device(pdev); + if (rc) { + dev_warn(&pdev->dev, "failed to enable device.\n"); + goto unmap_cfgtable; } - wmb(); - pci_write_config_word(pdev, 4, saved_config_space[2]); + pci_write_config_word(pdev, 4, command_register); /* Some devices (notably the HP Smart Array 5i Controller) need a little pause here */ 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"); + 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"); + rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_READY); + if (rc) { + dev_warn(&pdev->dev, + "failed waiting for board to become ready\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 @@ -4589,8 +4531,6 @@ static __devinit int cciss_init_reset_devices(struct pci_dev *pdev) return 0; /* just try to do the kdump anyhow. */ if (rc) return -ENODEV; - if (cciss_reset_msi(pdev)) - return -ENODEV; /* Now try to get the controller to respond to a no-op */ for (i = 0; i < CCISS_POST_RESET_NOOP_RETRIES; i++) { @@ -4986,7 +4926,8 @@ static void __exit cciss_cleanup(void) } } kthread_stop(cciss_scan_thread); - remove_proc_entry("driver/cciss", NULL); + if (proc_cciss) + remove_proc_entry("driver/cciss", NULL); bus_unregister(&cciss_bus_type); } diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h index 7915dd4..579f749 100644 --- a/drivers/block/cciss.h +++ b/drivers/block/cciss.h @@ -200,10 +200,14 @@ struct ctlr_info * the above. */ #define CCISS_BOARD_READY_WAIT_SECS (120) +#define CCISS_BOARD_NOT_READY_WAIT_SECS (10) #define CCISS_BOARD_READY_POLL_INTERVAL_MSECS (100) #define CCISS_BOARD_READY_ITERATIONS \ ((CCISS_BOARD_READY_WAIT_SECS * 1000) / \ CCISS_BOARD_READY_POLL_INTERVAL_MSECS) +#define CCISS_BOARD_NOT_READY_ITERATIONS \ + ((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_RETRIES (12) diff --git a/drivers/block/cciss_scsi.c b/drivers/block/cciss_scsi.c index 575495f..727d022 100644 --- a/drivers/block/cciss_scsi.c +++ b/drivers/block/cciss_scsi.c @@ -62,8 +62,8 @@ static int cciss_scsi_proc_info( int length, /* length of data in buffer */ int func); /* 0 == read, 1 == write */ -static int cciss_scsi_queue_command (struct scsi_cmnd *cmd, - void (* done)(struct scsi_cmnd *)); +static int cciss_scsi_queue_command (struct Scsi_Host *h, + struct scsi_cmnd *cmd); static int cciss_eh_device_reset_handler(struct scsi_cmnd *); static int cciss_eh_abort_handler(struct scsi_cmnd *); @@ -1406,7 +1406,7 @@ static void cciss_scatter_gather(ctlr_info_t *h, CommandList_struct *c, static int -cciss_scsi_queue_command (struct scsi_cmnd *cmd, void (* done)(struct scsi_cmnd *)) +cciss_scsi_queue_command_lck(struct scsi_cmnd *cmd, void (*done)(struct scsi_cmnd *)) { ctlr_info_t *h; int rc; @@ -1504,6 +1504,8 @@ cciss_scsi_queue_command (struct scsi_cmnd *cmd, void (* done)(struct scsi_cmnd return 0; } +static DEF_SCSI_QCMD(cciss_scsi_queue_command) + static void cciss_unregister_scsi(ctlr_info_t *h) { struct cciss_scsi_adapter_data_t *sa; diff --git a/drivers/block/cpqarray.c b/drivers/block/cpqarray.c index d53b029..946dad4 100644 --- a/drivers/block/cpqarray.c +++ b/drivers/block/cpqarray.c @@ -35,7 +35,7 @@ #include <linux/seq_file.h> #include <linux/init.h> #include <linux/hdreg.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/spinlock.h> #include <linux/blkdev.h> #include <linux/genhd.h> @@ -68,6 +68,7 @@ MODULE_LICENSE("GPL"); #define CPQARRAY_DMA_MASK 0xFFFFFFFF /* 32 bit DMA */ +static DEFINE_MUTEX(cpqarray_mutex); static int nr_ctlr; static ctlr_info_t *hba[MAX_CTLR]; @@ -845,9 +846,9 @@ static int ida_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&cpqarray_mutex); ret = ida_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&cpqarray_mutex); return ret; } @@ -859,10 +860,10 @@ static int ida_release(struct gendisk *disk, fmode_t mode) { ctlr_info_t *host; - lock_kernel(); + mutex_lock(&cpqarray_mutex); host = get_host(disk); host->usage_count--; - unlock_kernel(); + mutex_unlock(&cpqarray_mutex); return 0; } @@ -1217,9 +1218,9 @@ static int ida_ioctl(struct block_device *bdev, fmode_t mode, { int ret; - lock_kernel(); + mutex_lock(&cpqarray_mutex); ret = ida_locked_ioctl(bdev, mode, cmd, param); - unlock_kernel(); + mutex_unlock(&cpqarray_mutex); return ret; } diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index 9400845..ba95cba 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -78,11 +78,10 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev, init_completion(&md_io.event); md_io.error = 0; - if ((rw & WRITE) && !test_bit(MD_NO_BARRIER, &mdev->flags)) - rw |= REQ_HARDBARRIER; + if ((rw & WRITE) && !test_bit(MD_NO_FUA, &mdev->flags)) + rw |= REQ_FUA; rw |= REQ_UNPLUG | REQ_SYNC; - retry: bio = bio_alloc(GFP_NOIO, 1); bio->bi_bdev = bdev->md_bdev; bio->bi_sector = sector; @@ -100,17 +99,6 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev, wait_for_completion(&md_io.event); ok = bio_flagged(bio, BIO_UPTODATE) && md_io.error == 0; - /* check for unsupported barrier op. - * would rather check on EOPNOTSUPP, but that is not reliable. - * don't try again for ANY return value != 0 */ - if (unlikely((bio->bi_rw & REQ_HARDBARRIER) && !ok)) { - /* Try again with no barrier */ - dev_warn(DEV, "Barriers not supported on meta data device - disabling\n"); - set_bit(MD_NO_BARRIER, &mdev->flags); - rw &= ~REQ_HARDBARRIER; - bio_put(bio); - goto retry; - } out: bio_put(bio); return ok; @@ -284,18 +272,32 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) u32 xor_sum = 0; if (!get_ldev(mdev)) { - dev_err(DEV, "get_ldev() failed in w_al_write_transaction\n"); + dev_err(DEV, + "disk is %s, cannot start al transaction (-%d +%d)\n", + drbd_disk_str(mdev->state.disk), evicted, new_enr); complete(&((struct update_al_work *)w)->event); return 1; } /* do we have to do a bitmap write, first? * TODO reduce maximum latency: * submit both bios, then wait for both, - * instead of doing two synchronous sector writes. */ + * instead of doing two synchronous sector writes. + * 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); - mutex_lock(&mdev->md_io_mutex); /* protects md_io_page, al_tr_cycle, ... */ + /* The bitmap write may have failed, causing a state change. */ + if (mdev->state.disk < D_INCONSISTENT) { + dev_err(DEV, + "disk is %s, cannot write al transaction (-%d +%d)\n", + drbd_disk_str(mdev->state.disk), evicted, new_enr); + complete(&((struct update_al_work *)w)->event); + put_ldev(mdev); + return 1; + } + + mutex_lock(&mdev->md_io_mutex); /* protects md_io_buffer, al_tr_cycle, ... */ buffer = (struct al_transaction *)page_address(mdev->md_io_page); buffer->magic = __constant_cpu_to_be32(DRBD_MAGIC); @@ -739,7 +741,7 @@ void drbd_al_apply_to_bm(struct drbd_conf *mdev) unsigned int enr; unsigned long add = 0; char ppb[10]; - int i; + int i, tmp; wait_event(mdev->al_wait, lc_try_lock(mdev->act_log)); @@ -747,7 +749,9 @@ void drbd_al_apply_to_bm(struct drbd_conf *mdev) enr = lc_element_by_index(mdev->act_log, i)->lc_number; if (enr == LC_FREE) continue; - add += drbd_bm_ALe_set_all(mdev, enr); + tmp = drbd_bm_ALe_set_all(mdev, enr); + dynamic_dev_dbg(DEV, "AL: set %d bits in extent %u\n", tmp, enr); + add += tmp; } lc_unlock(mdev->act_log); @@ -965,29 +969,30 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, * ok, (capacity & 7) != 0 sometimes, but who cares... * we count rs_{total,left} in bits, not sectors. */ - spin_lock_irqsave(&mdev->al_lock, flags); count = drbd_bm_clear_bits(mdev, sbnr, ebnr); - if (count) { - /* we need the lock for drbd_try_clear_on_disk_bm */ - if (jiffies - mdev->rs_mark_time > HZ*10) { - /* should be rolling marks, - * but we estimate only anyways. */ - if (mdev->rs_mark_left != drbd_bm_total_weight(mdev) && + 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 = jiffies; - mdev->rs_mark_left = drbd_bm_total_weight(mdev); + mdev->rs_mark_time[next] = now; + mdev->rs_mark_left[next] = tw; + mdev->rs_last_mark = next; } } - if (get_ldev(mdev)) { - drbd_try_clear_on_disk_bm(mdev, sector, count, TRUE); - put_ldev(mdev); - } + spin_lock_irqsave(&mdev->al_lock, flags); + 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(), * lc_put() in drbd_try_clear_on_disk_bm(). */ wake_up = 1; + put_ldev(mdev); } - spin_unlock_irqrestore(&mdev->al_lock, flags); if (wake_up) wake_up(&mdev->al_wait); } @@ -1118,7 +1123,7 @@ static int _is_in_al(struct drbd_conf *mdev, unsigned int enr) * @mdev: DRBD device. * @sector: The sector number. * - * This functions sleeps on al_wait. Returns 1 on success, 0 if interrupted. + * This functions sleeps on al_wait. Returns 0 on success, -EINTR if interrupted. */ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector) { @@ -1129,10 +1134,10 @@ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector) sig = wait_event_interruptible(mdev->al_wait, (bm_ext = _bme_get(mdev, enr))); if (sig) - return 0; + return -EINTR; if (test_bit(BME_LOCKED, &bm_ext->flags)) - return 1; + return 0; for (i = 0; i < AL_EXT_PER_BM_SECT; i++) { sig = wait_event_interruptible(mdev->al_wait, @@ -1145,13 +1150,11 @@ int drbd_rs_begin_io(struct drbd_conf *mdev, sector_t sector) wake_up(&mdev->al_wait); } spin_unlock_irq(&mdev->al_lock); - return 0; + return -EINTR; } } - set_bit(BME_LOCKED, &bm_ext->flags); - - return 1; + return 0; } /** diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index e3f88d6..fd42832 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -569,7 +569,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) * * maybe bm_set should be atomic_t ? */ -static unsigned long _drbd_bm_total_weight(struct drbd_conf *mdev) +unsigned long _drbd_bm_total_weight(struct drbd_conf *mdev) { struct drbd_bitmap *b = mdev->bitmap; unsigned long s; diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 352441b..3803a03 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -114,11 +114,11 @@ struct drbd_conf; #define D_ASSERT(exp) if (!(exp)) \ dev_err(DEV, "ASSERT( " #exp " ) in %s:%d\n", __FILE__, __LINE__) -#define ERR_IF(exp) if (({ \ - int _b = (exp) != 0; \ - if (_b) dev_err(DEV, "%s: (%s) in %s:%d\n", \ - __func__, #exp, __FILE__, __LINE__); \ - _b; \ +#define ERR_IF(exp) if (({ \ + int _b = (exp) != 0; \ + if (_b) dev_err(DEV, "ASSERT FAILED: %s: (%s) in %s:%d\n", \ + __func__, #exp, __FILE__, __LINE__); \ + _b; \ })) /* Defines to control fault insertion */ @@ -337,13 +337,25 @@ static inline void bm_xfer_ctx_bit_to_word_offset(struct bm_xfer_ctx *c) * NOTE that the payload starts at a long aligned offset, * regardless of 32 or 64 bit arch! */ -struct p_header { +struct p_header80 { u32 magic; u16 command; u16 length; /* bytes of data after this header */ u8 payload[0]; } __packed; -/* 8 bytes. packet FIXED for the next century! */ + +/* Header for big packets, Used for data packets exceeding 64kB */ +struct p_header95 { + u16 magic; /* use DRBD_MAGIC_BIG here */ + u16 command; + u32 length; /* Use only 24 bits of that. Ignore the highest 8 bit. */ + u8 payload[0]; +} __packed; + +union p_header { + struct p_header80 h80; + struct p_header95 h95; +}; /* * short commands, packets without payload, plain p_header: @@ -362,12 +374,16 @@ struct p_header { */ /* these defines must not be changed without changing the protocol version */ -#define DP_HARDBARRIER 1 -#define DP_RW_SYNC 2 +#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_FUA 16 /* equals REQ_FUA */ +#define DP_FLUSH 32 /* equals REQ_FLUSH */ +#define DP_DISCARD 64 /* equals REQ_DISCARD */ struct p_data { - struct p_header head; + union p_header head; u64 sector; /* 64 bits sector number */ u64 block_id; /* to identify the request in protocol B&C */ u32 seq_num; @@ -383,7 +399,7 @@ struct p_data { * P_DATA_REQUEST, P_RS_DATA_REQUEST */ struct p_block_ack { - struct p_header head; + struct p_header80 head; u64 sector; u64 block_id; u32 blksize; @@ -392,7 +408,7 @@ struct p_block_ack { struct p_block_req { - struct p_header head; + struct p_header80 head; u64 sector; u64 block_id; u32 blksize; @@ -409,7 +425,7 @@ struct p_block_req { */ struct p_handshake { - struct p_header head; /* 8 bytes */ + struct p_header80 head; /* 8 bytes */ u32 protocol_min; u32 feature_flags; u32 protocol_max; @@ -424,19 +440,19 @@ struct p_handshake { /* 80 bytes, FIXED for the next century */ struct p_barrier { - struct p_header head; + struct p_header80 head; u32 barrier; /* barrier number _handle_ only */ u32 pad; /* to multiple of 8 Byte */ } __packed; struct p_barrier_ack { - struct p_header head; + struct p_header80 head; u32 barrier; u32 set_size; } __packed; struct p_rs_param { - struct p_header head; + struct p_header80 head; u32 rate; /* Since protocol version 88 and higher. */ @@ -444,20 +460,31 @@ struct p_rs_param { } __packed; struct p_rs_param_89 { - struct p_header head; + struct p_header80 head; u32 rate; /* protocol version 89: */ char verify_alg[SHARED_SECRET_MAX]; char csums_alg[SHARED_SECRET_MAX]; } __packed; +struct p_rs_param_95 { + struct p_header80 head; + u32 rate; + char verify_alg[SHARED_SECRET_MAX]; + char csums_alg[SHARED_SECRET_MAX]; + u32 c_plan_ahead; + u32 c_delay_target; + u32 c_fill_target; + u32 c_max_rate; +} __packed; + enum drbd_conn_flags { CF_WANT_LOSE = 1, CF_DRY_RUN = 2, }; struct p_protocol { - struct p_header head; + struct p_header80 head; u32 protocol; u32 after_sb_0p; u32 after_sb_1p; @@ -471,17 +498,17 @@ struct p_protocol { } __packed; struct p_uuids { - struct p_header head; + struct p_header80 head; u64 uuid[UI_EXTENDED_SIZE]; } __packed; struct p_rs_uuid { - struct p_header head; + struct p_header80 head; u64 uuid; } __packed; struct p_sizes { - struct p_header head; + struct p_header80 head; u64 d_size; /* size of disk */ u64 u_size; /* user requested size */ u64 c_size; /* current exported size */ @@ -491,18 +518,18 @@ struct p_sizes { } __packed; struct p_state { - struct p_header head; + struct p_header80 head; u32 state; } __packed; struct p_req_state { - struct p_header head; + struct p_header80 head; u32 mask; u32 val; } __packed; struct p_req_state_reply { - struct p_header head; + struct p_header80 head; u32 retcode; } __packed; @@ -517,7 +544,7 @@ struct p_drbd06_param { } __packed; struct p_discard { - struct p_header head; + struct p_header80 head; u64 block_id; u32 seq_num; u32 pad; @@ -533,7 +560,7 @@ enum drbd_bitmap_code { }; struct p_compressed_bm { - struct p_header head; + struct p_header80 head; /* (encoding & 0x0f): actual encoding, see enum drbd_bitmap_code * (encoding & 0x80): polarity (set/unset) of first runlength * ((encoding >> 4) & 0x07): pad_bits, number of trailing zero bits @@ -544,10 +571,10 @@ struct p_compressed_bm { u8 code[0]; } __packed; -struct p_delay_probe { - struct p_header head; - u32 seq_num; /* sequence number to match the two probe packets */ - u32 offset; /* usecs the probe got sent after the reference time point */ +struct p_delay_probe93 { + struct p_header80 head; + u32 seq_num; /* sequence number to match the two probe packets */ + u32 offset; /* usecs the probe got sent after the reference time point */ } __packed; /* DCBP: Drbd Compressed Bitmap Packet ... */ @@ -594,7 +621,7 @@ DCBP_set_pad_bits(struct p_compressed_bm *p, int n) * so we need to use the fixed size 4KiB page size * most architechtures have used for a long time. */ -#define BM_PACKET_PAYLOAD_BYTES (4096 - sizeof(struct p_header)) +#define BM_PACKET_PAYLOAD_BYTES (4096 - sizeof(struct p_header80)) #define BM_PACKET_WORDS (BM_PACKET_PAYLOAD_BYTES/sizeof(long)) #define BM_PACKET_VLI_BYTES_MAX (4096 - sizeof(struct p_compressed_bm)) #if (PAGE_SIZE < 4096) @@ -603,13 +630,14 @@ DCBP_set_pad_bits(struct p_compressed_bm *p, int n) #endif union p_polymorph { - struct p_header header; + union p_header header; struct p_handshake handshake; struct p_data data; struct p_block_ack block_ack; struct p_barrier barrier; struct p_barrier_ack barrier_ack; struct p_rs_param_89 rs_param_89; + struct p_rs_param_95 rs_param_95; struct p_protocol protocol; struct p_sizes sizes; struct p_uuids uuids; @@ -617,6 +645,8 @@ union p_polymorph { struct p_req_state req_state; struct p_req_state_reply req_state_reply; struct p_block_req block_req; + struct p_delay_probe93 delay_probe93; + struct p_rs_uuid rs_uuid; } __packed; /**********************************************************************/ @@ -697,7 +727,7 @@ struct drbd_tl_epoch { struct list_head requests; /* requests before */ struct drbd_tl_epoch *next; /* pointer to the next barrier */ unsigned int br_number; /* the barriers identifier. */ - int n_req; /* number of requests attached before this barrier */ + int n_writes; /* number of requests attached before this barrier */ }; struct drbd_request; @@ -719,17 +749,12 @@ struct drbd_epoch { /* drbd_epoch flag bits */ enum { - DE_BARRIER_IN_NEXT_EPOCH_ISSUED, - DE_BARRIER_IN_NEXT_EPOCH_DONE, - DE_CONTAINS_A_BARRIER, DE_HAVE_BARRIER_NUMBER, - DE_IS_FINISHING, }; enum epoch_event { EV_PUT, EV_GOT_BARRIER_NR, - EV_BARRIER_DONE, EV_BECAME_LAST, EV_CLEANUP = 32, /* used as flag */ }; @@ -747,7 +772,7 @@ struct digest_info { struct drbd_epoch_entry { struct drbd_work w; struct hlist_node colision; - struct drbd_epoch *epoch; + struct drbd_epoch *epoch; /* for writes */ struct drbd_conf *mdev; struct page *pages; atomic_t pending_bios; @@ -755,7 +780,10 @@ struct drbd_epoch_entry { /* see comments on ee flag bits below */ unsigned long flags; sector_t sector; - u64 block_id; + union { + u64 block_id; + struct digest_info *digest; + }; }; /* ee flag bits. @@ -768,11 +796,6 @@ enum { __EE_CALL_AL_COMPLETE_IO, __EE_MAY_SET_IN_SYNC, - /* This epoch entry closes an epoch using a barrier. - * On sucessful completion, the epoch is released, - * and the P_BARRIER_ACK send. */ - __EE_IS_BARRIER, - /* In case a barrier failed, * we need to resubmit without the barrier flag. */ __EE_RESUBMITTED, @@ -781,12 +804,15 @@ enum { * if any of those fail, we set this flag atomically * from the endio callback */ __EE_WAS_ERROR, + + /* This ee has a pointer to a digest instead of a block id */ + __EE_HAS_DIGEST, }; #define EE_CALL_AL_COMPLETE_IO (1<<__EE_CALL_AL_COMPLETE_IO) #define EE_MAY_SET_IN_SYNC (1<<__EE_MAY_SET_IN_SYNC) -#define EE_IS_BARRIER (1<<__EE_IS_BARRIER) #define EE_RESUBMITTED (1<<__EE_RESUBMITTED) #define EE_WAS_ERROR (1<<__EE_WAS_ERROR) +#define EE_HAS_DIGEST (1<<__EE_HAS_DIGEST) /* global flag bits */ enum { @@ -794,7 +820,6 @@ enum { SIGNAL_ASENDER, /* whether asender wants to be interrupted */ SEND_PING, /* whether asender should send a ping asap */ - STOP_SYNC_TIMER, /* tell timer to cancel itself */ UNPLUG_QUEUED, /* only relevant with kernel 2.4 */ UNPLUG_REMOTE, /* sending a "UnplugRemote" could help */ MD_DIRTY, /* current uuids and flags not yet on disk */ @@ -807,15 +832,15 @@ enum { * 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_BARRIER, /* meta data device does not support barriers, - so don't even try */ + MD_NO_FUA, /* Users wants us to not use FUA/FLUSH on meta data dev */ SUSPEND_IO, /* suspend application io */ BITMAP_IO, /* suspend application io; once no more io in flight, start bitmap io */ BITMAP_IO_QUEUED, /* Started bitmap IO */ + GO_DISKLESS, /* Disk is being detached, on io-error or admin request. */ + WAS_IO_ERROR, /* Local disk failed returned IO error */ RESYNC_AFTER_NEG, /* Resync after online grow after the attach&negotiate finished. */ NET_CONGESTED, /* The data socket is congested */ @@ -829,6 +854,8 @@ enum { * the peer, if it changed there as well. */ CONN_DRY_RUN, /* Expect disconnect after resync handshake. */ 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. */ }; struct drbd_bitmap; /* opaque for drbd_conf */ @@ -838,10 +865,6 @@ struct drbd_bitmap; /* opaque for drbd_conf */ /* THINK maybe we actually want to use the default "event/%s" worker threads * or similar in linux 2.6, which uses per cpu data and threads. - * - * To be general, this might need a spin_lock member. - * For now, please use the mdev->req_lock to protect list_head, - * see drbd_queue_work below. */ struct drbd_work_queue { struct list_head q; @@ -888,8 +911,6 @@ struct drbd_md { struct drbd_backing_dev { struct block_device *backing_bdev; struct block_device *md_bdev; - struct file *lo_file; - struct file *md_file; struct drbd_md md; struct disk_conf dc; /* The user provided config... */ sector_t known_size; /* last known size of that backing device */ @@ -912,7 +933,12 @@ enum write_ordering_e { WO_none, WO_drain_io, WO_bdev_flush, - WO_bio_barrier +}; + +struct fifo_buffer { + int *values; + unsigned int head_index; + unsigned int size; }; struct drbd_conf { @@ -936,9 +962,16 @@ struct drbd_conf { unsigned int ko_count; struct drbd_work resync_work, unplug_work, + go_diskless, md_sync_work; struct timer_list resync_timer; struct timer_list md_sync_timer; +#ifdef DRBD_DEBUG_MD_SYNC + struct { + unsigned int line; + const char* func; + } last_md_mark_dirty; +#endif /* Used after attach while negotiating new disk state. */ union drbd_state new_state_tmp; @@ -946,6 +979,7 @@ struct drbd_conf { union drbd_state state; wait_queue_head_t misc_wait; wait_queue_head_t state_wait; /* upon each state change. */ + wait_queue_head_t net_cnt_wait; unsigned int send_cnt; unsigned int recv_cnt; unsigned int read_cnt; @@ -974,12 +1008,16 @@ struct drbd_conf { unsigned long rs_start; /* cumulated time in PausedSyncX state [unit jiffies] */ unsigned long rs_paused; + /* skipped because csum was equal [unit BM_BLOCK_SIZE] */ + unsigned long rs_same_csum; +#define DRBD_SYNC_MARKS 8 +#define DRBD_SYNC_MARK_STEP (3*HZ) /* block not up-to-date at mark [unit BM_BLOCK_SIZE] */ - unsigned long rs_mark_left; + unsigned long rs_mark_left[DRBD_SYNC_MARKS]; /* marks's time [unit jiffies] */ - unsigned long rs_mark_time; - /* skipped because csum was equeal [unit BM_BLOCK_SIZE] */ - unsigned long rs_same_csum; + unsigned long rs_mark_time[DRBD_SYNC_MARKS]; + /* current index into rs_mark_{left,time} */ + int rs_last_mark; /* where does the admin want us to start? (sector) */ sector_t ov_start_sector; @@ -1012,10 +1050,10 @@ struct drbd_conf { spinlock_t epoch_lock; unsigned int epochs; enum write_ordering_e write_ordering; - struct list_head active_ee; /* IO in progress */ - struct list_head sync_ee; /* IO in progress */ + struct list_head active_ee; /* IO in progress (P_DATA gets written to disk) */ + struct list_head sync_ee; /* IO in progress (P_RS_DATA_REPLY gets written to disk) */ struct list_head done_ee; /* send ack */ - struct list_head read_ee; /* IO in progress */ + struct list_head read_ee; /* IO in progress (any read) */ struct list_head net_ee; /* zero-copy network send in progress */ struct hlist_head *ee_hash; /* is proteced by req_lock! */ unsigned int ee_hash_s; @@ -1026,7 +1064,8 @@ struct drbd_conf { int next_barrier_nr; struct hlist_head *app_reads_hash; /* is proteced by req_lock */ struct list_head resync_reads; - atomic_t pp_in_use; + atomic_t pp_in_use; /* allocated from page pool */ + atomic_t pp_in_use_by_net; /* sendpage()d, still referenced by tcp */ wait_queue_head_t ee_wait; struct page *md_io_page; /* one page buffer for md_io */ struct page *md_io_tmpp; /* for logical_block_size != 512 */ @@ -1054,6 +1093,15 @@ struct drbd_conf { u64 ed_uuid; /* UUID of the exposed data */ struct mutex state_mutex; char congestion_reason; /* Why we where congested... */ + atomic_t rs_sect_in; /* for incoming resync data rate, SyncTarget */ + atomic_t rs_sect_ev; /* for submitted resync data rate, both */ + int rs_last_sect_ev; /* counter to compare with */ + int rs_last_events; /* counter of read or write "events" (unit sectors) + * on the lower level device when we last looked. */ + 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 */ }; static inline struct drbd_conf *minor_to_mdev(unsigned int minor) @@ -1138,6 +1186,8 @@ 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, @@ -1150,12 +1200,12 @@ extern int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_f extern int _drbd_send_state(struct drbd_conf *mdev); extern int drbd_send_state(struct drbd_conf *mdev); extern int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, - enum drbd_packets cmd, struct p_header *h, + enum drbd_packets cmd, struct p_header80 *h, size_t size, unsigned msg_flags); #define USE_DATA_SOCKET 1 #define USE_META_SOCKET 0 extern int drbd_send_cmd(struct drbd_conf *mdev, int use_data_socket, - enum drbd_packets cmd, struct p_header *h, + enum drbd_packets cmd, struct p_header80 *h, size_t size); extern int drbd_send_cmd2(struct drbd_conf *mdev, enum drbd_packets cmd, char *data, size_t size); @@ -1167,7 +1217,7 @@ extern int drbd_send_ack(struct drbd_conf *mdev, enum drbd_packets cmd, extern int drbd_send_ack_rp(struct drbd_conf *mdev, enum drbd_packets cmd, struct p_block_req *rp); extern int drbd_send_ack_dp(struct drbd_conf *mdev, enum drbd_packets cmd, - struct p_data *dp); + 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_block(struct drbd_conf *mdev, enum drbd_packets cmd, @@ -1201,7 +1251,13 @@ extern void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local) extern void drbd_md_set_flag(struct drbd_conf *mdev, int flags) __must_hold(local); extern void drbd_md_clear_flag(struct drbd_conf *mdev, int flags)__must_hold(local); extern int drbd_md_test_flag(struct drbd_backing_dev *, int); +#ifndef DRBD_DEBUG_MD_SYNC extern void drbd_md_mark_dirty(struct drbd_conf *mdev); +#else +#define drbd_md_mark_dirty(m) drbd_md_mark_dirty_(m, __LINE__ , __func__ ) +extern void drbd_md_mark_dirty_(struct drbd_conf *mdev, + unsigned int line, const char *func); +#endif extern void drbd_queue_bitmap_io(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), void (*done)(struct drbd_conf *, int), @@ -1209,6 +1265,8 @@ extern void drbd_queue_bitmap_io(struct drbd_conf *mdev, 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); /* Meta data layout @@ -1264,6 +1322,8 @@ struct bm_extent { * Bit 1 ==> local node thinks this block needs to be synced. */ +#define SLEEP_TIME (HZ/10) + #define BM_BLOCK_SHIFT 12 /* 4k per bit */ #define BM_BLOCK_SIZE (1<<BM_BLOCK_SHIFT) /* (9+3) : 512 bytes @ 8 bits; representing 16M storage @@ -1335,11 +1395,13 @@ struct bm_extent { #endif /* Sector shift value for the "hash" functions of tl_hash and ee_hash tables. - * With a value of 6 all IO in one 32K block make it to the same slot of the + * With a value of 8 all IO in one 128K block make it to the same slot of the * hash table. */ -#define HT_SHIFT 6 +#define HT_SHIFT 8 #define DRBD_MAX_SEGMENT_SIZE (1U<<(9+HT_SHIFT)) +#define DRBD_MAX_SIZE_H80_PACKET (1 << 15) /* The old header only allows packets up to 32Kib data */ + /* Number of elements in the app_reads_hash */ #define APP_R_HSIZE 15 @@ -1369,6 +1431,7 @@ extern unsigned long drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_ /* 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); extern unsigned long _drbd_bm_find_next_zero(struct drbd_conf *mdev, unsigned long bm_fo); +extern unsigned long _drbd_bm_total_weight(struct drbd_conf *mdev); extern unsigned long drbd_bm_total_weight(struct drbd_conf *mdev); extern int drbd_bm_rs_done(struct drbd_conf *mdev); /* for receive_bitmap */ @@ -1421,7 +1484,8 @@ 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); -enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev); +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); /* drbd_worker.c */ @@ -1467,10 +1531,12 @@ 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 void resync_timer_fn(unsigned long data); /* drbd_receiver.c */ +extern int drbd_rs_should_slow_down(struct drbd_conf *mdev); 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); @@ -1479,7 +1545,10 @@ extern struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev, sector_t sector, unsigned int data_size, gfp_t gfp_mask) __must_hold(local); -extern void drbd_free_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e); +extern void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, + int is_net); +#define drbd_free_ee(m,e) drbd_free_some_ee(m, e, 0) +#define drbd_free_net_ee(m,e) drbd_free_some_ee(m, e, 1) extern void drbd_wait_ee_list_empty(struct drbd_conf *mdev, struct list_head *head); extern void _drbd_wait_ee_list_empty(struct drbd_conf *mdev, @@ -1487,6 +1556,7 @@ extern void _drbd_wait_ee_list_empty(struct drbd_conf *mdev, extern void drbd_set_recv_tcq(struct drbd_conf *mdev, int tcq_enabled); extern void _drbd_clear_done_ee(struct drbd_conf *mdev, struct list_head *to_be_freed); extern void drbd_flush_workqueue(struct drbd_conf *mdev); +extern void drbd_free_tl_hash(struct drbd_conf *mdev); /* yes, there is kernel_setsockopt, but only since 2.6.18. we don't need to * mess with get_fs/set_fs, we know we are KERNEL_DS always. */ @@ -1600,6 +1670,8 @@ void drbd_bcast_ee(struct drbd_conf *mdev, #define susp_MASK 1 #define user_isp_MASK 1 #define aftr_isp_MASK 1 +#define susp_nod_MASK 1 +#define susp_fen_MASK 1 #define NS(T, S) \ ({ union drbd_state mask; mask.i = 0; mask.T = T##_MASK; mask; }), \ @@ -1712,17 +1784,17 @@ static inline void __drbd_chk_io_error_(struct drbd_conf *mdev, int forcedetach, case EP_PASS_ON: if (!forcedetach) { if (__ratelimit(&drbd_ratelimit_state)) - dev_err(DEV, "Local IO failed in %s." - "Passing error on...\n", where); + dev_err(DEV, "Local IO failed in %s.\n", where); break; } /* NOTE fall through to detach case if forcedetach set */ case EP_DETACH: case EP_CALL_HELPER: + set_bit(WAS_IO_ERROR, &mdev->flags); if (mdev->state.disk > D_FAILED) { _drbd_set_state(_NS(mdev, disk, D_FAILED), CS_HARD, NULL); - dev_err(DEV, "Local IO failed in %s." - "Detaching...\n", where); + dev_err(DEV, + "Local IO failed in %s. Detaching...\n", where); } break; } @@ -1788,7 +1860,7 @@ static inline sector_t drbd_md_last_sector(struct drbd_backing_dev *bdev) static inline sector_t drbd_get_capacity(struct block_device *bdev) { /* return bdev ? get_capacity(bdev->bd_disk) : 0; */ - return bdev ? bdev->bd_inode->i_size >> 9 : 0; + return bdev ? i_size_read(bdev->bd_inode) >> 9 : 0; } /** @@ -1856,13 +1928,6 @@ static inline sector_t drbd_md_ss__(struct drbd_conf *mdev, } static inline void -_drbd_queue_work(struct drbd_work_queue *q, struct drbd_work *w) -{ - list_add_tail(&w->list, &q->q); - up(&q->s); -} - -static inline void drbd_queue_work_front(struct drbd_work_queue *q, struct drbd_work *w) { unsigned long flags; @@ -1899,19 +1964,19 @@ static inline void request_ping(struct drbd_conf *mdev) static inline int drbd_send_short_cmd(struct drbd_conf *mdev, enum drbd_packets cmd) { - struct p_header h; + struct p_header80 h; return drbd_send_cmd(mdev, USE_DATA_SOCKET, cmd, &h, sizeof(h)); } static inline int drbd_send_ping(struct drbd_conf *mdev) { - struct p_header h; + struct p_header80 h; return drbd_send_cmd(mdev, USE_META_SOCKET, P_PING, &h, sizeof(h)); } static inline int drbd_send_ping_ack(struct drbd_conf *mdev) { - struct p_header h; + struct p_header80 h; return drbd_send_cmd(mdev, USE_META_SOCKET, P_PING_ACK, &h, sizeof(h)); } @@ -2013,7 +2078,7 @@ static inline void inc_unacked(struct drbd_conf *mdev) static inline void put_net_conf(struct drbd_conf *mdev) { if (atomic_dec_and_test(&mdev->net_cnt)) - wake_up(&mdev->misc_wait); + wake_up(&mdev->net_cnt_wait); } /** @@ -2044,10 +2109,18 @@ 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); __release(local); - if (atomic_dec_and_test(&mdev->local_cnt)) + D_ASSERT(i >= 0); + if (i == 0) { + if (mdev->state.disk == D_DISKLESS) + /* even internal references gone, safe to destroy */ + drbd_ldev_destroy(mdev); + if (mdev->state.disk == D_FAILED) + /* all application IO references gone. */ + drbd_go_diskless(mdev); wake_up(&mdev->misc_wait); - D_ASSERT(atomic_read(&mdev->local_cnt) >= 0); + } } #ifndef __CHECKER__ @@ -2055,6 +2128,10 @@ static inline int _get_ldev_if_state(struct drbd_conf *mdev, enum drbd_disk_stat { int io_allowed; + /* never get a reference while D_DISKLESS */ + if (mdev->state.disk == D_DISKLESS) + return 0; + atomic_inc(&mdev->local_cnt); io_allowed = (mdev->state.disk >= mins); if (!io_allowed) @@ -2179,11 +2256,16 @@ static inline int drbd_state_is_stable(union drbd_state s) return 1; } +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) { int mxb = drbd_get_max_buffers(mdev); - if (mdev->state.susp) + if (is_susp(mdev->state)) return 0; if (test_bit(SUSPEND_IO, &mdev->flags)) return 0; @@ -2318,13 +2400,12 @@ static inline void drbd_md_flush(struct drbd_conf *mdev) { int r; - if (test_bit(MD_NO_BARRIER, &mdev->flags)) + if (test_bit(MD_NO_FUA, &mdev->flags)) return; - r = blkdev_issue_flush(mdev->ldev->md_bdev, GFP_KERNEL, NULL, - BLKDEV_IFL_WAIT); + r = blkdev_issue_flush(mdev->ldev->md_bdev, GFP_KERNEL, NULL); if (r) { - set_bit(MD_NO_BARRIER, &mdev->flags); + set_bit(MD_NO_FUA, &mdev->flags); dev_err(DEV, "meta data flush failed with status %d, disabling md-flushes\n", r); } } diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index fa650dd..29cd0dc 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -32,7 +32,7 @@ #include <asm/types.h> #include <net/sock.h> #include <linux/ctype.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/fs.h> #include <linux/file.h> #include <linux/proc_fs.h> @@ -64,6 +64,7 @@ struct after_state_chg_work { struct completion *done; }; +static DEFINE_MUTEX(drbd_main_mutex); int drbdd_init(struct drbd_thread *); int drbd_worker(struct drbd_thread *); int drbd_asender(struct drbd_thread *); @@ -77,6 +78,7 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, static int w_md_sync(struct drbd_conf *mdev, struct drbd_work *w, int unused); static void md_sync_timer_fn(unsigned long data); static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused); +static int w_go_diskless(struct drbd_conf *mdev, struct drbd_work *w, int unused); MODULE_AUTHOR("Philipp Reisner <phil@linbit.com>, " "Lars Ellenberg <lars@linbit.com>"); @@ -199,7 +201,7 @@ static int tl_init(struct drbd_conf *mdev) INIT_LIST_HEAD(&b->w.list); b->next = NULL; b->br_number = 4711; - b->n_req = 0; + b->n_writes = 0; b->w.cb = NULL; /* if this is != NULL, we need to dec_ap_pending in tl_clear */ mdev->oldest_tle = b; @@ -240,7 +242,7 @@ void _tl_add_barrier(struct drbd_conf *mdev, struct drbd_tl_epoch *new) INIT_LIST_HEAD(&new->w.list); new->w.cb = NULL; /* if this is != NULL, we need to dec_ap_pending in tl_clear */ new->next = NULL; - new->n_req = 0; + new->n_writes = 0; newest_before = mdev->newest_tle; /* never send a barrier number == 0, because that is special-cased @@ -284,9 +286,9 @@ void tl_release(struct drbd_conf *mdev, unsigned int barrier_nr, barrier_nr, b->br_number); goto bail; } - if (b->n_req != set_size) { - dev_err(DEV, "BAD! BarrierAck #%u received with n_req=%u, expected n_req=%u!\n", - barrier_nr, set_size, b->n_req); + if (b->n_writes != set_size) { + dev_err(DEV, "BAD! BarrierAck #%u received with n_writes=%u, expected n_writes=%u!\n", + barrier_nr, set_size, b->n_writes); goto bail; } @@ -333,6 +335,82 @@ 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. + * @what: The action/event to perform with all request objects + * + * @what might be one of connection_lost_while_pending, resend, fail_frozen_disk_io, + * restart_frozen_disk_io. + */ +static void _tl_restart(struct drbd_conf *mdev, enum drbd_req_event what) +{ + struct drbd_tl_epoch *b, *tmp, **pn; + struct list_head *le, *tle, carry_reads; + struct drbd_request *req; + int rv, n_writes, n_reads; + + b = mdev->oldest_tle; + pn = &mdev->oldest_tle; + while (b) { + n_writes = 0; + n_reads = 0; + INIT_LIST_HEAD(&carry_reads); + list_for_each_safe(le, tle, &b->requests) { + req = list_entry(le, struct drbd_request, tl_requests); + rv = _req_mod(req, what); + + n_writes += (rv & MR_WRITE) >> MR_WRITE_SHIFT; + n_reads += (rv & MR_READ) >> MR_READ_SHIFT; + } + tmp = b->next; + + if (n_writes) { + if (what == resend) { + b->n_writes = n_writes; + if (b->w.cb == NULL) { + b->w.cb = w_send_barrier; + inc_ap_pending(mdev); + set_bit(CREATE_BARRIER, &mdev->flags); + } + + drbd_queue_work(&mdev->data.work, &b->w); + } + pn = &b->next; + } else { + if (n_reads) + list_add(&carry_reads, &b->requests); + /* there could still be requests on that ring list, + * in case local io is still pending */ + list_del(&b->requests); + + /* dec_ap_pending corresponding to queue_barrier. + * the newest barrier may not have been queued yet, + * in which case w.cb is still NULL. */ + if (b->w.cb != NULL) + dec_ap_pending(mdev); + + if (b == mdev->newest_tle) { + /* recycle, but reinit! */ + D_ASSERT(tmp == NULL); + INIT_LIST_HEAD(&b->requests); + list_splice(&carry_reads, &b->requests); + INIT_LIST_HEAD(&b->w.list); + b->w.cb = NULL; + b->br_number = net_random(); + b->n_writes = 0; + + *pn = b; + break; + } + *pn = tmp; + kfree(b); + } + b = tmp; + list_splice(&carry_reads, &b->requests); + } +} + /** * tl_clear() - Clears all requests and &struct drbd_tl_epoch objects out of the TL @@ -344,48 +422,12 @@ bail: */ void tl_clear(struct drbd_conf *mdev) { - struct drbd_tl_epoch *b, *tmp; struct list_head *le, *tle; struct drbd_request *r; - int new_initial_bnr = net_random(); spin_lock_irq(&mdev->req_lock); - b = mdev->oldest_tle; - while (b) { - list_for_each_safe(le, tle, &b->requests) { - r = list_entry(le, struct drbd_request, tl_requests); - /* It would be nice to complete outside of spinlock. - * But this is easier for now. */ - _req_mod(r, connection_lost_while_pending); - } - tmp = b->next; - - /* there could still be requests on that ring list, - * in case local io is still pending */ - list_del(&b->requests); - - /* dec_ap_pending corresponding to queue_barrier. - * the newest barrier may not have been queued yet, - * in which case w.cb is still NULL. */ - if (b->w.cb != NULL) - dec_ap_pending(mdev); - - if (b == mdev->newest_tle) { - /* recycle, but reinit! */ - D_ASSERT(tmp == NULL); - INIT_LIST_HEAD(&b->requests); - INIT_LIST_HEAD(&b->w.list); - b->w.cb = NULL; - b->br_number = new_initial_bnr; - b->n_req = 0; - - mdev->oldest_tle = b; - break; - } - kfree(b); - b = tmp; - } + _tl_restart(mdev, connection_lost_while_pending); /* we expect this list to be empty. */ D_ASSERT(list_empty(&mdev->out_of_sequence_requests)); @@ -401,6 +443,15 @@ void tl_clear(struct drbd_conf *mdev) /* ensure bit indicating barrier is required is clear */ clear_bit(CREATE_BARRIER, &mdev->flags); + memset(mdev->app_reads_hash, 0, APP_R_HSIZE*sizeof(void *)); + + spin_unlock_irq(&mdev->req_lock); +} + +void tl_restart(struct drbd_conf *mdev, enum drbd_req_event what) +{ + spin_lock_irq(&mdev->req_lock); + _tl_restart(mdev, what); spin_unlock_irq(&mdev->req_lock); } @@ -455,7 +506,7 @@ 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 union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os, - union drbd_state ns, int *warn_sync_abort); + union drbd_state ns, const char **warn_sync_abort); int drbd_send_state_req(struct drbd_conf *, union drbd_state, union drbd_state); @@ -605,7 +656,7 @@ static void print_st(struct drbd_conf *mdev, char *name, union drbd_state ns) drbd_role_str(ns.peer), drbd_disk_str(ns.disk), drbd_disk_str(ns.pdsk), - ns.susp ? 's' : 'r', + is_susp(ns) ? 's' : 'r', ns.aftr_isp ? 'a' : '-', ns.peer_isp ? 'p' : '-', ns.user_isp ? 'u' : '-' @@ -763,7 +814,7 @@ static int is_valid_state_transition(struct drbd_conf *mdev, * to D_UNKNOWN. This rule and many more along those lines are in this function. */ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os, - union drbd_state ns, int *warn_sync_abort) + union drbd_state ns, const char **warn_sync_abort) { enum drbd_fencing_p fp; @@ -778,11 +829,21 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os.conn <= C_DISCONNECTING) ns.conn = os.conn; - /* After a network error (+C_TEAR_DOWN) only C_UNCONNECTED or C_DISCONNECTING can follow */ + /* After a network error (+C_TEAR_DOWN) only C_UNCONNECTED or C_DISCONNECTING can follow. + * If you try to go into some Sync* state, that shall fail (elsewhere). */ if (os.conn >= C_TIMEOUT && os.conn <= C_TEAR_DOWN && - ns.conn != C_UNCONNECTED && ns.conn != C_DISCONNECTING) + ns.conn != C_UNCONNECTED && ns.conn != C_DISCONNECTING && ns.conn <= C_TEAR_DOWN) ns.conn = os.conn; + /* we cannot fail (again) if we already detached */ + if (ns.disk == D_FAILED && os.disk == D_DISKLESS) + ns.disk = D_DISKLESS; + + /* if we are only D_ATTACHING yet, + * we can (and should) go directly to D_DISKLESS. */ + if (ns.disk == D_FAILED && os.disk == D_ATTACHING) + ns.disk = D_DISKLESS; + /* After C_DISCONNECTING only C_STANDALONE may follow */ if (os.conn == C_DISCONNECTING && ns.conn != C_STANDALONE) ns.conn = os.conn; @@ -798,14 +859,13 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state if (ns.conn == C_STANDALONE && ns.disk == D_DISKLESS && ns.role == R_SECONDARY) ns.aftr_isp = 0; - if (ns.conn <= C_DISCONNECTING && ns.disk == D_DISKLESS) - ns.pdsk = D_UNKNOWN; - /* Abort resync if a disk fails/detaches */ if (os.conn > C_CONNECTED && ns.conn > C_CONNECTED && (ns.disk <= D_FAILED || ns.pdsk <= D_FAILED)) { if (warn_sync_abort) - *warn_sync_abort = 1; + *warn_sync_abort = + os.conn == C_VERIFY_S || os.conn == C_VERIFY_T ? + "Online-verify" : "Resync"; ns.conn = C_CONNECTED; } @@ -876,7 +936,12 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state 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)) - ns.susp = 1; + ns.susp_fen = 1; /* Suspend IO while fence-peer handler runs (peer lost) */ + + if (mdev->sync_conf.on_no_data == OND_SUSPEND_IO && + (ns.role == R_PRIMARY && ns.disk < D_UP_TO_DATE && ns.pdsk < D_UP_TO_DATE) && + !(os.role == R_PRIMARY && os.disk < D_UP_TO_DATE && os.pdsk < D_UP_TO_DATE)) + ns.susp_nod = 1; /* Suspend IO while no data available (no accessible data available) */ if (ns.aftr_isp || ns.peer_isp || ns.user_isp) { if (ns.conn == C_SYNC_SOURCE) @@ -912,6 +977,12 @@ static void set_ov_position(struct drbd_conf *mdev, enum drbd_conns cs) } } +static void drbd_resume_al(struct drbd_conf *mdev) +{ + if (test_and_clear_bit(AL_SUSPENDED, &mdev->flags)) + dev_info(DEV, "Resumed AL updates\n"); +} + /** * __drbd_set_state() - Set a new DRBD state * @mdev: DRBD device. @@ -927,7 +998,7 @@ int __drbd_set_state(struct drbd_conf *mdev, { union drbd_state os; int rv = SS_SUCCESS; - int warn_sync_abort = 0; + const char *warn_sync_abort = NULL; struct after_state_chg_work *ascw; os = mdev->state; @@ -946,14 +1017,8 @@ int __drbd_set_state(struct drbd_conf *mdev, /* If the old state was illegal as well, then let this happen...*/ - if (is_valid_state(mdev, os) == rv) { - dev_err(DEV, "Considering state change from bad state. " - "Error would be: '%s'\n", - drbd_set_st_err_str(rv)); - print_st(mdev, "old", os); - print_st(mdev, "new", ns); + if (is_valid_state(mdev, os) == rv) rv = is_valid_state_transition(mdev, ns, os); - } } else rv = is_valid_state_transition(mdev, ns, os); } @@ -965,7 +1030,7 @@ int __drbd_set_state(struct drbd_conf *mdev, } if (warn_sync_abort) - dev_warn(DEV, "Resync aborted.\n"); + dev_warn(DEV, "%s aborted.\n", warn_sync_abort); { char *pbp, pb[300]; @@ -976,7 +1041,10 @@ int __drbd_set_state(struct drbd_conf *mdev, PSC(conn); PSC(disk); PSC(pdsk); - PSC(susp); + 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); @@ -997,16 +1065,18 @@ int __drbd_set_state(struct drbd_conf *mdev, !test_and_set_bit(CONFIG_PENDING, &mdev->flags)) set_bit(DEVICE_DYING, &mdev->flags); - mdev->state.i = ns.i; + /* if we are going -> D_FAILED or D_DISKLESS, grab one extra reference + * on the ldev here, to be sure the transition -> D_DISKLESS resp. + * drbd_ldev_destroy() won't happen before our corresponding + * after_state_ch works run, where we put_ldev again. */ + if ((os.disk != D_FAILED && ns.disk == D_FAILED) || + (os.disk != D_DISKLESS && ns.disk == D_DISKLESS)) + atomic_inc(&mdev->local_cnt); + + mdev->state = ns; wake_up(&mdev->misc_wait); wake_up(&mdev->state_wait); - /* post-state-change actions */ - if (os.conn >= C_SYNC_SOURCE && ns.conn <= C_CONNECTED) { - set_bit(STOP_SYNC_TIMER, &mdev->flags); - mod_timer(&mdev->resync_timer, jiffies); - } - /* aborted verify run. log the last position */ if ((os.conn == C_VERIFY_S || os.conn == C_VERIFY_T) && ns.conn < C_CONNECTED) { @@ -1019,41 +1089,42 @@ int __drbd_set_state(struct drbd_conf *mdev, if ((os.conn == C_PAUSED_SYNC_T || os.conn == C_PAUSED_SYNC_S) && (ns.conn == C_SYNC_TARGET || ns.conn == C_SYNC_SOURCE)) { dev_info(DEV, "Syncer continues.\n"); - mdev->rs_paused += (long)jiffies-(long)mdev->rs_mark_time; - if (ns.conn == C_SYNC_TARGET) { - if (!test_and_clear_bit(STOP_SYNC_TIMER, &mdev->flags)) - mod_timer(&mdev->resync_timer, jiffies); - /* This if (!test_bit) is only needed for the case - that a device that has ceased to used its timer, - i.e. it is already in drbd_resync_finished() gets - paused and resumed. */ - } + mdev->rs_paused += (long)jiffies + -(long)mdev->rs_mark_time[mdev->rs_last_mark]; + if (ns.conn == C_SYNC_TARGET) + mod_timer(&mdev->resync_timer, jiffies); } if ((os.conn == C_SYNC_TARGET || os.conn == C_SYNC_SOURCE) && (ns.conn == C_PAUSED_SYNC_T || ns.conn == C_PAUSED_SYNC_S)) { dev_info(DEV, "Resync suspended\n"); - mdev->rs_mark_time = jiffies; - if (ns.conn == C_PAUSED_SYNC_T) - set_bit(STOP_SYNC_TIMER, &mdev->flags); + mdev->rs_mark_time[mdev->rs_last_mark] = jiffies; } if (os.conn == C_CONNECTED && (ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T)) { + unsigned long now = jiffies; + int i; + mdev->ov_position = 0; - mdev->rs_total = - mdev->rs_mark_left = drbd_bm_bits(mdev); + 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); - mdev->rs_start = - mdev->rs_mark_time = jiffies; + mdev->rs_start = now; + mdev->rs_last_events = 0; + mdev->rs_last_sect_ev = 0; mdev->ov_last_oos_size = 0; 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_time[i] = now; + } + if (ns.conn == C_VERIFY_S) { dev_info(DEV, "Starting Online Verify from sector %llu\n", (unsigned long long)mdev->ov_position); @@ -1106,6 +1177,10 @@ int __drbd_set_state(struct drbd_conf *mdev, ns.conn <= C_TEAR_DOWN && ns.conn >= C_TIMEOUT) drbd_thread_restart_nowait(&mdev->receiver); + /* Resume AL writing if we get a connection */ + if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) + drbd_resume_al(mdev); + ascw = kmalloc(sizeof(*ascw), GFP_ATOMIC); if (ascw) { ascw->os = os; @@ -1164,6 +1239,8 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, union drbd_state ns, enum chg_state_flags flags) { enum drbd_fencing_p fp; + enum drbd_req_event what = nothing; + union drbd_state nsm = (union drbd_state){ .i = -1 }; if (os.conn != C_CONNECTED && ns.conn == C_CONNECTED) { clear_bit(CRASHED_PRIMARY, &mdev->flags); @@ -1187,17 +1264,48 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, /* Here we have the actions that are performed after a state change. This function might sleep */ - if (fp == FP_STONITH && ns.susp) { - /* case1: The outdate peer handler is successful: - * case2: The connection was established again: */ - if ((os.pdsk > D_OUTDATED && ns.pdsk <= D_OUTDATED) || - (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED)) { + 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.disk == D_ATTACHING && ns.disk > D_ATTACHING) + what = restart_frozen_disk_io, nsm.susp_nod = 0; + + } + + if (ns.susp_fen) { + /* case1: The outdate peer handler is successful: */ + if (os.pdsk > D_OUTDATED && ns.pdsk <= D_OUTDATED) { tl_clear(mdev); + if (test_bit(NEW_CUR_UUID, &mdev->flags)) { + drbd_uuid_new_current(mdev); + clear_bit(NEW_CUR_UUID, &mdev->flags); + } spin_lock_irq(&mdev->req_lock); - _drbd_set_state(_NS(mdev, susp, 0), CS_VERBOSE, NULL); + _drbd_set_state(_NS(mdev, susp_fen, 0), CS_VERBOSE, NULL); spin_unlock_irq(&mdev->req_lock); } + /* case2: The connection was established again: */ + if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) { + clear_bit(NEW_CUR_UUID, &mdev->flags); + what = resend; + nsm.susp_fen = 0; + } } + + if (what != nothing) { + spin_lock_irq(&mdev->req_lock); + _tl_restart(mdev, what); + nsm.i &= mdev->state.i; + _drbd_set_state(mdev, nsm, CS_VERBOSE, NULL); + spin_unlock_irq(&mdev->req_lock); + } + /* 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); @@ -1216,16 +1324,22 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, if (get_ldev(mdev)) { if ((ns.role == R_PRIMARY || ns.peer == R_PRIMARY) && mdev->ldev->md.uuid[UI_BITMAP] == 0 && ns.disk >= D_UP_TO_DATE) { - drbd_uuid_new_current(mdev); - drbd_send_uuids(mdev); + if (is_susp(mdev->state)) { + set_bit(NEW_CUR_UUID, &mdev->flags); + } else { + drbd_uuid_new_current(mdev); + drbd_send_uuids(mdev); + } } put_ldev(mdev); } } if (ns.pdsk < D_INCONSISTENT && get_ldev(mdev)) { - if (ns.peer == R_PRIMARY && mdev->ldev->md.uuid[UI_BITMAP] == 0) + if (ns.peer == R_PRIMARY && mdev->ldev->md.uuid[UI_BITMAP] == 0) { drbd_uuid_new_current(mdev); + drbd_send_uuids(mdev); + } /* D_DISKLESS Peer becomes secondary */ if (os.peer == R_PRIMARY && ns.peer == R_SECONDARY) @@ -1267,52 +1381,64 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, os.disk > D_INCONSISTENT && ns.disk == D_INCONSISTENT) drbd_queue_bitmap_io(mdev, &drbd_bmio_set_n_write, NULL, "set_n_write from invalidate"); - if (os.disk > D_FAILED && ns.disk == D_FAILED) { + /* first half of local IO error, failure to attach, + * or administrative detach */ + if (os.disk != D_FAILED && ns.disk == D_FAILED) { enum drbd_io_error_p eh; - - eh = EP_PASS_ON; - if (get_ldev_if_state(mdev, D_FAILED)) { - eh = mdev->ldev->dc.on_io_error; - put_ldev(mdev); - } + int was_io_error; + /* corresponding get_ldev was in __drbd_set_state, to serialize + * our cleanup here with the transition to D_DISKLESS, + * so it is safe to dreference ldev here. */ + eh = mdev->ldev->dc.on_io_error; + was_io_error = test_and_clear_bit(WAS_IO_ERROR, &mdev->flags); + + /* current state still has to be D_FAILED, + * there is only one way out: to D_DISKLESS, + * and that may only happen after our put_ldev below. */ + if (mdev->state.disk != D_FAILED) + dev_err(DEV, + "ASSERT FAILED: disk is %s during detach\n", + drbd_disk_str(mdev->state.disk)); + + if (drbd_send_state(mdev)) + dev_warn(DEV, "Notified peer that I am detaching my disk\n"); + else + dev_err(DEV, "Sending state for detaching disk failed\n"); drbd_rs_cancel_all(mdev); - /* since get_ldev() only works as long as disk>=D_INCONSISTENT, - and it is D_DISKLESS here, local_cnt can only go down, it can - not increase... It will reach zero */ - wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt)); - mdev->rs_total = 0; - mdev->rs_failed = 0; - atomic_set(&mdev->rs_pending_cnt, 0); - spin_lock_irq(&mdev->req_lock); - _drbd_set_state(_NS(mdev, disk, D_DISKLESS), CS_HARD, NULL); - spin_unlock_irq(&mdev->req_lock); + /* In case we want to get something to stable storage still, + * this may be the last chance. + * Following put_ldev may transition to D_DISKLESS. */ + drbd_md_sync(mdev); + put_ldev(mdev); - if (eh == EP_CALL_HELPER) + if (was_io_error && eh == EP_CALL_HELPER) drbd_khelper(mdev, "local-io-error"); } - if (os.disk > D_DISKLESS && ns.disk == D_DISKLESS) { + /* second half of local IO error, failure to attach, + * or administrative detach, + * after local_cnt references have reached zero again */ + if (os.disk != D_DISKLESS && ns.disk == D_DISKLESS) { + /* We must still be diskless, + * re-attach has to be serialized with this! */ + if (mdev->state.disk != D_DISKLESS) + dev_err(DEV, + "ASSERT FAILED: disk is %s while going diskless\n", + drbd_disk_str(mdev->state.disk)); - if (os.disk == D_FAILED) /* && ns.disk == D_DISKLESS*/ { - if (drbd_send_state(mdev)) - dev_warn(DEV, "Notified peer that my disk is broken.\n"); - else - dev_err(DEV, "Sending state in drbd_io_error() failed\n"); - } + mdev->rs_total = 0; + mdev->rs_failed = 0; + atomic_set(&mdev->rs_pending_cnt, 0); - wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt)); - lc_destroy(mdev->resync); - mdev->resync = NULL; - lc_destroy(mdev->act_log); - mdev->act_log = NULL; - __no_warn(local, - drbd_free_bc(mdev->ldev); - mdev->ldev = NULL;); - - if (mdev->md_io_tmpp) - __free_page(mdev->md_io_tmpp); + 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. */ + put_ldev(mdev); } /* Disks got bigger while they were detached */ @@ -1328,6 +1454,15 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, (os.user_isp && !ns.user_isp)) resume_next_sg(mdev); + /* sync target done with resync. Explicitly notify peer, even though + * it should (at least for non-empty resyncs) already know itself. */ + if (os.disk < D_UP_TO_DATE && os.conn >= C_SYNC_SOURCE && ns.conn == C_CONNECTED) + drbd_send_state(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); + /* Upon network connection, we need to start the receiver */ if (os.conn == C_STANDALONE && ns.conn == C_UNCONNECTED) drbd_thread_start(&mdev->receiver); @@ -1554,7 +1689,7 @@ void drbd_thread_current_set_cpu(struct drbd_conf *mdev) /* the appropriate socket mutex must be held already */ int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, - enum drbd_packets cmd, struct p_header *h, + enum drbd_packets cmd, struct p_header80 *h, size_t size, unsigned msg_flags) { int sent, ok; @@ -1564,7 +1699,7 @@ int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, h->magic = BE_DRBD_MAGIC; h->command = cpu_to_be16(cmd); - h->length = cpu_to_be16(size-sizeof(struct p_header)); + h->length = cpu_to_be16(size-sizeof(struct p_header80)); sent = drbd_send(mdev, sock, h, size, msg_flags); @@ -1579,7 +1714,7 @@ int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, * when we hold the appropriate socket mutex. */ int drbd_send_cmd(struct drbd_conf *mdev, int use_data_socket, - enum drbd_packets cmd, struct p_header *h, size_t size) + enum drbd_packets cmd, struct p_header80 *h, size_t size) { int ok = 0; struct socket *sock; @@ -1607,7 +1742,7 @@ int drbd_send_cmd(struct drbd_conf *mdev, int use_data_socket, int drbd_send_cmd2(struct drbd_conf *mdev, enum drbd_packets cmd, char *data, size_t size) { - struct p_header h; + struct p_header80 h; int ok; h.magic = BE_DRBD_MAGIC; @@ -1629,7 +1764,7 @@ int drbd_send_cmd2(struct drbd_conf *mdev, enum drbd_packets cmd, char *data, int drbd_send_sync_param(struct drbd_conf *mdev, struct syncer_conf *sc) { - struct p_rs_param_89 *p; + struct p_rs_param_95 *p; struct socket *sock; int size, rv; const int apv = mdev->agreed_pro_version; @@ -1637,7 +1772,8 @@ int drbd_send_sync_param(struct drbd_conf *mdev, struct syncer_conf *sc) size = apv <= 87 ? sizeof(struct p_rs_param) : apv == 88 ? sizeof(struct p_rs_param) + strlen(mdev->sync_conf.verify_alg) + 1 - : /* 89 */ sizeof(struct p_rs_param_89); + : apv <= 94 ? sizeof(struct p_rs_param_89) + : /* apv >= 95 */ sizeof(struct p_rs_param_95); /* used from admin command context and receiver/worker context. * to avoid kmalloc, grab the socket right here, @@ -1648,12 +1784,16 @@ int drbd_send_sync_param(struct drbd_conf *mdev, struct syncer_conf *sc) if (likely(sock != NULL)) { enum drbd_packets cmd = apv >= 89 ? P_SYNC_PARAM89 : P_SYNC_PARAM; - p = &mdev->data.sbuf.rs_param_89; + p = &mdev->data.sbuf.rs_param_95; /* initialize verify_alg and csums_alg */ memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX); p->rate = cpu_to_be32(sc->rate); + p->c_plan_ahead = cpu_to_be32(sc->c_plan_ahead); + p->c_delay_target = cpu_to_be32(sc->c_delay_target); + p->c_fill_target = cpu_to_be32(sc->c_fill_target); + p->c_max_rate = cpu_to_be32(sc->c_max_rate); if (apv >= 88) strcpy(p->verify_alg, mdev->sync_conf.verify_alg); @@ -1709,7 +1849,7 @@ int drbd_send_protocol(struct drbd_conf *mdev) strcpy(p->integrity_alg, mdev->net_conf->integrity_alg); rv = drbd_send_cmd(mdev, USE_DATA_SOCKET, P_PROTOCOL, - (struct p_header *)p, size); + (struct p_header80 *)p, size); kfree(p); return rv; } @@ -1735,7 +1875,7 @@ int _drbd_send_uuids(struct drbd_conf *mdev, u64 uuid_flags) put_ldev(mdev); return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_UUIDS, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); } int drbd_send_uuids(struct drbd_conf *mdev) @@ -1756,7 +1896,7 @@ int drbd_send_sync_uuid(struct drbd_conf *mdev, u64 val) p.uuid = cpu_to_be64(val); return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_SYNC_UUID, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); } int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags flags) @@ -1786,7 +1926,7 @@ int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags fl p.dds_flags = cpu_to_be16(flags); ok = drbd_send_cmd(mdev, USE_DATA_SOCKET, P_SIZES, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); return ok; } @@ -1811,7 +1951,7 @@ int drbd_send_state(struct drbd_conf *mdev) if (likely(sock != NULL)) { ok = _drbd_send_cmd(mdev, sock, P_STATE, - (struct p_header *)&p, sizeof(p), 0); + (struct p_header80 *)&p, sizeof(p), 0); } mutex_unlock(&mdev->data.mutex); @@ -1829,7 +1969,7 @@ int drbd_send_state_req(struct drbd_conf *mdev, p.val = cpu_to_be32(val.i); return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_STATE_CHG_REQ, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); } int drbd_send_sr_reply(struct drbd_conf *mdev, int retcode) @@ -1839,7 +1979,7 @@ int drbd_send_sr_reply(struct drbd_conf *mdev, int retcode) p.retcode = cpu_to_be32(retcode); return drbd_send_cmd(mdev, USE_META_SOCKET, P_STATE_CHG_REPLY, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); } int fill_bitmap_rle_bits(struct drbd_conf *mdev, @@ -1938,7 +2078,7 @@ int fill_bitmap_rle_bits(struct drbd_conf *mdev, enum { OK, FAILED, DONE } send_bitmap_rle_or_plain(struct drbd_conf *mdev, - struct p_header *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; @@ -1968,12 +2108,12 @@ send_bitmap_rle_or_plain(struct drbd_conf *mdev, if (len) drbd_bm_get_lel(mdev, c->word_offset, num_words, (unsigned long*)h->payload); ok = _drbd_send_cmd(mdev, mdev->data.socket, P_BITMAP, - h, sizeof(struct p_header) + len, 0); + h, sizeof(struct p_header80) + len, 0); c->word_offset += num_words; c->bit_offset = c->word_offset * BITS_PER_LONG; c->packets[1]++; - c->bytes[1] += sizeof(struct p_header) + len; + c->bytes[1] += sizeof(struct p_header80) + len; if (c->bit_offset > c->bm_bits) c->bit_offset = c->bm_bits; @@ -1989,14 +2129,14 @@ send_bitmap_rle_or_plain(struct drbd_conf *mdev, int _drbd_send_bitmap(struct drbd_conf *mdev) { struct bm_xfer_ctx c; - struct p_header *p; + struct p_header80 *p; int ret; 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_header *) __get_free_page(GFP_NOIO); + 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; @@ -2054,7 +2194,7 @@ int drbd_send_b_ack(struct drbd_conf *mdev, u32 barrier_nr, u32 set_size) if (mdev->state.conn < C_CONNECTED) return FALSE; ok = drbd_send_cmd(mdev, USE_META_SOCKET, P_BARRIER_ACK, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); return ok; } @@ -2082,17 +2222,18 @@ static int _drbd_send_ack(struct drbd_conf *mdev, enum drbd_packets cmd, if (!mdev->meta.socket || mdev->state.conn < C_CONNECTED) return FALSE; ok = drbd_send_cmd(mdev, USE_META_SOCKET, cmd, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); return ok; } +/* dp->sector and dp->block_id already/still in network byte order, + * data_size is payload size according to dp->head, + * and may need to be corrected for digest size. */ int drbd_send_ack_dp(struct drbd_conf *mdev, enum drbd_packets cmd, - struct p_data *dp) + struct p_data *dp, int data_size) { - const int header_size = sizeof(struct p_data) - - sizeof(struct p_header); - int data_size = ((struct p_header *)dp)->length - header_size; - + data_size -= (mdev->agreed_pro_version >= 87 && mdev->integrity_r_tfm) ? + crypto_hash_digestsize(mdev->integrity_r_tfm) : 0; return _drbd_send_ack(mdev, cmd, dp->sector, cpu_to_be32(data_size), dp->block_id); } @@ -2140,7 +2281,7 @@ int drbd_send_drequest(struct drbd_conf *mdev, int cmd, p.blksize = cpu_to_be32(size); ok = drbd_send_cmd(mdev, USE_DATA_SOCKET, cmd, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); return ok; } @@ -2158,7 +2299,7 @@ int drbd_send_drequest_csum(struct drbd_conf *mdev, p.head.magic = BE_DRBD_MAGIC; p.head.command = cpu_to_be16(cmd); - p.head.length = cpu_to_be16(sizeof(p) - sizeof(struct p_header) + digest_size); + p.head.length = cpu_to_be16(sizeof(p) - sizeof(struct p_header80) + digest_size); mutex_lock(&mdev->data.mutex); @@ -2180,7 +2321,7 @@ int drbd_send_ov_request(struct drbd_conf *mdev, sector_t sector, int size) p.blksize = cpu_to_be32(size); ok = drbd_send_cmd(mdev, USE_DATA_SOCKET, P_OV_REQUEST, - (struct p_header *)&p, sizeof(p)); + (struct p_header80 *)&p, sizeof(p)); return ok; } @@ -2332,6 +2473,18 @@ static int _drbd_send_zc_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e) return 1; } +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; +} + /* Used to send write requests * R_PRIMARY -> Peer (P_DATA) */ @@ -2349,30 +2502,25 @@ int drbd_send_dblock(struct drbd_conf *mdev, struct drbd_request *req) dgs = (mdev->agreed_pro_version >= 87 && mdev->integrity_w_tfm) ? crypto_hash_digestsize(mdev->integrity_w_tfm) : 0; - p.head.magic = BE_DRBD_MAGIC; - p.head.command = cpu_to_be16(P_DATA); - p.head.length = - cpu_to_be16(sizeof(p) - sizeof(struct p_header) + dgs + req->size); + if (req->size <= DRBD_MAX_SIZE_H80_PACKET) { + p.head.h80.magic = BE_DRBD_MAGIC; + p.head.h80.command = cpu_to_be16(P_DATA); + p.head.h80.length = + cpu_to_be16(sizeof(p) - sizeof(union p_header) + dgs + req->size); + } else { + p.head.h95.magic = BE_DRBD_MAGIC_BIG; + p.head.h95.command = cpu_to_be16(P_DATA); + p.head.h95.length = + cpu_to_be32(sizeof(p) - sizeof(union p_header) + dgs + req->size); + } p.sector = cpu_to_be64(req->sector); p.block_id = (unsigned long)req; p.seq_num = cpu_to_be32(req->seq_num = atomic_add_return(1, &mdev->packet_seq)); - dp_flags = 0; - /* NOTE: no need to check if barriers supported here as we would - * not pass the test in make_request_common in that case - */ - if (req->master_bio->bi_rw & REQ_HARDBARRIER) { - dev_err(DEV, "ASSERT FAILED would have set DP_HARDBARRIER\n"); - /* dp_flags |= DP_HARDBARRIER; */ - } - if (req->master_bio->bi_rw & REQ_SYNC) - dp_flags |= DP_RW_SYNC; - /* for now handle SYNCIO and UNPLUG - * as if they still were one and the same flag */ - if (req->master_bio->bi_rw & REQ_UNPLUG) - dp_flags |= DP_RW_SYNC; + dp_flags = bio_flags_to_wire(mdev, req->master_bio->bi_rw); + if (mdev->state.conn >= C_SYNC_SOURCE && mdev->state.conn <= C_PAUSED_SYNC_T) dp_flags |= DP_MAY_SET_IN_SYNC; @@ -2413,10 +2561,17 @@ int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, dgs = (mdev->agreed_pro_version >= 87 && mdev->integrity_w_tfm) ? crypto_hash_digestsize(mdev->integrity_w_tfm) : 0; - p.head.magic = BE_DRBD_MAGIC; - p.head.command = cpu_to_be16(cmd); - p.head.length = - cpu_to_be16(sizeof(p) - sizeof(struct p_header) + dgs + e->size); + if (e->size <= DRBD_MAX_SIZE_H80_PACKET) { + p.head.h80.magic = BE_DRBD_MAGIC; + p.head.h80.command = cpu_to_be16(cmd); + p.head.h80.length = + cpu_to_be16(sizeof(p) - sizeof(struct p_header80) + dgs + e->size); + } else { + p.head.h95.magic = BE_DRBD_MAGIC_BIG; + p.head.h95.command = cpu_to_be16(cmd); + p.head.h95.length = + cpu_to_be32(sizeof(p) - sizeof(struct p_header80) + dgs + e->size); + } p.sector = cpu_to_be64(e->sector); p.block_id = e->block_id; @@ -2429,8 +2584,7 @@ int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, if (!drbd_get_data_sock(mdev)) return 0; - ok = sizeof(p) == drbd_send(mdev, mdev->data.socket, &p, - sizeof(p), dgs ? MSG_MORE : 0); + ok = sizeof(p) == drbd_send(mdev, mdev->data.socket, &p, sizeof(p), dgs ? MSG_MORE : 0); if (ok && dgs) { dgb = mdev->int_dig_out; drbd_csum_ee(mdev, mdev->integrity_w_tfm, e, dgb); @@ -2536,7 +2690,7 @@ static int drbd_open(struct block_device *bdev, fmode_t mode) unsigned long flags; int rv = 0; - lock_kernel(); + mutex_lock(&drbd_main_mutex); spin_lock_irqsave(&mdev->req_lock, flags); /* to have a stable mdev->state.role * and no race with updating open_cnt */ @@ -2551,7 +2705,7 @@ static int drbd_open(struct block_device *bdev, fmode_t mode) if (!rv) mdev->open_cnt++; spin_unlock_irqrestore(&mdev->req_lock, flags); - unlock_kernel(); + mutex_unlock(&drbd_main_mutex); return rv; } @@ -2559,9 +2713,9 @@ static int drbd_open(struct block_device *bdev, fmode_t mode) static int drbd_release(struct gendisk *gd, fmode_t mode) { struct drbd_conf *mdev = gd->private_data; - lock_kernel(); + mutex_lock(&drbd_main_mutex); mdev->open_cnt--; - unlock_kernel(); + mutex_unlock(&drbd_main_mutex); return 0; } @@ -2605,7 +2759,13 @@ static void drbd_set_defaults(struct drbd_conf *mdev) /* .verify_alg = */ {}, 0, /* .cpu_mask = */ {}, 0, /* .csums_alg = */ {}, 0, - /* .use_rle = */ 0 + /* .use_rle = */ 0, + /* .on_no_data = */ DRBD_ON_NO_DATA_DEF, + /* .c_plan_ahead = */ DRBD_C_PLAN_AHEAD_DEF, + /* .c_delay_target = */ DRBD_C_DELAY_TARGET_DEF, + /* .c_fill_target = */ DRBD_C_FILL_TARGET_DEF, + /* .c_max_rate = */ DRBD_C_MAX_RATE_DEF, + /* .c_min_rate = */ DRBD_C_MIN_RATE_DEF }; /* Have to use that way, because the layout differs between @@ -2616,7 +2776,9 @@ static void drbd_set_defaults(struct drbd_conf *mdev) .conn = C_STANDALONE, .disk = D_DISKLESS, .pdsk = D_UNKNOWN, - .susp = 0 + .susp = 0, + .susp_nod = 0, + .susp_fen = 0 } }; } @@ -2627,11 +2789,6 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) drbd_set_defaults(mdev); - /* for now, we do NOT yet support it, - * even though we start some framework - * to eventually support barriers */ - set_bit(NO_BARRIER_SUPP, &mdev->flags); - atomic_set(&mdev->ap_bio_cnt, 0); atomic_set(&mdev->ap_pending_cnt, 0); atomic_set(&mdev->rs_pending_cnt, 0); @@ -2640,6 +2797,9 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) atomic_set(&mdev->net_cnt, 0); atomic_set(&mdev->packet_seq, 0); atomic_set(&mdev->pp_in_use, 0); + atomic_set(&mdev->pp_in_use_by_net, 0); + atomic_set(&mdev->rs_sect_in, 0); + atomic_set(&mdev->rs_sect_ev, 0); mutex_init(&mdev->md_io_mutex); mutex_init(&mdev->data.mutex); @@ -2666,11 +2826,13 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) INIT_LIST_HEAD(&mdev->meta.work.q); INIT_LIST_HEAD(&mdev->resync_work.list); 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->bm_io_work.w.list); mdev->resync_work.cb = w_resync_inactive; 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; init_timer(&mdev->resync_timer); @@ -2682,6 +2844,7 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) init_waitqueue_head(&mdev->misc_wait); init_waitqueue_head(&mdev->state_wait); + init_waitqueue_head(&mdev->net_cnt_wait); init_waitqueue_head(&mdev->ee_wait); init_waitqueue_head(&mdev->al_wait); init_waitqueue_head(&mdev->seq_wait); @@ -2691,12 +2854,13 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) drbd_thread_init(mdev, &mdev->asender, drbd_asender); mdev->agreed_pro_version = PRO_VERSION_MAX; - mdev->write_ordering = WO_bio_barrier; + mdev->write_ordering = WO_bdev_flush; mdev->resync_wenr = LC_FREE; } void drbd_mdev_cleanup(struct drbd_conf *mdev) { + int i; if (mdev->receiver.t_state != None) dev_err(DEV, "ASSERT FAILED: receiver t_state == %d expected 0.\n", mdev->receiver.t_state); @@ -2713,9 +2877,13 @@ void drbd_mdev_cleanup(struct drbd_conf *mdev) mdev->p_size = mdev->rs_start = mdev->rs_total = - mdev->rs_failed = - mdev->rs_mark_left = - mdev->rs_mark_time = 0; + mdev->rs_failed = 0; + mdev->rs_last_events = 0; + mdev->rs_last_sect_ev = 0; + for (i = 0; i < DRBD_SYNC_MARKS; i++) { + mdev->rs_mark_left[i] = 0; + mdev->rs_mark_time[i] = 0; + } D_ASSERT(mdev->net_conf == NULL); drbd_set_my_capacity(mdev, 0); @@ -2726,6 +2894,7 @@ void drbd_mdev_cleanup(struct drbd_conf *mdev) } drbd_free_resources(mdev); + clear_bit(AL_SUSPENDED, &mdev->flags); /* * currently we drbd_init_ee only on module load, so @@ -2741,7 +2910,7 @@ void drbd_mdev_cleanup(struct drbd_conf *mdev) D_ASSERT(list_empty(&mdev->meta.work.q)); D_ASSERT(list_empty(&mdev->resync_work.list)); D_ASSERT(list_empty(&mdev->unplug_work.list)); - + D_ASSERT(list_empty(&mdev->go_diskless.list)); } @@ -2824,7 +2993,7 @@ static int drbd_create_mempools(void) drbd_ee_mempool = mempool_create(number, mempool_alloc_slab, mempool_free_slab, drbd_ee_cache); - if (drbd_request_mempool == NULL) + if (drbd_ee_mempool == NULL) goto Enomem; /* drbd's page pool */ @@ -3203,11 +3372,8 @@ void drbd_free_bc(struct drbd_backing_dev *ldev) if (ldev == NULL) return; - bd_release(ldev->backing_bdev); - bd_release(ldev->md_bdev); - - fput(ldev->lo_file); - fput(ldev->md_file); + blkdev_put(ldev->backing_bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL); + blkdev_put(ldev->md_bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL); kfree(ldev); } @@ -3280,9 +3446,10 @@ void drbd_md_sync(struct drbd_conf *mdev) sector_t sector; int i; + del_timer(&mdev->md_sync_timer); + /* timer may be rearmed by drbd_md_mark_dirty() now. */ if (!test_and_clear_bit(MD_DIRTY, &mdev->flags)) return; - del_timer(&mdev->md_sync_timer); /* We use here D_FAILED and not D_ATTACHING because we try to write * metadata even if we detach due to a disk failure! */ @@ -3310,12 +3477,9 @@ void drbd_md_sync(struct drbd_conf *mdev) D_ASSERT(drbd_md_ss__(mdev, mdev->ldev) == mdev->ldev->md.md_offset); sector = mdev->ldev->md.md_offset; - if (drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) { - clear_bit(MD_DIRTY, &mdev->flags); - } else { + 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); } @@ -3402,6 +3566,28 @@ 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. @@ -3410,19 +3596,31 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev) * the meta-data super block. This function sets MD_DIRTY, and starts a * timer that ensures that within five seconds you have to call drbd_md_sync(). */ +#ifdef DEBUG +void drbd_md_mark_dirty_(struct drbd_conf *mdev, unsigned int line, const char *func) +{ + if (!test_and_set_bit(MD_DIRTY, &mdev->flags)) { + mod_timer(&mdev->md_sync_timer, jiffies + HZ); + mdev->last_md_mark_dirty.line = line; + mdev->last_md_mark_dirty.func = func; + } +} +#else void drbd_md_mark_dirty(struct drbd_conf *mdev) { - set_bit(MD_DIRTY, &mdev->flags); - mod_timer(&mdev->md_sync_timer, jiffies + 5*HZ); + if (!test_and_set_bit(MD_DIRTY, &mdev->flags)) + mod_timer(&mdev->md_sync_timer, jiffies + 5*HZ); } - +#endif 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) @@ -3437,6 +3635,7 @@ 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); } @@ -3446,6 +3645,7 @@ 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); } @@ -3464,9 +3664,12 @@ void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local) 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); + /* get it to stable storage _now_ */ + drbd_md_sync(mdev); } void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local) @@ -3478,6 +3681,8 @@ 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"); @@ -3485,6 +3690,7 @@ void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local) mdev->ldev->md.uuid[UI_BITMAP] = val; mdev->ldev->md.uuid[UI_BITMAP] &= ~((u64)1); + debug_drbd_uuid(mdev, UI_BITMAP); } drbd_md_mark_dirty(mdev); } @@ -3527,6 +3733,7 @@ int drbd_bmio_clear_n_write(struct drbd_conf *mdev) { int rv = -EIO; + drbd_resume_al(mdev); if (get_ldev_if_state(mdev, D_ATTACHING)) { drbd_bm_clear_all(mdev); rv = drbd_bm_write(mdev); @@ -3559,6 +3766,41 @@ static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused) return 1; } +void drbd_ldev_destroy(struct drbd_conf *mdev) +{ + lc_destroy(mdev->resync); + mdev->resync = NULL; + lc_destroy(mdev->act_log); + mdev->act_log = NULL; + __no_warn(local, + drbd_free_bc(mdev->ldev); + mdev->ldev = NULL;); + + if (mdev->md_io_tmpp) { + __free_page(mdev->md_io_tmpp); + mdev->md_io_tmpp = NULL; + } + clear_bit(GO_DISKLESS, &mdev->flags); +} + +static int w_go_diskless(struct drbd_conf *mdev, struct drbd_work *w, int unused) +{ + D_ASSERT(mdev->state.disk == D_FAILED); + /* we cannot assert local_cnt == 0 here, as get_ldev_if_state will + * inc/dec it frequently. Once we are D_DISKLESS, no one will touch + * the protected members anymore, though, so once put_ldev reaches zero + * again, it will be safe to free them. */ + drbd_force_state(mdev, NS(disk, D_DISKLESS)); + return 1; +} + +void drbd_go_diskless(struct drbd_conf *mdev) +{ + D_ASSERT(mdev->state.disk == D_FAILED); + if (!test_and_set_bit(GO_DISKLESS, &mdev->flags)) + drbd_queue_work(&mdev->data.work, &mdev->go_diskless); +} + /** * drbd_queue_bitmap_io() - Queues an IO operation on the whole bitmap * @mdev: DRBD device. @@ -3655,8 +3897,11 @@ static void md_sync_timer_fn(unsigned long data) static int w_md_sync(struct drbd_conf *mdev, struct drbd_work *w, int unused) { dev_warn(DEV, "md_sync_timer expired! Worker calls drbd_md_sync().\n"); +#ifdef DEBUG + dev_warn(DEV, "last md_mark_dirty: %s:%u\n", + mdev->last_md_mark_dirty.func, mdev->last_md_mark_dirty.line); +#endif drbd_md_sync(mdev); - return 1; } diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index 73131c5..8cbfaa6 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -33,10 +33,13 @@ #include <linux/blkpg.h> #include <linux/cpumask.h> #include "drbd_int.h" +#include "drbd_req.h" #include "drbd_wrappers.h" #include <asm/unaligned.h> #include <linux/drbd_tag_magic.h> #include <linux/drbd_limits.h> +#include <linux/compiler.h> +#include <linux/kthread.h> static unsigned short *tl_add_blob(unsigned short *, enum drbd_tags, const void *, int); static unsigned short *tl_add_str(unsigned short *, enum drbd_tags, const char *); @@ -169,6 +172,10 @@ int drbd_khelper(struct drbd_conf *mdev, char *cmd) put_net_conf(mdev); } + /* The helper may take some time. + * write out any unsynced meta data changes now */ + drbd_md_sync(mdev); + dev_info(DEV, "helper command: %s %s %s\n", usermode_helper, cmd, mb); drbd_bcast_ev_helper(mdev, cmd); @@ -202,12 +209,10 @@ enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev) put_ldev(mdev); } else { dev_warn(DEV, "Not fencing peer, I'm not even Consistent myself.\n"); - return mdev->state.pdsk; + nps = mdev->state.pdsk; + goto out; } - if (fp == FP_STONITH) - _drbd_request_state(mdev, NS(susp, 1), CS_WAIT_COMPLETE); - r = drbd_khelper(mdev, "fence-peer"); switch ((r>>8) & 0xff) { @@ -252,9 +257,36 @@ enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev) dev_info(DEV, "fence-peer helper returned %d (%s)\n", (r>>8) & 0xff, ex_to_string); + +out: + if (mdev->state.susp_fen && nps >= D_UNKNOWN) { + /* The handler was not successful... unfreeze here, the + state engine can not unfreeze... */ + _drbd_request_state(mdev, NS(susp_fen, 0), CS_VERBOSE); + } + return nps; } +static int _try_outdate_peer_async(void *data) +{ + struct drbd_conf *mdev = (struct drbd_conf *)data; + enum drbd_disk_state nps; + + nps = drbd_try_outdate_peer(mdev); + drbd_request_state(mdev, NS(pdsk, nps)); + + return 0; +} + +void drbd_try_outdate_peer_async(struct drbd_conf *mdev) +{ + struct task_struct *opa; + + opa = kthread_run(_try_outdate_peer_async, mdev, "drbd%d_a_helper", mdev_to_minor(mdev)); + if (IS_ERR(opa)) + 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) { @@ -394,6 +426,39 @@ int drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) return r; } +static struct drbd_conf *ensure_mdev(int minor, int create) +{ + struct drbd_conf *mdev; + + if (minor >= minor_count) + return NULL; + + mdev = minor_to_mdev(minor); + + if (!mdev && create) { + struct gendisk *disk = NULL; + mdev = drbd_new_device(minor); + + spin_lock_irq(&drbd_pp_lock); + if (minor_table[minor] == NULL) { + minor_table[minor] = mdev; + disk = mdev->vdisk; + mdev = NULL; + } /* else: we lost the race */ + spin_unlock_irq(&drbd_pp_lock); + + if (disk) /* we won the race above */ + /* in case we ever add a drbd_delete_device(), + * don't forget the del_gendisk! */ + add_disk(disk); + else /* we lost the race above */ + drbd_free_mdev(mdev); + + mdev = minor_to_mdev(minor); + } + + return mdev; +} static int drbd_nl_primary(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) @@ -494,6 +559,8 @@ char *ppsize(char *buf, unsigned long long size) void drbd_suspend_io(struct drbd_conf *mdev) { set_bit(SUSPEND_IO, &mdev->flags); + if (is_susp(mdev->state)) + return; wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); } @@ -713,9 +780,6 @@ void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int max_seg_s) __mu blk_queue_segment_boundary(q, PAGE_SIZE-1); blk_stack_limits(&q->limits, &b->limits, 0); - if (b->merge_bvec_fn) - dev_warn(DEV, "Backing device's merge_bvec_fn() = %p\n", - b->merge_bvec_fn); dev_info(DEV, "max_segment_size ( = BIO size ) = %u\n", queue_max_segment_size(q)); if (q->backing_dev_info.ra_pages != b->backing_dev_info.ra_pages) { @@ -729,14 +793,16 @@ void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int max_seg_s) __mu /* serialize deconfig (worker exiting, doing cleanup) * and reconfig (drbdsetup disk, drbdsetup net) * - * wait for a potentially exiting worker, then restart it, - * or start a new one. + * Wait for a potentially exiting worker, then restart it, + * or start a new one. Flush any pending work, there may still be an + * after_state_change queued. */ static void drbd_reconfig_start(struct drbd_conf *mdev) { wait_event(mdev->state_wait, !test_and_set_bit(CONFIG_PENDING, &mdev->flags)); wait_event(mdev->state_wait, !test_bit(DEVICE_DYING, &mdev->flags)); drbd_thread_start(&mdev->worker); + drbd_flush_workqueue(mdev); } /* if still unconfigured, stops worker again. @@ -756,6 +822,29 @@ static void drbd_reconfig_done(struct drbd_conf *mdev) wake_up(&mdev->state_wait); } +/* Make sure IO is suspended before calling this function(). */ +static void drbd_suspend_al(struct drbd_conf *mdev) +{ + int s = 0; + + if (lc_try_lock(mdev->act_log)) { + drbd_al_shrink(mdev); + lc_unlock(mdev->act_log); + } else { + dev_warn(DEV, "Failed to lock al in drbd_suspend_al()\n"); + return; + } + + spin_lock_irq(&mdev->req_lock); + if (mdev->state.conn < C_CONNECTED) + s = !test_and_set_bit(AL_SUSPENDED, &mdev->flags); + + spin_unlock_irq(&mdev->req_lock); + + if (s) + dev_info(DEV, "Suspended AL updates\n"); +} + /* does always return 0; * interesting return code is in reply->ret_code */ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, @@ -766,9 +855,10 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp sector_t max_possible_sectors; sector_t min_md_device_sectors; struct drbd_backing_dev *nbc = NULL; /* new_backing_conf */ - struct inode *inode, *inode2; + struct block_device *bdev; struct lru_cache *resync_lru = NULL; union drbd_state ns, os; + unsigned int max_seg_s; int rv; int cp_discovered = 0; int logical_block_size; @@ -780,6 +870,11 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp retcode = ERR_DISK_CONFIGURED; goto fail; } + /* It may just now have detached because of IO error. Make sure + * drbd_ldev_destroy is done already, we may end up here very fast, + * e.g. if someone calls attach from the on-io-error handler, + * to realize a "hot spare" feature (not that I'd recommend that) */ + wait_event(mdev->misc_wait, !atomic_read(&mdev->local_cnt)); /* allocation not in the IO path, cqueue thread context */ nbc = kzalloc(sizeof(struct drbd_backing_dev), GFP_KERNEL); @@ -803,46 +898,49 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp goto fail; } - nbc->lo_file = filp_open(nbc->dc.backing_dev, O_RDWR, 0); - if (IS_ERR(nbc->lo_file)) { - dev_err(DEV, "open(\"%s\") failed with %ld\n", nbc->dc.backing_dev, - PTR_ERR(nbc->lo_file)); - nbc->lo_file = NULL; - retcode = ERR_OPEN_DISK; - goto fail; + if (get_net_conf(mdev)) { + int prot = mdev->net_conf->wire_protocol; + put_net_conf(mdev); + if (nbc->dc.fencing == FP_STONITH && prot == DRBD_PROT_A) { + retcode = ERR_STONITH_AND_PROT_A; + goto fail; + } } - inode = nbc->lo_file->f_dentry->d_inode; - - if (!S_ISBLK(inode->i_mode)) { - retcode = ERR_DISK_NOT_BDEV; + bdev = blkdev_get_by_path(nbc->dc.backing_dev, + FMODE_READ | FMODE_WRITE | FMODE_EXCL, mdev); + if (IS_ERR(bdev)) { + dev_err(DEV, "open(\"%s\") failed with %ld\n", nbc->dc.backing_dev, + PTR_ERR(bdev)); + retcode = ERR_OPEN_DISK; goto fail; } + nbc->backing_bdev = bdev; - nbc->md_file = filp_open(nbc->dc.meta_dev, O_RDWR, 0); - if (IS_ERR(nbc->md_file)) { + /* + * meta_dev_idx >= 0: external fixed size, possibly multiple + * drbd sharing one meta device. TODO in that case, paranoia + * check that [md_bdev, meta_dev_idx] is not yet used by some + * other drbd minor! (if you use drbd.conf + drbdadm, that + * should check it for you already; but if you don't, or + * someone fooled it, we need to double check here) + */ + bdev = blkdev_get_by_path(nbc->dc.meta_dev, + FMODE_READ | FMODE_WRITE | FMODE_EXCL, + (nbc->dc.meta_dev_idx < 0) ? + (void *)mdev : (void *)drbd_m_holder); + if (IS_ERR(bdev)) { dev_err(DEV, "open(\"%s\") failed with %ld\n", nbc->dc.meta_dev, - PTR_ERR(nbc->md_file)); - nbc->md_file = NULL; + PTR_ERR(bdev)); retcode = ERR_OPEN_MD_DISK; goto fail; } + nbc->md_bdev = bdev; - inode2 = nbc->md_file->f_dentry->d_inode; - - if (!S_ISBLK(inode2->i_mode)) { - retcode = ERR_MD_NOT_BDEV; - goto fail; - } - - nbc->backing_bdev = inode->i_bdev; - if (bd_claim(nbc->backing_bdev, mdev)) { - printk(KERN_ERR "drbd: bd_claim(%p,%p); failed [%p;%p;%u]\n", - nbc->backing_bdev, mdev, - nbc->backing_bdev->bd_holder, - nbc->backing_bdev->bd_contains->bd_holder, - nbc->backing_bdev->bd_holders); - retcode = ERR_BDCLAIM_DISK; + if ((nbc->backing_bdev == nbc->md_bdev) != + (nbc->dc.meta_dev_idx == DRBD_MD_INDEX_INTERNAL || + nbc->dc.meta_dev_idx == DRBD_MD_INDEX_FLEX_INT)) { + retcode = ERR_MD_IDX_INVALID; goto fail; } @@ -851,28 +949,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp offsetof(struct bm_extent, lce)); if (!resync_lru) { retcode = ERR_NOMEM; - goto release_bdev_fail; - } - - /* meta_dev_idx >= 0: external fixed size, - * possibly multiple drbd sharing one meta device. - * TODO in that case, paranoia check that [md_bdev, meta_dev_idx] is - * not yet used by some other drbd minor! - * (if you use drbd.conf + drbdadm, - * that should check it for you already; but if you don't, or someone - * fooled it, we need to double check here) */ - nbc->md_bdev = inode2->i_bdev; - if (bd_claim(nbc->md_bdev, (nbc->dc.meta_dev_idx < 0) ? (void *)mdev - : (void *) drbd_m_holder)) { - retcode = ERR_BDCLAIM_MD_DISK; - goto release_bdev_fail; - } - - if ((nbc->backing_bdev == nbc->md_bdev) != - (nbc->dc.meta_dev_idx == DRBD_MD_INDEX_INTERNAL || - nbc->dc.meta_dev_idx == DRBD_MD_INDEX_FLEX_INT)) { - retcode = ERR_MD_IDX_INVALID; - goto release_bdev2_fail; + goto fail; } /* RT - for drbd_get_max_capacity() DRBD_MD_INDEX_FLEX_INT */ @@ -883,7 +960,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp (unsigned long long) drbd_get_max_capacity(nbc), (unsigned long long) nbc->dc.disk_size); retcode = ERR_DISK_TO_SMALL; - goto release_bdev2_fail; + goto fail; } if (nbc->dc.meta_dev_idx < 0) { @@ -900,7 +977,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp dev_warn(DEV, "refusing attach: md-device too small, " "at least %llu sectors needed for this meta-disk type\n", (unsigned long long) min_md_device_sectors); - goto release_bdev2_fail; + goto fail; } /* Make sure the new disk is big enough @@ -908,7 +985,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp if (drbd_get_max_capacity(nbc) < drbd_get_capacity(mdev->this_bdev)) { retcode = ERR_DISK_TO_SMALL; - goto release_bdev2_fail; + goto fail; } nbc->known_size = drbd_get_capacity(nbc->backing_bdev); @@ -924,14 +1001,14 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp drbd_suspend_io(mdev); /* also wait for the last barrier ack. */ - wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_pending_cnt)); + wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_pending_cnt) || is_susp(mdev->state)); /* and for any other previously queued work */ drbd_flush_workqueue(mdev); retcode = _drbd_request_state(mdev, NS(disk, D_ATTACHING), CS_VERBOSE); drbd_resume_io(mdev); if (retcode < SS_SUCCESS) - goto release_bdev2_fail; + goto fail; if (!get_ldev_if_state(mdev, D_ATTACHING)) goto force_diskless; @@ -999,9 +1076,9 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp /* Reset the "barriers don't work" bits here, then force meta data to * be written, to ensure we determine if barriers are supported. */ if (nbc->dc.no_md_flush) - set_bit(MD_NO_BARRIER, &mdev->flags); + set_bit(MD_NO_FUA, &mdev->flags); else - clear_bit(MD_NO_BARRIER, &mdev->flags); + clear_bit(MD_NO_FUA, &mdev->flags); /* Point of no return reached. * Devices and memory are no longer released by error cleanup below. @@ -1013,15 +1090,16 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp nbc = NULL; resync_lru = NULL; - mdev->write_ordering = WO_bio_barrier; - drbd_bump_write_ordering(mdev, WO_bio_barrier); + mdev->write_ordering = WO_bdev_flush; + drbd_bump_write_ordering(mdev, WO_bdev_flush); if (drbd_md_test_flag(mdev->ldev, MDF_CRASHED_PRIMARY)) set_bit(CRASHED_PRIMARY, &mdev->flags); else clear_bit(CRASHED_PRIMARY, &mdev->flags); - if (drbd_md_test_flag(mdev->ldev, MDF_PRIMARY_IND)) { + if (drbd_md_test_flag(mdev->ldev, MDF_PRIMARY_IND) && + !(mdev->state.role == R_PRIMARY && mdev->state.susp_nod)) { set_bit(CRASHED_PRIMARY, &mdev->flags); cp_discovered = 1; } @@ -1031,7 +1109,20 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp mdev->read_cnt = 0; mdev->writ_cnt = 0; - drbd_setup_queue_param(mdev, DRBD_MAX_SEGMENT_SIZE); + 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); /* If I am currently not R_PRIMARY, * but meta data primary indicator is set, @@ -1079,6 +1170,9 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp drbd_al_to_on_disk_bm(mdev); } + if (_drbd_bm_total_weight(mdev) == drbd_bm_bits(mdev)) + drbd_suspend_al(mdev); /* IO is still suspended here... */ + spin_lock_irq(&mdev->req_lock); os = mdev->state; ns.i = os.i; @@ -1146,20 +1240,16 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp force_diskless_dec: put_ldev(mdev); force_diskless: - drbd_force_state(mdev, NS(disk, D_DISKLESS)); + drbd_force_state(mdev, NS(disk, D_FAILED)); drbd_md_sync(mdev); - release_bdev2_fail: - if (nbc) - bd_release(nbc->md_bdev); - release_bdev_fail: - if (nbc) - bd_release(nbc->backing_bdev); fail: if (nbc) { - if (nbc->lo_file) - fput(nbc->lo_file); - if (nbc->md_file) - fput(nbc->md_file); + if (nbc->backing_bdev) + blkdev_put(nbc->backing_bdev, + FMODE_READ | FMODE_WRITE | FMODE_EXCL); + if (nbc->md_bdev) + blkdev_put(nbc->md_bdev, + FMODE_READ | FMODE_WRITE | FMODE_EXCL); kfree(nbc); } lc_destroy(resync_lru); @@ -1169,10 +1259,19 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp return 0; } +/* Detaching the disk is a process in multiple stages. First we need to lock + * out application IO, in-flight IO, IO stuck in drbd_al_begin_io. + * Then we transition to D_DISKLESS, and wait for put_ldev() to return all + * internal references as well. + * Only then we have finally detached. */ static int drbd_nl_detach(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { + 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)); + drbd_resume_io(mdev); return 0; } @@ -1235,7 +1334,16 @@ static int drbd_nl_net_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, && (new_conf->wire_protocol != DRBD_PROT_C)) { retcode = ERR_NOT_PROTO_C; goto fail; - }; + } + + if (get_ldev(mdev)) { + enum drbd_fencing_p fp = mdev->ldev->dc.fencing; + put_ldev(mdev); + if (new_conf->wire_protocol == DRBD_PROT_A && fp == FP_STONITH) { + retcode = ERR_STONITH_AND_PROT_A; + goto fail; + } + } if (mdev->state.role == R_PRIMARY && new_conf->want_lose) { retcode = ERR_DISCARD; @@ -1350,6 +1458,7 @@ static int drbd_nl_net_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, } } + drbd_flush_workqueue(mdev); spin_lock_irq(&mdev->req_lock); if (mdev->net_conf != NULL) { retcode = ERR_NET_CONFIGURED; @@ -1388,10 +1497,9 @@ static int drbd_nl_net_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, mdev->int_dig_out=int_dig_out; mdev->int_dig_in=int_dig_in; mdev->int_dig_vv=int_dig_vv; + retcode = _drbd_set_state(_NS(mdev, conn, C_UNCONNECTED), CS_VERBOSE, NULL); spin_unlock_irq(&mdev->req_lock); - retcode = _drbd_request_state(mdev, NS(conn, C_UNCONNECTED), CS_VERBOSE); - kobject_uevent(&disk_to_dev(mdev->vdisk)->kobj, KOBJ_CHANGE); reply->ret_code = retcode; drbd_reconfig_done(mdev); @@ -1546,6 +1654,8 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n struct crypto_hash *csums_tfm = NULL; struct syncer_conf sc; cpumask_var_t new_cpu_mask; + int *rs_plan_s = NULL; + int fifo_size; if (!zalloc_cpumask_var(&new_cpu_mask, GFP_KERNEL)) { retcode = ERR_NOMEM; @@ -1557,6 +1667,12 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n sc.rate = DRBD_RATE_DEF; sc.after = DRBD_AFTER_DEF; sc.al_extents = DRBD_AL_EXTENTS_DEF; + sc.on_no_data = DRBD_ON_NO_DATA_DEF; + sc.c_plan_ahead = DRBD_C_PLAN_AHEAD_DEF; + sc.c_delay_target = DRBD_C_DELAY_TARGET_DEF; + sc.c_fill_target = DRBD_C_FILL_TARGET_DEF; + sc.c_max_rate = DRBD_C_MAX_RATE_DEF; + sc.c_min_rate = DRBD_C_MIN_RATE_DEF; } else memcpy(&sc, &mdev->sync_conf, sizeof(struct syncer_conf)); @@ -1634,6 +1750,12 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n } #undef AL_MAX + /* to avoid spurious errors when configuring minors before configuring + * the minors they depend on: if necessary, first create the minor we + * depend on */ + if (sc.after >= 0) + ensure_mdev(sc.after, 1); + /* most sanity checks done, try to assign the new sync-after * dependency. need to hold the global lock in there, * to avoid a race in the dependency loop check. */ @@ -1641,6 +1763,16 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n if (retcode != NO_ERROR) goto fail; + fifo_size = (sc.c_plan_ahead * 10 * SLEEP_TIME) / HZ; + if (fifo_size != mdev->rs_plan_s.size && fifo_size > 0) { + rs_plan_s = kzalloc(sizeof(int) * fifo_size, GFP_KERNEL); + if (!rs_plan_s) { + dev_err(DEV, "kmalloc of fifo_buffer failed"); + retcode = ERR_NOMEM; + goto fail; + } + } + /* ok, assign the rest of it as well. * lock against receive_SyncParam() */ spin_lock(&mdev->peer_seq_lock); @@ -1657,6 +1789,15 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n mdev->verify_tfm = verify_tfm; verify_tfm = NULL; } + + if (fifo_size != mdev->rs_plan_s.size) { + kfree(mdev->rs_plan_s.values); + mdev->rs_plan_s.values = rs_plan_s; + mdev->rs_plan_s.size = fifo_size; + mdev->rs_planed = 0; + rs_plan_s = NULL; + } + spin_unlock(&mdev->peer_seq_lock); if (get_ldev(mdev)) { @@ -1688,6 +1829,7 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n kobject_uevent(&disk_to_dev(mdev->vdisk)->kobj, KOBJ_CHANGE); fail: + kfree(rs_plan_s); free_cpumask_var(new_cpu_mask); crypto_free_hash(csums_tfm); crypto_free_hash(verify_tfm); @@ -1721,12 +1863,38 @@ static int drbd_nl_invalidate(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl return 0; } +static int drbd_bmio_set_susp_al(struct drbd_conf *mdev) +{ + int rv; + + rv = drbd_bmio_set_n_write(mdev); + drbd_suspend_al(mdev); + return rv; +} + static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { + int retcode; - reply->ret_code = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S)); + retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S), CS_ORDERED); + + if (retcode < SS_SUCCESS) { + if (retcode == SS_NEED_CONNECTION && mdev->state.role == R_PRIMARY) { + /* The peer will get a resync upon connect anyways. Just make that + 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")) + retcode = ERR_IO_MD_DISK; + } + } else + retcode = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S)); + } + reply->ret_code = retcode; return 0; } @@ -1765,7 +1933,20 @@ static int drbd_nl_suspend_io(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl static int drbd_nl_resume_io(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { - reply->ret_code = drbd_request_state(mdev, NS(susp, 0)); + if (test_bit(NEW_CUR_UUID, &mdev->flags)) { + drbd_uuid_new_current(mdev); + clear_bit(NEW_CUR_UUID, &mdev->flags); + } + drbd_suspend_io(mdev); + reply->ret_code = drbd_request_state(mdev, NS3(susp, 0, susp_nod, 0, susp_fen, 0)); + if (reply->ret_code == SS_SUCCESS) { + if (mdev->state.conn < C_CONNECTED) + tl_clear(mdev); + if (mdev->state.disk == D_DISKLESS || mdev->state.disk == D_FAILED) + tl_restart(mdev, fail_frozen_disk_io); + } + drbd_resume_io(mdev); + return 0; } @@ -1941,40 +2122,6 @@ out: return 0; } -static struct drbd_conf *ensure_mdev(struct drbd_nl_cfg_req *nlp) -{ - struct drbd_conf *mdev; - - if (nlp->drbd_minor >= minor_count) - return NULL; - - mdev = minor_to_mdev(nlp->drbd_minor); - - if (!mdev && (nlp->flags & DRBD_NL_CREATE_DEVICE)) { - struct gendisk *disk = NULL; - mdev = drbd_new_device(nlp->drbd_minor); - - spin_lock_irq(&drbd_pp_lock); - if (minor_table[nlp->drbd_minor] == NULL) { - minor_table[nlp->drbd_minor] = mdev; - disk = mdev->vdisk; - mdev = NULL; - } /* else: we lost the race */ - spin_unlock_irq(&drbd_pp_lock); - - if (disk) /* we won the race above */ - /* in case we ever add a drbd_delete_device(), - * don't forget the del_gendisk! */ - add_disk(disk); - else /* we lost the race above */ - drbd_free_mdev(mdev); - - mdev = minor_to_mdev(nlp->drbd_minor); - } - - return mdev; -} - struct cn_handler_struct { int (*function)(struct drbd_conf *, struct drbd_nl_cfg_req *, @@ -2035,7 +2182,8 @@ static void drbd_connector_callback(struct cn_msg *req, struct netlink_skb_parms goto fail; } - mdev = ensure_mdev(nlp); + mdev = ensure_mdev(nlp->drbd_minor, + (nlp->flags & DRBD_NL_CREATE_DEVICE)); if (!mdev) { retcode = ERR_MINOR_INVALID; goto fail; diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c index be3374b..7e6ac30 100644 --- a/drivers/block/drbd/drbd_proc.c +++ b/drivers/block/drbd/drbd_proc.c @@ -57,6 +57,7 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) unsigned long db, dt, dbdt, rt, rs_left; unsigned int res; int i, x, y; + int stalled = 0; drbd_get_syncer_progress(mdev, &rs_left, &res); @@ -90,18 +91,17 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) * db: blocks written from mark until now * rt: remaining time */ - dt = (jiffies - mdev->rs_mark_time) / HZ; - - if (dt > 20) { - /* if we made no update to rs_mark_time for too long, - * we are stalled. show that. */ - seq_printf(seq, "stalled\n"); - return; - } + /* 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. */ + 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)) + stalled = 1; if (!dt) dt++; - db = mdev->rs_mark_left - rs_left; + db = mdev->rs_mark_left[i] - rs_left; rt = (dt * (rs_left / (db/100+1)))/100; /* seconds */ seq_printf(seq, "finish: %lu:%02lu:%02lu", @@ -118,7 +118,7 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) /* mean speed since syncer started * we do account for PausedSync periods */ dt = (jiffies - mdev->rs_start - mdev->rs_paused) / HZ; - if (dt <= 0) + if (dt == 0) dt = 1; db = mdev->rs_total - rs_left; dbdt = Bit2KB(db/dt); @@ -128,7 +128,14 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) else seq_printf(seq, " (%ld)", dbdt); - seq_printf(seq, " K/sec\n"); + 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); + } + seq_printf(seq, " K/sec%s\n", stalled ? " (stalled)" : ""); } static void resync_dump_detail(struct seq_file *seq, struct lc_element *e) @@ -151,7 +158,6 @@ static int drbd_seq_show(struct seq_file *seq, void *v) [WO_none] = 'n', [WO_drain_io] = 'd', [WO_bdev_flush] = 'f', - [WO_bio_barrier] = 'b', }; seq_printf(seq, "version: " REL_VERSION " (api:%d/proto:%d-%d)\n%s\n", @@ -196,7 +202,7 @@ static int drbd_seq_show(struct seq_file *seq, void *v) seq_printf(seq, "%2d: cs:Unconfigured\n", i); } else { seq_printf(seq, - "%2d: cs:%s ro:%s/%s ds:%s/%s %c %c%c%c%c%c\n" + "%2d: cs:%s ro:%s/%s ds:%s/%s %c %c%c%c%c%c%c\n" " ns:%u nr:%u dw:%u dr:%u al:%u bm:%u " "lo:%d pe:%d ua:%d ap:%d ep:%d wo:%c", i, sn, @@ -206,11 +212,12 @@ static int drbd_seq_show(struct seq_file *seq, void *v) drbd_disk_str(mdev->state.pdsk), (mdev->net_conf == NULL ? ' ' : (mdev->net_conf->wire_protocol - DRBD_PROT_A+'A')), - mdev->state.susp ? 's' : 'r', + is_susp(mdev->state) ? 's' : 'r', mdev->state.aftr_isp ? 'a' : '-', mdev->state.peer_isp ? 'p' : '-', mdev->state.user_isp ? 'u' : '-', mdev->congestion_reason ?: '-', + test_bit(AL_SUSPENDED, &mdev->flags) ? 's' : '-', mdev->send_cnt/2, mdev->recv_cnt/2, mdev->writ_cnt/2, diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 081522d..24487d4 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -36,7 +36,6 @@ #include <linux/memcontrol.h> #include <linux/mm_inline.h> #include <linux/slab.h> -#include <linux/smp_lock.h> #include <linux/pkt_sched.h> #define __KERNEL_SYSCALLS__ #include <linux/unistd.h> @@ -49,11 +48,6 @@ #include "drbd_vli.h" -struct flush_work { - struct drbd_work w; - struct drbd_epoch *epoch; -}; - enum finish_epoch { FE_STILL_LIVE, FE_DESTROYED, @@ -66,16 +60,6 @@ static int drbd_do_auth(struct drbd_conf *mdev); static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *, struct drbd_epoch *, enum epoch_event); static int e_end_block(struct drbd_conf *, struct drbd_work *, int); -static struct drbd_epoch *previous_epoch(struct drbd_conf *mdev, struct drbd_epoch *epoch) -{ - struct drbd_epoch *prev; - spin_lock(&mdev->epoch_lock); - prev = list_entry(epoch->list.prev, struct drbd_epoch, list); - if (prev == epoch || prev == mdev->current_epoch) - prev = NULL; - spin_unlock(&mdev->epoch_lock); - return prev; -} #define GFP_TRY (__GFP_HIGHMEM | __GFP_NOWARN) @@ -241,7 +225,7 @@ static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev) spin_unlock_irq(&mdev->req_lock); list_for_each_entry_safe(e, t, &reclaimed, w.list) - drbd_free_ee(mdev, e); + drbd_free_net_ee(mdev, e); } /** @@ -298,9 +282,11 @@ static struct page *drbd_pp_alloc(struct drbd_conf *mdev, unsigned number, bool * Is also used from inside an other spin_lock_irq(&mdev->req_lock); * Either links the page chain back to the global pool, * or returns all pages to the system. */ -static void drbd_pp_free(struct drbd_conf *mdev, struct page *page) +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) i = page_chain_free(page); else { @@ -311,10 +297,10 @@ static void drbd_pp_free(struct drbd_conf *mdev, struct page *page) drbd_pp_vacant += i; spin_unlock(&drbd_pp_lock); } - atomic_sub(i, &mdev->pp_in_use); - i = atomic_read(&mdev->pp_in_use); + i = atomic_sub_return(i, a); if (i < 0) - dev_warn(DEV, "ASSERTION FAILED: pp_in_use: %d < 0\n", i); + dev_warn(DEV, "ASSERTION FAILED: %s: %d < 0\n", + is_net ? "pp_in_use_by_net" : "pp_in_use", i); wake_up(&drbd_pp_wait); } @@ -365,7 +351,6 @@ struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev, e->size = data_size; e->flags = 0; e->sector = sector; - e->sector = sector; e->block_id = id; return e; @@ -375,9 +360,11 @@ struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev, return NULL; } -void drbd_free_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e) +void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, int is_net) { - drbd_pp_free(mdev, e->pages); + if (e->flags & EE_HAS_DIGEST) + 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)); mempool_free(e, drbd_ee_mempool); @@ -388,13 +375,14 @@ int drbd_release_ee(struct drbd_conf *mdev, struct list_head *list) LIST_HEAD(work_list); struct drbd_epoch_entry *e, *t; int count = 0; + int is_net = list == &mdev->net_ee; spin_lock_irq(&mdev->req_lock); list_splice_init(list, &work_list); spin_unlock_irq(&mdev->req_lock); list_for_each_entry_safe(e, t, &work_list, w.list) { - drbd_free_ee(mdev, e); + drbd_free_some_ee(mdev, e, is_net); count++; } return count; @@ -423,7 +411,7 @@ static int drbd_process_done_ee(struct drbd_conf *mdev) spin_unlock_irq(&mdev->req_lock); list_for_each_entry_safe(e, t, &reclaimed, w.list) - drbd_free_ee(mdev, e); + drbd_free_net_ee(mdev, e); /* possible callbacks here: * e_end_block, and e_end_resync_block, e_send_discard_ack. @@ -719,14 +707,14 @@ out: static int drbd_send_fp(struct drbd_conf *mdev, struct socket *sock, enum drbd_packets cmd) { - struct p_header *h = (struct p_header *) &mdev->data.sbuf.header; + struct p_header80 *h = &mdev->data.sbuf.header.h80; return _drbd_send_cmd(mdev, sock, cmd, h, sizeof(*h), 0); } static enum drbd_packets drbd_recv_fp(struct drbd_conf *mdev, struct socket *sock) { - struct p_header *h = (struct p_header *) &mdev->data.sbuf.header; + struct p_header80 *h = &mdev->data.rbuf.header.h80; int rr; rr = drbd_recv_short(mdev, sock, h, sizeof(*h), 0); @@ -776,9 +764,6 @@ static int drbd_connect(struct drbd_conf *mdev) D_ASSERT(!mdev->data.socket); - if (test_and_clear_bit(CREATE_BARRIER, &mdev->flags)) - dev_err(DEV, "CREATE_BARRIER flag was set in drbd_connect - now cleared!\n"); - if (drbd_request_state(mdev, NS(conn, C_WF_CONNECTION)) < SS_SUCCESS) return -2; @@ -927,6 +912,11 @@ 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)) return -1; drbd_send_sync_param(mdev, &mdev->sync_conf); @@ -946,22 +936,28 @@ out_release_sockets: return -1; } -static int drbd_recv_header(struct drbd_conf *mdev, struct p_header *h) +static int drbd_recv_header(struct drbd_conf *mdev, enum drbd_packets *cmd, unsigned int *packet_size) { + union p_header *h = &mdev->data.rbuf.header; int r; 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; - }; - h->command = be16_to_cpu(h->command); - h->length = be16_to_cpu(h->length); - if (unlikely(h->magic != BE_DRBD_MAGIC)) { - dev_err(DEV, "magic?? on data m: 0x%lx c: %d l: %d\n", - (long)be32_to_cpu(h->magic), - h->command, h->length); + } + + if (likely(h->h80.magic == BE_DRBD_MAGIC)) { + *cmd = be16_to_cpu(h->h80.command); + *packet_size = be16_to_cpu(h->h80.length); + } else if (h->h95.magic == BE_DRBD_MAGIC_BIG) { + *cmd = be16_to_cpu(h->h95.command); + *packet_size = be32_to_cpu(h->h95.length); + } else { + dev_err(DEV, "magic?? on data m: 0x%08x c: %d l: %d\n", + be32_to_cpu(h->h80.magic), + be16_to_cpu(h->h80.command), + be16_to_cpu(h->h80.length)); return FALSE; } mdev->last_received = jiffies; @@ -969,13 +965,13 @@ static int drbd_recv_header(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static enum finish_epoch drbd_flush_after_epoch(struct drbd_conf *mdev, struct drbd_epoch *epoch) +static void drbd_flush(struct drbd_conf *mdev) { int rv; if (mdev->write_ordering >= WO_bdev_flush && get_ldev(mdev)) { rv = blkdev_issue_flush(mdev->ldev->backing_bdev, GFP_KERNEL, - NULL, BLKDEV_IFL_WAIT); + NULL); if (rv) { dev_err(DEV, "local disk flush failed with status %d\n", rv); /* would rather check on EOPNOTSUPP, but that is not reliable. @@ -985,24 +981,6 @@ static enum finish_epoch drbd_flush_after_epoch(struct drbd_conf *mdev, struct d } put_ldev(mdev); } - - return drbd_may_finish_epoch(mdev, epoch, EV_BARRIER_DONE); -} - -static int w_flush(struct drbd_conf *mdev, struct drbd_work *w, int cancel) -{ - struct flush_work *fw = (struct flush_work *)w; - struct drbd_epoch *epoch = fw->epoch; - - kfree(w); - - if (!test_and_set_bit(DE_BARRIER_IN_NEXT_EPOCH_ISSUED, &epoch->flags)) - drbd_flush_after_epoch(mdev, epoch); - - drbd_may_finish_epoch(mdev, epoch, EV_PUT | - (mdev->state.conn < C_CONNECTED ? EV_CLEANUP : 0)); - - return 1; } /** @@ -1015,15 +993,13 @@ static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *mdev, struct drbd_epoch *epoch, enum epoch_event ev) { - int finish, epoch_size; + int epoch_size; struct drbd_epoch *next_epoch; - int schedule_flush = 0; enum finish_epoch rv = FE_STILL_LIVE; spin_lock(&mdev->epoch_lock); do { next_epoch = NULL; - finish = 0; epoch_size = atomic_read(&epoch->epoch_size); @@ -1033,16 +1009,6 @@ static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *mdev, break; case EV_GOT_BARRIER_NR: set_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags); - - /* Special case: If we just switched from WO_bio_barrier to - WO_bdev_flush we should not finish the current epoch */ - if (test_bit(DE_CONTAINS_A_BARRIER, &epoch->flags) && epoch_size == 1 && - mdev->write_ordering != WO_bio_barrier && - epoch == mdev->current_epoch) - clear_bit(DE_CONTAINS_A_BARRIER, &epoch->flags); - break; - case EV_BARRIER_DONE: - set_bit(DE_BARRIER_IN_NEXT_EPOCH_DONE, &epoch->flags); break; case EV_BECAME_LAST: /* nothing to do*/ @@ -1051,23 +1017,7 @@ static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *mdev, if (epoch_size != 0 && atomic_read(&epoch->active) == 0 && - test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags) && - epoch->list.prev == &mdev->current_epoch->list && - !test_bit(DE_IS_FINISHING, &epoch->flags)) { - /* Nearly all conditions are met to finish that epoch... */ - if (test_bit(DE_BARRIER_IN_NEXT_EPOCH_DONE, &epoch->flags) || - mdev->write_ordering == WO_none || - (epoch_size == 1 && test_bit(DE_CONTAINS_A_BARRIER, &epoch->flags)) || - ev & EV_CLEANUP) { - finish = 1; - set_bit(DE_IS_FINISHING, &epoch->flags); - } else if (!test_bit(DE_BARRIER_IN_NEXT_EPOCH_ISSUED, &epoch->flags) && - mdev->write_ordering == WO_bio_barrier) { - atomic_inc(&epoch->active); - schedule_flush = 1; - } - } - if (finish) { + test_bit(DE_HAVE_BARRIER_NUMBER, &epoch->flags)) { if (!(ev & EV_CLEANUP)) { spin_unlock(&mdev->epoch_lock); drbd_send_b_ack(mdev, epoch->barrier_nr, epoch_size); @@ -1090,6 +1040,7 @@ static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *mdev, /* atomic_set(&epoch->active, 0); is already zero */ if (rv == FE_STILL_LIVE) rv = FE_RECYCLED; + wake_up(&mdev->ee_wait); } } @@ -1101,22 +1052,6 @@ static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *mdev, spin_unlock(&mdev->epoch_lock); - if (schedule_flush) { - struct flush_work *fw; - fw = kmalloc(sizeof(*fw), GFP_ATOMIC); - if (fw) { - fw->w.cb = w_flush; - fw->epoch = epoch; - drbd_queue_work(&mdev->data.work, &fw->w); - } else { - dev_warn(DEV, "Could not kmalloc a flush_work obj\n"); - set_bit(DE_BARRIER_IN_NEXT_EPOCH_ISSUED, &epoch->flags); - /* That is not a recursion, only one level */ - drbd_may_finish_epoch(mdev, epoch, EV_BARRIER_DONE); - drbd_may_finish_epoch(mdev, epoch, EV_PUT); - } - } - return rv; } @@ -1132,19 +1067,16 @@ void drbd_bump_write_ordering(struct drbd_conf *mdev, enum write_ordering_e wo) [WO_none] = "none", [WO_drain_io] = "drain", [WO_bdev_flush] = "flush", - [WO_bio_barrier] = "barrier", }; pwo = mdev->write_ordering; wo = min(pwo, wo); - if (wo == WO_bio_barrier && mdev->ldev->dc.no_disk_barrier) - wo = WO_bdev_flush; if (wo == WO_bdev_flush && mdev->ldev->dc.no_disk_flush) wo = WO_drain_io; if (wo == WO_drain_io && mdev->ldev->dc.no_disk_drain) wo = WO_none; mdev->write_ordering = wo; - if (pwo != mdev->write_ordering || wo == WO_bio_barrier) + if (pwo != mdev->write_ordering || wo == WO_bdev_flush) dev_info(DEV, "Method to ensure write ordering: %s\n", write_ordering_str[mdev->write_ordering]); } @@ -1180,7 +1112,7 @@ next_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, REQ_HARDBARRIER) */ + * (REQ_UNPLUG) */ bio->bi_rw = rw; bio->bi_private = e; bio->bi_end_io = drbd_endio_sec; @@ -1214,11 +1146,6 @@ next_bio: bio->bi_rw &= ~REQ_UNPLUG; drbd_generic_make_request(mdev, fault_type, bio); - - /* strip off REQ_HARDBARRIER, - * unless it is the first or last bio */ - if (bios && bios->bi_next) - bios->bi_rw &= ~REQ_HARDBARRIER; } while (bios); maybe_kick_lo(mdev); return 0; @@ -1232,53 +1159,12 @@ fail: return -ENOMEM; } -/** - * w_e_reissue() - Worker callback; Resubmit a bio, without REQ_HARDBARRIER set - * @mdev: DRBD device. - * @w: work object. - * @cancel: The connection will be closed anyways (unused in this callback) - */ -int w_e_reissue(struct drbd_conf *mdev, struct drbd_work *w, int cancel) __releases(local) -{ - struct drbd_epoch_entry *e = (struct drbd_epoch_entry *)w; - /* We leave DE_CONTAINS_A_BARRIER and EE_IS_BARRIER in place, - (and DE_BARRIER_IN_NEXT_EPOCH_ISSUED in the previous Epoch) - so that we can finish that epoch in drbd_may_finish_epoch(). - That is necessary if we already have a long chain of Epochs, before - we realize that REQ_HARDBARRIER is actually not supported */ - - /* As long as the -ENOTSUPP on the barrier is reported immediately - that will never trigger. If it is reported late, we will just - print that warning and continue correctly for all future requests - with WO_bdev_flush */ - if (previous_epoch(mdev, e->epoch)) - dev_warn(DEV, "Write ordering was not enforced (one time event)\n"); - - /* we still have a local reference, - * get_ldev was done in receive_Data. */ - - e->w.cb = e_end_block; - if (drbd_submit_ee(mdev, e, WRITE, DRBD_FAULT_DT_WR) != 0) { - /* drbd_submit_ee fails for one reason only: - * if was not able to allocate sufficient bios. - * requeue, try again later. */ - e->w.cb = w_e_reissue; - drbd_queue_work(&mdev->data.work, &e->w); - } - return 1; -} - -static int receive_Barrier(struct drbd_conf *mdev, struct p_header *h) +static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - int rv, issue_flush; - struct p_barrier *p = (struct p_barrier *)h; + int rv; + struct p_barrier *p = &mdev->data.rbuf.barrier; struct drbd_epoch *epoch; - ERR_IF(h->length != (sizeof(*p)-sizeof(*h))) return FALSE; - - rv = drbd_recv(mdev, h->payload, h->length); - ERR_IF(rv != h->length) return FALSE; - inc_unacked(mdev); if (mdev->net_conf->wire_protocol != DRBD_PROT_C) @@ -1293,44 +1179,40 @@ static int receive_Barrier(struct drbd_conf *mdev, struct p_header *h) * Therefore we must send the barrier_ack after the barrier request was * completed. */ switch (mdev->write_ordering) { - case WO_bio_barrier: case WO_none: if (rv == FE_RECYCLED) return TRUE; - break; + + /* receiver context, in the writeout path of the other node. + * avoid potential distributed deadlock */ + epoch = kmalloc(sizeof(struct drbd_epoch), GFP_NOIO); + if (epoch) + break; + else + dev_warn(DEV, "Allocation of an epoch failed, slowing down\n"); + /* Fall through */ case WO_bdev_flush: case WO_drain_io: - if (rv == FE_STILL_LIVE) { - set_bit(DE_BARRIER_IN_NEXT_EPOCH_ISSUED, &mdev->current_epoch->flags); - drbd_wait_ee_list_empty(mdev, &mdev->active_ee); - rv = drbd_flush_after_epoch(mdev, mdev->current_epoch); - } - if (rv == FE_RECYCLED) - return TRUE; - - /* The asender will send all the ACKs and barrier ACKs out, since - all EEs moved from the active_ee to the done_ee. We need to - provide a new epoch object for the EEs that come in soon */ - break; - } - - /* receiver context, in the writeout path of the other node. - * avoid potential distributed deadlock */ - epoch = kmalloc(sizeof(struct drbd_epoch), GFP_NOIO); - if (!epoch) { - dev_warn(DEV, "Allocation of an epoch failed, slowing down\n"); - issue_flush = !test_and_set_bit(DE_BARRIER_IN_NEXT_EPOCH_ISSUED, &mdev->current_epoch->flags); drbd_wait_ee_list_empty(mdev, &mdev->active_ee); - if (issue_flush) { - rv = drbd_flush_after_epoch(mdev, mdev->current_epoch); - if (rv == FE_RECYCLED) - return TRUE; + drbd_flush(mdev); + + if (atomic_read(&mdev->current_epoch->epoch_size)) { + epoch = kmalloc(sizeof(struct drbd_epoch), GFP_NOIO); + if (epoch) + break; } - drbd_wait_ee_list_empty(mdev, &mdev->done_ee); + epoch = mdev->current_epoch; + wait_event(mdev->ee_wait, atomic_read(&epoch->epoch_size) == 0); + + D_ASSERT(atomic_read(&epoch->active) == 0); + D_ASSERT(epoch->flags == 0); return TRUE; + default: + dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering); + return FALSE; } epoch->flags = 0; @@ -1457,7 +1339,7 @@ static int drbd_drain_block(struct drbd_conf *mdev, int data_size) data_size -= rr; } kunmap(page); - drbd_pp_free(mdev, page); + drbd_pp_free(mdev, page, 0); return rv; } @@ -1562,30 +1444,29 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si list_add(&e->w.list, &mdev->sync_ee); spin_unlock_irq(&mdev->req_lock); + atomic_add(data_size >> 9, &mdev->rs_sect_ev); if (drbd_submit_ee(mdev, e, WRITE, DRBD_FAULT_RS_WR) == 0) 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? */ + spin_lock_irq(&mdev->req_lock); + list_del(&e->w.list); + spin_unlock_irq(&mdev->req_lock); + drbd_free_ee(mdev, e); fail: put_ldev(mdev); return FALSE; } -static int receive_DataReply(struct drbd_conf *mdev, struct p_header *h) +static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { struct drbd_request *req; sector_t sector; - unsigned int header_size, data_size; int ok; - struct p_data *p = (struct p_data *)h; - - header_size = sizeof(*p) - sizeof(*h); - data_size = h->length - header_size; - - ERR_IF(data_size == 0) return FALSE; - - if (drbd_recv(mdev, h->payload, header_size) != header_size) - return FALSE; + struct p_data *p = &mdev->data.rbuf.data; sector = be64_to_cpu(p->sector); @@ -1611,20 +1492,11 @@ static int receive_DataReply(struct drbd_conf *mdev, struct p_header *h) return ok; } -static int receive_RSDataReply(struct drbd_conf *mdev, struct p_header *h) +static int receive_RSDataReply(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { sector_t sector; - unsigned int header_size, data_size; int ok; - struct p_data *p = (struct p_data *)h; - - header_size = sizeof(*p) - sizeof(*h); - data_size = h->length - header_size; - - ERR_IF(data_size == 0) return FALSE; - - if (drbd_recv(mdev, h->payload, header_size) != header_size) - return FALSE; + struct p_data *p = &mdev->data.rbuf.data; sector = be64_to_cpu(p->sector); D_ASSERT(p->block_id == ID_SYNCER); @@ -1640,9 +1512,11 @@ static int receive_RSDataReply(struct drbd_conf *mdev, struct p_header *h) ok = drbd_drain_block(mdev, data_size); - drbd_send_ack_dp(mdev, P_NEG_ACK, p); + drbd_send_ack_dp(mdev, P_NEG_ACK, p, data_size); } + atomic_add(data_size >> 9, &mdev->rs_sect_in); + return ok; } @@ -1653,15 +1527,8 @@ static int e_end_block(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { struct drbd_epoch_entry *e = (struct drbd_epoch_entry *)w; sector_t sector = e->sector; - struct drbd_epoch *epoch; int ok = 1, pcmd; - if (e->flags & EE_IS_BARRIER) { - epoch = previous_epoch(mdev, e->epoch); - if (epoch) - drbd_may_finish_epoch(mdev, epoch, EV_BARRIER_DONE + (cancel ? EV_CLEANUP : 0)); - } - if (mdev->net_conf->wire_protocol == DRBD_PROT_C) { if (likely((e->flags & EE_WAS_ERROR) == 0)) { pcmd = (mdev->state.conn >= C_SYNC_SOURCE && @@ -1765,24 +1632,27 @@ 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) +{ + 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; +} + /* mirrored write */ -static int receive_Data(struct drbd_conf *mdev, struct p_header *h) +static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { sector_t sector; struct drbd_epoch_entry *e; - struct p_data *p = (struct p_data *)h; - int header_size, data_size; + struct p_data *p = &mdev->data.rbuf.data; int rw = WRITE; u32 dp_flags; - header_size = sizeof(*p) - sizeof(*h); - data_size = h->length - header_size; - - ERR_IF(data_size == 0) return FALSE; - - if (drbd_recv(mdev, h->payload, header_size) != header_size) - return FALSE; - if (!get_ldev(mdev)) { if (__ratelimit(&drbd_ratelimit_state)) dev_err(DEV, "Can not write mirrored data block " @@ -1792,7 +1662,7 @@ static int receive_Data(struct drbd_conf *mdev, struct p_header *h) mdev->peer_seq++; spin_unlock(&mdev->peer_seq_lock); - drbd_send_ack_dp(mdev, P_NEG_ACK, p); + drbd_send_ack_dp(mdev, P_NEG_ACK, p, data_size); atomic_inc(&mdev->current_epoch->epoch_size); return drbd_drain_block(mdev, data_size); } @@ -1815,36 +1685,11 @@ static int receive_Data(struct drbd_conf *mdev, struct p_header *h) e->epoch = mdev->current_epoch; atomic_inc(&e->epoch->epoch_size); atomic_inc(&e->epoch->active); - - if (mdev->write_ordering == WO_bio_barrier && atomic_read(&e->epoch->epoch_size) == 1) { - struct drbd_epoch *epoch; - /* Issue a barrier if we start a new epoch, and the previous epoch - was not a epoch containing a single request which already was - a Barrier. */ - epoch = list_entry(e->epoch->list.prev, struct drbd_epoch, list); - if (epoch == e->epoch) { - set_bit(DE_CONTAINS_A_BARRIER, &e->epoch->flags); - rw |= REQ_HARDBARRIER; - e->flags |= EE_IS_BARRIER; - } else { - if (atomic_read(&epoch->epoch_size) > 1 || - !test_bit(DE_CONTAINS_A_BARRIER, &epoch->flags)) { - set_bit(DE_BARRIER_IN_NEXT_EPOCH_ISSUED, &epoch->flags); - set_bit(DE_CONTAINS_A_BARRIER, &e->epoch->flags); - rw |= REQ_HARDBARRIER; - e->flags |= EE_IS_BARRIER; - } - } - } spin_unlock(&mdev->epoch_lock); dp_flags = be32_to_cpu(p->dp_flags); - if (dp_flags & DP_HARDBARRIER) { - dev_err(DEV, "ASSERT FAILED would have submitted barrier request\n"); - /* rw |= REQ_HARDBARRIER; */ - } - if (dp_flags & DP_RW_SYNC) - rw |= REQ_SYNC | REQ_UNPLUG; + rw |= write_flags_to_bio(mdev, dp_flags); + if (dp_flags & DP_MAY_SET_IN_SYNC) e->flags |= EE_MAY_SET_IN_SYNC; @@ -1997,16 +1842,27 @@ static int receive_Data(struct drbd_conf *mdev, struct p_header *h) break; } - if (mdev->state.pdsk == D_DISKLESS) { + if (mdev->state.pdsk < D_INCONSISTENT) { /* In case we have the only disk of the cluster, */ drbd_set_out_of_sync(mdev, e->sector, e->size); e->flags |= EE_CALL_AL_COMPLETE_IO; + e->flags &= ~EE_MAY_SET_IN_SYNC; drbd_al_begin_io(mdev, e->sector); } if (drbd_submit_ee(mdev, e, rw, DRBD_FAULT_DT_WR) == 0) 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? */ + spin_lock_irq(&mdev->req_lock); + list_del(&e->w.list); + hlist_del_init(&e->colision); + 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 @@ -2016,20 +1872,64 @@ out_interrupted: return FALSE; } -static int receive_DataRequest(struct drbd_conf *mdev, struct p_header *h) +/* We may throttle resync, if the lower device seems to be busy, + * and current sync rate is above c_min_rate. + * + * To decide whether or not the lower device is busy, we use a scheme similar + * to MD RAID is_mddev_idle(): if the partition stats reveal "significant" + * (more than 64 sectors) of activity we cannot account for with our own resync + * activity, it obviously is "busy". + * + * 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) +{ + struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk; + unsigned long db, dt, dbdt; + int curr_events; + int throttle = 0; + + /* feature disabled? */ + if (mdev->sync_conf.c_min_rate == 0) + return 0; + + 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; + + mdev->rs_last_events = curr_events; + + /* 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; + + dt = ((long)jiffies - (long)mdev->rs_mark_time[i]) / HZ; + if (!dt) + dt++; + db = mdev->rs_mark_left[i] - rs_left; + dbdt = Bit2KB(db/dt); + + if (dbdt > mdev->sync_conf.c_min_rate) + throttle = 1; + } + return throttle; +} + + +static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int digest_size) { sector_t sector; const sector_t capacity = drbd_get_capacity(mdev->this_bdev); struct drbd_epoch_entry *e; struct digest_info *di = NULL; - int size, digest_size; + int size, verb; unsigned int fault_type; - struct p_block_req *p = - (struct p_block_req *)h; - const int brps = sizeof(*p)-sizeof(*h); - - if (drbd_recv(mdev, h->payload, brps) != brps) - return FALSE; + struct p_block_req *p = &mdev->data.rbuf.block_req; sector = be64_to_cpu(p->sector); size = be32_to_cpu(p->blksize); @@ -2046,12 +1946,31 @@ static int receive_DataRequest(struct drbd_conf *mdev, struct p_header *h) } if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) { - if (__ratelimit(&drbd_ratelimit_state)) + verb = 1; + switch (cmd) { + case P_DATA_REQUEST: + drbd_send_ack_rp(mdev, P_NEG_DREPLY, p); + break; + case P_RS_DATA_REQUEST: + case P_CSUM_RS_REQUEST: + case P_OV_REQUEST: + drbd_send_ack_rp(mdev, P_NEG_RS_DREPLY , p); + break; + case P_OV_REPLY: + verb = 0; + dec_rs_pending(mdev); + drbd_send_ack_ex(mdev, P_OV_RESULT, sector, size, ID_IN_SYNC); + break; + default: + dev_err(DEV, "unexpected command (%s) in receive_DataRequest\n", + cmdname(cmd)); + } + if (verb && __ratelimit(&drbd_ratelimit_state)) dev_err(DEV, "Can not satisfy peer's read request, " "no local data.\n"); - drbd_send_ack_rp(mdev, h->command == P_DATA_REQUEST ? P_NEG_DREPLY : - P_NEG_RS_DREPLY , p); - return drbd_drain_block(mdev, h->length - brps); + + /* drain possibly payload */ + return drbd_drain_block(mdev, digest_size); } /* GFP_NOIO, because we must not cause arbitrary write-out: in a DRBD @@ -2063,31 +1982,21 @@ static int receive_DataRequest(struct drbd_conf *mdev, struct p_header *h) return FALSE; } - switch (h->command) { + switch (cmd) { case P_DATA_REQUEST: e->w.cb = w_e_end_data_req; fault_type = DRBD_FAULT_DT_RD; - break; + /* application IO, don't drbd_rs_begin_io */ + goto submit; + case P_RS_DATA_REQUEST: e->w.cb = w_e_end_rsdata_req; fault_type = DRBD_FAULT_RS_RD; - /* Eventually this should become asynchronously. Currently it - * blocks the whole receiver just to delay the reading of a - * resync data block. - * the drbd_work_queue mechanism is made for this... - */ - if (!drbd_rs_begin_io(mdev, sector)) { - /* we have been interrupted, - * probably connection lost! */ - D_ASSERT(signal_pending(current)); - goto out_free_e; - } break; case P_OV_REPLY: case P_CSUM_RS_REQUEST: fault_type = DRBD_FAULT_RS_RD; - digest_size = h->length - brps ; di = kmalloc(sizeof(*di) + digest_size, GFP_NOIO); if (!di) goto out_free_e; @@ -2095,31 +2004,25 @@ static int receive_DataRequest(struct drbd_conf *mdev, struct p_header *h) di->digest_size = digest_size; di->digest = (((char *)di)+sizeof(struct digest_info)); + e->digest = di; + e->flags |= EE_HAS_DIGEST; + if (drbd_recv(mdev, di->digest, digest_size) != digest_size) goto out_free_e; - e->block_id = (u64)(unsigned long)di; - if (h->command == P_CSUM_RS_REQUEST) { + if (cmd == P_CSUM_RS_REQUEST) { D_ASSERT(mdev->agreed_pro_version >= 89); e->w.cb = w_e_end_csum_rs_req; - } else if (h->command == P_OV_REPLY) { + } else if (cmd == P_OV_REPLY) { e->w.cb = w_e_end_ov_reply; dec_rs_pending(mdev); - break; - } - - if (!drbd_rs_begin_io(mdev, sector)) { - /* we have been interrupted, probably connection lost! */ - D_ASSERT(signal_pending(current)); - goto out_free_e; + /* drbd_rs_begin_io done when we sent this request, + * but accounting still needs to be done. */ + goto submit_for_resync; } break; case P_OV_REQUEST: - if (mdev->state.conn >= C_CONNECTED && - mdev->state.conn != C_VERIFY_T) - dev_warn(DEV, "ASSERT FAILED: got P_OV_REQUEST while being %s\n", - drbd_conn_str(mdev->state.conn)); if (mdev->ov_start_sector == ~(sector_t)0 && mdev->agreed_pro_version >= 90) { mdev->ov_start_sector = sector; @@ -2130,37 +2033,63 @@ static int receive_DataRequest(struct drbd_conf *mdev, struct p_header *h) } e->w.cb = w_e_end_ov_req; fault_type = DRBD_FAULT_RS_RD; - /* Eventually this should become asynchronous. Currently it - * blocks the whole receiver just to delay the reading of a - * resync data block. - * the drbd_work_queue mechanism is made for this... - */ - if (!drbd_rs_begin_io(mdev, sector)) { - /* we have been interrupted, - * probably connection lost! */ - D_ASSERT(signal_pending(current)); - goto out_free_e; - } break; - default: dev_err(DEV, "unexpected command (%s) in receive_DataRequest\n", - cmdname(h->command)); + cmdname(cmd)); fault_type = DRBD_FAULT_MAX; + goto out_free_e; } - spin_lock_irq(&mdev->req_lock); - list_add(&e->w.list, &mdev->read_ee); - spin_unlock_irq(&mdev->req_lock); + /* Throttle, drbd_rs_begin_io and submit should become asynchronous + * wrt the receiver, but it is not as straightforward as it may seem. + * Various places in the resync start and stop logic assume resync + * requests are processed in order, requeuing this on the worker thread + * introduces a bunch of new code for synchronization between threads. + * + * Unlimited throttling before drbd_rs_begin_io may stall the resync + * "forever", throttling after drbd_rs_begin_io will lock that extent + * for application writes for the same time. For now, just throttle + * here, where the rest of the code expects the receiver to sleep for + * a while, anyways. + */ + + /* Throttle before drbd_rs_begin_io, as that locks out application IO; + * this defers syncer requests for some time, before letting at least + * on request through. The resync controller on the receiving side + * will adapt to the incoming rate accordingly. + * + * We cannot throttle here if remote is Primary/SyncTarget: + * 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)) + goto out_free_e; +submit_for_resync: + atomic_add(size >> 9, &mdev->rs_sect_ev); + +submit: inc_unacked(mdev); + spin_lock_irq(&mdev->req_lock); + list_add_tail(&e->w.list, &mdev->read_ee); + spin_unlock_irq(&mdev->req_lock); if (drbd_submit_ee(mdev, e, READ, fault_type) == 0) 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? */ + spin_lock_irq(&mdev->req_lock); + list_del(&e->w.list); + spin_unlock_irq(&mdev->req_lock); + /* no drbd_rs_complete_io(), we are dropping the connection anyways */ + out_free_e: - kfree(di); put_ldev(mdev); drbd_free_ee(mdev, e); return FALSE; @@ -2699,20 +2628,13 @@ static int cmp_after_sb(enum drbd_after_sb_p peer, enum drbd_after_sb_p self) return 1; } -static int receive_protocol(struct drbd_conf *mdev, struct p_header *h) +static int receive_protocol(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - struct p_protocol *p = (struct p_protocol *)h; - int header_size, data_size; + struct p_protocol *p = &mdev->data.rbuf.protocol; int p_proto, p_after_sb_0p, p_after_sb_1p, p_after_sb_2p; int p_want_lose, p_two_primaries, cf; char p_integrity_alg[SHARED_SECRET_MAX] = ""; - header_size = sizeof(*p) - sizeof(*h); - data_size = h->length - header_size; - - if (drbd_recv(mdev, h->payload, header_size) != header_size) - return FALSE; - p_proto = be32_to_cpu(p->protocol); p_after_sb_0p = be32_to_cpu(p->after_sb_0p); p_after_sb_1p = be32_to_cpu(p->after_sb_1p); @@ -2805,39 +2727,46 @@ struct crypto_hash *drbd_crypto_alloc_digest_safe(const struct drbd_conf *mdev, return tfm; } -static int receive_SyncParam(struct drbd_conf *mdev, struct p_header *h) +static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int packet_size) { int ok = TRUE; - struct p_rs_param_89 *p = (struct p_rs_param_89 *)h; + 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; struct crypto_hash *csums_tfm = NULL; const int apv = mdev->agreed_pro_version; + int *rs_plan_s = NULL; + int fifo_size = 0; exp_max_sz = apv <= 87 ? sizeof(struct p_rs_param) : apv == 88 ? sizeof(struct p_rs_param) + SHARED_SECRET_MAX - : /* 89 */ sizeof(struct p_rs_param_89); + : apv <= 94 ? sizeof(struct p_rs_param_89) + : /* apv >= 95 */ sizeof(struct p_rs_param_95); - if (h->length > exp_max_sz) { + if (packet_size > exp_max_sz) { dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n", - h->length, exp_max_sz); + packet_size, exp_max_sz); return FALSE; } if (apv <= 88) { - header_size = sizeof(struct p_rs_param) - sizeof(*h); - data_size = h->length - header_size; - } else /* apv >= 89 */ { - header_size = sizeof(struct p_rs_param_89) - sizeof(*h); - data_size = h->length - header_size; + header_size = sizeof(struct p_rs_param) - sizeof(struct p_header80); + data_size = packet_size - header_size; + } else if (apv <= 94) { + header_size = sizeof(struct p_rs_param_89) - sizeof(struct p_header80); + data_size = packet_size - header_size; + D_ASSERT(data_size == 0); + } else { + header_size = sizeof(struct p_rs_param_95) - sizeof(struct p_header80); + data_size = packet_size - header_size; D_ASSERT(data_size == 0); } /* initialize verify_alg and csums_alg */ memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX); - if (drbd_recv(mdev, h->payload, header_size) != header_size) + if (drbd_recv(mdev, &p->head.payload, header_size) != header_size) return FALSE; mdev->sync_conf.rate = be32_to_cpu(p->rate); @@ -2896,6 +2825,22 @@ static int receive_SyncParam(struct drbd_conf *mdev, struct p_header *h) } } + if (apv > 94) { + mdev->sync_conf.rate = be32_to_cpu(p->rate); + mdev->sync_conf.c_plan_ahead = be32_to_cpu(p->c_plan_ahead); + mdev->sync_conf.c_delay_target = be32_to_cpu(p->c_delay_target); + mdev->sync_conf.c_fill_target = be32_to_cpu(p->c_fill_target); + mdev->sync_conf.c_max_rate = be32_to_cpu(p->c_max_rate); + + fifo_size = (mdev->sync_conf.c_plan_ahead * 10 * SLEEP_TIME) / HZ; + if (fifo_size != mdev->rs_plan_s.size && fifo_size > 0) { + rs_plan_s = kzalloc(sizeof(int) * fifo_size, GFP_KERNEL); + if (!rs_plan_s) { + dev_err(DEV, "kmalloc of fifo_buffer failed"); + goto disconnect; + } + } + } spin_lock(&mdev->peer_seq_lock); /* lock against drbd_nl_syncer_conf() */ @@ -2913,6 +2858,12 @@ static int receive_SyncParam(struct drbd_conf *mdev, struct p_header *h) mdev->csums_tfm = csums_tfm; dev_info(DEV, "using csums-alg: \"%s\"\n", p->csums_alg); } + if (fifo_size != mdev->rs_plan_s.size) { + kfree(mdev->rs_plan_s.values); + mdev->rs_plan_s.values = rs_plan_s; + mdev->rs_plan_s.size = fifo_size; + mdev->rs_planed = 0; + } spin_unlock(&mdev->peer_seq_lock); } @@ -2946,19 +2897,15 @@ static void warn_if_differ_considerably(struct drbd_conf *mdev, (unsigned long long)a, (unsigned long long)b); } -static int receive_sizes(struct drbd_conf *mdev, struct p_header *h) +static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - struct p_sizes *p = (struct p_sizes *)h; + 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; - ERR_IF(h->length != (sizeof(*p)-sizeof(*h))) return FALSE; - if (drbd_recv(mdev, h->payload, h->length) != h->length) - return FALSE; - p_size = be64_to_cpu(p->d_size); p_usize = be64_to_cpu(p->u_size); @@ -2972,7 +2919,6 @@ static int receive_sizes(struct drbd_conf *mdev, struct p_header *h) * we still need to figure out whether we accept that. */ mdev->p_size = p_size; -#define min_not_zero(l, r) (l == 0) ? r : ((r == 0) ? l : min(l, r)) if (get_ldev(mdev)) { warn_if_differ_considerably(mdev, "lower level device sizes", p_size, drbd_get_max_capacity(mdev->ldev)); @@ -3029,6 +2975,8 @@ static int receive_sizes(struct drbd_conf *mdev, struct p_header *h) 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; @@ -3062,16 +3010,12 @@ static int receive_sizes(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int receive_uuids(struct drbd_conf *mdev, struct p_header *h) +static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - struct p_uuids *p = (struct p_uuids *)h; + struct p_uuids *p = &mdev->data.rbuf.uuids; u64 *p_uuid; int i; - ERR_IF(h->length != (sizeof(*p)-sizeof(*h))) return FALSE; - if (drbd_recv(mdev, h->payload, h->length) != h->length) - return FALSE; - p_uuid = kmalloc(sizeof(u64)*UI_EXTENDED_SIZE, GFP_NOIO); for (i = UI_CURRENT; i < UI_EXTENDED_SIZE; i++) @@ -3107,6 +3051,11 @@ static int receive_uuids(struct drbd_conf *mdev, struct p_header *h) drbd_md_sync(mdev); } 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]); } /* Before we test for the disk state, we should wait until an eventually @@ -3150,16 +3099,12 @@ static union drbd_state convert_state(union drbd_state ps) return ms; } -static int receive_req_state(struct drbd_conf *mdev, struct p_header *h) +static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - struct p_req_state *p = (struct p_req_state *)h; + struct p_req_state *p = &mdev->data.rbuf.req_state; union drbd_state mask, val; int rv; - ERR_IF(h->length != (sizeof(*p)-sizeof(*h))) return FALSE; - if (drbd_recv(mdev, h->payload, h->length) != h->length) - return FALSE; - mask.i = be32_to_cpu(p->mask); val.i = be32_to_cpu(p->val); @@ -3180,20 +3125,14 @@ static int receive_req_state(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int receive_state(struct drbd_conf *mdev, struct p_header *h) +static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - struct p_state *p = (struct p_state *)h; - enum drbd_conns nconn, oconn; - union drbd_state ns, peer_state; + struct p_state *p = &mdev->data.rbuf.state; + union drbd_state os, ns, peer_state; enum drbd_disk_state real_peer_disk; + enum chg_state_flags cs_flags; int rv; - ERR_IF(h->length != (sizeof(*p)-sizeof(*h))) - return FALSE; - - if (drbd_recv(mdev, h->payload, h->length) != h->length) - return FALSE; - peer_state.i = be32_to_cpu(p->state); real_peer_disk = peer_state.disk; @@ -3204,40 +3143,74 @@ static int receive_state(struct drbd_conf *mdev, struct p_header *h) spin_lock_irq(&mdev->req_lock); retry: - oconn = nconn = mdev->state.conn; + os = ns = mdev->state; spin_unlock_irq(&mdev->req_lock); - if (nconn == C_WF_REPORT_PARAMS) - nconn = C_CONNECTED; + /* peer says his disk is uptodate, while we think it is inconsistent, + * and this happens while we think we have a sync going on. */ + if (os.pdsk == D_INCONSISTENT && real_peer_disk == D_UP_TO_DATE && + os.conn > C_CONNECTED && os.disk == D_UP_TO_DATE) { + /* If we are (becoming) SyncSource, but peer is still in sync + * preparation, ignore its uptodate-ness to avoid flapping, it + * will change to inconsistent once the peer reaches active + * syncing states. + * It may have changed syncer-paused flags, however, so we + * cannot ignore this completely. */ + if (peer_state.conn > C_CONNECTED && + peer_state.conn < C_SYNC_SOURCE) + real_peer_disk = D_INCONSISTENT; + + /* if peer_state changes to connected at the same time, + * it explicitly notifies us that it finished resync. + * Maybe we should finish it up, too? */ + else if (os.conn >= C_SYNC_SOURCE && + peer_state.conn == C_CONNECTED) { + if (drbd_bm_total_weight(mdev) <= mdev->rs_failed) + drbd_resync_finished(mdev); + return TRUE; + } + } + + /* peer says his disk is inconsistent, while we think it is uptodate, + * and this happens while the peer still thinks we have a sync going on, + * but we think we are already done with the sync. + * We ignore this to avoid flapping pdsk. + * This should not happen, if the peer is a recent version of drbd. */ + if (os.pdsk == D_UP_TO_DATE && real_peer_disk == D_INCONSISTENT && + os.conn == C_CONNECTED && peer_state.conn > C_SYNC_SOURCE) + real_peer_disk = D_UP_TO_DATE; + + if (ns.conn == C_WF_REPORT_PARAMS) + ns.conn = C_CONNECTED; if (mdev->p_uuid && peer_state.disk >= D_NEGOTIATING && get_ldev_if_state(mdev, D_NEGOTIATING)) { int cr; /* consider resync */ /* if we established a new connection */ - cr = (oconn < C_CONNECTED); + cr = (os.conn < C_CONNECTED); /* if we had an established connection * and one of the nodes newly attaches a disk */ - cr |= (oconn == C_CONNECTED && + cr |= (os.conn == C_CONNECTED && (peer_state.disk == D_NEGOTIATING || - mdev->state.disk == D_NEGOTIATING)); + os.disk == D_NEGOTIATING)); /* if we have both been inconsistent, and the peer has been * forced to be UpToDate with --overwrite-data */ cr |= test_bit(CONSIDER_RESYNC, &mdev->flags); /* if we had been plain connected, and the admin requested to * start a sync by "invalidate" or "invalidate-remote" */ - cr |= (oconn == C_CONNECTED && + cr |= (os.conn == C_CONNECTED && (peer_state.conn >= C_STARTING_SYNC_S && peer_state.conn <= C_WF_BITMAP_T)); if (cr) - nconn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk); + ns.conn = drbd_sync_handshake(mdev, peer_state.role, real_peer_disk); put_ldev(mdev); - if (nconn == C_MASK) { - nconn = C_CONNECTED; + if (ns.conn == C_MASK) { + ns.conn = C_CONNECTED; if (mdev->state.disk == D_NEGOTIATING) { - drbd_force_state(mdev, NS(disk, D_DISKLESS)); + drbd_force_state(mdev, NS(disk, D_FAILED)); } else if (peer_state.disk == D_NEGOTIATING) { dev_err(DEV, "Disk attach process on the peer node was aborted.\n"); peer_state.disk = D_DISKLESS; @@ -3245,7 +3218,7 @@ static int receive_state(struct drbd_conf *mdev, struct p_header *h) } else { if (test_and_clear_bit(CONN_DRY_RUN, &mdev->flags)) return FALSE; - D_ASSERT(oconn == C_WF_REPORT_PARAMS); + D_ASSERT(os.conn == C_WF_REPORT_PARAMS); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); return FALSE; } @@ -3253,18 +3226,28 @@ static int receive_state(struct drbd_conf *mdev, struct p_header *h) } spin_lock_irq(&mdev->req_lock); - if (mdev->state.conn != oconn) + if (mdev->state.i != os.i) goto retry; clear_bit(CONSIDER_RESYNC, &mdev->flags); - ns.i = mdev->state.i; - ns.conn = nconn; ns.peer = peer_state.role; ns.pdsk = real_peer_disk; ns.peer_isp = (peer_state.aftr_isp | peer_state.user_isp); - if ((nconn == C_CONNECTED || nconn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING) + if ((ns.conn == C_CONNECTED || ns.conn == C_WF_BITMAP_S) && ns.disk == D_NEGOTIATING) ns.disk = mdev->new_state_tmp.disk; - - rv = _drbd_set_state(mdev, ns, CS_VERBOSE | CS_HARD, NULL); + cs_flags = CS_VERBOSE + (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED ? 0 : CS_HARD); + if (ns.pdsk == D_CONSISTENT && is_susp(ns) && ns.conn == C_CONNECTED && os.conn < C_CONNECTED && + test_bit(NEW_CUR_UUID, &mdev->flags)) { + /* Do not allow tl_restart(resend) for a rebooted peer. We can only allow this + for temporal network outages! */ + spin_unlock_irq(&mdev->req_lock); + dev_err(DEV, "Aborting Connect, can not thaw IO with an only Consistent peer\n"); + tl_clear(mdev); + 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; + } + rv = _drbd_set_state(mdev, ns, cs_flags, NULL); ns = mdev->state; spin_unlock_irq(&mdev->req_lock); @@ -3273,8 +3256,8 @@ static int receive_state(struct drbd_conf *mdev, struct p_header *h) return FALSE; } - if (oconn > C_WF_REPORT_PARAMS) { - if (nconn > C_CONNECTED && peer_state.conn <= C_CONNECTED && + if (os.conn > C_WF_REPORT_PARAMS) { + if (ns.conn > C_CONNECTED && peer_state.conn <= C_CONNECTED && peer_state.disk != D_NEGOTIATING ) { /* we want resync, peer has not yet decided to sync... */ /* Nowadays only used when forcing a node into primary role and @@ -3291,9 +3274,9 @@ static int receive_state(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int receive_sync_uuid(struct drbd_conf *mdev, struct p_header *h) +static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - struct p_rs_uuid *p = (struct p_rs_uuid *)h; + struct p_rs_uuid *p = &mdev->data.rbuf.rs_uuid; wait_event(mdev->misc_wait, mdev->state.conn == C_WF_SYNC_UUID || @@ -3302,10 +3285,6 @@ static int receive_sync_uuid(struct drbd_conf *mdev, struct p_header *h) /* D_ASSERT( mdev->state.conn == C_WF_SYNC_UUID ); */ - ERR_IF(h->length != (sizeof(*p)-sizeof(*h))) return FALSE; - if (drbd_recv(mdev, h->payload, h->length) != h->length) - return FALSE; - /* Here the _drbd_uuid_ functions are right, current should _not_ be rotated into the history */ if (get_ldev_if_state(mdev, D_NEGOTIATING)) { @@ -3324,14 +3303,14 @@ static int receive_sync_uuid(struct drbd_conf *mdev, struct p_header *h) enum receive_bitmap_ret { OK, DONE, FAILED }; static enum receive_bitmap_ret -receive_bitmap_plain(struct drbd_conf *mdev, struct p_header *h, - unsigned long *buffer, struct bm_xfer_ctx *c) +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); - if (want != h->length) { - dev_err(DEV, "%s:want (%u) != h->length (%u)\n", __func__, want, h->length); + if (want != data_size) { + dev_err(DEV, "%s:want (%u) != data_size (%u)\n", __func__, want, data_size); return FAILED; } if (want == 0) @@ -3360,7 +3339,7 @@ recv_bm_rle_bits(struct drbd_conf *mdev, u64 tmp; unsigned long s = c->bit_offset; unsigned long e; - int len = p->head.length - (sizeof(*p) - sizeof(p->head)); + int len = be16_to_cpu(p->head.length) - (sizeof(*p) - sizeof(p->head)); int toggle = DCBP_get_start(p); int have; int bits; @@ -3429,7 +3408,7 @@ void INFO_bm_xfer_stats(struct drbd_conf *mdev, const char *direction, struct bm_xfer_ctx *c) { /* what would it take to transfer it "plaintext" */ - unsigned plain = sizeof(struct p_header) * + unsigned plain = sizeof(struct p_header80) * ((c->bm_words+BM_PACKET_WORDS-1)/BM_PACKET_WORDS+1) + c->bm_words * sizeof(long); unsigned total = c->bytes[0] + c->bytes[1]; @@ -3467,12 +3446,13 @@ void INFO_bm_xfer_stats(struct drbd_conf *mdev, in order to be agnostic to the 32 vs 64 bits issue. returns 0 on failure, 1 if we successfully received it. */ -static int receive_bitmap(struct drbd_conf *mdev, struct p_header *h) +static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { struct bm_xfer_ctx c; void *buffer; enum receive_bitmap_ret ret; int ok = FALSE; + struct p_header80 *h = &mdev->data.rbuf.header.h80; wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); @@ -3492,39 +3472,39 @@ static int receive_bitmap(struct drbd_conf *mdev, struct p_header *h) }; do { - if (h->command == P_BITMAP) { - ret = receive_bitmap_plain(mdev, h, buffer, &c); - } else if (h->command == P_COMPRESSED_BITMAP) { + if (cmd == P_BITMAP) { + ret = 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! */ struct p_compressed_bm *p; - if (h->length > BM_PACKET_PAYLOAD_BYTES) { + if (data_size > BM_PACKET_PAYLOAD_BYTES) { dev_err(DEV, "ReportCBitmap packet too large\n"); goto out; } /* use the page buff */ p = buffer; memcpy(p, h, sizeof(*h)); - if (drbd_recv(mdev, p->head.payload, h->length) != h->length) + if (drbd_recv(mdev, p->head.payload, data_size) != data_size) goto out; - if (p->head.length <= (sizeof(*p) - sizeof(p->head))) { - dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", p->head.length); + if (data_size <= (sizeof(*p) - sizeof(p->head))) { + dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", data_size); return FAILED; } ret = decode_bitmap_c(mdev, p, &c); } else { - dev_warn(DEV, "receive_bitmap: h->command neither ReportBitMap nor ReportCBitMap (is 0x%x)", h->command); + dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", cmd); goto out; } - c.packets[h->command == P_BITMAP]++; - c.bytes[h->command == P_BITMAP] += sizeof(struct p_header) + h->length; + c.packets[cmd == P_BITMAP]++; + c.bytes[cmd == P_BITMAP] += sizeof(struct p_header80) + data_size; if (ret != OK) break; - if (!drbd_recv_header(mdev, h)) + if (!drbd_recv_header(mdev, &cmd, &data_size)) goto out; } while (ret == OK); if (ret == FAILED) @@ -3555,17 +3535,16 @@ static int receive_bitmap(struct drbd_conf *mdev, struct p_header *h) return ok; } -static int receive_skip_(struct drbd_conf *mdev, struct p_header *h, int silent) +static int receive_skip(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { /* TODO zero copy sink :) */ static char sink[128]; int size, want, r; - if (!silent) - dev_warn(DEV, "skipping unknown optional packet type %d, l: %d!\n", - h->command, h->length); + dev_warn(DEV, "skipping unknown optional packet type %d, l: %d!\n", + cmd, data_size); - size = h->length; + size = data_size; while (size > 0) { want = min_t(int, size, sizeof(sink)); r = drbd_recv(mdev, sink, want); @@ -3575,17 +3554,7 @@ static int receive_skip_(struct drbd_conf *mdev, struct p_header *h, int silent) return size == 0; } -static int receive_skip(struct drbd_conf *mdev, struct p_header *h) -{ - return receive_skip_(mdev, h, 0); -} - -static int receive_skip_silent(struct drbd_conf *mdev, struct p_header *h) -{ - return receive_skip_(mdev, h, 1); -} - -static int receive_UnplugRemote(struct drbd_conf *mdev, struct p_header *h) +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); @@ -3597,108 +3566,96 @@ static int receive_UnplugRemote(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -typedef int (*drbd_cmd_handler_f)(struct drbd_conf *, struct p_header *); - -static drbd_cmd_handler_f drbd_default_handler[] = { - [P_DATA] = receive_Data, - [P_DATA_REPLY] = receive_DataReply, - [P_RS_DATA_REPLY] = receive_RSDataReply, - [P_BARRIER] = receive_Barrier, - [P_BITMAP] = receive_bitmap, - [P_COMPRESSED_BITMAP] = receive_bitmap, - [P_UNPLUG_REMOTE] = receive_UnplugRemote, - [P_DATA_REQUEST] = receive_DataRequest, - [P_RS_DATA_REQUEST] = receive_DataRequest, - [P_SYNC_PARAM] = receive_SyncParam, - [P_SYNC_PARAM89] = receive_SyncParam, - [P_PROTOCOL] = receive_protocol, - [P_UUIDS] = receive_uuids, - [P_SIZES] = receive_sizes, - [P_STATE] = receive_state, - [P_STATE_CHG_REQ] = receive_req_state, - [P_SYNC_UUID] = receive_sync_uuid, - [P_OV_REQUEST] = receive_DataRequest, - [P_OV_REPLY] = receive_DataRequest, - [P_CSUM_RS_REQUEST] = receive_DataRequest, - [P_DELAY_PROBE] = receive_skip_silent, +typedef int (*drbd_cmd_handler_f)(struct drbd_conf *, enum drbd_packets cmd, unsigned int to_receive); + +struct data_cmd { + int expect_payload; + size_t pkt_size; + drbd_cmd_handler_f function; +}; + +static struct data_cmd drbd_cmd_handler[] = { + [P_DATA] = { 1, sizeof(struct p_data), receive_Data }, + [P_DATA_REPLY] = { 1, sizeof(struct p_data), receive_DataReply }, + [P_RS_DATA_REPLY] = { 1, sizeof(struct p_data), receive_RSDataReply } , + [P_BARRIER] = { 0, sizeof(struct p_barrier), receive_Barrier } , + [P_BITMAP] = { 1, sizeof(struct p_header80), receive_bitmap } , + [P_COMPRESSED_BITMAP] = { 1, sizeof(struct p_header80), receive_bitmap } , + [P_UNPLUG_REMOTE] = { 0, sizeof(struct p_header80), receive_UnplugRemote }, + [P_DATA_REQUEST] = { 0, sizeof(struct p_block_req), receive_DataRequest }, + [P_RS_DATA_REQUEST] = { 0, sizeof(struct p_block_req), receive_DataRequest }, + [P_SYNC_PARAM] = { 1, sizeof(struct p_header80), receive_SyncParam }, + [P_SYNC_PARAM89] = { 1, sizeof(struct p_header80), receive_SyncParam }, + [P_PROTOCOL] = { 1, sizeof(struct p_protocol), receive_protocol }, + [P_UUIDS] = { 0, sizeof(struct p_uuids), receive_uuids }, + [P_SIZES] = { 0, sizeof(struct p_sizes), receive_sizes }, + [P_STATE] = { 0, sizeof(struct p_state), receive_state }, + [P_STATE_CHG_REQ] = { 0, sizeof(struct p_req_state), receive_req_state }, + [P_SYNC_UUID] = { 0, sizeof(struct p_rs_uuid), receive_sync_uuid }, + [P_OV_REQUEST] = { 0, sizeof(struct p_block_req), receive_DataRequest }, + [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 }, /* anything missing from this table is in * the asender_tbl, see get_asender_cmd */ - [P_MAX_CMD] = NULL, + [P_MAX_CMD] = { 0, 0, NULL }, }; -static drbd_cmd_handler_f *drbd_cmd_handler = drbd_default_handler; -static drbd_cmd_handler_f *drbd_opt_cmd_handler; +/* All handler functions that expect a sub-header get that sub-heder in + mdev->data.rbuf.header.head.payload. + + Usually in mdev->data.rbuf.header.head the callback can find the usual + p_header, but they may not rely on that. Since there is also p_header95 ! + */ static void drbdd(struct drbd_conf *mdev) { - drbd_cmd_handler_f handler; - struct p_header *header = &mdev->data.rbuf.header; + union p_header *header = &mdev->data.rbuf.header; + unsigned int packet_size; + enum drbd_packets cmd; + size_t shs; /* sub header size */ + int rv; while (get_t_state(&mdev->receiver) == Running) { drbd_thread_current_set_cpu(mdev); - if (!drbd_recv_header(mdev, header)) { - drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); - break; - } + if (!drbd_recv_header(mdev, &cmd, &packet_size)) + goto err_out; - if (header->command < P_MAX_CMD) - handler = drbd_cmd_handler[header->command]; - else if (P_MAY_IGNORE < header->command - && header->command < P_MAX_OPT_CMD) - handler = drbd_opt_cmd_handler[header->command-P_MAY_IGNORE]; - else if (header->command > P_MAX_OPT_CMD) - handler = receive_skip; - else - handler = NULL; + if (unlikely(cmd >= P_MAX_CMD || !drbd_cmd_handler[cmd].function)) { + dev_err(DEV, "unknown packet type %d, l: %d!\n", cmd, packet_size); + goto err_out; + } - if (unlikely(!handler)) { - dev_err(DEV, "unknown packet type %d, l: %d!\n", - header->command, header->length); - drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); - break; + shs = drbd_cmd_handler[cmd].pkt_size - sizeof(union p_header); + if (packet_size - shs > 0 && !drbd_cmd_handler[cmd].expect_payload) { + dev_err(DEV, "No payload expected %s l:%d\n", cmdname(cmd), packet_size); + goto err_out; } - if (unlikely(!handler(mdev, header))) { - dev_err(DEV, "error receiving %s, l: %d!\n", - cmdname(header->command), header->length); - drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); - break; + + 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); + goto err_out; + } } - } -} -static void drbd_fail_pending_reads(struct drbd_conf *mdev) -{ - struct hlist_head *slot; - struct hlist_node *pos; - struct hlist_node *tmp; - struct drbd_request *req; - int i; + rv = drbd_cmd_handler[cmd].function(mdev, cmd, packet_size - shs); - /* - * Application READ requests - */ - spin_lock_irq(&mdev->req_lock); - for (i = 0; i < APP_R_HSIZE; i++) { - slot = mdev->app_reads_hash+i; - hlist_for_each_entry_safe(req, pos, tmp, slot, colision) { - /* it may (but should not any longer!) - * be on the work queue; if that assert triggers, - * we need to also grab the - * spin_lock_irq(&mdev->data.work.q_lock); - * and list_del_init here. */ - D_ASSERT(list_empty(&req->w.list)); - /* It would be nice to complete outside of spinlock. - * But this is easier for now. */ - _req_mod(req, connection_lost_while_pending); + if (unlikely(!rv)) { + dev_err(DEV, "error receiving %s, l: %d!\n", + cmdname(cmd), packet_size); + goto err_out; } } - for (i = 0; i < APP_R_HSIZE; i++) - if (!hlist_empty(mdev->app_reads_hash+i)) - dev_warn(DEV, "ASSERT FAILED: app_reads_hash[%d].first: " - "%p, should be NULL\n", i, mdev->app_reads_hash[i].first); - memset(mdev->app_reads_hash, 0, APP_R_HSIZE*sizeof(void *)); - spin_unlock_irq(&mdev->req_lock); + if (0) { + err_out: + drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); + } + /* If we leave here, we probably want to update at least the + * "Connected" indicator on stable storage. Do so explicitly here. */ + drbd_md_sync(mdev); } void drbd_flush_workqueue(struct drbd_conf *mdev) @@ -3711,6 +3668,36 @@ void drbd_flush_workqueue(struct drbd_conf *mdev) wait_for_completion(&barr.done); } +void drbd_free_tl_hash(struct drbd_conf *mdev) +{ + struct hlist_head *h; + + spin_lock_irq(&mdev->req_lock); + + if (!mdev->tl_hash || mdev->state.conn != C_STANDALONE) { + spin_unlock_irq(&mdev->req_lock); + return; + } + /* paranoia code */ + for (h = mdev->ee_hash; h < mdev->ee_hash + mdev->ee_hash_s; h++) + if (h->first) + dev_err(DEV, "ASSERT FAILED ee_hash[%u].first == %p, expected NULL\n", + (int)(h - mdev->ee_hash), h->first); + kfree(mdev->ee_hash); + mdev->ee_hash = NULL; + mdev->ee_hash_s = 0; + + /* paranoia code */ + for (h = mdev->tl_hash; h < mdev->tl_hash + mdev->tl_hash_s; h++) + if (h->first) + dev_err(DEV, "ASSERT FAILED tl_hash[%u] == %p, expected NULL\n", + (int)(h - mdev->tl_hash), h->first); + kfree(mdev->tl_hash); + mdev->tl_hash = NULL; + mdev->tl_hash_s = 0; + spin_unlock_irq(&mdev->req_lock); +} + static void drbd_disconnect(struct drbd_conf *mdev) { enum drbd_fencing_p fp; @@ -3728,6 +3715,7 @@ static void drbd_disconnect(struct drbd_conf *mdev) drbd_thread_stop(&mdev->asender); drbd_free_sock(mdev); + /* wait for current activity to cease. */ spin_lock_irq(&mdev->req_lock); _drbd_wait_ee_list_empty(mdev, &mdev->active_ee); _drbd_wait_ee_list_empty(mdev, &mdev->sync_ee); @@ -3752,7 +3740,6 @@ static void drbd_disconnect(struct drbd_conf *mdev) /* make sure syncer is stopped and w_resume_next_sg queued */ del_timer_sync(&mdev->resync_timer); - set_bit(STOP_SYNC_TIMER, &mdev->flags); resync_timer_fn((unsigned long)mdev); /* wait for all w_e_end_data_req, w_e_end_rsdata_req, w_send_barrier, @@ -3767,11 +3754,9 @@ static void drbd_disconnect(struct drbd_conf *mdev) kfree(mdev->p_uuid); mdev->p_uuid = NULL; - if (!mdev->state.susp) + if (!is_susp(mdev->state)) tl_clear(mdev); - drbd_fail_pending_reads(mdev); - dev_info(DEV, "Connection closed\n"); drbd_md_sync(mdev); @@ -3782,12 +3767,8 @@ static void drbd_disconnect(struct drbd_conf *mdev) put_ldev(mdev); } - if (mdev->state.role == R_PRIMARY) { - if (fp >= FP_RESOURCE && mdev->state.pdsk >= D_UNKNOWN) { - enum drbd_disk_state nps = drbd_try_outdate_peer(mdev); - drbd_request_state(mdev, NS(pdsk, nps)); - } - } + if (mdev->state.role == R_PRIMARY && fp >= FP_RESOURCE && mdev->state.pdsk >= D_UNKNOWN) + drbd_try_outdate_peer_async(mdev); spin_lock_irq(&mdev->req_lock); os = mdev->state; @@ -3800,32 +3781,14 @@ static void drbd_disconnect(struct drbd_conf *mdev) spin_unlock_irq(&mdev->req_lock); if (os.conn == C_DISCONNECTING) { - struct hlist_head *h; - wait_event(mdev->misc_wait, atomic_read(&mdev->net_cnt) == 0); - - /* 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) == 0); + wait_event(mdev->net_cnt_wait, atomic_read(&mdev->net_cnt) == 0); - spin_lock_irq(&mdev->req_lock); - /* paranoia code */ - for (h = mdev->ee_hash; h < mdev->ee_hash + mdev->ee_hash_s; h++) - if (h->first) - dev_err(DEV, "ASSERT FAILED ee_hash[%u].first == %p, expected NULL\n", - (int)(h - mdev->ee_hash), h->first); - kfree(mdev->ee_hash); - mdev->ee_hash = NULL; - mdev->ee_hash_s = 0; - - /* paranoia code */ - for (h = mdev->tl_hash; h < mdev->tl_hash + mdev->tl_hash_s; h++) - if (h->first) - dev_err(DEV, "ASSERT FAILED tl_hash[%u] == %p, expected NULL\n", - (int)(h - mdev->tl_hash), h->first); - kfree(mdev->tl_hash); - mdev->tl_hash = NULL; - mdev->tl_hash_s = 0; - spin_unlock_irq(&mdev->req_lock); + 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; @@ -3845,6 +3808,9 @@ static void drbd_disconnect(struct drbd_conf *mdev) i = drbd_release_ee(mdev, &mdev->net_ee); if (i) dev_info(DEV, "net_ee not empty, killed %u entries\n", i); + i = atomic_read(&mdev->pp_in_use_by_net); + if (i) + dev_info(DEV, "pp_in_use_by_net = %d, expected 0\n", i); i = atomic_read(&mdev->pp_in_use); if (i) dev_info(DEV, "pp_in_use = %d, expected 0\n", i); @@ -3888,7 +3854,7 @@ static int drbd_send_handshake(struct drbd_conf *mdev) p->protocol_min = cpu_to_be32(PRO_VERSION_MIN); p->protocol_max = cpu_to_be32(PRO_VERSION_MAX); ok = _drbd_send_cmd( mdev, mdev->data.socket, P_HAND_SHAKE, - (struct p_header *)p, sizeof(*p), 0 ); + (struct p_header80 *)p, sizeof(*p), 0 ); mutex_unlock(&mdev->data.mutex); return ok; } @@ -3904,27 +3870,28 @@ static int drbd_do_handshake(struct drbd_conf *mdev) { /* ASSERT current == mdev->receiver ... */ struct p_handshake *p = &mdev->data.rbuf.handshake; - const int expect = sizeof(struct p_handshake) - -sizeof(struct p_header); + const int expect = sizeof(struct p_handshake) - sizeof(struct p_header80); + unsigned int length; + enum drbd_packets cmd; int rv; rv = drbd_send_handshake(mdev); if (!rv) return 0; - rv = drbd_recv_header(mdev, &p->head); + rv = drbd_recv_header(mdev, &cmd, &length); if (!rv) return 0; - if (p->head.command != P_HAND_SHAKE) { + if (cmd != P_HAND_SHAKE) { dev_err(DEV, "expected HandShake packet, received: %s (0x%04x)\n", - cmdname(p->head.command), p->head.command); + cmdname(cmd), cmd); return -1; } - if (p->head.length != expect) { + if (length != expect) { dev_err(DEV, "expected HandShake length: %u, received: %u\n", - expect, p->head.length); + expect, length); return -1; } @@ -3982,10 +3949,11 @@ static int drbd_do_auth(struct drbd_conf *mdev) char *response = NULL; char *right_response = NULL; char *peers_ch = NULL; - struct p_header p; unsigned int key_len = strlen(mdev->net_conf->shared_secret); unsigned int resp_size; struct hash_desc desc; + enum drbd_packets cmd; + unsigned int length; int rv; desc.tfm = mdev->cram_hmac_tfm; @@ -4005,33 +3973,33 @@ static int drbd_do_auth(struct drbd_conf *mdev) if (!rv) goto fail; - rv = drbd_recv_header(mdev, &p); + rv = drbd_recv_header(mdev, &cmd, &length); if (!rv) goto fail; - if (p.command != P_AUTH_CHALLENGE) { + if (cmd != P_AUTH_CHALLENGE) { dev_err(DEV, "expected AuthChallenge packet, received: %s (0x%04x)\n", - cmdname(p.command), p.command); + cmdname(cmd), cmd); rv = 0; goto fail; } - if (p.length > CHALLENGE_LEN*2) { + if (length > CHALLENGE_LEN * 2) { dev_err(DEV, "expected AuthChallenge payload too big.\n"); rv = -1; goto fail; } - peers_ch = kmalloc(p.length, GFP_NOIO); + peers_ch = kmalloc(length, GFP_NOIO); if (peers_ch == NULL) { dev_err(DEV, "kmalloc of peers_ch failed\n"); rv = -1; goto fail; } - rv = drbd_recv(mdev, peers_ch, p.length); + rv = drbd_recv(mdev, peers_ch, length); - if (rv != p.length) { + if (rv != length) { dev_err(DEV, "short read AuthChallenge: l=%u\n", rv); rv = 0; goto fail; @@ -4046,7 +4014,7 @@ static int drbd_do_auth(struct drbd_conf *mdev) } sg_init_table(&sg, 1); - sg_set_buf(&sg, peers_ch, p.length); + sg_set_buf(&sg, peers_ch, length); rv = crypto_hash_digest(&desc, &sg, sg.length, response); if (rv) { @@ -4059,18 +4027,18 @@ static int drbd_do_auth(struct drbd_conf *mdev) if (!rv) goto fail; - rv = drbd_recv_header(mdev, &p); + rv = drbd_recv_header(mdev, &cmd, &length); if (!rv) goto fail; - if (p.command != P_AUTH_RESPONSE) { + if (cmd != P_AUTH_RESPONSE) { dev_err(DEV, "expected AuthResponse packet, received: %s (0x%04x)\n", - cmdname(p.command), p.command); + cmdname(cmd), cmd); rv = 0; goto fail; } - if (p.length != resp_size) { + if (length != resp_size) { dev_err(DEV, "expected AuthResponse payload of wrong size\n"); rv = 0; goto fail; @@ -4155,7 +4123,7 @@ int drbdd_init(struct drbd_thread *thi) /* ********* acknowledge sender ******** */ -static int got_RqSReply(struct drbd_conf *mdev, struct p_header *h) +static int got_RqSReply(struct drbd_conf *mdev, struct p_header80 *h) { struct p_req_state_reply *p = (struct p_req_state_reply *)h; @@ -4173,13 +4141,13 @@ static int got_RqSReply(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int got_Ping(struct drbd_conf *mdev, struct p_header *h) +static int got_Ping(struct drbd_conf *mdev, struct p_header80 *h) { return drbd_send_ping_ack(mdev); } -static int got_PingAck(struct drbd_conf *mdev, struct p_header *h) +static int got_PingAck(struct drbd_conf *mdev, struct p_header80 *h) { /* restore idle timeout */ mdev->meta.socket->sk->sk_rcvtimeo = mdev->net_conf->ping_int*HZ; @@ -4189,7 +4157,7 @@ static int got_PingAck(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int got_IsInSync(struct drbd_conf *mdev, struct p_header *h) +static int got_IsInSync(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); @@ -4199,11 +4167,15 @@ static int got_IsInSync(struct drbd_conf *mdev, struct p_header *h) update_peer_seq(mdev, be32_to_cpu(p->seq_num)); - drbd_rs_complete_io(mdev, sector); - drbd_set_in_sync(mdev, sector, blksize); - /* rs_same_csums is supposed to count in units of BM_BLOCK_SIZE */ - mdev->rs_same_csum += (blksize >> BM_BLOCK_SHIFT); + if (get_ldev(mdev)) { + drbd_rs_complete_io(mdev, sector); + drbd_set_in_sync(mdev, sector, blksize); + /* rs_same_csums is supposed to count in units of BM_BLOCK_SIZE */ + mdev->rs_same_csum += (blksize >> BM_BLOCK_SHIFT); + put_ldev(mdev); + } dec_rs_pending(mdev); + atomic_add(blksize >> 9, &mdev->rs_sect_in); return TRUE; } @@ -4259,7 +4231,7 @@ static int validate_req_change_req_state(struct drbd_conf *mdev, return TRUE; } -static int got_BlockAck(struct drbd_conf *mdev, struct p_header *h) +static int got_BlockAck(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); @@ -4299,7 +4271,7 @@ static int got_BlockAck(struct drbd_conf *mdev, struct p_header *h) _ack_id_to_req, __func__ , what); } -static int got_NegAck(struct drbd_conf *mdev, struct p_header *h) +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); @@ -4319,7 +4291,7 @@ static int got_NegAck(struct drbd_conf *mdev, struct p_header *h) _ack_id_to_req, __func__ , neg_acked); } -static int got_NegDReply(struct drbd_conf *mdev, struct p_header *h) +static int got_NegDReply(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); @@ -4332,7 +4304,7 @@ static int got_NegDReply(struct drbd_conf *mdev, struct p_header *h) _ar_id_to_req, __func__ , neg_acked); } -static int got_NegRSDReply(struct drbd_conf *mdev, struct p_header *h) +static int got_NegRSDReply(struct drbd_conf *mdev, struct p_header80 *h) { sector_t sector; int size; @@ -4354,7 +4326,7 @@ static int got_NegRSDReply(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int got_BarrierAck(struct drbd_conf *mdev, struct p_header *h) +static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) { struct p_barrier_ack *p = (struct p_barrier_ack *)h; @@ -4363,7 +4335,7 @@ static int got_BarrierAck(struct drbd_conf *mdev, struct p_header *h) return TRUE; } -static int got_OVResult(struct drbd_conf *mdev, struct p_header *h) +static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) { struct p_block_ack *p = (struct p_block_ack *)h; struct drbd_work *w; @@ -4380,6 +4352,9 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header *h) else ov_oos_print(mdev); + if (!get_ldev(mdev)) + return TRUE; + drbd_rs_complete_io(mdev, sector); dec_rs_pending(mdev); @@ -4394,18 +4369,18 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header *h) drbd_resync_finished(mdev); } } + put_ldev(mdev); return TRUE; } -static int got_something_to_ignore_m(struct drbd_conf *mdev, struct p_header *h) +static int got_skip(struct drbd_conf *mdev, struct p_header80 *h) { - /* IGNORE */ return TRUE; } struct asender_cmd { size_t pkt_size; - int (*process)(struct drbd_conf *mdev, struct p_header *h); + int (*process)(struct drbd_conf *mdev, struct p_header80 *h); }; static struct asender_cmd *get_asender_cmd(int cmd) @@ -4414,8 +4389,8 @@ static struct asender_cmd *get_asender_cmd(int cmd) /* anything missing from this table is in * the drbd_cmd_handler (drbd_default_handler) table, * see the beginning of drbdd() */ - [P_PING] = { sizeof(struct p_header), got_Ping }, - [P_PING_ACK] = { sizeof(struct p_header), got_PingAck }, + [P_PING] = { sizeof(struct p_header80), got_Ping }, + [P_PING_ACK] = { sizeof(struct p_header80), got_PingAck }, [P_RECV_ACK] = { sizeof(struct p_block_ack), got_BlockAck }, [P_WRITE_ACK] = { sizeof(struct p_block_ack), got_BlockAck }, [P_RS_WRITE_ACK] = { sizeof(struct p_block_ack), got_BlockAck }, @@ -4427,7 +4402,7 @@ static struct asender_cmd *get_asender_cmd(int cmd) [P_BARRIER_ACK] = { sizeof(struct p_barrier_ack), got_BarrierAck }, [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_probe), got_something_to_ignore_m }, + [P_DELAY_PROBE] = { sizeof(struct p_delay_probe93), got_skip }, [P_MAX_CMD] = { 0, NULL }, }; if (cmd > P_MAX_CMD || asender_tbl[cmd].process == NULL) @@ -4438,13 +4413,13 @@ static struct asender_cmd *get_asender_cmd(int cmd) int drbd_asender(struct drbd_thread *thi) { struct drbd_conf *mdev = thi->mdev; - struct p_header *h = &mdev->meta.rbuf.header; + struct p_header80 *h = &mdev->meta.rbuf.header.h80; struct asender_cmd *cmd = NULL; int rv, len; void *buf = h; int received = 0; - int expect = sizeof(struct p_header); + int expect = sizeof(struct p_header80); int empty; sprintf(current->comm, "drbd%d_asender", mdev_to_minor(mdev)); @@ -4468,10 +4443,8 @@ int drbd_asender(struct drbd_thread *thi) while (1) { clear_bit(SIGNAL_ASENDER, &mdev->flags); flush_signals(current); - if (!drbd_process_done_ee(mdev)) { - dev_err(DEV, "process_done_ee() = NOT_OK\n"); + if (!drbd_process_done_ee(mdev)) goto reconnect; - } /* to avoid race with newly queued ACKs */ set_bit(SIGNAL_ASENDER, &mdev->flags); spin_lock_irq(&mdev->req_lock); @@ -4530,21 +4503,23 @@ int drbd_asender(struct drbd_thread *thi) if (received == expect && cmd == NULL) { if (unlikely(h->magic != BE_DRBD_MAGIC)) { - dev_err(DEV, "magic?? on meta m: 0x%lx c: %d l: %d\n", - (long)be32_to_cpu(h->magic), - h->command, h->length); + dev_err(DEV, "magic?? on meta m: 0x%08x c: %d l: %d\n", + be32_to_cpu(h->magic), + be16_to_cpu(h->command), + be16_to_cpu(h->length)); goto reconnect; } cmd = get_asender_cmd(be16_to_cpu(h->command)); len = be16_to_cpu(h->length); if (unlikely(cmd == NULL)) { - dev_err(DEV, "unknown command?? on meta m: 0x%lx c: %d l: %d\n", - (long)be32_to_cpu(h->magic), - h->command, h->length); + dev_err(DEV, "unknown command?? on meta m: 0x%08x c: %d l: %d\n", + be32_to_cpu(h->magic), + be16_to_cpu(h->command), + be16_to_cpu(h->length)); goto disconnect; } expect = cmd->pkt_size; - ERR_IF(len != expect-sizeof(struct p_header)) + ERR_IF(len != expect-sizeof(struct p_header80)) goto reconnect; } if (received == expect) { @@ -4554,7 +4529,7 @@ int drbd_asender(struct drbd_thread *thi) buf = h; received = 0; - expect = sizeof(struct p_header); + expect = sizeof(struct p_header80); cmd = NULL; } } @@ -4562,10 +4537,12 @@ int drbd_asender(struct drbd_thread *thi) if (0) { reconnect: drbd_force_state(mdev, NS(conn, C_NETWORK_FAILURE)); + drbd_md_sync(mdev); } if (0) { disconnect: drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); + drbd_md_sync(mdev); } clear_bit(SIGNAL_ASENDER, &mdev->flags); diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index f761d98..11a75d3 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -59,17 +59,19 @@ static void _drbd_end_io_acct(struct drbd_conf *mdev, struct drbd_request *req) static void _req_is_done(struct drbd_conf *mdev, struct drbd_request *req, const int rw) { const unsigned long s = req->rq_state; + + /* remove it from the transfer log. + * well, only if it had been there in the first + * place... if it had not (local only or conflicting + * and never sent), it should still be "empty" as + * initialized in drbd_req_new(), so we can list_del() it + * here unconditionally */ + list_del(&req->tl_requests); + /* if it was a write, we may have to set the corresponding * bit(s) out-of-sync first. If it had a local part, we need to * release the reference to the activity log. */ if (rw == WRITE) { - /* remove it from the transfer log. - * well, only if it had been there in the first - * place... if it had not (local only or conflicting - * and never sent), it should still be "empty" as - * initialized in drbd_req_new(), so we can list_del() it - * here unconditionally */ - list_del(&req->tl_requests); /* Set out-of-sync unless both OK flags are set * (local only or remote failed). * Other places where we set out-of-sync: @@ -92,7 +94,8 @@ static void _req_is_done(struct drbd_conf *mdev, struct drbd_request *req, const */ if (s & RQ_LOCAL_MASK) { if (get_ldev_if_state(mdev, D_FAILED)) { - drbd_al_complete_io(mdev, req->sector); + if (s & RQ_IN_ACT_LOG) + drbd_al_complete_io(mdev, req->sector); put_ldev(mdev); } else if (__ratelimit(&drbd_ratelimit_state)) { dev_warn(DEV, "Should have called drbd_al_complete_io(, %llu), " @@ -255,7 +258,7 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m) if (!hlist_unhashed(&req->colision)) hlist_del(&req->colision); else - D_ASSERT((s & RQ_NET_MASK) == 0); + D_ASSERT((s & (RQ_NET_MASK & ~RQ_NET_DONE)) == 0); /* for writes we need to do some extra housekeeping */ if (rw == WRITE) @@ -280,6 +283,14 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m) * protocol A or B, barrier ack still pending... */ } +static void _req_may_be_done_not_susp(struct drbd_request *req, struct bio_and_error *m) +{ + struct drbd_conf *mdev = req->mdev; + + if (!is_susp(mdev->state)) + _req_may_be_done(req, m); +} + /* * checks whether there was an overlapping request * or ee already registered. @@ -380,10 +391,11 @@ out_conflict: * and it enforces that we have to think in a very structured manner * about the "events" that may happen to a request during its life time ... */ -void __req_mod(struct drbd_request *req, enum drbd_req_event what, +int __req_mod(struct drbd_request *req, enum drbd_req_event what, struct bio_and_error *m) { struct drbd_conf *mdev = req->mdev; + int rv = 0; m->bio = NULL; switch (what) { @@ -420,7 +432,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state |= (RQ_LOCAL_COMPLETED|RQ_LOCAL_OK); req->rq_state &= ~RQ_LOCAL_PENDING; - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); put_ldev(mdev); break; @@ -429,7 +441,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state &= ~RQ_LOCAL_PENDING; __drbd_chk_io_error(mdev, FALSE); - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); put_ldev(mdev); break; @@ -437,7 +449,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, /* it is legal to fail READA */ req->rq_state |= RQ_LOCAL_COMPLETED; req->rq_state &= ~RQ_LOCAL_PENDING; - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); put_ldev(mdev); break; @@ -455,7 +467,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, /* no point in retrying if there is no good remote data, * or we have no connection. */ if (mdev->state.pdsk != D_UP_TO_DATE) { - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); break; } @@ -517,11 +529,9 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, D_ASSERT(test_bit(CREATE_BARRIER, &mdev->flags) == 0); req->epoch = mdev->newest_tle->br_number; - list_add_tail(&req->tl_requests, - &mdev->newest_tle->requests); /* increment size of current epoch */ - mdev->newest_tle->n_req++; + mdev->newest_tle->n_writes++; /* queue work item to send data */ D_ASSERT(req->rq_state & RQ_NET_PENDING); @@ -530,7 +540,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, drbd_queue_work(&mdev->data.work, &req->w); /* close the epoch, in case it outgrew the limit */ - if (mdev->newest_tle->n_req >= mdev->net_conf->max_epoch_size) + if (mdev->newest_tle->n_writes >= mdev->net_conf->max_epoch_size) queue_barrier(mdev); break; @@ -543,7 +553,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state &= ~RQ_NET_QUEUED; /* if we did it right, tl_clear should be scheduled only after * this, so this should not be necessary! */ - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); break; case handed_over_to_network: @@ -568,7 +578,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, * "completed_ok" events came in, once we return from * _drbd_send_zc_bio (drbd_send_dblock), we have to check * whether it is done already, and end it. */ - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); break; case read_retry_remote_canceled: @@ -584,7 +594,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, /* if it is still queued, we may not complete it here. * it will be canceled soon. */ if (!(req->rq_state & RQ_NET_QUEUED)) - _req_may_be_done(req, m); + _req_may_be_done(req, m); /* Allowed while state.susp */ break; case write_acked_by_peer_and_sis: @@ -619,7 +629,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, D_ASSERT(req->rq_state & RQ_NET_PENDING); dec_ap_pending(mdev); req->rq_state &= ~RQ_NET_PENDING; - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); break; case neg_acked: @@ -629,11 +639,50 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state &= ~(RQ_NET_OK|RQ_NET_PENDING); req->rq_state |= RQ_NET_DONE; - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); /* else: done by handed_over_to_network */ break; + case fail_frozen_disk_io: + if (!(req->rq_state & RQ_LOCAL_COMPLETED)) + break; + + _req_may_be_done(req, m); /* Allowed while state.susp */ + break; + + case restart_frozen_disk_io: + if (!(req->rq_state & RQ_LOCAL_COMPLETED)) + break; + + req->rq_state &= ~RQ_LOCAL_COMPLETED; + + rv = MR_READ; + if (bio_data_dir(req->master_bio) == WRITE) + rv = MR_WRITE; + + get_ldev(mdev); + req->w.cb = w_restart_disk_io; + drbd_queue_work(&mdev->data.work, &req->w); + break; + + case resend: + /* If RQ_NET_OK is already set, we got a P_WRITE_ACK or P_RECV_ACK + before the connection loss (B&C only); only P_BARRIER_ACK was missing. + Trowing them out of the TL here by pretending we got a BARRIER_ACK + We ensure that the peer was not rebooted */ + if (!(req->rq_state & RQ_NET_OK)) { + if (req->w.cb) { + drbd_queue_work(&mdev->data.work, &req->w); + rv = req->rq_state & RQ_WRITE ? MR_WRITE : MR_READ; + } + break; + } + /* else, fall through to barrier_acked */ + case barrier_acked: + if (!(req->rq_state & RQ_WRITE)) + break; + if (req->rq_state & RQ_NET_PENDING) { /* barrier came in before all requests have been acked. * this is bad, because if the connection is lost now, @@ -643,7 +692,7 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, } D_ASSERT(req->rq_state & RQ_NET_SENT); req->rq_state |= RQ_NET_DONE; - _req_may_be_done(req, m); + _req_may_be_done(req, m); /* Allowed while state.susp */ break; case data_received: @@ -651,9 +700,11 @@ void __req_mod(struct drbd_request *req, enum drbd_req_event what, dec_ap_pending(mdev); req->rq_state &= ~RQ_NET_PENDING; req->rq_state |= (RQ_NET_OK|RQ_NET_DONE); - _req_may_be_done(req, m); + _req_may_be_done_not_susp(req, m); break; }; + + return rv; } /* we may do a local read if: @@ -752,15 +803,18 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) * resync extent to finish, and, if necessary, pulls in the target * extent into the activity log, which involves further disk io because * of transactional on-disk meta data updates. */ - if (rw == WRITE && local) + if (rw == WRITE && local && !test_bit(AL_SUSPENDED, &mdev->flags)) { + req->rq_state |= RQ_IN_ACT_LOG; 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)); - if (!(local || remote) && !mdev->state.susp) { - dev_err(DEV, "IO ERROR: neither local nor remote disk\n"); + if (!(local || remote) && !is_susp(mdev->state)) { + if (__ratelimit(&drbd_ratelimit_state)) + dev_err(DEV, "IO ERROR: neither local nor remote disk\n"); goto fail_free_complete; } @@ -785,7 +839,7 @@ allocate_barrier: /* GOOD, everything prepared, grab the spin_lock */ spin_lock_irq(&mdev->req_lock); - if (mdev->state.susp) { + 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 @@ -867,30 +921,10 @@ allocate_barrier: /* check this request on the collision detection hash tables. * if we have a conflict, just complete it here. * THINK do we want to check reads, too? (I don't think so...) */ - if (rw == WRITE && _req_conflicts(req)) { - /* this is a conflicting request. - * even though it may have been only _partially_ - * overlapping with one of the currently pending requests, - * without even submitting or sending it, we will - * pretend that it was successfully served right now. - */ - if (local) { - bio_put(req->private_bio); - req->private_bio = NULL; - drbd_al_complete_io(mdev, req->sector); - put_ldev(mdev); - local = 0; - } - if (remote) - dec_ap_pending(mdev); - _drbd_end_io_acct(mdev, req); - /* THINK: do we want to fail it (-EIO), or pretend success? */ - bio_endio(req->master_bio, 0); - req->master_bio = NULL; - dec_ap_bio(mdev); - drbd_req_free(req); - remote = 0; - } + if (rw == WRITE && _req_conflicts(req)) + goto fail_conflicting; + + list_add_tail(&req->tl_requests, &mdev->newest_tle->requests); /* NOTE remote first: to get the concurrent write detection right, * we must register the request before start of local IO. */ @@ -909,12 +943,21 @@ allocate_barrier: if (local) { req->private_bio->bi_bdev = mdev->ldev->backing_bdev; - if (FAULT_ACTIVE(mdev, rw == WRITE ? DRBD_FAULT_DT_WR - : rw == READ ? DRBD_FAULT_DT_RD - : DRBD_FAULT_DT_RA)) + /* State may have changed since we grabbed our reference on the + * mdev->ldev member. Double check, and short-circuit to endio. + * In case the last activity log transaction failed to get on + * 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)) + bio_endio(req->private_bio, -EIO); + else + generic_make_request(req->private_bio); + put_ldev(mdev); + } else bio_endio(req->private_bio, -EIO); - else - generic_make_request(req->private_bio); } /* we need to plug ALWAYS since we possibly need to kick lo_dev. @@ -923,6 +966,21 @@ allocate_barrier: return 0; +fail_conflicting: + /* this is a conflicting request. + * even though it may have been only _partially_ + * overlapping with one of the currently pending requests, + * without even submitting or sending it, we will + * pretend that it was successfully served right now. + */ + _drbd_end_io_acct(mdev, req); + spin_unlock_irq(&mdev->req_lock); + if (remote) + dec_ap_pending(mdev); + /* THINK: do we want to fail it (-EIO), or pretend success? + * this pretends success. */ + err = 0; + fail_free_complete: if (rw == WRITE && local) drbd_al_complete_io(mdev, sector); @@ -961,21 +1019,6 @@ static int drbd_fail_request_early(struct drbd_conf *mdev, int is_write) return 1; } - /* - * Paranoia: we might have been primary, but sync target, or - * even diskless, then lost the connection. - * This should have been handled (panic? suspend?) somewhere - * else. But maybe it was not, so check again here. - * Caution: as long as we do not have a read/write lock on mdev, - * to serialize state changes, this is racy, since we may lose - * the connection *after* we test for the cstate. - */ - if (mdev->state.disk < D_UP_TO_DATE && mdev->state.pdsk < D_UP_TO_DATE) { - if (__ratelimit(&drbd_ratelimit_state)) - dev_err(DEV, "Sorry, I have no access to good data anymore.\n"); - return 1; - } - return 0; } @@ -989,20 +1032,6 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) return 0; } - /* Reject barrier requests if we know the underlying device does - * not support them. - * XXX: Need to get this info from peer as well some how so we - * XXX: reject if EITHER side/data/metadata area does not support them. - * - * because of those XXX, this is not yet enabled, - * i.e. in drbd_init_set_defaults we set the NO_BARRIER_SUPP bit. - */ - if (unlikely(bio->bi_rw & REQ_HARDBARRIER) && test_bit(NO_BARRIER_SUPP, &mdev->flags)) { - /* dev_warn(DEV, "Rejecting barrier request as underlying device does not support\n"); */ - bio_endio(bio, -EOPNOTSUPP); - return 0; - } - /* * what we "blindly" assume: */ diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h index 02d575d..ab2bd09 100644 --- a/drivers/block/drbd/drbd_req.h +++ b/drivers/block/drbd/drbd_req.h @@ -104,6 +104,9 @@ enum drbd_req_event { read_ahead_completed_with_error, write_completed_with_error, completed_ok, + resend, + fail_frozen_disk_io, + restart_frozen_disk_io, nothing, /* for tracing only */ }; @@ -183,6 +186,12 @@ enum drbd_req_state_bits { /* keep this last, its for the RQ_NET_MASK */ __RQ_NET_MAX, + + /* Set when this is a write, clear for a read */ + __RQ_WRITE, + + /* Should call drbd_al_complete_io() for this request... */ + __RQ_IN_ACT_LOG, }; #define RQ_LOCAL_PENDING (1UL << __RQ_LOCAL_PENDING) @@ -201,6 +210,16 @@ enum drbd_req_state_bits { /* 0x1f8 */ #define RQ_NET_MASK (((1UL << __RQ_NET_MAX)-1) & ~RQ_LOCAL_MASK) +#define RQ_WRITE (1UL << __RQ_WRITE) +#define RQ_IN_ACT_LOG (1UL << __RQ_IN_ACT_LOG) + +/* For waking up the frozen transfer log mod_req() has to return if the request + should be counted in the epoch object*/ +#define MR_WRITE_SHIFT 0 +#define MR_WRITE (1 << MR_WRITE_SHIFT) +#define MR_READ_SHIFT 1 +#define MR_READ (1 << MR_READ_SHIFT) + /* epoch entries */ static inline struct hlist_head *ee_hash_slot(struct drbd_conf *mdev, sector_t sector) @@ -244,30 +263,36 @@ static inline struct drbd_request *_ar_id_to_req(struct drbd_conf *mdev, return NULL; } +static inline void drbd_req_make_private_bio(struct drbd_request *req, struct bio *bio_src) +{ + struct bio *bio; + bio = bio_clone(bio_src, GFP_NOIO); /* XXX cannot fail?? */ + + req->private_bio = bio; + + bio->bi_private = req; + bio->bi_end_io = drbd_endio_pri; + bio->bi_next = NULL; +} + static inline struct drbd_request *drbd_req_new(struct drbd_conf *mdev, struct bio *bio_src) { - struct bio *bio; struct drbd_request *req = mempool_alloc(drbd_request_mempool, GFP_NOIO); if (likely(req)) { - bio = bio_clone(bio_src, GFP_NOIO); /* XXX cannot fail?? */ + drbd_req_make_private_bio(req, bio_src); - req->rq_state = 0; + req->rq_state = bio_data_dir(bio_src) == WRITE ? RQ_WRITE : 0; req->mdev = mdev; req->master_bio = bio_src; - req->private_bio = bio; req->epoch = 0; - req->sector = bio->bi_sector; - req->size = bio->bi_size; + req->sector = bio_src->bi_sector; + req->size = bio_src->bi_size; req->start_time = jiffies; INIT_HLIST_NODE(&req->colision); INIT_LIST_HEAD(&req->tl_requests); INIT_LIST_HEAD(&req->w.list); - - bio->bi_private = req; - bio->bi_end_io = drbd_endio_pri; - bio->bi_next = NULL; } return req; } @@ -292,36 +317,44 @@ struct bio_and_error { extern void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m); -extern void __req_mod(struct drbd_request *req, enum drbd_req_event what, +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); /* 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. */ -static inline void _req_mod(struct drbd_request *req, enum drbd_req_event what) +static inline int _req_mod(struct drbd_request *req, enum drbd_req_event what) { struct drbd_conf *mdev = req->mdev; struct bio_and_error m; + int rv; /* __req_mod possibly frees req, do not touch req after that! */ - __req_mod(req, what, &m); + rv = __req_mod(req, what, &m); if (m.bio) complete_master_bio(mdev, &m); + + return rv; } /* completion of master bio is outside of spinlock. - * If you need it irqsave, do it your self! */ -static inline void req_mod(struct drbd_request *req, + * If you need it irqsave, do it your self! + * Which means: don't use from bio endio callback. */ +static inline int req_mod(struct drbd_request *req, enum drbd_req_event what) { struct drbd_conf *mdev = req->mdev; struct bio_and_error m; + int rv; + spin_lock_irq(&mdev->req_lock); - __req_mod(req, what, &m); + rv = __req_mod(req, what, &m); spin_unlock_irq(&mdev->req_lock); if (m.bio) complete_master_bio(mdev, &m); + + return rv; } #endif diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index ca4a16c..34f224b 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -26,7 +26,6 @@ #include <linux/module.h> #include <linux/drbd.h> #include <linux/sched.h> -#include <linux/smp_lock.h> #include <linux/wait.h> #include <linux/mm.h> #include <linux/memcontrol.h> @@ -39,8 +38,6 @@ #include "drbd_int.h" #include "drbd_req.h" -#define SLEEP_TIME (HZ/10) - static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel); @@ -104,12 +101,6 @@ void drbd_endio_read_sec_final(struct drbd_epoch_entry *e) __releases(local) put_ldev(mdev); } -static int is_failed_barrier(int ee_flags) -{ - return (ee_flags & (EE_IS_BARRIER|EE_WAS_ERROR|EE_RESUBMITTED)) - == (EE_IS_BARRIER|EE_WAS_ERROR); -} - /* writes on behalf of the partner, or resync writes, * "submitted" by the receiver, final stage. */ static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(local) @@ -121,21 +112,6 @@ static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(lo int is_syncer_req; int do_al_complete_io; - /* if this is a failed barrier request, disable use of barriers, - * and schedule for resubmission */ - if (is_failed_barrier(e->flags)) { - drbd_bump_write_ordering(mdev, WO_bdev_flush); - spin_lock_irqsave(&mdev->req_lock, flags); - list_del(&e->w.list); - e->flags = (e->flags & ~EE_WAS_ERROR) | EE_RESUBMITTED; - e->w.cb = w_e_reissue; - /* put_ldev actually happens below, once we come here again. */ - __release(local); - spin_unlock_irqrestore(&mdev->req_lock, flags); - drbd_queue_work(&mdev->data.work, &e->w); - return; - } - D_ASSERT(e->block_id != ID_VACANT); /* after we moved e to done_ee, @@ -246,6 +222,7 @@ void drbd_endio_pri(struct bio *bio, int error) bio_put(req->private_bio); req->private_bio = ERR_PTR(error); + /* not req_mod(), we need irqsave here! */ spin_lock_irqsave(&mdev->req_lock, flags); __req_mod(req, what, &m); spin_unlock_irqrestore(&mdev->req_lock, flags); @@ -376,54 +353,145 @@ static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size) struct drbd_epoch_entry *e; if (!get_ldev(mdev)) - return 0; + return -EIO; + + if (drbd_rs_should_slow_down(mdev)) + goto defer; /* GFP_TRY, because if there is no memory available right now, this may * be rescheduled for later. It is "only" background resync, after all. */ e = drbd_alloc_ee(mdev, DRBD_MAGIC+0xbeef, sector, size, GFP_TRY); if (!e) - goto fail; + goto defer; + e->w.cb = w_e_send_csum; spin_lock_irq(&mdev->req_lock); list_add(&e->w.list, &mdev->read_ee); spin_unlock_irq(&mdev->req_lock); - e->w.cb = w_e_send_csum; + atomic_add(size >> 9, &mdev->rs_sect_ev); if (drbd_submit_ee(mdev, e, READ, DRBD_FAULT_RS_RD) == 0) - return 1; + 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? */ + spin_lock_irq(&mdev->req_lock); + list_del(&e->w.list); + spin_unlock_irq(&mdev->req_lock); drbd_free_ee(mdev, e); -fail: +defer: put_ldev(mdev); - return 2; + return -EAGAIN; } void resync_timer_fn(unsigned long data) { - unsigned long flags; struct drbd_conf *mdev = (struct drbd_conf *) data; int queue; - spin_lock_irqsave(&mdev->req_lock, flags); - - if (likely(!test_and_clear_bit(STOP_SYNC_TIMER, &mdev->flags))) { - queue = 1; - if (mdev->state.conn == C_VERIFY_S) - mdev->resync_work.cb = w_make_ov_request; - else - mdev->resync_work.cb = w_make_resync_request; - } else { + queue = 1; + switch (mdev->state.conn) { + case C_VERIFY_S: + mdev->resync_work.cb = w_make_ov_request; + break; + case C_SYNC_TARGET: + mdev->resync_work.cb = w_make_resync_request; + break; + default: queue = 0; mdev->resync_work.cb = w_resync_inactive; } - spin_unlock_irqrestore(&mdev->req_lock, flags); - /* harmless race: list_empty outside data.work.q_lock */ if (list_empty(&mdev->resync_work.list) && queue) drbd_queue_work(&mdev->data.work, &mdev->resync_work); } +static void fifo_set(struct fifo_buffer *fb, int value) +{ + int i; + + for (i = 0; i < fb->size; i++) + fb->values[i] = value; +} + +static int fifo_push(struct fifo_buffer *fb, int value) +{ + int ov; + + ov = fb->values[fb->head_index]; + fb->values[fb->head_index++] = value; + + if (fb->head_index >= fb->size) + fb->head_index = 0; + + return ov; +} + +static void fifo_add_val(struct fifo_buffer *fb, int value) +{ + int i; + + for (i = 0; i < fb->size; i++) + fb->values[i] += value; +} + +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 */ + int req_sect; /* Number of sectors to request in this turn */ + int correction; /* Number of sectors more we need in the proxy*/ + int cps; /* correction per invocation of drbd_rs_controller() */ + int steps; /* Number of time steps to plan ahead */ + int curr_corr; + int max_sect; + + sect_in = atomic_xchg(&mdev->rs_sect_in, 0); /* Number of sectors that came in */ + mdev->rs_in_flight -= sect_in; + + spin_lock(&mdev->peer_seq_lock); /* get an atomic view on mdev->rs_plan_s */ + + steps = mdev->rs_plan_s.size; /* (mdev->sync_conf.c_plan_ahead * 10 * SLEEP_TIME) / HZ; */ + + if (mdev->rs_in_flight + sect_in == 0) { /* At start of resync */ + want = ((mdev->sync_conf.rate * 2 * SLEEP_TIME) / HZ) * steps; + } else { /* normal path */ + want = mdev->sync_conf.c_fill_target ? mdev->sync_conf.c_fill_target : + sect_in * mdev->sync_conf.c_delay_target * HZ / (SLEEP_TIME * 10); + } + + correction = want - mdev->rs_in_flight - mdev->rs_planed; + + /* Plan ahead */ + cps = correction / steps; + fifo_add_val(&mdev->rs_plan_s, cps); + mdev->rs_planed += cps * steps; + + /* What we do in this step */ + curr_corr = fifo_push(&mdev->rs_plan_s, 0); + spin_unlock(&mdev->peer_seq_lock); + mdev->rs_planed -= curr_corr; + + req_sect = sect_in + curr_corr; + if (req_sect < 0) + req_sect = 0; + + max_sect = (mdev->sync_conf.c_max_rate * 2 * SLEEP_TIME) / HZ; + if (req_sect > max_sect) + req_sect = max_sect; + + /* + dev_warn(DEV, "si=%u if=%d wa=%u co=%d st=%d cps=%d pl=%d cc=%d rs=%d\n", + sect_in, mdev->rs_in_flight, want, correction, + steps, cps, mdev->rs_planed, curr_corr, req_sect); + */ + + return req_sect; +} + int w_make_resync_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { @@ -431,8 +499,9 @@ int w_make_resync_request(struct drbd_conf *mdev, sector_t sector; const sector_t capacity = drbd_get_capacity(mdev->this_bdev); int max_segment_size; - int number, i, size, pe, mx; + int number, rollback_i, size, pe, mx; int align, queued, sndbuf; + int i = 0; if (unlikely(cancel)) return 1; @@ -446,6 +515,12 @@ int w_make_resync_request(struct drbd_conf *mdev, 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); + return 1; + } + if (!get_ldev(mdev)) { /* Since we only need to access mdev->rsync a get_ldev_if_state(mdev,D_FAILED) would be sufficient, but @@ -458,11 +533,25 @@ int w_make_resync_request(struct drbd_conf *mdev, /* 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) : DRBD_MAX_SEGMENT_SIZE; + 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; - number = SLEEP_TIME * mdev->sync_conf.rate / ((BM_BLOCK_SIZE / 1024) * HZ); - pe = atomic_read(&mdev->rs_pending_cnt); + 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); + } + + /* 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)) + goto requeue; mutex_lock(&mdev->data.mutex); if (mdev->data.socket) @@ -476,6 +565,7 @@ int w_make_resync_request(struct drbd_conf *mdev, 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; } @@ -526,6 +616,7 @@ next_sector: * be prepared for all stripe sizes of software RAIDs. */ align = 1; + rollback_i = i; for (;;) { if (size + BM_BLOCK_SIZE > max_segment_size) break; @@ -561,14 +652,19 @@ next_sector: size = (capacity-sector)<<9; if (mdev->agreed_pro_version >= 89 && mdev->csums_tfm) { switch (read_for_csum(mdev, sector, size)) { - case 0: /* Disk failure*/ + case -EIO: /* Disk failure */ put_ldev(mdev); return 0; - case 2: /* Allocation failed */ + case -EAGAIN: /* allocation failed, or ldev busy */ drbd_rs_complete_io(mdev, sector); mdev->bm_resync_fo = BM_SECT_TO_BIT(sector); + i = rollback_i; goto requeue; - /* case 1: everything ok */ + case 0: + /* everything ok */ + break; + default: + BUG(); } } else { inc_rs_pending(mdev); @@ -595,6 +691,7 @@ next_sector: } requeue: + mdev->rs_in_flight += (i << (BM_BLOCK_SHIFT - 9)); mod_timer(&mdev->resync_timer, jiffies + SLEEP_TIME); put_ldev(mdev); return 1; @@ -670,6 +767,14 @@ static int w_resync_finished(struct drbd_conf *mdev, struct drbd_work *w, int ca return 1; } +static void ping_peer(struct drbd_conf *mdev) +{ + clear_bit(GOT_PING_ACK, &mdev->flags); + request_ping(mdev); + wait_event(mdev->misc_wait, + test_bit(GOT_PING_ACK, &mdev->flags) || mdev->state.conn < C_CONNECTED); +} + int drbd_resync_finished(struct drbd_conf *mdev) { unsigned long db, dt, dbdt; @@ -709,6 +814,8 @@ int drbd_resync_finished(struct drbd_conf *mdev) if (!get_ldev(mdev)) goto out; + ping_peer(mdev); + spin_lock_irq(&mdev->req_lock); os = mdev->state; @@ -801,8 +908,10 @@ out: mdev->rs_paused = 0; mdev->ov_start_sector = 0; + drbd_md_sync(mdev); + if (test_and_clear_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags)) { - dev_warn(DEV, "Writing the whole bitmap, due to failed kmalloc\n"); + dev_info(DEV, "Writing the whole bitmap\n"); drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished"); } @@ -817,9 +926,13 @@ static void move_to_net_ee_or_free(struct drbd_conf *mdev, struct drbd_epoch_ent { if (drbd_ee_has_active_page(e)) { /* This might happen if sendpage() has not finished */ + int i = (e->size + PAGE_SIZE -1) >> PAGE_SHIFT; + atomic_add(i, &mdev->pp_in_use_by_net); + atomic_sub(i, &mdev->pp_in_use); spin_lock_irq(&mdev->req_lock); list_add_tail(&e->w.list, &mdev->net_ee); spin_unlock_irq(&mdev->req_lock); + wake_up(&drbd_pp_wait); } else drbd_free_ee(mdev, e); } @@ -926,9 +1039,12 @@ int w_e_end_csum_rs_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return 1; } - drbd_rs_complete_io(mdev, e->sector); + if (get_ldev(mdev)) { + drbd_rs_complete_io(mdev, e->sector); + put_ldev(mdev); + } - di = (struct digest_info *)(unsigned long)e->block_id; + di = e->digest; if (likely((e->flags & EE_WAS_ERROR) == 0)) { /* quick hack to try to avoid a race against reconfiguration. @@ -952,7 +1068,9 @@ int w_e_end_csum_rs_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) ok = drbd_send_ack(mdev, P_RS_IS_IN_SYNC, e); } else { inc_rs_pending(mdev); - e->block_id = ID_SYNCER; + e->block_id = ID_SYNCER; /* By setting block_id, digest pointer becomes invalid! */ + e->flags &= ~EE_HAS_DIGEST; /* This e no longer has a digest pointer */ + kfree(di); ok = drbd_send_block(mdev, P_RS_DATA_REPLY, e); } } else { @@ -962,9 +1080,6 @@ int w_e_end_csum_rs_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) } dec_unacked(mdev); - - kfree(di); - move_to_net_ee_or_free(mdev, e); if (unlikely(!ok)) @@ -1034,9 +1149,12 @@ int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) /* after "cancel", because after drbd_disconnect/drbd_rs_cancel_all * the resync lru has been cleaned up already */ - drbd_rs_complete_io(mdev, e->sector); + if (get_ldev(mdev)) { + drbd_rs_complete_io(mdev, e->sector); + put_ldev(mdev); + } - di = (struct digest_info *)(unsigned long)e->block_id; + di = e->digest; if (likely((e->flags & EE_WAS_ERROR) == 0)) { digest_size = crypto_hash_digestsize(mdev->verify_tfm); @@ -1055,9 +1173,6 @@ int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) } dec_unacked(mdev); - - kfree(di); - if (!eq) drbd_ov_oos_found(mdev, e->sector, e->size); else @@ -1108,7 +1223,7 @@ int w_send_barrier(struct drbd_conf *mdev, struct drbd_work *w, int cancel) * dec_ap_pending will be done in got_BarrierAck * or (on connection loss) in w_clear_epoch. */ ok = _drbd_send_cmd(mdev, mdev->data.socket, P_BARRIER, - (struct p_header *)p, sizeof(*p), 0); + (struct p_header80 *)p, sizeof(*p), 0); drbd_put_data_sock(mdev); return ok; @@ -1173,6 +1288,24 @@ int w_send_read_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return ok; } +int w_restart_disk_io(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +{ + struct drbd_request *req = container_of(w, struct drbd_request, w); + + if (bio_data_dir(req->master_bio) == WRITE && req->rq_state & RQ_IN_ACT_LOG) + drbd_al_begin_io(mdev, req->sector); + /* Calling drbd_al_begin_io() out of the worker might deadlocks + theoretically. Practically it can not deadlock, since this is + only used when unfreezing IOs. All the extents of the requests + that made it into the TL are already active */ + + drbd_req_make_private_bio(req, req->master_bio); + req->private_bio->bi_bdev = mdev->ldev->backing_bdev; + generic_make_request(req->private_bio); + + return 1; +} + static int _drbd_may_sync_now(struct drbd_conf *mdev) { struct drbd_conf *odev = mdev; @@ -1298,14 +1431,6 @@ int drbd_alter_sa(struct drbd_conf *mdev, int na) return retcode; } -static void ping_peer(struct drbd_conf *mdev) -{ - clear_bit(GOT_PING_ACK, &mdev->flags); - request_ping(mdev); - wait_event(mdev->misc_wait, - test_bit(GOT_PING_ACK, &mdev->flags) || mdev->state.conn < C_CONNECTED); -} - /** * drbd_start_resync() - Start the resync process * @mdev: DRBD device. @@ -1379,13 +1504,21 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) r = SS_UNKNOWN_ERROR; if (r == SS_SUCCESS) { - mdev->rs_total = - mdev->rs_mark_left = drbd_bm_total_weight(mdev); + unsigned long tw = drbd_bm_total_weight(mdev); + unsigned long now = jiffies; + int i; + mdev->rs_failed = 0; mdev->rs_paused = 0; - mdev->rs_start = - mdev->rs_mark_time = jiffies; mdev->rs_same_csum = 0; + mdev->rs_last_events = 0; + mdev->rs_last_sect_ev = 0; + mdev->rs_total = tw; + mdev->rs_start = now; + for (i = 0; i < DRBD_SYNC_MARKS; i++) { + mdev->rs_mark_left[i] = tw; + mdev->rs_mark_time[i] = now; + } _drbd_pause_after(mdev); } write_unlock_irq(&global_state_lock); @@ -1397,12 +1530,31 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) (unsigned long) mdev->rs_total << (BM_BLOCK_SHIFT-10), (unsigned long) mdev->rs_total); - if (mdev->rs_total == 0) { - /* Peer still reachable? Beware of failing before-resync-target handlers! */ - ping_peer(mdev); + if (mdev->agreed_pro_version < 95 && mdev->rs_total == 0) { + /* This still has a race (about when exactly the peers + * detect connection loss) that can lead to a full sync + * on next handshake. In 8.3.9 we fixed this with explicit + * resync-finished notifications, but the fix + * introduces a protocol change. Sleeping for some + * time longer than the ping interval + timeout on the + * SyncSource, to give the SyncTarget the chance to + * detect connection loss, then waiting for a ping + * response (implicit in drbd_resync_finished) reduces + * the race considerably, but does not solve it. */ + if (side == C_SYNC_SOURCE) + schedule_timeout_interruptible( + mdev->net_conf->ping_int * HZ + + mdev->net_conf->ping_timeo*HZ/9); 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); /* ns.conn may already be != mdev->state.conn, * we may have been paused in between, or become paused until * the timer triggers. diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 709e69c..b9ba04f 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -178,7 +178,6 @@ static int print_unex = 1; #include <linux/slab.h> #include <linux/mm.h> #include <linux/bio.h> -#include <linux/smp_lock.h> #include <linux/string.h> #include <linux/jiffies.h> #include <linux/fcntl.h> @@ -199,6 +198,7 @@ static int print_unex = 1; * It's been recommended that take about 1/4 of the default speed * in some more extreme cases. */ +static DEFINE_MUTEX(floppy_mutex); static int slow_floppy; #include <asm/dma.h> @@ -258,8 +258,8 @@ static int irqdma_allocated; #include <linux/completion.h> static struct request *current_req; -static struct request_queue *floppy_queue; static void do_fd_request(struct request_queue *q); +static int set_next_request(void); #ifndef fd_get_dma_residue #define fd_get_dma_residue() get_dma_residue(FLOPPY_DMA) @@ -413,6 +413,7 @@ static struct gendisk *disks[N_DRIVE]; static struct block_device *opened_bdev[N_DRIVE]; static DEFINE_MUTEX(open_lock); static struct floppy_raw_cmd *raw_cmd, default_raw_cmd; +static int fdc_queue; /* * This struct defines the different floppy types. @@ -895,8 +896,8 @@ static void unlock_fdc(void) del_timer(&fd_timeout); cont = NULL; clear_bit(0, &fdc_busy); - if (current_req || blk_peek_request(floppy_queue)) - do_fd_request(floppy_queue); + if (current_req || set_next_request()) + do_fd_request(current_req->q); spin_unlock_irqrestore(&floppy_lock, flags); wake_up(&fdc_wait); } @@ -2248,8 +2249,8 @@ static void floppy_end_request(struct request *req, int error) * logical buffer */ static void request_done(int uptodate) { - struct request_queue *q = floppy_queue; struct request *req = current_req; + struct request_queue *q; unsigned long flags; int block; char msg[sizeof("request done ") + sizeof(int) * 3]; @@ -2263,6 +2264,8 @@ static void request_done(int uptodate) return; } + q = req->q; + if (uptodate) { /* maintain values for invalidation on geometry * change */ @@ -2816,6 +2819,28 @@ static int make_raw_rw_request(void) return 2; } +/* + * Round-robin between our available drives, doing one request from each + */ +static int set_next_request(void) +{ + struct request_queue *q; + int old_pos = fdc_queue; + + do { + q = disks[fdc_queue]->queue; + if (++fdc_queue == N_DRIVE) + fdc_queue = 0; + if (q) { + current_req = blk_fetch_request(q); + if (current_req) + break; + } + } while (fdc_queue != old_pos); + + return current_req != NULL; +} + static void redo_fd_request(void) { int drive; @@ -2827,17 +2852,17 @@ static void redo_fd_request(void) do_request: if (!current_req) { - struct request *req; + int pending; + + spin_lock_irq(&floppy_lock); + pending = set_next_request(); + spin_unlock_irq(&floppy_lock); - spin_lock_irq(floppy_queue->queue_lock); - req = blk_fetch_request(floppy_queue); - spin_unlock_irq(floppy_queue->queue_lock); - if (!req) { + if (!pending) { do_floppy = NULL; unlock_fdc(); return; } - current_req = req; } drive = (long)current_req->rq_disk->private_data; set_fdc(drive); @@ -3558,9 +3583,9 @@ static int fd_ioctl(struct block_device *bdev, fmode_t mode, { int ret; - lock_kernel(); + mutex_lock(&floppy_mutex); ret = fd_locked_ioctl(bdev, mode, cmd, param); - unlock_kernel(); + mutex_unlock(&floppy_mutex); return ret; } @@ -3621,7 +3646,7 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) { int drive = (long)disk->private_data; - lock_kernel(); + mutex_lock(&floppy_mutex); mutex_lock(&open_lock); if (UDRS->fd_ref < 0) UDRS->fd_ref = 0; @@ -3632,7 +3657,7 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) if (!UDRS->fd_ref) opened_bdev[drive] = NULL; mutex_unlock(&open_lock); - unlock_kernel(); + mutex_unlock(&floppy_mutex); return 0; } @@ -3650,7 +3675,7 @@ static int floppy_open(struct block_device *bdev, fmode_t mode) int res = -EBUSY; char *tmp; - lock_kernel(); + mutex_lock(&floppy_mutex); mutex_lock(&open_lock); old_dev = UDRS->fd_device; if (opened_bdev[drive] && opened_bdev[drive] != bdev) @@ -3727,7 +3752,7 @@ static int floppy_open(struct block_device *bdev, fmode_t mode) goto out; } mutex_unlock(&open_lock); - unlock_kernel(); + mutex_unlock(&floppy_mutex); return 0; out: if (UDRS->fd_ref < 0) @@ -3738,7 +3763,7 @@ out: opened_bdev[drive] = NULL; out2: mutex_unlock(&open_lock); - unlock_kernel(); + mutex_unlock(&floppy_mutex); return res; } @@ -4170,6 +4195,13 @@ static int __init floppy_init(void) goto out_put_disk; } + disks[dr]->queue = blk_init_queue(do_fd_request, &floppy_lock); + if (!disks[dr]->queue) { + err = -ENOMEM; + goto out_put_disk; + } + + blk_queue_max_hw_sectors(disks[dr]->queue, 64); disks[dr]->major = FLOPPY_MAJOR; disks[dr]->first_minor = TOMINOR(dr); disks[dr]->fops = &floppy_fops; @@ -4188,13 +4220,6 @@ static int __init floppy_init(void) if (err) goto out_unreg_blkdev; - floppy_queue = blk_init_queue(do_fd_request, &floppy_lock); - if (!floppy_queue) { - err = -ENOMEM; - goto out_unreg_driver; - } - blk_queue_max_hw_sectors(floppy_queue, 64); - blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE, floppy_find, NULL, NULL); @@ -4322,7 +4347,6 @@ static int __init floppy_init(void) /* to be cleaned up... */ disks[drive]->private_data = (void *)(long)drive; - disks[drive]->queue = floppy_queue; disks[drive]->flags |= GENHD_FL_REMOVABLE; disks[drive]->driverfs_dev = &floppy_device[drive].dev; add_disk(disks[drive]); @@ -4333,19 +4357,19 @@ static int __init floppy_init(void) out_unreg_platform_dev: platform_device_unregister(&floppy_device[drive]); out_flush_work: - flush_scheduled_work(); + flush_work_sync(&floppy_work); if (atomic_read(&usage_count)) floppy_release_irq_and_dma(); out_unreg_region: blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); - blk_cleanup_queue(floppy_queue); -out_unreg_driver: platform_driver_unregister(&floppy_driver); out_unreg_blkdev: unregister_blkdev(FLOPPY_MAJOR, "fd"); out_put_disk: while (dr--) { del_timer(&motor_off_timer[dr]); + if (disks[dr]->queue) + blk_cleanup_queue(disks[dr]->queue); put_disk(disks[dr]); } return err; @@ -4403,7 +4427,7 @@ static int floppy_grab_irq_and_dma(void) * We might have scheduled a free_irq(), wait it to * drain first: */ - flush_scheduled_work(); + flush_work_sync(&floppy_work); if (fd_request_irq()) { DPRINT("Unable to grab IRQ%d for the floppy driver\n", @@ -4554,12 +4578,12 @@ static void __exit floppy_module_exit(void) device_remove_file(&floppy_device[drive].dev, &dev_attr_cmos); platform_device_unregister(&floppy_device[drive]); } + blk_cleanup_queue(disks[drive]->queue); put_disk(disks[drive]); } del_timer_sync(&fd_timeout); del_timer_sync(&fd_timer); - blk_cleanup_queue(floppy_queue); if (atomic_read(&usage_count)) floppy_release_irq_and_dma(); diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 91797bb..44e18c0 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -67,16 +67,18 @@ #include <linux/compat.h> #include <linux/suspend.h> #include <linux/freezer.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/writeback.h> #include <linux/buffer_head.h> /* for invalidate_bdev() */ #include <linux/completion.h> #include <linux/highmem.h> #include <linux/kthread.h> #include <linux/splice.h> +#include <linux/sysfs.h> #include <asm/uaccess.h> +static DEFINE_MUTEX(loop_mutex); static LIST_HEAD(loop_devices); static DEFINE_MUTEX(loop_devices_mutex); @@ -99,8 +101,8 @@ static int transfer_none(struct loop_device *lo, int cmd, else memcpy(raw_buf, loop_buf, size); - kunmap_atomic(raw_buf, KM_USER0); kunmap_atomic(loop_buf, KM_USER1); + kunmap_atomic(raw_buf, KM_USER0); cond_resched(); return 0; } @@ -128,8 +130,8 @@ static int transfer_xor(struct loop_device *lo, int cmd, for (i = 0; i < size; i++) *out++ = *in++ ^ key[(i & 511) % keysize]; - kunmap_atomic(raw_buf, KM_USER0); kunmap_atomic(loop_buf, KM_USER1); + kunmap_atomic(raw_buf, KM_USER0); cond_resched(); return 0; } @@ -393,11 +395,7 @@ lo_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf, struct loop_device *lo = p->lo; struct page *page = buf->page; sector_t IV; - int size, ret; - - ret = buf->ops->confirm(pipe, buf); - if (unlikely(ret)) - return ret; + int size; IV = ((sector_t) page->index << (PAGE_CACHE_SHIFT - 9)) + (buf->offset >> 9); @@ -477,17 +475,11 @@ static int do_bio_filebacked(struct loop_device *lo, struct bio *bio) pos = ((loff_t) bio->bi_sector << 9) + lo->lo_offset; if (bio_rw(bio) == WRITE) { - bool barrier = !!(bio->bi_rw & REQ_HARDBARRIER); struct file *file = lo->lo_backing_file; - if (barrier) { - if (unlikely(!file->f_op->fsync)) { - ret = -EOPNOTSUPP; - goto out; - } - + if (bio->bi_rw & REQ_FLUSH) { ret = vfs_fsync(file, 0); - if (unlikely(ret)) { + if (unlikely(ret && ret != -EINVAL)) { ret = -EIO; goto out; } @@ -495,9 +487,9 @@ static int do_bio_filebacked(struct loop_device *lo, struct bio *bio) ret = lo_send(lo, bio, pos); - if (barrier && !ret) { + if ((bio->bi_rw & REQ_FUA) && !ret) { ret = vfs_fsync(file, 0); - if (unlikely(ret)) + if (unlikely(ret && ret != -EINVAL)) ret = -EIO; } } else @@ -737,6 +729,103 @@ static inline int is_loop_device(struct file *file) return i && S_ISBLK(i->i_mode) && MAJOR(i->i_rdev) == LOOP_MAJOR; } +/* loop sysfs attributes */ + +static ssize_t loop_attr_show(struct device *dev, char *page, + ssize_t (*callback)(struct loop_device *, char *)) +{ + struct loop_device *l, *lo = NULL; + + mutex_lock(&loop_devices_mutex); + list_for_each_entry(l, &loop_devices, lo_list) + if (disk_to_dev(l->lo_disk) == dev) { + lo = l; + break; + } + mutex_unlock(&loop_devices_mutex); + + return lo ? callback(lo, page) : -EIO; +} + +#define LOOP_ATTR_RO(_name) \ +static ssize_t loop_attr_##_name##_show(struct loop_device *, char *); \ +static ssize_t loop_attr_do_show_##_name(struct device *d, \ + struct device_attribute *attr, char *b) \ +{ \ + return loop_attr_show(d, b, loop_attr_##_name##_show); \ +} \ +static struct device_attribute loop_attr_##_name = \ + __ATTR(_name, S_IRUGO, loop_attr_do_show_##_name, NULL); + +static ssize_t loop_attr_backing_file_show(struct loop_device *lo, char *buf) +{ + ssize_t ret; + char *p = NULL; + + mutex_lock(&lo->lo_ctl_mutex); + if (lo->lo_backing_file) + p = d_path(&lo->lo_backing_file->f_path, buf, PAGE_SIZE - 1); + mutex_unlock(&lo->lo_ctl_mutex); + + if (IS_ERR_OR_NULL(p)) + ret = PTR_ERR(p); + else { + ret = strlen(p); + memmove(buf, p, ret); + buf[ret++] = '\n'; + buf[ret] = 0; + } + + return ret; +} + +static ssize_t loop_attr_offset_show(struct loop_device *lo, char *buf) +{ + return sprintf(buf, "%llu\n", (unsigned long long)lo->lo_offset); +} + +static ssize_t loop_attr_sizelimit_show(struct loop_device *lo, char *buf) +{ + return sprintf(buf, "%llu\n", (unsigned long long)lo->lo_sizelimit); +} + +static ssize_t loop_attr_autoclear_show(struct loop_device *lo, char *buf) +{ + int autoclear = (lo->lo_flags & LO_FLAGS_AUTOCLEAR); + + return sprintf(buf, "%s\n", autoclear ? "1" : "0"); +} + +LOOP_ATTR_RO(backing_file); +LOOP_ATTR_RO(offset); +LOOP_ATTR_RO(sizelimit); +LOOP_ATTR_RO(autoclear); + +static struct attribute *loop_attrs[] = { + &loop_attr_backing_file.attr, + &loop_attr_offset.attr, + &loop_attr_sizelimit.attr, + &loop_attr_autoclear.attr, + NULL, +}; + +static struct attribute_group loop_attribute_group = { + .name = "loop", + .attrs= loop_attrs, +}; + +static int loop_sysfs_init(struct loop_device *lo) +{ + return sysfs_create_group(&disk_to_dev(lo->lo_disk)->kobj, + &loop_attribute_group); +} + +static void loop_sysfs_exit(struct loop_device *lo) +{ + sysfs_remove_group(&disk_to_dev(lo->lo_disk)->kobj, + &loop_attribute_group); +} + static int loop_set_fd(struct loop_device *lo, fmode_t mode, struct block_device *bdev, unsigned int arg) { @@ -832,10 +921,11 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, lo->lo_queue->unplug_fn = loop_unplug; if (!(lo_flags & LO_FLAGS_READ_ONLY) && file->f_op->fsync) - blk_queue_ordered(lo->lo_queue, QUEUE_ORDERED_DRAIN); + blk_queue_flush(lo->lo_queue, REQ_FLUSH); set_capacity(lo->lo_disk, size); bd_set_size(bdev, size << 9); + loop_sysfs_init(lo); /* let user-space know about the new size */ kobject_uevent(&disk_to_dev(bdev->bd_disk)->kobj, KOBJ_CHANGE); @@ -854,6 +944,7 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, return 0; out_clr: + loop_sysfs_exit(lo); lo->lo_thread = NULL; lo->lo_device = NULL; lo->lo_backing_file = NULL; @@ -948,6 +1039,7 @@ static int loop_clr_fd(struct loop_device *lo, struct block_device *bdev) if (bdev) invalidate_bdev(bdev); set_capacity(lo->lo_disk, 0); + loop_sysfs_exit(lo); if (bdev) { bd_set_size(bdev, 0); /* let user-space know about this change */ @@ -1409,11 +1501,11 @@ static int lo_open(struct block_device *bdev, fmode_t mode) { struct loop_device *lo = bdev->bd_disk->private_data; - lock_kernel(); + mutex_lock(&loop_mutex); mutex_lock(&lo->lo_ctl_mutex); lo->lo_refcnt++; mutex_unlock(&lo->lo_ctl_mutex); - unlock_kernel(); + mutex_unlock(&loop_mutex); return 0; } @@ -1423,7 +1515,7 @@ static int lo_release(struct gendisk *disk, fmode_t mode) struct loop_device *lo = disk->private_data; int err; - lock_kernel(); + mutex_lock(&loop_mutex); mutex_lock(&lo->lo_ctl_mutex); if (--lo->lo_refcnt) @@ -1448,7 +1540,7 @@ static int lo_release(struct gendisk *disk, fmode_t mode) out: mutex_unlock(&lo->lo_ctl_mutex); out_unlocked: - lock_kernel(); + mutex_unlock(&loop_mutex); return 0; } diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c index 0daa422..a32fb41 100644 --- a/drivers/block/nbd.c +++ b/drivers/block/nbd.c @@ -24,7 +24,7 @@ #include <linux/errno.h> #include <linux/file.h> #include <linux/ioctl.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/compiler.h> #include <linux/err.h> #include <linux/kernel.h> @@ -53,6 +53,7 @@ #define DBG_BLKDEV 0x0100 #define DBG_RX 0x0200 #define DBG_TX 0x0400 +static DEFINE_MUTEX(nbd_mutex); static unsigned int debugflags; #endif /* NDEBUG */ @@ -717,11 +718,11 @@ static int nbd_ioctl(struct block_device *bdev, fmode_t mode, dprintk(DBG_IOCTL, "%s: nbd_ioctl cmd=%s(0x%x) arg=%lu\n", lo->disk->disk_name, ioctl_cmd_to_ascii(cmd), cmd, arg); - lock_kernel(); + mutex_lock(&nbd_mutex); mutex_lock(&lo->tx_lock); error = __nbd_ioctl(bdev, lo, cmd, arg); mutex_unlock(&lo->tx_lock); - unlock_kernel(); + mutex_unlock(&nbd_mutex); return error; } diff --git a/drivers/block/osdblk.c b/drivers/block/osdblk.c index 2284b4f..87311eb 100644 --- a/drivers/block/osdblk.c +++ b/drivers/block/osdblk.c @@ -310,8 +310,7 @@ static void osdblk_rq_fn(struct request_queue *q) break; /* filter out block requests we don't understand */ - if (rq->cmd_type != REQ_TYPE_FS && - !(rq->cmd_flags & REQ_HARDBARRIER)) { + if (rq->cmd_type != REQ_TYPE_FS) { blk_end_request_all(rq, 0); continue; } @@ -439,7 +438,7 @@ static int osdblk_init_disk(struct osdblk_device *osdev) blk_queue_stack_limits(q, osd_request_queue(osdev->osd)); blk_queue_prep_rq(q, blk_queue_start_tag); - blk_queue_ordered(q, QUEUE_ORDERED_DRAIN_FLUSH); + blk_queue_flush(q, REQ_FLUSH); disk->queue = q; diff --git a/drivers/block/paride/pcd.c b/drivers/block/paride/pcd.c index 76f8565..62cec6a 100644 --- a/drivers/block/paride/pcd.c +++ b/drivers/block/paride/pcd.c @@ -138,9 +138,10 @@ enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_DLY}; #include <linux/cdrom.h> #include <linux/spinlock.h> #include <linux/blkdev.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <asm/uaccess.h> +static DEFINE_MUTEX(pcd_mutex); static DEFINE_SPINLOCK(pcd_lock); module_param(verbose, bool, 0644); @@ -227,9 +228,9 @@ static int pcd_block_open(struct block_device *bdev, fmode_t mode) struct pcd_unit *cd = bdev->bd_disk->private_data; int ret; - lock_kernel(); + mutex_lock(&pcd_mutex); ret = cdrom_open(&cd->info, bdev, mode); - unlock_kernel(); + mutex_unlock(&pcd_mutex); return ret; } @@ -237,9 +238,9 @@ static int pcd_block_open(struct block_device *bdev, fmode_t mode) static int pcd_block_release(struct gendisk *disk, fmode_t mode) { struct pcd_unit *cd = disk->private_data; - lock_kernel(); + mutex_lock(&pcd_mutex); cdrom_release(&cd->info, mode); - unlock_kernel(); + mutex_unlock(&pcd_mutex); return 0; } @@ -249,9 +250,9 @@ static int pcd_block_ioctl(struct block_device *bdev, fmode_t mode, struct pcd_unit *cd = bdev->bd_disk->private_data; int ret; - lock_kernel(); + mutex_lock(&pcd_mutex); ret = cdrom_ioctl(&cd->info, bdev, mode, cmd, arg); - unlock_kernel(); + mutex_unlock(&pcd_mutex); return ret; } diff --git a/drivers/block/paride/pd.c b/drivers/block/paride/pd.c index 985f0d4..c0ee155 100644 --- a/drivers/block/paride/pd.c +++ b/drivers/block/paride/pd.c @@ -153,10 +153,11 @@ enum {D_PRT, D_PRO, D_UNI, D_MOD, D_GEO, D_SBY, D_DLY, D_SLV}; #include <linux/blkdev.h> #include <linux/blkpg.h> #include <linux/kernel.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <asm/uaccess.h> #include <linux/workqueue.h> +static DEFINE_MUTEX(pd_mutex); static DEFINE_SPINLOCK(pd_lock); module_param(verbose, bool, 0); @@ -736,14 +737,14 @@ static int pd_open(struct block_device *bdev, fmode_t mode) { struct pd_unit *disk = bdev->bd_disk->private_data; - lock_kernel(); + mutex_lock(&pd_mutex); disk->access++; if (disk->removable) { pd_special_command(disk, pd_media_check); pd_special_command(disk, pd_door_lock); } - unlock_kernel(); + mutex_unlock(&pd_mutex); return 0; } @@ -771,10 +772,10 @@ static int pd_ioctl(struct block_device *bdev, fmode_t mode, switch (cmd) { case CDROMEJECT: - lock_kernel(); + mutex_lock(&pd_mutex); if (disk->access == 1) pd_special_command(disk, pd_eject); - unlock_kernel(); + mutex_unlock(&pd_mutex); return 0; default: return -EINVAL; @@ -785,10 +786,10 @@ static int pd_release(struct gendisk *p, fmode_t mode) { struct pd_unit *disk = p->private_data; - lock_kernel(); + mutex_lock(&pd_mutex); if (!--disk->access && disk->removable) pd_special_command(disk, pd_door_unlock); - unlock_kernel(); + mutex_unlock(&pd_mutex); return 0; } diff --git a/drivers/block/paride/pf.c b/drivers/block/paride/pf.c index 4457b49..635f25d 100644 --- a/drivers/block/paride/pf.c +++ b/drivers/block/paride/pf.c @@ -152,9 +152,10 @@ enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_LUN, D_DLY}; #include <linux/spinlock.h> #include <linux/blkdev.h> #include <linux/blkpg.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <asm/uaccess.h> +static DEFINE_MUTEX(pf_mutex); static DEFINE_SPINLOCK(pf_spin_lock); module_param(verbose, bool, 0644); @@ -302,7 +303,7 @@ static int pf_open(struct block_device *bdev, fmode_t mode) struct pf_unit *pf = bdev->bd_disk->private_data; int ret; - lock_kernel(); + mutex_lock(&pf_mutex); pf_identify(pf); ret = -ENODEV; @@ -318,7 +319,7 @@ static int pf_open(struct block_device *bdev, fmode_t mode) if (pf->removable) pf_lock(pf, 1); out: - unlock_kernel(); + mutex_unlock(&pf_mutex); return ret; } @@ -349,9 +350,9 @@ static int pf_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, u if (pf->access != 1) return -EBUSY; - lock_kernel(); + mutex_lock(&pf_mutex); pf_eject(pf); - unlock_kernel(); + mutex_unlock(&pf_mutex); return 0; } @@ -360,9 +361,9 @@ static int pf_release(struct gendisk *disk, fmode_t mode) { struct pf_unit *pf = disk->private_data; - lock_kernel(); + mutex_lock(&pf_mutex); if (pf->access <= 0) { - unlock_kernel(); + mutex_unlock(&pf_mutex); return -EINVAL; } @@ -371,7 +372,7 @@ static int pf_release(struct gendisk *disk, fmode_t mode) if (!pf->access && pf->removable) pf_lock(pf, 0); - unlock_kernel(); + mutex_unlock(&pf_mutex); return 0; } diff --git a/drivers/block/paride/pg.c b/drivers/block/paride/pg.c index c397b3d..6b9a200 100644 --- a/drivers/block/paride/pg.c +++ b/drivers/block/paride/pg.c @@ -162,7 +162,7 @@ enum {D_PRT, D_PRO, D_UNI, D_MOD, D_SLV, D_DLY}; #include <linux/pg.h> #include <linux/device.h> #include <linux/sched.h> /* current, TASK_* */ -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/jiffies.h> #include <asm/uaccess.h> @@ -193,6 +193,7 @@ module_param_array(drive3, int, NULL, 0); #define ATAPI_IDENTIFY 0x12 +static DEFINE_MUTEX(pg_mutex); static int pg_open(struct inode *inode, struct file *file); static int pg_release(struct inode *inode, struct file *file); static ssize_t pg_read(struct file *filp, char __user *buf, @@ -234,6 +235,7 @@ static const struct file_operations pg_fops = { .write = pg_write, .open = pg_open, .release = pg_release, + .llseek = noop_llseek, }; static void pg_init_units(void) @@ -518,7 +520,7 @@ static int pg_open(struct inode *inode, struct file *file) struct pg *dev = &devices[unit]; int ret = 0; - lock_kernel(); + mutex_lock(&pg_mutex); if ((unit >= PG_UNITS) || (!dev->present)) { ret = -ENODEV; goto out; @@ -547,7 +549,7 @@ static int pg_open(struct inode *inode, struct file *file) file->private_data = dev; out: - unlock_kernel(); + mutex_unlock(&pg_mutex); return ret; } diff --git a/drivers/block/paride/pt.c b/drivers/block/paride/pt.c index bc5825f..7179f79 100644 --- a/drivers/block/paride/pt.c +++ b/drivers/block/paride/pt.c @@ -146,7 +146,7 @@ static int (*drives[4])[6] = {&drive0, &drive1, &drive2, &drive3}; #include <linux/mtio.h> #include <linux/device.h> #include <linux/sched.h> /* current, TASK_*, schedule_timeout() */ -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <asm/uaccess.h> @@ -189,6 +189,7 @@ module_param_array(drive3, int, NULL, 0); #define ATAPI_MODE_SENSE 0x1a #define ATAPI_LOG_SENSE 0x4d +static DEFINE_MUTEX(pt_mutex); static int pt_open(struct inode *inode, struct file *file); static long pt_ioctl(struct file *file, unsigned int cmd, unsigned long arg); static int pt_release(struct inode *inode, struct file *file); @@ -239,6 +240,7 @@ static const struct file_operations pt_fops = { .unlocked_ioctl = pt_ioctl, .open = pt_open, .release = pt_release, + .llseek = noop_llseek, }; /* sysfs class support */ @@ -650,9 +652,9 @@ static int pt_open(struct inode *inode, struct file *file) struct pt_unit *tape = pt + unit; int err; - lock_kernel(); + mutex_lock(&pt_mutex); if (unit >= PT_UNITS || (!tape->present)) { - unlock_kernel(); + mutex_unlock(&pt_mutex); return -ENODEV; } @@ -681,12 +683,12 @@ static int pt_open(struct inode *inode, struct file *file) } file->private_data = tape; - unlock_kernel(); + mutex_unlock(&pt_mutex); return 0; out: atomic_inc(&tape->available); - unlock_kernel(); + mutex_unlock(&pt_mutex); return err; } @@ -704,15 +706,15 @@ static long pt_ioctl(struct file *file, unsigned int cmd, unsigned long arg) switch (mtop.mt_op) { case MTREW: - lock_kernel(); + mutex_lock(&pt_mutex); pt_rewind(tape); - unlock_kernel(); + mutex_unlock(&pt_mutex); return 0; case MTWEOF: - lock_kernel(); + mutex_lock(&pt_mutex); pt_write_fm(tape); - unlock_kernel(); + mutex_unlock(&pt_mutex); return 0; default: diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c index 37a2bb5..77d70ee 100644 --- a/drivers/block/pktcdvd.c +++ b/drivers/block/pktcdvd.c @@ -57,7 +57,6 @@ #include <linux/seq_file.h> #include <linux/miscdevice.h> #include <linux/freezer.h> -#include <linux/smp_lock.h> #include <linux/mutex.h> #include <linux/slab.h> #include <scsi/scsi_cmnd.h> @@ -86,6 +85,7 @@ #define ZONE(sector, pd) (((sector) + (pd)->offset) & ~((pd)->settings.size - 1)) +static DEFINE_MUTEX(pktcdvd_mutex); static struct pktcdvd_device *pkt_devs[MAX_WRITERS]; static struct proc_dir_entry *pkt_proc; static int pktdev_major; @@ -753,7 +753,6 @@ static int pkt_generic_packet(struct pktcdvd_device *pd, struct packet_command * rq->timeout = 60*HZ; rq->cmd_type = REQ_TYPE_BLOCK_PC; - rq->cmd_flags |= REQ_HARDBARRIER; if (cgc->quiet) rq->cmd_flags |= REQ_QUIET; @@ -2297,15 +2296,12 @@ static int pkt_open_dev(struct pktcdvd_device *pd, fmode_t write) * so bdget() can't fail. */ bdget(pd->bdev->bd_dev); - if ((ret = blkdev_get(pd->bdev, FMODE_READ))) + if ((ret = blkdev_get(pd->bdev, FMODE_READ | FMODE_EXCL, pd))) goto out; - if ((ret = bd_claim(pd->bdev, pd))) - goto out_putdev; - if ((ret = pkt_get_last_written(pd, &lba))) { printk(DRIVER_NAME": pkt_get_last_written failed\n"); - goto out_unclaim; + goto out_putdev; } set_capacity(pd->disk, lba << 2); @@ -2315,7 +2311,7 @@ static int pkt_open_dev(struct pktcdvd_device *pd, fmode_t write) q = bdev_get_queue(pd->bdev); if (write) { if ((ret = pkt_open_write(pd))) - goto out_unclaim; + goto out_putdev; /* * Some CDRW drives can not handle writes larger than one packet, * even if the size is a multiple of the packet size. @@ -2330,23 +2326,21 @@ static int pkt_open_dev(struct pktcdvd_device *pd, fmode_t write) } if ((ret = pkt_set_segment_merging(pd, q))) - goto out_unclaim; + goto out_putdev; if (write) { if (!pkt_grow_pktlist(pd, CONFIG_CDROM_PKTCDVD_BUFFERS)) { printk(DRIVER_NAME": not enough memory for buffers\n"); ret = -ENOMEM; - goto out_unclaim; + goto out_putdev; } printk(DRIVER_NAME": %lukB available on disc\n", lba << 1); } return 0; -out_unclaim: - bd_release(pd->bdev); out_putdev: - blkdev_put(pd->bdev, FMODE_READ); + blkdev_put(pd->bdev, FMODE_READ | FMODE_EXCL); out: return ret; } @@ -2363,8 +2357,7 @@ static void pkt_release_dev(struct pktcdvd_device *pd, int flush) pkt_lock_door(pd, 0); pkt_set_speed(pd, MAX_SPEED, MAX_SPEED); - bd_release(pd->bdev); - blkdev_put(pd->bdev, FMODE_READ); + blkdev_put(pd->bdev, FMODE_READ | FMODE_EXCL); pkt_shrink_pktlist(pd); } @@ -2383,7 +2376,7 @@ static int pkt_open(struct block_device *bdev, fmode_t mode) VPRINTK(DRIVER_NAME": entering open\n"); - lock_kernel(); + mutex_lock(&pktcdvd_mutex); mutex_lock(&ctl_mutex); pd = pkt_find_dev_from_minor(MINOR(bdev->bd_dev)); if (!pd) { @@ -2411,7 +2404,7 @@ static int pkt_open(struct block_device *bdev, fmode_t mode) } mutex_unlock(&ctl_mutex); - unlock_kernel(); + mutex_unlock(&pktcdvd_mutex); return 0; out_dec: @@ -2419,7 +2412,7 @@ out_dec: out: VPRINTK(DRIVER_NAME": failed open (%d)\n", ret); mutex_unlock(&ctl_mutex); - unlock_kernel(); + mutex_unlock(&pktcdvd_mutex); return ret; } @@ -2428,7 +2421,7 @@ static int pkt_close(struct gendisk *disk, fmode_t mode) struct pktcdvd_device *pd = disk->private_data; int ret = 0; - lock_kernel(); + mutex_lock(&pktcdvd_mutex); mutex_lock(&ctl_mutex); pd->refcnt--; BUG_ON(pd->refcnt < 0); @@ -2437,7 +2430,7 @@ static int pkt_close(struct gendisk *disk, fmode_t mode) pkt_release_dev(pd, flush); } mutex_unlock(&ctl_mutex); - unlock_kernel(); + mutex_unlock(&pktcdvd_mutex); return ret; } @@ -2734,7 +2727,7 @@ static int pkt_new_dev(struct pktcdvd_device *pd, dev_t dev) bdev = bdget(dev); if (!bdev) return -ENOMEM; - ret = blkdev_get(bdev, FMODE_READ | FMODE_NDELAY); + ret = blkdev_get(bdev, FMODE_READ | FMODE_NDELAY, NULL); if (ret) return ret; @@ -2773,7 +2766,7 @@ static int pkt_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, VPRINTK("pkt_ioctl: cmd %x, dev %d:%d\n", cmd, MAJOR(bdev->bd_dev), MINOR(bdev->bd_dev)); - lock_kernel(); + mutex_lock(&pktcdvd_mutex); switch (cmd) { case CDROMEJECT: /* @@ -2798,7 +2791,7 @@ static int pkt_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, VPRINTK(DRIVER_NAME": Unknown ioctl for %s (%x)\n", pd->name, cmd); ret = -ENOTTY; } - unlock_kernel(); + mutex_unlock(&pktcdvd_mutex); return ret; } @@ -3046,6 +3039,7 @@ static const struct file_operations pkt_ctl_fops = { .compat_ioctl = pkt_ctl_compat_ioctl, #endif .owner = THIS_MODULE, + .llseek = no_llseek, }; static struct miscdevice pkt_misc = { diff --git a/drivers/block/ps3disk.c b/drivers/block/ps3disk.c index 03688c2..8e1ce2e 100644 --- a/drivers/block/ps3disk.c +++ b/drivers/block/ps3disk.c @@ -468,7 +468,7 @@ static int __devinit ps3disk_probe(struct ps3_system_bus_device *_dev) blk_queue_dma_alignment(queue, dev->blk_size-1); blk_queue_logical_block_size(queue, dev->blk_size); - blk_queue_ordered(queue, QUEUE_ORDERED_DRAIN_FLUSH); + blk_queue_flush(queue, REQ_FLUSH); blk_queue_max_segments(queue, -1); blk_queue_max_segment_size(queue, dev->bounce_size); diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c new file mode 100644 index 0000000..e1e38b1 --- /dev/null +++ b/drivers/block/rbd.c @@ -0,0 +1,2046 @@ +/* + rbd.c -- Export ceph rados objects as a Linux block device + + + based on drivers/block/osdblk.c: + + Copyright 2009 Red Hat, Inc. + + 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. + + 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. + + You should have received a copy of the GNU General Public License + along with this program; see the file COPYING. If not, write to + the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. + + + + For usage instructions, please refer to: + + Documentation/ABI/testing/sysfs-bus-rbd + + */ + +#include <linux/ceph/libceph.h> +#include <linux/ceph/osd_client.h> +#include <linux/ceph/mon_client.h> +#include <linux/ceph/decode.h> + +#include <linux/kernel.h> +#include <linux/device.h> +#include <linux/module.h> +#include <linux/fs.h> +#include <linux/blkdev.h> + +#include "rbd_types.h" + +#define DRV_NAME "rbd" +#define DRV_NAME_LONG "rbd (rados block device)" + +#define RBD_MINORS_PER_MAJOR 256 /* max minors per blkdev */ + +#define RBD_MAX_MD_NAME_LEN (96 + sizeof(RBD_SUFFIX)) +#define RBD_MAX_POOL_NAME_LEN 64 +#define RBD_MAX_SNAP_NAME_LEN 32 +#define RBD_MAX_OPT_LEN 1024 + +#define RBD_SNAP_HEAD_NAME "-" + +#define DEV_NAME_LEN 32 + +/* + * block device image metadata (in-memory version) + */ +struct rbd_image_header { + u64 image_size; + char block_name[32]; + __u8 obj_order; + __u8 crypt_type; + __u8 comp_type; + struct rw_semaphore snap_rwsem; + struct ceph_snap_context *snapc; + size_t snap_names_len; + u64 snap_seq; + u32 total_snaps; + + char *snap_names; + u64 *snap_sizes; +}; + +/* + * an instance of the client. multiple devices may share a client. + */ +struct rbd_client { + struct ceph_client *client; + struct kref kref; + struct list_head node; +}; + +/* + * a single io request + */ +struct rbd_request { + struct request *rq; /* blk layer request */ + struct bio *bio; /* cloned bio */ + struct page **pages; /* list of used pages */ + u64 len; +}; + +struct rbd_snap { + struct device dev; + const char *name; + size_t size; + struct list_head node; + u64 id; +}; + +/* + * a single device + */ +struct rbd_device { + int id; /* blkdev unique id */ + + int major; /* blkdev assigned major */ + struct gendisk *disk; /* blkdev's gendisk and rq */ + struct request_queue *q; + + struct ceph_client *client; + struct rbd_client *rbd_client; + + char name[DEV_NAME_LEN]; /* blkdev name, e.g. rbd3 */ + + spinlock_t lock; /* queue lock */ + + struct rbd_image_header header; + char obj[RBD_MAX_OBJ_NAME_LEN]; /* rbd image name */ + int obj_len; + char obj_md_name[RBD_MAX_MD_NAME_LEN]; /* hdr nm. */ + char pool_name[RBD_MAX_POOL_NAME_LEN]; + int poolid; + + char snap_name[RBD_MAX_SNAP_NAME_LEN]; + u32 cur_snap; /* index+1 of current snapshot within snap context + 0 - for the head */ + int read_only; + + struct list_head node; + + /* list of snapshots */ + struct list_head snaps; + + /* sysfs related */ + struct device dev; +}; + +static struct bus_type rbd_bus_type = { + .name = "rbd", +}; + +static spinlock_t node_lock; /* protects client get/put */ + +static DEFINE_MUTEX(ctl_mutex); /* Serialize open/close/setup/teardown */ +static LIST_HEAD(rbd_dev_list); /* devices */ +static LIST_HEAD(rbd_client_list); /* clients */ + +static int __rbd_init_snaps_header(struct rbd_device *rbd_dev); +static void rbd_dev_release(struct device *dev); +static ssize_t rbd_snap_rollback(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t size); +static ssize_t rbd_snap_add(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count); +static void __rbd_remove_snap_dev(struct rbd_device *rbd_dev, + struct rbd_snap *snap);; + + +static struct rbd_device *dev_to_rbd(struct device *dev) +{ + return container_of(dev, struct rbd_device, dev); +} + +static struct device *rbd_get_dev(struct rbd_device *rbd_dev) +{ + return get_device(&rbd_dev->dev); +} + +static void rbd_put_dev(struct rbd_device *rbd_dev) +{ + put_device(&rbd_dev->dev); +} + +static int rbd_open(struct block_device *bdev, fmode_t mode) +{ + struct gendisk *disk = bdev->bd_disk; + struct rbd_device *rbd_dev = disk->private_data; + + rbd_get_dev(rbd_dev); + + set_device_ro(bdev, rbd_dev->read_only); + + if ((mode & FMODE_WRITE) && rbd_dev->read_only) + return -EROFS; + + return 0; +} + +static int rbd_release(struct gendisk *disk, fmode_t mode) +{ + struct rbd_device *rbd_dev = disk->private_data; + + rbd_put_dev(rbd_dev); + + return 0; +} + +static const struct block_device_operations rbd_bd_ops = { + .owner = THIS_MODULE, + .open = rbd_open, + .release = rbd_release, +}; + +/* + * Initialize an rbd client instance. + * We own *opt. + */ +static struct rbd_client *rbd_client_create(struct ceph_options *opt) +{ + struct rbd_client *rbdc; + int ret = -ENOMEM; + + dout("rbd_client_create\n"); + rbdc = kmalloc(sizeof(struct rbd_client), GFP_KERNEL); + if (!rbdc) + goto out_opt; + + kref_init(&rbdc->kref); + INIT_LIST_HEAD(&rbdc->node); + + rbdc->client = ceph_create_client(opt, rbdc); + if (IS_ERR(rbdc->client)) + goto out_rbdc; + opt = NULL; /* Now rbdc->client is responsible for opt */ + + ret = ceph_open_session(rbdc->client); + if (ret < 0) + goto out_err; + + spin_lock(&node_lock); + list_add_tail(&rbdc->node, &rbd_client_list); + spin_unlock(&node_lock); + + dout("rbd_client_create created %p\n", rbdc); + return rbdc; + +out_err: + ceph_destroy_client(rbdc->client); +out_rbdc: + kfree(rbdc); +out_opt: + if (opt) + ceph_destroy_options(opt); + return ERR_PTR(ret); +} + +/* + * Find a ceph client with specific addr and configuration. + */ +static struct rbd_client *__rbd_client_find(struct ceph_options *opt) +{ + struct rbd_client *client_node; + + if (opt->flags & CEPH_OPT_NOSHARE) + return NULL; + + list_for_each_entry(client_node, &rbd_client_list, node) + if (ceph_compare_options(opt, client_node->client) == 0) + return client_node; + return NULL; +} + +/* + * Get a ceph client with specific addr and configuration, if one does + * not exist create it. + */ +static int rbd_get_client(struct rbd_device *rbd_dev, const char *mon_addr, + char *options) +{ + struct rbd_client *rbdc; + struct ceph_options *opt; + int ret; + + ret = ceph_parse_options(&opt, options, mon_addr, + mon_addr + strlen(mon_addr), NULL, NULL); + if (ret < 0) + return ret; + + spin_lock(&node_lock); + rbdc = __rbd_client_find(opt); + if (rbdc) { + ceph_destroy_options(opt); + + /* using an existing client */ + kref_get(&rbdc->kref); + rbd_dev->rbd_client = rbdc; + rbd_dev->client = rbdc->client; + spin_unlock(&node_lock); + return 0; + } + spin_unlock(&node_lock); + + rbdc = rbd_client_create(opt); + if (IS_ERR(rbdc)) + return PTR_ERR(rbdc); + + rbd_dev->rbd_client = rbdc; + rbd_dev->client = rbdc->client; + return 0; +} + +/* + * Destroy ceph client + */ +static void rbd_client_release(struct kref *kref) +{ + struct rbd_client *rbdc = container_of(kref, struct rbd_client, kref); + + dout("rbd_release_client %p\n", rbdc); + spin_lock(&node_lock); + list_del(&rbdc->node); + spin_unlock(&node_lock); + + ceph_destroy_client(rbdc->client); + kfree(rbdc); +} + +/* + * Drop reference to ceph client node. If it's not referenced anymore, release + * it. + */ +static void rbd_put_client(struct rbd_device *rbd_dev) +{ + kref_put(&rbd_dev->rbd_client->kref, rbd_client_release); + rbd_dev->rbd_client = NULL; + rbd_dev->client = NULL; +} + + +/* + * Create a new header structure, translate header format from the on-disk + * header. + */ +static int rbd_header_from_disk(struct rbd_image_header *header, + struct rbd_image_header_ondisk *ondisk, + int allocated_snaps, + gfp_t gfp_flags) +{ + int i; + u32 snap_count = le32_to_cpu(ondisk->snap_count); + int ret = -ENOMEM; + + init_rwsem(&header->snap_rwsem); + header->snap_names_len = le64_to_cpu(ondisk->snap_names_len); + header->snapc = kmalloc(sizeof(struct ceph_snap_context) + + snap_count * + sizeof(struct rbd_image_snap_ondisk), + gfp_flags); + if (!header->snapc) + return -ENOMEM; + if (snap_count) { + header->snap_names = kmalloc(header->snap_names_len, + GFP_KERNEL); + if (!header->snap_names) + goto err_snapc; + header->snap_sizes = kmalloc(snap_count * sizeof(u64), + GFP_KERNEL); + if (!header->snap_sizes) + goto err_names; + } else { + header->snap_names = NULL; + header->snap_sizes = NULL; + } + memcpy(header->block_name, ondisk->block_name, + sizeof(ondisk->block_name)); + + header->image_size = le64_to_cpu(ondisk->image_size); + header->obj_order = ondisk->options.order; + header->crypt_type = ondisk->options.crypt_type; + header->comp_type = ondisk->options.comp_type; + + atomic_set(&header->snapc->nref, 1); + header->snap_seq = le64_to_cpu(ondisk->snap_seq); + header->snapc->num_snaps = snap_count; + header->total_snaps = snap_count; + + if (snap_count && + allocated_snaps == snap_count) { + for (i = 0; i < snap_count; i++) { + header->snapc->snaps[i] = + le64_to_cpu(ondisk->snaps[i].id); + header->snap_sizes[i] = + le64_to_cpu(ondisk->snaps[i].image_size); + } + + /* copy snapshot names */ + memcpy(header->snap_names, &ondisk->snaps[i], + header->snap_names_len); + } + + return 0; + +err_names: + kfree(header->snap_names); +err_snapc: + kfree(header->snapc); + return ret; +} + +static int snap_index(struct rbd_image_header *header, int snap_num) +{ + return header->total_snaps - snap_num; +} + +static u64 cur_snap_id(struct rbd_device *rbd_dev) +{ + struct rbd_image_header *header = &rbd_dev->header; + + if (!rbd_dev->cur_snap) + return 0; + + return header->snapc->snaps[snap_index(header, rbd_dev->cur_snap)]; +} + +static int snap_by_name(struct rbd_image_header *header, const char *snap_name, + u64 *seq, u64 *size) +{ + int i; + char *p = header->snap_names; + + for (i = 0; i < header->total_snaps; i++, p += strlen(p) + 1) { + if (strcmp(snap_name, p) == 0) + break; + } + if (i == header->total_snaps) + return -ENOENT; + if (seq) + *seq = header->snapc->snaps[i]; + + if (size) + *size = header->snap_sizes[i]; + + return i; +} + +static int rbd_header_set_snap(struct rbd_device *dev, + const char *snap_name, + u64 *size) +{ + struct rbd_image_header *header = &dev->header; + struct ceph_snap_context *snapc = header->snapc; + int ret = -ENOENT; + + down_write(&header->snap_rwsem); + + if (!snap_name || + !*snap_name || + strcmp(snap_name, "-") == 0 || + strcmp(snap_name, RBD_SNAP_HEAD_NAME) == 0) { + if (header->total_snaps) + snapc->seq = header->snap_seq; + else + snapc->seq = 0; + dev->cur_snap = 0; + dev->read_only = 0; + if (size) + *size = header->image_size; + } else { + ret = snap_by_name(header, snap_name, &snapc->seq, size); + if (ret < 0) + goto done; + + dev->cur_snap = header->total_snaps - ret; + dev->read_only = 1; + } + + ret = 0; +done: + up_write(&header->snap_rwsem); + return ret; +} + +static void rbd_header_free(struct rbd_image_header *header) +{ + kfree(header->snapc); + kfree(header->snap_names); + kfree(header->snap_sizes); +} + +/* + * get the actual striped segment name, offset and length + */ +static u64 rbd_get_segment(struct rbd_image_header *header, + const char *block_name, + u64 ofs, u64 len, + char *seg_name, u64 *segofs) +{ + u64 seg = ofs >> header->obj_order; + + if (seg_name) + snprintf(seg_name, RBD_MAX_SEG_NAME_LEN, + "%s.%012llx", block_name, seg); + + ofs = ofs & ((1 << header->obj_order) - 1); + len = min_t(u64, len, (1 << header->obj_order) - ofs); + + if (segofs) + *segofs = ofs; + + return len; +} + +/* + * bio helpers + */ + +static void bio_chain_put(struct bio *chain) +{ + struct bio *tmp; + + while (chain) { + tmp = chain; + chain = chain->bi_next; + bio_put(tmp); + } +} + +/* + * zeros a bio chain, starting at specific offset + */ +static void zero_bio_chain(struct bio *chain, int start_ofs) +{ + struct bio_vec *bv; + unsigned long flags; + void *buf; + int i; + int pos = 0; + + while (chain) { + bio_for_each_segment(bv, chain, i) { + if (pos + bv->bv_len > start_ofs) { + int remainder = max(start_ofs - pos, 0); + buf = bvec_kmap_irq(bv, &flags); + memset(buf + remainder, 0, + bv->bv_len - remainder); + bvec_kunmap_irq(buf, &flags); + } + pos += bv->bv_len; + } + + chain = chain->bi_next; + } +} + +/* + * bio_chain_clone - clone a chain of bios up to a certain length. + * might return a bio_pair that will need to be released. + */ +static struct bio *bio_chain_clone(struct bio **old, struct bio **next, + struct bio_pair **bp, + int len, gfp_t gfpmask) +{ + struct bio *tmp, *old_chain = *old, *new_chain = NULL, *tail = NULL; + int total = 0; + + if (*bp) { + bio_pair_release(*bp); + *bp = NULL; + } + + while (old_chain && (total < len)) { + tmp = bio_kmalloc(gfpmask, old_chain->bi_max_vecs); + if (!tmp) + goto err_out; + + if (total + old_chain->bi_size > len) { + struct bio_pair *bp; + + /* + * this split can only happen with a single paged bio, + * split_bio will BUG_ON if this is not the case + */ + dout("bio_chain_clone split! total=%d remaining=%d" + "bi_size=%d\n", + (int)total, (int)len-total, + (int)old_chain->bi_size); + + /* split the bio. We'll release it either in the next + call, or it will have to be released outside */ + bp = bio_split(old_chain, (len - total) / 512ULL); + if (!bp) + goto err_out; + + __bio_clone(tmp, &bp->bio1); + + *next = &bp->bio2; + } else { + __bio_clone(tmp, old_chain); + *next = old_chain->bi_next; + } + + tmp->bi_bdev = NULL; + gfpmask &= ~__GFP_WAIT; + tmp->bi_next = NULL; + + if (!new_chain) { + new_chain = tail = tmp; + } else { + tail->bi_next = tmp; + tail = tmp; + } + old_chain = old_chain->bi_next; + + total += tmp->bi_size; + } + + BUG_ON(total < len); + + if (tail) + tail->bi_next = NULL; + + *old = old_chain; + + return new_chain; + +err_out: + dout("bio_chain_clone with err\n"); + bio_chain_put(new_chain); + return NULL; +} + +/* + * helpers for osd request op vectors. + */ +static int rbd_create_rw_ops(struct ceph_osd_req_op **ops, + int num_ops, + int opcode, + u32 payload_len) +{ + *ops = kzalloc(sizeof(struct ceph_osd_req_op) * (num_ops + 1), + GFP_NOIO); + if (!*ops) + return -ENOMEM; + (*ops)[0].op = opcode; + /* + * op extent offset and length will be set later on + * in calc_raw_layout() + */ + (*ops)[0].payload_len = payload_len; + return 0; +} + +static void rbd_destroy_ops(struct ceph_osd_req_op *ops) +{ + kfree(ops); +} + +/* + * Send ceph osd request + */ +static int rbd_do_request(struct request *rq, + struct rbd_device *dev, + struct ceph_snap_context *snapc, + u64 snapid, + const char *obj, u64 ofs, u64 len, + struct bio *bio, + struct page **pages, + int num_pages, + int flags, + struct ceph_osd_req_op *ops, + int num_reply, + void (*rbd_cb)(struct ceph_osd_request *req, + struct ceph_msg *msg)) +{ + struct ceph_osd_request *req; + struct ceph_file_layout *layout; + int ret; + u64 bno; + struct timespec mtime = CURRENT_TIME; + struct rbd_request *req_data; + 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; + + dout("rbd_do_request len=%lld ofs=%lld\n", len, ofs); + + down_read(&header->snap_rwsem); + + req = ceph_osdc_alloc_request(&dev->client->osdc, flags, + snapc, + ops, + false, + GFP_NOIO, pages, bio); + if (IS_ERR(req)) { + up_read(&header->snap_rwsem); + ret = PTR_ERR(req); + goto done_pages; + } + + req->r_callback = rbd_cb; + + req_data->rq = rq; + req_data->bio = bio; + req_data->pages = pages; + req_data->len = len; + + req->r_priv = req_data; + + reqhead = req->r_request->front.iov_base; + reqhead->snapid = cpu_to_le64(CEPH_NOSNAP); + + strncpy(req->r_oid, obj, sizeof(req->r_oid)); + req->r_oid_len = strlen(req->r_oid); + + layout = &req->r_file_layout; + memset(layout, 0, sizeof(*layout)); + layout->fl_stripe_unit = cpu_to_le32(1 << RBD_MAX_OBJ_ORDER); + layout->fl_stripe_count = cpu_to_le32(1); + layout->fl_object_size = cpu_to_le32(1 << RBD_MAX_OBJ_ORDER); + layout->fl_pg_preferred = cpu_to_le32(-1); + layout->fl_pg_pool = cpu_to_le32(dev->poolid); + ceph_calc_raw_layout(&dev->client->osdc, layout, snapid, + ofs, &len, &bno, req, ops); + + ceph_osdc_build_request(req, ofs, &len, + ops, + snapc, + &mtime, + req->r_oid, req->r_oid_len); + up_read(&header->snap_rwsem); + + 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); + ceph_osdc_put_request(req); + } + return ret; + +done_err: + bio_chain_put(req_data->bio); + ceph_osdc_put_request(req); +done_pages: + kfree(req_data); +done: + if (rq) + blk_end_request(rq, ret, len); + return ret; +} + +/* + * Ceph osd op callback + */ +static void rbd_req_cb(struct ceph_osd_request *req, struct ceph_msg *msg) +{ + struct rbd_request *req_data = req->r_priv; + struct ceph_osd_reply_head *replyhead; + struct ceph_osd_op *op; + __s32 rc; + u64 bytes; + int read_op; + + /* parse reply */ + replyhead = msg->front.iov_base; + WARN_ON(le32_to_cpu(replyhead->num_ops) == 0); + op = (void *)(replyhead + 1); + rc = le32_to_cpu(replyhead->result); + bytes = le64_to_cpu(op->extent.length); + read_op = (le32_to_cpu(op->op) == CEPH_OSD_OP_READ); + + dout("rbd_req_cb bytes=%lld readop=%d rc=%d\n", bytes, read_op, rc); + + if (rc == -ENOENT && read_op) { + zero_bio_chain(req_data->bio, 0); + rc = 0; + } else if (rc == 0 && read_op && bytes < req_data->len) { + zero_bio_chain(req_data->bio, bytes); + bytes = req_data->len; + } + + blk_end_request(req_data->rq, rc, bytes); + + if (req_data->bio) + bio_chain_put(req_data->bio); + + ceph_osdc_put_request(req); + kfree(req_data); +} + +/* + * Do a synchronous ceph osd operation + */ +static int rbd_req_sync_op(struct rbd_device *dev, + struct ceph_snap_context *snapc, + u64 snapid, + int opcode, + int flags, + struct ceph_osd_req_op *orig_ops, + int num_reply, + const char *obj, + u64 ofs, u64 len, + char *buf) +{ + int ret; + struct page **pages; + int num_pages; + struct ceph_osd_req_op *ops = orig_ops; + u32 payload_len; + + num_pages = calc_pages_for(ofs , len); + pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL); + if (IS_ERR(pages)) + return PTR_ERR(pages); + + if (!orig_ops) { + payload_len = (flags & CEPH_OSD_FLAG_WRITE ? len : 0); + ret = rbd_create_rw_ops(&ops, 1, opcode, payload_len); + if (ret < 0) + goto done; + + if ((flags & CEPH_OSD_FLAG_WRITE) && buf) { + ret = ceph_copy_to_page_vector(pages, buf, ofs, len); + if (ret < 0) + goto done_ops; + } + } + + ret = rbd_do_request(NULL, dev, snapc, snapid, + obj, ofs, len, NULL, + pages, num_pages, + flags, + ops, + 2, + NULL); + if (ret < 0) + goto done_ops; + + if ((flags & CEPH_OSD_FLAG_READ) && buf) + ret = ceph_copy_from_page_vector(pages, buf, ofs, ret); + +done_ops: + if (!orig_ops) + rbd_destroy_ops(ops); +done: + ceph_release_page_vector(pages, num_pages); + return ret; +} + +/* + * Do an asynchronous ceph osd operation + */ +static int rbd_do_op(struct request *rq, + struct rbd_device *rbd_dev , + struct ceph_snap_context *snapc, + u64 snapid, + int opcode, int flags, int num_reply, + u64 ofs, u64 len, + struct bio *bio) +{ + char *seg_name; + u64 seg_ofs; + u64 seg_len; + int ret; + struct ceph_osd_req_op *ops; + u32 payload_len; + + seg_name = kmalloc(RBD_MAX_SEG_NAME_LEN + 1, GFP_NOIO); + if (!seg_name) + return -ENOMEM; + + seg_len = rbd_get_segment(&rbd_dev->header, + rbd_dev->header.block_name, + ofs, len, + seg_name, &seg_ofs); + + payload_len = (flags & CEPH_OSD_FLAG_WRITE ? seg_len : 0); + + ret = rbd_create_rw_ops(&ops, 1, opcode, payload_len); + if (ret < 0) + goto done; + + /* we've taken care of segment sizes earlier when we + cloned the bios. We should never have a segment + truncated at this point */ + BUG_ON(seg_len < len); + + ret = rbd_do_request(rq, rbd_dev, snapc, snapid, + seg_name, seg_ofs, seg_len, + bio, + NULL, 0, + flags, + ops, + num_reply, + rbd_req_cb); +done: + kfree(seg_name); + return ret; +} + +/* + * Request async osd write + */ +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) +{ + 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); +} + +/* + * Request async osd read + */ +static int rbd_req_read(struct request *rq, + struct rbd_device *rbd_dev, + u64 snapid, + u64 ofs, u64 len, + struct bio *bio) +{ + return rbd_do_op(rq, rbd_dev, NULL, + (snapid ? snapid : CEPH_NOSNAP), + CEPH_OSD_OP_READ, + CEPH_OSD_FLAG_READ, + 2, + ofs, len, bio); +} + +/* + * Request sync osd read + */ +static int rbd_req_sync_read(struct rbd_device *dev, + struct ceph_snap_context *snapc, + u64 snapid, + const char *obj, + u64 ofs, u64 len, + char *buf) +{ + 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); +} + +/* + * Request sync osd read + */ +static int rbd_req_sync_rollback_obj(struct rbd_device *dev, + u64 snapid, + const char *obj) +{ + struct ceph_osd_req_op *ops; + int ret = rbd_create_rw_ops(&ops, 1, CEPH_OSD_OP_ROLLBACK, 0); + if (ret < 0) + return ret; + + ops[0].snap.snapid = snapid; + + ret = rbd_req_sync_op(dev, NULL, + CEPH_NOSNAP, + 0, + CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, + ops, + 1, obj, 0, 0, NULL); + + rbd_destroy_ops(ops); + + if (ret < 0) + return ret; + + return ret; +} + +/* + * Request sync osd read + */ +static int rbd_req_sync_exec(struct rbd_device *dev, + const char *obj, + const char *cls, + const char *method, + const char *data, + int len) +{ + struct ceph_osd_req_op *ops; + int cls_len = strlen(cls); + int method_len = strlen(method); + int ret = rbd_create_rw_ops(&ops, 1, CEPH_OSD_OP_CALL, + cls_len + method_len + len); + if (ret < 0) + return ret; + + ops[0].cls.class_name = cls; + ops[0].cls.class_len = (__u8)cls_len; + ops[0].cls.method_name = method; + ops[0].cls.method_len = (__u8)method_len; + ops[0].cls.argc = 0; + ops[0].cls.indata = data; + ops[0].cls.indata_len = len; + + ret = rbd_req_sync_op(dev, NULL, + CEPH_NOSNAP, + 0, + CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK, + ops, + 1, obj, 0, 0, NULL); + + rbd_destroy_ops(ops); + + dout("cls_exec returned %d\n", ret); + return ret; +} + +/* + * block device queue callback + */ +static void rbd_rq_fn(struct request_queue *q) +{ + struct rbd_device *rbd_dev = q->queuedata; + struct request *rq; + struct bio_pair *bp = NULL; + + rq = blk_fetch_request(q); + + while (1) { + struct bio *bio; + struct bio *rq_bio, *next_bio = NULL; + bool do_write; + int size, op_size = 0; + u64 ofs; + + /* peek at request from block layer */ + if (!rq) + break; + + dout("fetched request\n"); + + /* filter out block requests we don't understand */ + if ((rq->cmd_type != REQ_TYPE_FS)) { + __blk_end_request_all(rq, 0); + goto next; + } + + /* deduce our operation (read, write) */ + do_write = (rq_data_dir(rq) == WRITE); + + size = blk_rq_bytes(rq); + ofs = blk_rq_pos(rq) * 512ULL; + rq_bio = rq->bio; + if (do_write && rbd_dev->read_only) { + __blk_end_request_all(rq, -EROFS); + goto next; + } + + spin_unlock_irq(q->queue_lock); + + dout("%s 0x%x bytes at 0x%llx\n", + do_write ? "write" : "read", + size, blk_rq_pos(rq) * 512ULL); + + do { + /* a bio clone to be passed down to OSD req */ + dout("rq->bio->bi_vcnt=%d\n", rq->bio->bi_vcnt); + op_size = rbd_get_segment(&rbd_dev->header, + rbd_dev->header.block_name, + ofs, size, + NULL, NULL); + 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; + } + + /* init OSD command: write or read */ + if (do_write) + rbd_req_write(rq, rbd_dev, + rbd_dev->header.snapc, + ofs, + op_size, bio); + else + rbd_req_read(rq, rbd_dev, + cur_snap_id(rbd_dev), + ofs, + op_size, bio); + + size -= op_size; + ofs += op_size; + + rq_bio = next_bio; + } while (size > 0); + + if (bp) + bio_pair_release(bp); + + spin_lock_irq(q->queue_lock); +next: + rq = blk_fetch_request(q); + } +} + +/* + * a queue callback. Makes sure that we don't create a bio that spans across + * multiple osd objects. One exception would be with a single page bios, + * which we handle later at bio_chain_clone + */ +static int rbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bmd, + struct bio_vec *bvec) +{ + struct rbd_device *rbd_dev = q->queuedata; + unsigned int chunk_sectors = 1 << (rbd_dev->header.obj_order - 9); + sector_t sector = bmd->bi_sector + get_start_sect(bmd->bi_bdev); + unsigned int bio_sectors = bmd->bi_size >> 9; + int max; + + max = (chunk_sectors - ((sector & (chunk_sectors - 1)) + + bio_sectors)) << 9; + if (max < 0) + max = 0; /* bio_add cannot handle a negative return */ + if (max <= bvec->bv_len && bio_sectors == 0) + return bvec->bv_len; + return max; +} + +static void rbd_free_disk(struct rbd_device *rbd_dev) +{ + struct gendisk *disk = rbd_dev->disk; + + if (!disk) + return; + + rbd_header_free(&rbd_dev->header); + + if (disk->flags & GENHD_FL_UP) + del_gendisk(disk); + if (disk->queue) + blk_cleanup_queue(disk->queue); + put_disk(disk); +} + +/* + * reload the ondisk the header + */ +static int rbd_read_header(struct rbd_device *rbd_dev, + struct rbd_image_header *header) +{ + ssize_t rc; + struct rbd_image_header_ondisk *dh; + int snap_count = 0; + u64 snap_names_len = 0; + + while (1) { + int len = sizeof(*dh) + + snap_count * sizeof(struct rbd_image_snap_ondisk) + + snap_names_len; + + rc = -ENOMEM; + dh = kmalloc(len, GFP_KERNEL); + if (!dh) + return -ENOMEM; + + rc = rbd_req_sync_read(rbd_dev, + NULL, CEPH_NOSNAP, + rbd_dev->obj_md_name, + 0, len, + (char *)dh); + if (rc < 0) + goto out_dh; + + rc = rbd_header_from_disk(header, dh, snap_count, GFP_KERNEL); + if (rc < 0) + goto out_dh; + + if (snap_count != header->total_snaps) { + snap_count = header->total_snaps; + snap_names_len = header->snap_names_len; + rbd_header_free(header); + kfree(dh); + continue; + } + break; + } + +out_dh: + kfree(dh); + return rc; +} + +/* + * create a snapshot + */ +static int rbd_header_add_snap(struct rbd_device *dev, + const char *snap_name, + gfp_t gfp_flags) +{ + int name_len = strlen(snap_name); + u64 new_snapid; + int ret; + void *data, *data_start, *data_end; + + /* we should create a snapshot only if we're pointing at the head */ + if (dev->cur_snap) + return -EINVAL; + + ret = ceph_monc_create_snapid(&dev->client->monc, dev->poolid, + &new_snapid); + dout("created snapid=%lld\n", new_snapid); + if (ret < 0) + return ret; + + data = kmalloc(name_len + 16, gfp_flags); + if (!data) + return -ENOMEM; + + data_start = data; + data_end = 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); + + ret = rbd_req_sync_exec(dev, dev->obj_md_name, "rbd", "snap_add", + data_start, data - data_start); + + kfree(data_start); + + if (ret < 0) + return ret; + + dev->header.snapc->seq = new_snapid; + + return 0; +bad: + return -ERANGE; +} + +static void __rbd_remove_all_snaps(struct rbd_device *rbd_dev) +{ + struct rbd_snap *snap; + + while (!list_empty(&rbd_dev->snaps)) { + snap = list_first_entry(&rbd_dev->snaps, struct rbd_snap, node); + __rbd_remove_snap_dev(rbd_dev, snap); + } +} + +/* + * only read the first part of the ondisk header, without the snaps info + */ +static int __rbd_update_snaps(struct rbd_device *rbd_dev) +{ + int ret; + struct rbd_image_header h; + u64 snap_seq; + + ret = rbd_read_header(rbd_dev, &h); + if (ret < 0) + return ret; + + down_write(&rbd_dev->header.snap_rwsem); + + snap_seq = rbd_dev->header.snapc->seq; + + kfree(rbd_dev->header.snapc); + kfree(rbd_dev->header.snap_names); + kfree(rbd_dev->header.snap_sizes); + + rbd_dev->header.total_snaps = h.total_snaps; + rbd_dev->header.snapc = h.snapc; + 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; + + ret = __rbd_init_snaps_header(rbd_dev); + + up_write(&rbd_dev->header.snap_rwsem); + + return ret; +} + +static int rbd_init_disk(struct rbd_device *rbd_dev) +{ + struct gendisk *disk; + struct request_queue *q; + int rc; + u64 total_size = 0; + + /* contact OSD, request size info about the object being mapped */ + rc = rbd_read_header(rbd_dev, &rbd_dev->header); + if (rc) + return rc; + + /* no need to lock here, as rbd_dev is not registered yet */ + rc = __rbd_init_snaps_header(rbd_dev); + if (rc) + return rc; + + rc = rbd_header_set_snap(rbd_dev, rbd_dev->snap_name, &total_size); + if (rc) + return rc; + + /* create gendisk info */ + rc = -ENOMEM; + disk = alloc_disk(RBD_MINORS_PER_MAJOR); + if (!disk) + goto out; + + sprintf(disk->disk_name, DRV_NAME "%d", rbd_dev->id); + disk->major = rbd_dev->major; + disk->first_minor = 0; + disk->fops = &rbd_bd_ops; + disk->private_data = rbd_dev; + + /* init rq */ + rc = -ENOMEM; + q = blk_init_queue(rbd_rq_fn, &rbd_dev->lock); + if (!q) + goto out_disk; + blk_queue_merge_bvec(q, rbd_merge_bvec); + disk->queue = q; + + q->queuedata = rbd_dev; + + rbd_dev->disk = disk; + rbd_dev->q = q; + + /* finally, announce the disk to the world */ + set_capacity(disk, total_size / 512ULL); + add_disk(disk); + + pr_info("%s: added with size 0x%llx\n", + disk->disk_name, (unsigned long long)total_size); + return 0; + +out_disk: + put_disk(disk); +out: + return rc; +} + +/* + sysfs +*/ + +static ssize_t rbd_size_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + + return sprintf(buf, "%llu\n", (unsigned long long)rbd_dev->header.image_size); +} + +static ssize_t rbd_major_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + + return sprintf(buf, "%d\n", rbd_dev->major); +} + +static ssize_t rbd_client_id_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + + return sprintf(buf, "client%lld\n", ceph_client_id(rbd_dev->client)); +} + +static ssize_t rbd_pool_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + + return sprintf(buf, "%s\n", rbd_dev->pool_name); +} + +static ssize_t rbd_name_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + + return sprintf(buf, "%s\n", rbd_dev->obj); +} + +static ssize_t rbd_snap_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + + return sprintf(buf, "%s\n", rbd_dev->snap_name); +} + +static ssize_t rbd_image_refresh(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t size) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + int rc; + int ret = size; + + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + + rc = __rbd_update_snaps(rbd_dev); + if (rc < 0) + ret = rc; + + mutex_unlock(&ctl_mutex); + return ret; +} + +static DEVICE_ATTR(size, S_IRUGO, rbd_size_show, NULL); +static DEVICE_ATTR(major, S_IRUGO, rbd_major_show, NULL); +static DEVICE_ATTR(client_id, S_IRUGO, rbd_client_id_show, NULL); +static DEVICE_ATTR(pool, S_IRUGO, rbd_pool_show, NULL); +static DEVICE_ATTR(name, S_IRUGO, rbd_name_show, NULL); +static DEVICE_ATTR(refresh, S_IWUSR, NULL, rbd_image_refresh); +static DEVICE_ATTR(current_snap, S_IRUGO, rbd_snap_show, NULL); +static DEVICE_ATTR(create_snap, S_IWUSR, NULL, rbd_snap_add); +static DEVICE_ATTR(rollback_snap, S_IWUSR, NULL, rbd_snap_rollback); + +static struct attribute *rbd_attrs[] = { + &dev_attr_size.attr, + &dev_attr_major.attr, + &dev_attr_client_id.attr, + &dev_attr_pool.attr, + &dev_attr_name.attr, + &dev_attr_current_snap.attr, + &dev_attr_refresh.attr, + &dev_attr_create_snap.attr, + &dev_attr_rollback_snap.attr, + NULL +}; + +static struct attribute_group rbd_attr_group = { + .attrs = rbd_attrs, +}; + +static const struct attribute_group *rbd_attr_groups[] = { + &rbd_attr_group, + NULL +}; + +static void rbd_sysfs_dev_release(struct device *dev) +{ +} + +static struct device_type rbd_device_type = { + .name = "rbd", + .groups = rbd_attr_groups, + .release = rbd_sysfs_dev_release, +}; + + +/* + sysfs - snapshots +*/ + +static ssize_t rbd_snap_size_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev); + + return sprintf(buf, "%lld\n", (long long)snap->size); +} + +static ssize_t rbd_snap_id_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev); + + return sprintf(buf, "%lld\n", (long long)snap->id); +} + +static DEVICE_ATTR(snap_size, S_IRUGO, rbd_snap_size_show, NULL); +static DEVICE_ATTR(snap_id, S_IRUGO, rbd_snap_id_show, NULL); + +static struct attribute *rbd_snap_attrs[] = { + &dev_attr_snap_size.attr, + &dev_attr_snap_id.attr, + NULL, +}; + +static struct attribute_group rbd_snap_attr_group = { + .attrs = rbd_snap_attrs, +}; + +static void rbd_snap_dev_release(struct device *dev) +{ + struct rbd_snap *snap = container_of(dev, struct rbd_snap, dev); + kfree(snap->name); + kfree(snap); +} + +static const struct attribute_group *rbd_snap_attr_groups[] = { + &rbd_snap_attr_group, + NULL +}; + +static struct device_type rbd_snap_device_type = { + .groups = rbd_snap_attr_groups, + .release = rbd_snap_dev_release, +}; + +static void __rbd_remove_snap_dev(struct rbd_device *rbd_dev, + struct rbd_snap *snap) +{ + list_del(&snap->node); + device_unregister(&snap->dev); +} + +static int rbd_register_snap_dev(struct rbd_device *rbd_dev, + struct rbd_snap *snap, + struct device *parent) +{ + struct device *dev = &snap->dev; + int ret; + + dev->type = &rbd_snap_device_type; + dev->parent = parent; + dev->release = rbd_snap_dev_release; + dev_set_name(dev, "snap_%s", snap->name); + ret = device_register(dev); + + return ret; +} + +static int __rbd_add_snap_dev(struct rbd_device *rbd_dev, + int i, const char *name, + struct rbd_snap **snapp) +{ + int ret; + struct rbd_snap *snap = kzalloc(sizeof(*snap), GFP_KERNEL); + if (!snap) + return -ENOMEM; + snap->name = kstrdup(name, GFP_KERNEL); + snap->size = rbd_dev->header.snap_sizes[i]; + snap->id = rbd_dev->header.snapc->snaps[i]; + if (device_is_registered(&rbd_dev->dev)) { + ret = rbd_register_snap_dev(rbd_dev, snap, + &rbd_dev->dev); + if (ret < 0) + goto err; + } + *snapp = snap; + return 0; +err: + kfree(snap->name); + kfree(snap); + return ret; +} + +/* + * search for the previous snap in a null delimited string list + */ +const char *rbd_prev_snap_name(const char *name, const char *start) +{ + if (name < start + 2) + return NULL; + + name -= 2; + while (*name) { + if (name == start) + return start; + name--; + } + return name + 1; +} + +/* + * compare the old list of snapshots that we have to what's in the header + * and update it accordingly. Note that the header holds the snapshots + * in a reverse order (from newest to oldest) and we need to go from + * older to new so that we don't get a duplicate snap name when + * doing the process (e.g., removed snapshot and recreated a new + * one with the same name. + */ +static int __rbd_init_snaps_header(struct rbd_device *rbd_dev) +{ + const char *name, *first_name; + int i = rbd_dev->header.total_snaps; + struct rbd_snap *snap, *old_snap = NULL; + int ret; + struct list_head *p, *n; + + first_name = rbd_dev->header.snap_names; + name = first_name + rbd_dev->header.snap_names_len; + + list_for_each_prev_safe(p, n, &rbd_dev->snaps) { + u64 cur_id; + + old_snap = list_entry(p, struct rbd_snap, node); + + if (i) + cur_id = rbd_dev->header.snapc->snaps[i - 1]; + + if (!i || old_snap->id < cur_id) { + /* old_snap->id was skipped, thus was removed */ + __rbd_remove_snap_dev(rbd_dev, old_snap); + continue; + } + if (old_snap->id == cur_id) { + /* we have this snapshot already */ + i--; + name = rbd_prev_snap_name(name, first_name); + continue; + } + for (; i > 0; + i--, name = rbd_prev_snap_name(name, first_name)) { + if (!name) { + WARN_ON(1); + return -EINVAL; + } + cur_id = rbd_dev->header.snapc->snaps[i]; + /* snapshot removal? handle it above */ + if (cur_id >= old_snap->id) + break; + /* a new snapshot */ + ret = __rbd_add_snap_dev(rbd_dev, i - 1, name, &snap); + if (ret < 0) + return ret; + + /* note that we add it backward so using n and not p */ + list_add(&snap->node, n); + p = &snap->node; + } + } + /* we're done going over the old snap list, just add what's left */ + for (; i > 0; i--) { + name = rbd_prev_snap_name(name, first_name); + if (!name) { + WARN_ON(1); + return -EINVAL; + } + ret = __rbd_add_snap_dev(rbd_dev, i - 1, name, &snap); + if (ret < 0) + return ret; + list_add(&snap->node, &rbd_dev->snaps); + } + + return 0; +} + + +static void rbd_root_dev_release(struct device *dev) +{ +} + +static struct device rbd_root_dev = { + .init_name = "rbd", + .release = rbd_root_dev_release, +}; + +static int rbd_bus_add_dev(struct rbd_device *rbd_dev) +{ + int ret = -ENOMEM; + struct device *dev; + struct rbd_snap *snap; + + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + dev = &rbd_dev->dev; + + dev->bus = &rbd_bus_type; + dev->type = &rbd_device_type; + dev->parent = &rbd_root_dev; + dev->release = rbd_dev_release; + dev_set_name(dev, "%d", rbd_dev->id); + ret = device_register(dev); + if (ret < 0) + goto done_free; + + list_for_each_entry(snap, &rbd_dev->snaps, node) { + ret = rbd_register_snap_dev(rbd_dev, snap, + &rbd_dev->dev); + if (ret < 0) + break; + } + + mutex_unlock(&ctl_mutex); + return 0; +done_free: + mutex_unlock(&ctl_mutex); + return ret; +} + +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) +{ + struct ceph_osd_client *osdc; + struct rbd_device *rbd_dev; + ssize_t rc = -ENOMEM; + int irc, new_id = 0; + struct list_head *tmp; + char *mon_dev_name; + char *options; + + if (!try_module_get(THIS_MODULE)) + return -ENODEV; + + mon_dev_name = kmalloc(RBD_MAX_OPT_LEN, GFP_KERNEL); + if (!mon_dev_name) + goto err_out_mod; + + options = kmalloc(RBD_MAX_OPT_LEN, GFP_KERNEL); + if (!options) + goto err_mon_dev; + + /* new rbd_device object */ + rbd_dev = kzalloc(sizeof(*rbd_dev), GFP_KERNEL); + if (!rbd_dev) + goto err_out_opt; + + /* static rbd_device initialization */ + spin_lock_init(&rbd_dev->lock); + INIT_LIST_HEAD(&rbd_dev->node); + INIT_LIST_HEAD(&rbd_dev->snaps); + + /* generate unique id: find highest unique id, add one */ + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + + list_for_each(tmp, &rbd_dev_list) { + struct rbd_device *rbd_dev; + + rbd_dev = list_entry(tmp, struct rbd_device, node); + if (rbd_dev->id >= new_id) + new_id = rbd_dev->id + 1; + } + + rbd_dev->id = new_id; + + /* add to global list */ + list_add_tail(&rbd_dev->node, &rbd_dev_list); + + /* parse add command */ + if (sscanf(buf, "%" __stringify(RBD_MAX_OPT_LEN) "s " + "%" __stringify(RBD_MAX_OPT_LEN) "s " + "%" __stringify(RBD_MAX_POOL_NAME_LEN) "s " + "%" __stringify(RBD_MAX_OBJ_NAME_LEN) "s" + "%" __stringify(RBD_MAX_SNAP_NAME_LEN) "s", + mon_dev_name, options, rbd_dev->pool_name, + rbd_dev->obj, rbd_dev->snap_name) < 4) { + rc = -EINVAL; + goto err_out_slot; + } + + if (rbd_dev->snap_name[0] == 0) + rbd_dev->snap_name[0] = '-'; + + rbd_dev->obj_len = strlen(rbd_dev->obj); + snprintf(rbd_dev->obj_md_name, sizeof(rbd_dev->obj_md_name), "%s%s", + rbd_dev->obj, RBD_SUFFIX); + + /* initialize rest of new object */ + snprintf(rbd_dev->name, DEV_NAME_LEN, DRV_NAME "%d", rbd_dev->id); + rc = rbd_get_client(rbd_dev, mon_dev_name, options); + if (rc < 0) + goto err_out_slot; + + mutex_unlock(&ctl_mutex); + + /* pick the pool */ + osdc = &rbd_dev->client->osdc; + rc = ceph_pg_poolid_by_name(osdc->osdmap, rbd_dev->pool_name); + if (rc < 0) + goto err_out_client; + rbd_dev->poolid = rc; + + /* register our block device */ + irc = register_blkdev(0, rbd_dev->name); + if (irc < 0) { + rc = irc; + goto err_out_client; + } + rbd_dev->major = irc; + + rc = rbd_bus_add_dev(rbd_dev); + if (rc) + goto err_out_blkdev; + + /* set up and announce blkdev mapping */ + rc = rbd_init_disk(rbd_dev); + if (rc) + goto err_out_bus; + + return count; + +err_out_bus: + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + list_del_init(&rbd_dev->node); + mutex_unlock(&ctl_mutex); + + /* this will also clean up rest of rbd_dev stuff */ + + rbd_bus_del_dev(rbd_dev); + kfree(options); + kfree(mon_dev_name); + return rc; + +err_out_blkdev: + unregister_blkdev(rbd_dev->major, rbd_dev->name); +err_out_client: + rbd_put_client(rbd_dev); + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); +err_out_slot: + list_del_init(&rbd_dev->node); + mutex_unlock(&ctl_mutex); + + kfree(rbd_dev); +err_out_opt: + kfree(options); +err_mon_dev: + kfree(mon_dev_name); +err_out_mod: + dout("Error adding device %s\n", buf); + module_put(THIS_MODULE); + return rc; +} + +static struct rbd_device *__rbd_get_dev(unsigned long id) +{ + struct list_head *tmp; + struct rbd_device *rbd_dev; + + list_for_each(tmp, &rbd_dev_list) { + rbd_dev = list_entry(tmp, struct rbd_device, node); + if (rbd_dev->id == id) + return rbd_dev; + } + return NULL; +} + +static void rbd_dev_release(struct device *dev) +{ + struct rbd_device *rbd_dev = + container_of(dev, struct rbd_device, dev); + + rbd_put_client(rbd_dev); + + /* clean up and free blkdev */ + rbd_free_disk(rbd_dev); + unregister_blkdev(rbd_dev->major, rbd_dev->name); + kfree(rbd_dev); + + /* release module ref */ + module_put(THIS_MODULE); +} + +static ssize_t rbd_remove(struct bus_type *bus, + const char *buf, + size_t count) +{ + struct rbd_device *rbd_dev = NULL; + int target_id, rc; + unsigned long ul; + int ret = count; + + rc = strict_strtoul(buf, 10, &ul); + if (rc) + return rc; + + /* convert to int; abort if we lost anything in the conversion */ + target_id = (int) ul; + if (target_id != ul) + return -EINVAL; + + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + + rbd_dev = __rbd_get_dev(target_id); + if (!rbd_dev) { + ret = -ENOENT; + goto done; + } + + list_del_init(&rbd_dev->node); + + __rbd_remove_all_snaps(rbd_dev); + rbd_bus_del_dev(rbd_dev); + +done: + mutex_unlock(&ctl_mutex); + return ret; +} + +static ssize_t rbd_snap_add(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + int ret; + char *name = kmalloc(count + 1, GFP_KERNEL); + if (!name) + return -ENOMEM; + + snprintf(name, count, "%s", buf); + + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + + ret = rbd_header_add_snap(rbd_dev, + name, GFP_KERNEL); + if (ret < 0) + goto done_unlock; + + ret = __rbd_update_snaps(rbd_dev); + if (ret < 0) + goto done_unlock; + + ret = count; +done_unlock: + mutex_unlock(&ctl_mutex); + kfree(name); + return ret; +} + +static ssize_t rbd_snap_rollback(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count) +{ + struct rbd_device *rbd_dev = dev_to_rbd(dev); + int ret; + u64 snapid; + u64 cur_ofs; + char *seg_name = NULL; + char *snap_name = kmalloc(count + 1, GFP_KERNEL); + ret = -ENOMEM; + if (!snap_name) + return ret; + + /* parse snaps add command */ + snprintf(snap_name, count, "%s", buf); + seg_name = kmalloc(RBD_MAX_SEG_NAME_LEN + 1, GFP_NOIO); + if (!seg_name) + goto done; + + mutex_lock_nested(&ctl_mutex, SINGLE_DEPTH_NESTING); + + ret = snap_by_name(&rbd_dev->header, snap_name, &snapid, NULL); + if (ret < 0) + goto done_unlock; + + dout("snapid=%lld\n", snapid); + + cur_ofs = 0; + while (cur_ofs < rbd_dev->header.image_size) { + cur_ofs += rbd_get_segment(&rbd_dev->header, + rbd_dev->obj, + cur_ofs, (u64)-1, + seg_name, NULL); + dout("seg_name=%s\n", seg_name); + + ret = rbd_req_sync_rollback_obj(rbd_dev, snapid, seg_name); + if (ret < 0) + pr_warning("could not roll back obj %s err=%d\n", + seg_name, ret); + } + + ret = __rbd_update_snaps(rbd_dev); + if (ret < 0) + goto done_unlock; + + ret = count; + +done_unlock: + mutex_unlock(&ctl_mutex); +done: + kfree(seg_name); + kfree(snap_name); + + return ret; +} + +static struct bus_attribute rbd_bus_attrs[] = { + __ATTR(add, S_IWUSR, NULL, rbd_add), + __ATTR(remove, S_IWUSR, NULL, rbd_remove), + __ATTR_NULL +}; + +/* + * create control files in sysfs + * /sys/bus/rbd/... + */ +static int rbd_sysfs_init(void) +{ + int ret; + + rbd_bus_type.bus_attrs = rbd_bus_attrs; + + ret = bus_register(&rbd_bus_type); + if (ret < 0) + return ret; + + ret = device_register(&rbd_root_dev); + + return ret; +} + +static void rbd_sysfs_cleanup(void) +{ + device_unregister(&rbd_root_dev); + bus_unregister(&rbd_bus_type); +} + +int __init rbd_init(void) +{ + int rc; + + rc = rbd_sysfs_init(); + if (rc) + return rc; + spin_lock_init(&node_lock); + pr_info("loaded " DRV_NAME_LONG "\n"); + return 0; +} + +void __exit rbd_exit(void) +{ + rbd_sysfs_cleanup(); +} + +module_init(rbd_init); +module_exit(rbd_exit); + +MODULE_AUTHOR("Sage Weil <sage@newdream.net>"); +MODULE_AUTHOR("Yehuda Sadeh <yehuda@hq.newdream.net>"); +MODULE_DESCRIPTION("rados block device"); + +/* following authorship retained from original osdblk.c */ +MODULE_AUTHOR("Jeff Garzik <jeff@garzik.org>"); + +MODULE_LICENSE("GPL"); diff --git a/drivers/block/rbd_types.h b/drivers/block/rbd_types.h new file mode 100644 index 0000000..fc6c678 --- /dev/null +++ b/drivers/block/rbd_types.h @@ -0,0 +1,73 @@ +/* + * Ceph - scalable distributed file system + * + * Copyright (C) 2004-2010 Sage Weil <sage@newdream.net> + * + * This is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License version 2.1, as published by the Free Software + * Foundation. See file COPYING. + * + */ + +#ifndef CEPH_RBD_TYPES_H +#define CEPH_RBD_TYPES_H + +#include <linux/types.h> + +/* + * rbd image 'foo' consists of objects + * foo.rbd - image metadata + * foo.00000000 + * foo.00000001 + * ... - data + */ + +#define RBD_SUFFIX ".rbd" +#define RBD_DIRECTORY "rbd_directory" +#define RBD_INFO "rbd_info" + +#define RBD_DEFAULT_OBJ_ORDER 22 /* 4MB */ +#define RBD_MIN_OBJ_ORDER 16 +#define RBD_MAX_OBJ_ORDER 30 + +#define RBD_MAX_OBJ_NAME_LEN 96 +#define RBD_MAX_SEG_NAME_LEN 128 + +#define RBD_COMP_NONE 0 +#define RBD_CRYPT_NONE 0 + +#define RBD_HEADER_TEXT "<<< Rados Block Device Image >>>\n" +#define RBD_HEADER_SIGNATURE "RBD" +#define RBD_HEADER_VERSION "001.005" + +struct rbd_info { + __le64 max_id; +} __attribute__ ((packed)); + +struct rbd_image_snap_ondisk { + __le64 id; + __le64 image_size; +} __attribute__((packed)); + +struct rbd_image_header_ondisk { + char text[40]; + char block_name[24]; + char signature[4]; + char version[8]; + struct { + __u8 order; + __u8 crypt_type; + __u8 comp_type; + __u8 unused; + } __attribute__((packed)) options; + __le64 image_size; + __le64 snap_seq; + __le32 snap_count; + __le32 reserved; + __le64 snap_names_len; + struct rbd_image_snap_ondisk snaps[0]; +} __attribute__((packed)); + + +#endif diff --git a/drivers/block/swim.c b/drivers/block/swim.c index 2e46815..75333d0 100644 --- a/drivers/block/swim.c +++ b/drivers/block/swim.c @@ -20,7 +20,7 @@ #include <linux/fd.h> #include <linux/slab.h> #include <linux/blkdev.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/hdreg.h> #include <linux/kernel.h> #include <linux/delay.h> @@ -222,6 +222,7 @@ extern int swim_read_sector_header(struct swim __iomem *base, extern int swim_read_sector_data(struct swim __iomem *base, unsigned char *data); +static DEFINE_MUTEX(swim_mutex); static inline void set_swim_mode(struct swim __iomem *base, int enable) { struct iwm __iomem *iwm_base; @@ -666,9 +667,9 @@ static int floppy_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&swim_mutex); ret = floppy_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&swim_mutex); return ret; } @@ -678,7 +679,7 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) struct floppy_state *fs = disk->private_data; struct swim __iomem *base = fs->swd->base; - lock_kernel(); + mutex_lock(&swim_mutex); if (fs->ref_count < 0) fs->ref_count = 0; else if (fs->ref_count > 0) @@ -686,7 +687,7 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) if (fs->ref_count == 0) swim_motor(base, OFF); - unlock_kernel(); + mutex_unlock(&swim_mutex); return 0; } @@ -704,9 +705,9 @@ static int floppy_ioctl(struct block_device *bdev, fmode_t mode, case FDEJECT: if (fs->ref_count != 1) return -EBUSY; - lock_kernel(); + mutex_lock(&swim_mutex); err = floppy_eject(fs); - unlock_kernel(); + mutex_unlock(&swim_mutex); return err; case FDGETPRM: diff --git a/drivers/block/swim3.c b/drivers/block/swim3.c index cc6a386..bf3a5b8 100644 --- a/drivers/block/swim3.c +++ b/drivers/block/swim3.c @@ -25,7 +25,7 @@ #include <linux/ioctl.h> #include <linux/blkdev.h> #include <linux/interrupt.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/module.h> #include <linux/spinlock.h> #include <asm/io.h> @@ -36,6 +36,7 @@ #include <asm/machdep.h> #include <asm/pmac_feature.h> +static DEFINE_MUTEX(swim3_mutex); static struct request_queue *swim3_queue; static struct gendisk *disks[2]; static struct request *fd_req; @@ -873,9 +874,9 @@ static int floppy_ioctl(struct block_device *bdev, fmode_t mode, { int ret; - lock_kernel(); + mutex_lock(&swim3_mutex); ret = floppy_locked_ioctl(bdev, mode, cmd, param); - unlock_kernel(); + mutex_unlock(&swim3_mutex); return ret; } @@ -953,9 +954,9 @@ static int floppy_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&swim3_mutex); ret = floppy_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&swim3_mutex); return ret; } @@ -964,13 +965,13 @@ static int floppy_release(struct gendisk *disk, fmode_t mode) { struct floppy_state *fs = disk->private_data; struct swim3 __iomem *sw = fs->swim3; - lock_kernel(); + mutex_lock(&swim3_mutex); if (fs->ref_count > 0 && --fs->ref_count == 0) { swim3_action(fs, MOTOR_OFF); out_8(&sw->control_bic, 0xff); swim3_select(fs, RELAX); } - unlock_kernel(); + mutex_unlock(&swim3_mutex); return 0; } diff --git a/drivers/block/ub.c b/drivers/block/ub.c index c48e148..9ae3bb7 100644 --- a/drivers/block/ub.c +++ b/drivers/block/ub.c @@ -28,7 +28,7 @@ #include <linux/timer.h> #include <linux/scatterlist.h> #include <linux/slab.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <scsi/scsi.h> #define DRV_NAME "ub" @@ -248,6 +248,7 @@ struct ub_completion { spinlock_t lock; }; +static DEFINE_MUTEX(ub_mutex); static inline void ub_init_completion(struct ub_completion *x) { x->done = 0; @@ -396,7 +397,7 @@ static int ub_probe_lun(struct ub_dev *sc, int lnum); #else static const struct usb_device_id ub_usb_ids[] = { - { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_SCSI, US_PR_BULK) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, USB_SC_SCSI, USB_PR_BULK) }, { } }; @@ -1715,9 +1716,9 @@ static int ub_bd_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&ub_mutex); ret = ub_bd_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&ub_mutex); return ret; } @@ -1730,9 +1731,9 @@ static int ub_bd_release(struct gendisk *disk, fmode_t mode) struct ub_lun *lun = disk->private_data; struct ub_dev *sc = lun->udev; - lock_kernel(); + mutex_lock(&ub_mutex); ub_put(sc); - unlock_kernel(); + mutex_unlock(&ub_mutex); return 0; } @@ -1747,9 +1748,9 @@ static int ub_bd_ioctl(struct block_device *bdev, fmode_t mode, void __user *usermem = (void __user *) arg; int ret; - lock_kernel(); + mutex_lock(&ub_mutex); ret = scsi_cmd_ioctl(disk->queue, disk, mode, cmd, usermem); - unlock_kernel(); + mutex_unlock(&ub_mutex); return ret; } diff --git a/drivers/block/viodasd.c b/drivers/block/viodasd.c index f651e51..e2ff697 100644 --- a/drivers/block/viodasd.c +++ b/drivers/block/viodasd.c @@ -41,7 +41,7 @@ #include <linux/errno.h> #include <linux/init.h> #include <linux/string.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/dma-mapping.h> #include <linux/completion.h> #include <linux/device.h> @@ -73,6 +73,7 @@ enum { MAX_DISK_NAME = FIELD_SIZEOF(struct gendisk, disk_name) }; +static DEFINE_MUTEX(viodasd_mutex); static DEFINE_SPINLOCK(viodasd_spinlock); #define VIOMAXREQ 16 @@ -180,9 +181,9 @@ static int viodasd_unlocked_open(struct block_device *bdev, fmode_t mode) { int ret; - lock_kernel(); + mutex_lock(&viodasd_mutex); ret = viodasd_open(bdev, mode); - unlock_kernel(); + mutex_unlock(&viodasd_mutex); return ret; } @@ -196,7 +197,7 @@ static int viodasd_release(struct gendisk *disk, fmode_t mode) struct viodasd_device *d = disk->private_data; HvLpEvent_Rc hvrc; - lock_kernel(); + mutex_lock(&viodasd_mutex); /* Send the event to OS/400. We DON'T expect a response */ hvrc = HvCallEvent_signalLpEventFast(viopath_hostLp, HvLpEvent_Type_VirtualIo, @@ -210,7 +211,7 @@ static int viodasd_release(struct gendisk *disk, fmode_t mode) if (hvrc != 0) pr_warning("HV close call failed %d\n", (int)hvrc); - unlock_kernel(); + mutex_unlock(&viodasd_mutex); return 0; } diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c index 1101e25..6ecf89c 100644 --- a/drivers/block/virtio_blk.c +++ b/drivers/block/virtio_blk.c @@ -2,7 +2,6 @@ #include <linux/spinlock.h> #include <linux/slab.h> #include <linux/blkdev.h> -#include <linux/smp_lock.h> #include <linux/hdreg.h> #include <linux/virtio.h> #include <linux/virtio_blk.h> @@ -128,9 +127,6 @@ static bool do_req(struct request_queue *q, struct virtio_blk *vblk, } } - if (vbr->req->cmd_flags & REQ_HARDBARRIER) - vbr->out_hdr.type |= VIRTIO_BLK_T_BARRIER; - sg_set_buf(&vblk->sg[out++], &vbr->out_hdr, sizeof(vbr->out_hdr)); /* @@ -222,8 +218,8 @@ static int virtblk_get_id(struct gendisk *disk, char *id_str) return err; } -static int virtblk_locked_ioctl(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long data) +static int virtblk_ioctl(struct block_device *bdev, fmode_t mode, + unsigned int cmd, unsigned long data) { struct gendisk *disk = bdev->bd_disk; struct virtio_blk *vblk = disk->private_data; @@ -238,18 +234,6 @@ static int virtblk_locked_ioctl(struct block_device *bdev, fmode_t mode, (void __user *)data); } -static int virtblk_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long param) -{ - int ret; - - lock_kernel(); - ret = virtblk_locked_ioctl(bdev, mode, cmd, param); - unlock_kernel(); - - return ret; -} - /* We provide getgeo only to please some old bootloader/partitioning tools */ static int virtblk_getgeo(struct block_device *bd, struct hd_geometry *geo) { @@ -392,31 +376,9 @@ static int __devinit virtblk_probe(struct virtio_device *vdev) vblk->disk->driverfs_dev = &vdev->dev; index++; - if (virtio_has_feature(vdev, VIRTIO_BLK_F_FLUSH)) { - /* - * If the FLUSH feature is supported we do have support for - * flushing a volatile write cache on the host. Use that - * to implement write barrier support. - */ - blk_queue_ordered(q, QUEUE_ORDERED_DRAIN_FLUSH); - } else if (virtio_has_feature(vdev, VIRTIO_BLK_F_BARRIER)) { - /* - * If the BARRIER feature is supported the host expects us - * to order request by tags. This implies there is not - * volatile write cache on the host, and that the host - * never re-orders outstanding I/O. This feature is not - * useful for real life scenarious and deprecated. - */ - blk_queue_ordered(q, QUEUE_ORDERED_TAG); - } else { - /* - * If the FLUSH feature is not supported we must assume that - * the host does not perform any kind of volatile write - * caching. We still need to drain the queue to provider - * proper barrier semantics. - */ - blk_queue_ordered(q, QUEUE_ORDERED_DRAIN); - } + /* configure queue flush support */ + if (virtio_has_feature(vdev, VIRTIO_BLK_F_FLUSH)) + blk_queue_flush(q, REQ_FLUSH); /* If disk is read-only in the host, the guest should obey */ if (virtio_has_feature(vdev, VIRTIO_BLK_F_RO)) @@ -535,9 +497,9 @@ static const struct virtio_device_id id_table[] = { }; static unsigned int features[] = { - VIRTIO_BLK_F_BARRIER, VIRTIO_BLK_F_SEG_MAX, VIRTIO_BLK_F_SIZE_MAX, - VIRTIO_BLK_F_GEOMETRY, VIRTIO_BLK_F_RO, VIRTIO_BLK_F_BLK_SIZE, - VIRTIO_BLK_F_SCSI, VIRTIO_BLK_F_FLUSH, VIRTIO_BLK_F_TOPOLOGY + VIRTIO_BLK_F_SEG_MAX, VIRTIO_BLK_F_SIZE_MAX, VIRTIO_BLK_F_GEOMETRY, + VIRTIO_BLK_F_RO, VIRTIO_BLK_F_BLK_SIZE, VIRTIO_BLK_F_SCSI, + VIRTIO_BLK_F_FLUSH, VIRTIO_BLK_F_TOPOLOGY }; /* diff --git a/drivers/block/xd.c b/drivers/block/xd.c index d5a3cd7..4abd2bc 100644 --- a/drivers/block/xd.c +++ b/drivers/block/xd.c @@ -46,7 +46,7 @@ #include <linux/init.h> #include <linux/wait.h> #include <linux/blkdev.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/blkpg.h> #include <linux/delay.h> #include <linux/io.h> @@ -58,6 +58,7 @@ #include "xd.h" +static DEFINE_MUTEX(xd_mutex); static void __init do_xd_setup (int *integers); #ifdef MODULE static int xd[5] = { -1,-1,-1,-1, }; @@ -381,9 +382,9 @@ static int xd_ioctl(struct block_device *bdev, fmode_t mode, { int ret; - lock_kernel(); + mutex_lock(&xd_mutex); ret = xd_locked_ioctl(bdev, mode, cmd, param); - unlock_kernel(); + mutex_unlock(&xd_mutex); return ret; } diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index ab735a6..d7aa39e 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -41,7 +41,7 @@ #include <linux/cdrom.h> #include <linux/module.h> #include <linux/slab.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/scatterlist.h> #include <xen/xen.h> @@ -65,13 +65,14 @@ enum blkif_state { struct blk_shadow { struct blkif_request req; - unsigned long request; + struct request *request; unsigned long frame[BLKIF_MAX_SEGMENTS_PER_REQUEST]; }; +static DEFINE_MUTEX(blkfront_mutex); static const struct block_device_operations xlvbd_block_fops; -#define BLK_RING_SIZE __RING_SIZE((struct blkif_sring *)0, PAGE_SIZE) +#define BLK_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE) /* * We have one of these per vbd, whether ide, scsi or 'other'. They @@ -95,7 +96,7 @@ struct blkfront_info struct gnttab_free_callback callback; struct blk_shadow shadow[BLK_RING_SIZE]; unsigned long shadow_free; - int feature_barrier; + unsigned int feature_flush; int is_ready; }; @@ -135,7 +136,7 @@ static void add_id_to_freelist(struct blkfront_info *info, unsigned long id) { info->shadow[id].req.id = info->shadow_free; - info->shadow[id].request = 0; + info->shadow[id].request = NULL; info->shadow_free = id; } @@ -244,14 +245,11 @@ static int blkif_ioctl(struct block_device *bdev, fmode_t mode, } /* - * blkif_queue_request + * 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. * - * request block io - * - * id: for guest use only. - * operation: BLKIF_OP_{READ,WRITE,PROBE} - * buffer: buffer to read/write into. this should be a - * virtual address in the guest os. + * @req: a request struct */ static int blkif_queue_request(struct request *req) { @@ -280,7 +278,7 @@ static int blkif_queue_request(struct request *req) /* Fill out a communications ring structure. */ ring_req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt); id = get_id_from_freelist(info); - info->shadow[id].request = (unsigned long)req; + info->shadow[id].request = req; ring_req->id = id; ring_req->sector_number = (blkif_sector_t)blk_rq_pos(req); @@ -288,8 +286,18 @@ static int blkif_queue_request(struct request *req) ring_req->operation = rq_data_dir(req) ? BLKIF_OP_WRITE : BLKIF_OP_READ; - if (req->cmd_flags & REQ_HARDBARRIER) + + 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 + * 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->nr_segments = blk_rq_map_sg(req->q, req, info->sg); BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST); @@ -418,26 +426,12 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size) } -static int xlvbd_barrier(struct blkfront_info *info) +static void xlvbd_flush(struct blkfront_info *info) { - int err; - const char *barrier; - - switch (info->feature_barrier) { - case QUEUE_ORDERED_DRAIN: barrier = "enabled (drain)"; break; - case QUEUE_ORDERED_TAG: barrier = "enabled (tag)"; break; - case QUEUE_ORDERED_NONE: barrier = "disabled"; break; - default: return -EINVAL; - } - - err = blk_queue_ordered(info->rq, info->feature_barrier); - - if (err) - return err; - + blk_queue_flush(info->rq, info->feature_flush); printk(KERN_INFO "blkfront: %s: barriers %s\n", - info->gd->disk_name, barrier); - return 0; + info->gd->disk_name, + info->feature_flush ? "enabled" : "disabled"); } @@ -516,7 +510,7 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity, info->rq = gd->queue; info->gd = gd; - xlvbd_barrier(info); + xlvbd_flush(info); if (vdisk_info & VDISK_READONLY) set_disk_ro(gd, 1); @@ -553,7 +547,7 @@ static void xlvbd_release_gendisk(struct blkfront_info *info) spin_unlock_irqrestore(&blkif_io_lock, flags); /* Flush gnttab callback work. Must be done with no locks held. */ - flush_scheduled_work(); + flush_work_sync(&info->work); del_gendisk(info->gd); @@ -602,7 +596,7 @@ static void blkif_free(struct blkfront_info *info, int suspend) spin_unlock_irq(&blkif_io_lock); /* Flush gnttab callback work. Must be done with no locks held. */ - flush_scheduled_work(); + flush_work_sync(&info->work); /* Free resources associated with old device channel. */ if (info->ring_ref != GRANT_INVALID_REF) { @@ -649,7 +643,7 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) bret = RING_GET_RESPONSE(&info->ring, i); id = bret->id; - req = (struct request *)info->shadow[id].request; + req = info->shadow[id].request; blkif_completion(&info->shadow[id]); @@ -662,8 +656,18 @@ static irqreturn_t blkif_interrupt(int irq, void *dev_id) printk(KERN_WARNING "blkfront: %s: write barrier op failed\n", info->gd->disk_name); error = -EOPNOTSUPP; - info->feature_barrier = QUEUE_ORDERED_NONE; - xlvbd_barrier(info); + } + 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", + info->gd->disk_name); + error = -EOPNOTSUPP; + } + if (unlikely(error)) { + if (error == -EOPNOTSUPP) + error = 0; + info->feature_flush = 0; + xlvbd_flush(info); } /* fall through */ case BLKIF_OP_READ: @@ -914,7 +918,7 @@ static int blkif_recover(struct blkfront_info *info) /* Stage 3: Find pending requests and requeue them. */ for (i = 0; i < BLK_RING_SIZE; i++) { /* Not in use? */ - if (copy[i].request == 0) + if (!copy[i].request) continue; /* Grab a request slot and copy shadow state into it. */ @@ -931,9 +935,7 @@ static int blkif_recover(struct blkfront_info *info) req->seg[j].gref, info->xbdev->otherend_id, pfn_to_mfn(info->shadow[req->id].frame[j]), - rq_data_dir( - (struct request *) - info->shadow[req->id].request)); + rq_data_dir(info->shadow[req->id].request)); info->shadow[req->id].req = *req; info->ring.req_prod_pvt++; @@ -1076,20 +1078,14 @@ static void blkfront_connect(struct blkfront_info *info) /* * If there's no "feature-barrier" defined, then it means * we're dealing with a very old backend which writes - * synchronously; draining will do what needs to get done. - * - * If there are barriers, then we can do full queued writes - * with tagged barriers. + * synchronously; nothing to do. * - * If barriers are not supported, then there's no much we can - * do, so just set ordering to NONE. + * If there are barriers, then we use flush. */ - if (err) - info->feature_barrier = QUEUE_ORDERED_DRAIN; - else if (barrier) - info->feature_barrier = QUEUE_ORDERED_TAG; - else - info->feature_barrier = QUEUE_ORDERED_NONE; + info->feature_flush = 0; + + if (!err && barrier) + info->feature_flush = REQ_FLUSH | REQ_FUA; err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size); if (err) { @@ -1125,6 +1121,8 @@ static void blkback_changed(struct xenbus_device *dev, case XenbusStateInitialising: case XenbusStateInitWait: case XenbusStateInitialised: + case XenbusStateReconfiguring: + case XenbusStateReconfigured: case XenbusStateUnknown: case XenbusStateClosed: break; @@ -1201,7 +1199,7 @@ static int blkif_open(struct block_device *bdev, fmode_t mode) struct blkfront_info *info; int err = 0; - lock_kernel(); + mutex_lock(&blkfront_mutex); info = disk->private_data; if (!info) { @@ -1219,7 +1217,7 @@ static int blkif_open(struct block_device *bdev, fmode_t mode) mutex_unlock(&info->mutex); out: - unlock_kernel(); + mutex_unlock(&blkfront_mutex); return err; } @@ -1229,7 +1227,7 @@ static int blkif_release(struct gendisk *disk, fmode_t mode) struct block_device *bdev; struct xenbus_device *xbdev; - lock_kernel(); + mutex_lock(&blkfront_mutex); bdev = bdget_disk(disk, 0); bdput(bdev); @@ -1263,7 +1261,7 @@ static int blkif_release(struct gendisk *disk, fmode_t mode) } out: - unlock_kernel(); + mutex_unlock(&blkfront_mutex); return 0; } diff --git a/drivers/block/xsysace.c b/drivers/block/xsysace.c index 057413b..829161e 100644 --- a/drivers/block/xsysace.c +++ b/drivers/block/xsysace.c @@ -89,7 +89,7 @@ #include <linux/delay.h> #include <linux/slab.h> #include <linux/blkdev.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/ata.h> #include <linux/hdreg.h> #include <linux/platform_device.h> @@ -214,6 +214,7 @@ struct ace_device { u16 cf_id[ATA_ID_WORDS]; }; +static DEFINE_MUTEX(xsysace_mutex); static int ace_major; /* --------------------------------------------------------------------- @@ -903,13 +904,13 @@ static int ace_open(struct block_device *bdev, fmode_t mode) dev_dbg(ace->dev, "ace_open() users=%i\n", ace->users + 1); - lock_kernel(); + mutex_lock(&xsysace_mutex); spin_lock_irqsave(&ace->lock, flags); ace->users++; spin_unlock_irqrestore(&ace->lock, flags); check_disk_change(bdev); - unlock_kernel(); + mutex_unlock(&xsysace_mutex); return 0; } @@ -922,7 +923,7 @@ static int ace_release(struct gendisk *disk, fmode_t mode) dev_dbg(ace->dev, "ace_release() users=%i\n", ace->users - 1); - lock_kernel(); + mutex_lock(&xsysace_mutex); spin_lock_irqsave(&ace->lock, flags); ace->users--; if (ace->users == 0) { @@ -930,7 +931,7 @@ static int ace_release(struct gendisk *disk, fmode_t mode) ace_out(ace, ACE_CTRL, val & ~ACE_CTRL_LOCKREQ); } spin_unlock_irqrestore(&ace->lock, flags); - unlock_kernel(); + mutex_unlock(&xsysace_mutex); return 0; } @@ -1224,7 +1225,8 @@ ace_of_probe(struct platform_device *op, const struct of_device_id *match) bus_width = ACE_BUS_WIDTH_8; /* Call the bus-independant setup code */ - return ace_alloc(&op->dev, id ? *id : 0, physaddr, irq, bus_width); + return ace_alloc(&op->dev, id ? be32_to_cpup(id) : 0, + physaddr, irq, bus_width); } static int __devexit ace_of_remove(struct platform_device *op) diff --git a/drivers/block/z2ram.c b/drivers/block/z2ram.c index d75b2bb..a22e3f8 100644 --- a/drivers/block/z2ram.c +++ b/drivers/block/z2ram.c @@ -33,7 +33,7 @@ #include <linux/module.h> #include <linux/blkdev.h> #include <linux/bitops.h> -#include <linux/smp_lock.h> +#include <linux/mutex.h> #include <linux/slab.h> #include <asm/setup.h> @@ -57,6 +57,7 @@ extern struct mem_info m68k_memory[NUM_MEMINFO]; #define Z2RAM_CHUNK1024 ( Z2RAM_CHUNKSIZE >> 10 ) +static DEFINE_MUTEX(z2ram_mutex); static u_long *z2ram_map = NULL; static u_long z2ram_size = 0; static int z2_count = 0; @@ -79,8 +80,10 @@ static void do_z2_request(struct request_queue *q) int err = 0; if (start + len > z2ram_size) { - printk( KERN_ERR DEVICE_NAME ": bad access: block=%lu, count=%u\n", - blk_rq_pos(req), blk_rq_cur_sectors(req)); + pr_err(DEVICE_NAME ": bad access: block=%llu, " + "count=%u\n", + (unsigned long long)blk_rq_pos(req), + blk_rq_cur_sectors(req)); err = -EIO; goto done; } @@ -154,7 +157,7 @@ static int z2_open(struct block_device *bdev, fmode_t mode) device = MINOR(bdev->bd_dev); - lock_kernel(); + mutex_lock(&z2ram_mutex); if ( current_device != -1 && current_device != device ) { rc = -EBUSY; @@ -296,25 +299,25 @@ static int z2_open(struct block_device *bdev, fmode_t mode) set_capacity(z2ram_gendisk, z2ram_size >> 9); } - unlock_kernel(); + mutex_unlock(&z2ram_mutex); return 0; err_out_kfree: kfree(z2ram_map); err_out: - unlock_kernel(); + mutex_unlock(&z2ram_mutex); return rc; } static int z2_release(struct gendisk *disk, fmode_t mode) { - lock_kernel(); + mutex_lock(&z2ram_mutex); if ( current_device == -1 ) { - unlock_kernel(); + mutex_unlock(&z2ram_mutex); return 0; } - unlock_kernel(); + mutex_unlock(&z2ram_mutex); /* * FIXME: unmap memory */ |