aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/md
Commit message (Collapse)AuthorAgeFilesLines
* [SCSI] scsi_dh: Remove hardware handler infrastructure from dmChandra Seetharaman2008-06-054-278/+1
| | | | | | | | | This patch just removes infrastructure that provided support for hardware handlers in the dm layer as it is not needed anymore. Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Acked-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
* [SCSI] scsi_dh: Remove hardware handlers from dmChandra Seetharaman2008-06-055-1315/+0
| | | | | | | | | | | This patch removes the 3 hardware handlers that currently exist under dm as the functionality is moved to SCSI layer in the earlier patches. [jejb: removed more makefile hunks and rejection fixes] Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Acked-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
* [SCSI] scsi_dh: Remove dm_pg_init_completeChandra Seetharaman2008-06-051-41/+0
| | | | | | | | | | This patch just removes the dm layer's path initialization completion routine. This is separated from the other patch(scsi_dh: Use SCSI device handler in dm-multipath) Just to make that patch more readable. Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Acked-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
* [SCSI] scsi_dh: Add a single threaded workqueue for initializing pathsChandra Seetharaman2008-06-051-7/+34
| | | | | | | | | | | | | | | | | | | | | | | | Before this patch set (SCSI hardware handlers), initialization of a path was done asynchronously. Doing that requires a workqueue in each device/hardware handler module and leads to unneccessary complication in the device handler code, making it difficult to read the code and follow the state diagram. Moving that workqueue to this level makes the device handler code simpler. Hence, the workqueue is moved to dm level. A new workqueue is added instead of adding it to the existing workqueue (kmpathd) for the following reasons: 1. Device activation has to happen faster, stacking them along with the other workqueue might lead to unnecessary delay in the activation of the path. 2. The effect could be felt the other way too. i.e the current events that are handled by the existing workqueue might get a delayed response. Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Acked-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
* [SCSI] scsi_dh: Use SCSI device handler in dm-multipathChandra Seetharaman2008-06-052-51/+81
| | | | | | | | | | | | | | | | | | | | | | | This patch converts dm-mpath to use scsi device handlers instead of dm's hardware handlers. This patch does not add any new functionality. Old behaviors remain and userspace tools work as is except that arguments supplied with hardware handler are ignored. One behavioral exception is: Activation of a path is synchronous in this patch, opposed to the older behavior of being asynchronous (changed in patch 07: scsi_dh: Add a single threaded workqueue for initializing a path) Note: There is no need to get a reference for the device handler module (as it was done in the dm hardware handler case) here as the reference is held when the device was first found. Instead we check and make sure that support for the specified device is present at table load time. Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Signed-off-by: Mike Christie <michaelc@cs.wisc.edu> Acked-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
* md: restart recovery cleanly after device failure.NeilBrown2008-05-245-16/+43
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When we get any IO error during a recovery (rebuilding a spare), we abort the recovery and restart it. For RAID6 (and multi-drive RAID1) it may not be best to restart at the beginning: when multiple failures can be tolerated, the recovery may be able to continue and re-doing all that has already been done doesn't make sense. We already have the infrastructure to record where a recovery is up to and restart from there, but it is not being used properly. This is because: - We sometimes abort with MD_RECOVERY_ERR rather than just MD_RECOVERY_INTR, which causes the recovery not be be checkpointed. - We remove spares and then re-added them which loses important state information. The distinction between MD_RECOVERY_ERR and MD_RECOVERY_INTR really isn't needed. If there is an error, the relevant drive will be marked as Faulty, and that is enough to ensure correct handling of the error. So we first remove MD_RECOVERY_ERR, changing some of the uses of it to MD_RECOVERY_INTR. Then we cause the attempt to remove a non-faulty device from an array to fail (unless recovery is impossible as the array is too degraded). Then when remove_and_add_spares attempts to remove the devices on which recovery can continue, it will fail, they will remain in place, and recovery will continue on them as desired. Issue: If we are halfway through rebuilding a spare and another drive fails, and a new spare is immediately available, do we want to: 1/ complete the current rebuild, then go back and rebuild the new spare or 2/ restart the rebuild from the start and rebuild both devices in parallel. Both options can be argued for. The code currently takes option 2 as a/ this requires least code change b/ this results in a minimally-degraded array in minimal time. Cc: "Eivind Sarto" <ivan@kasenna.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: allow parallel resync of md-devices.Bernd Schubert2008-05-241-4/+36
| | | | | | | | | | | | | | In some configurations, a raid6 resync can be limited by CPU speed (Calculating P and Q and moving data) rather than by device speed. In these cases there is nothing to be gained byt serialising resync of arrays that share a device, and doing the resync in parallel can provide benefit. So add a sysfs tunable to flag an array as being allowed to resync in parallel with other arrays that use (a different part of) the same device. Signed-off-by: Bernd Schubert <bs@q-leap.de> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: notify userspace on 'stop' eventsDan Williams2008-05-241-0/+2
| | | | | | | | | | | | | | | | This additional notification to 'array_state' is needed to allow the monitor application to learn about stop events via sysfs. The sysfs_notify("sync_action") call that comes at the end of do_md_stop() (via md_new_event) is insufficient since the 'sync_action' attribute has been removed by this point. (Seems like a sysfs-notify-on-removal patch is a better fix. Currently removal updates the event count but does not wake up waiters) Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: notify userspace on 'write-pending' changes to array_stateNeilBrown2008-05-241-1/+10
| | | | | | | | | | | | | | When an array enters write pending, 'array_state' changes, so we must be sure to sysfs_notify. Also, when waiting for user-space to acknowledge 'write-pending' by marking the metadata as dirty, we don't want to wait for MD_CHANGE_DEVS to be cleared as that might not happen. So explicity test for the bits that we are really interested in. Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: raid1: Fix restoration of bio between failed read and write.NeilBrown2008-05-241-2/+13
| | | | | | | | | | | | | | | | | | | | | When performing a "recovery" or "check" pass on a RAID1 array, we read from each device and possible, if there is a difference or a read error, write back to some devices. We use the same 'bio' for both read and write, resetting various fields between the two operations. We forgot to reset bv_offset and bv_len however. These are often left unchanged, but in the case where there is an IO error one or two sectors into a page, they are changed. This results in correctable errors not being corrected properly. It does not result in any data corruption. Cc: "Fairbanks, David" <David.Fairbanks@stratus.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: md: raid5 rate limit error printkBernd Schubert2008-05-241-12/+22
| | | | | | | | | | | | | | | | | | | Last night we had scsi problems and a hardware raid unit was offlined during heavy i/o. While this happened we got for about 3 minutes a huge number messages like these Apr 12 03:36:07 pfs1n14 kernel: [197510.696595] raid5:md7: read error not correctable (sector 2993096568 on sdj2). I guess the high error rate is responsible for not scheduling other events - during this time the system was not pingable and in the end also other devices run into scsi command timeouts causing problems on these unrelated devices as well. Signed-off-by: Bernd Schubert <bernd-schubert@gmx.de> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: kill file_path wrapperChristoph Hellwig2008-05-242-15/+6
| | | | | | | | | Kill the trivial and rather pointless file_path wrapper around d_path. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: fix possible oops when removing a bitmap from an active arrayNeilBrown2008-05-241-1/+3
| | | | | | | | | | | | | | | | It is possible to add a write-intent bitmap to an active array, or remove the bitmap that is there. When we do with the 'quiesce' the array, which causes make_request to block in "wait_barrier()". However we are sampling the value of "mddev->bitmap" before the wait_barrier call, and using it afterwards. This can result in using a bitmap structure that has been freed. Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* Remove blkdev warning triggered by using mdNeil Brown2008-05-146-2/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | As setting and clearing queue flags now requires that we hold a spinlock on the queue, and as blk_queue_stack_limits is called without that lock, get the lock inside blk_queue_stack_limits. For blk_queue_stack_limits to be able to find the right lock, each md personality needs to set q->queue_lock to point to the appropriate lock. Those personalities which didn't previously use a spin_lock, us q->__queue_lock. So always initialise that lock when allocated. With this in place, setting/clearing of the QUEUE_FLAG_PLUGGED bit will no longer cause warnings as it will be clear that the proper lock is held. Thanks to Dan Williams for review and fixing the silly bugs. Signed-off-by: NeilBrown <neilb@suse.de> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Alistair John Strachan <alistair@devzero.co.uk> Cc: Nick Piggin <npiggin@suse.de> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Jacek Luczak <difrost.kernel@gmail.com> Cc: Prakash Punnoor <prakash@punnoor.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: fix raid5 'repair' operationsDan Williams2008-05-131-12/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | commit bd2ab67030e9116f1e4aae1289220255412b37fd "md: close a livelock window in handle_parity_checks5" introduced a bug in handling 'repair' operations. After a repair operation completes we clear the state bits tracking this operation. However, they are cleared too early and this results in the code deciding to re-run the parity check operation. Since we have done the repair in memory the second check does not find a mismatch and thus does not do a writeback. Test results: $ echo repair > /sys/block/md0/md/sync_action $ cat /sys/block/md0/md/mismatch_cnt 51072 $ echo repair > /sys/block/md0/md/sync_action $ cat /sys/block/md0/md/mismatch_cnt 0 (also fix incorrect indentation) Cc: <stable@kernel.org> Tested-by: George Spelvin <linux@horizon.com> Acked-by: NeilBrown <neilb@suse.de> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* misc: fix integer as NULL pointer warningsHarvey Harrison2008-05-081-1/+1
| | | | | | | | | | | | drivers/md/raid10.c:889:17: warning: Using plain integer as NULL pointer drivers/media/video/cx18/cx18-driver.c:616:12: warning: Using plain integer as NULL pointer sound/oss/kahlua.c:70:12: warning: Using plain integer as NULL pointer Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Cc: Neil Brown <neilb@suse.de> Cc: Mauro Carvalho Chehab <mchehab@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: support blocking writes to an array on device failureDan Williams2008-04-304-7/+115
| | | | | | | | | | | | | | | | | | | | Allows a userspace metadata handler to take action upon detecting a device failure. Based on an original patch by Neil Brown. Changes: -added blocked_wait waitqueue to rdev -don't qualify Blocked with Faulty always let userspace block writes -added md_wait_for_blocked_rdev to wait for the block device to be clear, if userspace misses the notification another one is sent every 5 seconds -set MD_RECOVERY_NEEDED after clearing "blocked" -kill DoBlock flag, just test mddev->external Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: prevent duplicates in bind_rdev_to_arrayDan Williams2008-04-301-0/+5
| | | | | | | | | | | Found when trying to reassemble an active externally managed array. Without this check we hit the more noisy "sysfs duplicate" warning in the later call to kobject_add. Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: remove a stray command from a copy and paste error in resync_start_storeDan Williams2008-04-301-1/+0
| | | | | | | Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: fix up switching md arrays between read-only and read-writeNeilBrown2008-04-301-4/+10
| | | | | | | | | | | | | When setting an array to 'readonly' or to 'active' via sysfs, we must make the appropriate set_disk_ro call too. Also when switching to "read_auto" (which is like readonly, but blocks on the first write so that metadata can be marked 'dirty') we need to be more careful about what state we are changing from. Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: fix 'safemode' handling for external metadata.NeilBrown2008-04-301-11/+19
| | | | | | | | | | | | | | | | | | | | | | | | 'safemode' relates to marking an array as 'clean' if there has been no write traffic for a while (a couple of seconds), to reduce the chance of the array being found dirty on reboot. ->safemode is set to '1' when there have been no write for a while, and it gets set to '0' when the superblock is updates with the 'clean' flag set. This requires a few fixes for 'external' metadata: - When an array is set to 'clean' via sysfs, 'safemode' must be cleared. - when we write to an array that has 'safemode' set (there must have been some delay in updating the metadata), we need to clear safemode. - Don't try to update external metadata in md_check_recovery for safemode transitions - it won't work. Also, don't try to support "immediate safe mode" (safemode==2) for external metadata, it cannot really work (the safemode timeout can be set very low if this is really needed). Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: reinitialise more mddev fields in do_md_stop.NeilBrown2008-04-301-0/+25
| | | | | | | | | | | | I keep finding problems where an mddev gets reused and some fields has a value from a previous usage that confuses the new usage. So clear all fields that could possible need clearing when calling do_md_stop. Also initialise the 'level' of a new array to LEVEL_NONE (which isn't 0). Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: skip all metadata update processing when using external metadata.NeilBrown2008-04-301-0/+2
| | | | | | | | | All the metadata update processing for external metadata is on in user-space or through the sysfs interfaces, so make "md_update_sb" a no-op in that case. Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: fix use after free when removing rdev via sysfsDan Williams2008-04-301-1/+1
| | | | | | | | | | | rdev->mddev is no longer valid upon return from entry->store() when the 'remove' command is given. Cc: <stable@kernel.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* dm: use unlocked variants of queue flag check/setJens Axboe2008-04-291-5/+3
| | | | | | | dm.c already provides mutual exclusion through ->map_lock. Signed-off-by: Jens Axboe <jens.axboe@oracle.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-blockLinus Torvalds2008-04-295-7/+7
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * 'for-linus' of git://git.kernel.dk/linux-2.6-block: block: Skip I/O merges when disabled block: add large command support block: replace sizeof(rq->cmd) with BLK_MAX_CDB ide: use blk_rq_init() to initialize the request block: use blk_rq_init() to initialize the request block: rename and export rq_init() block: no need to initialize rq->cmd with blk_get_request block: no need to initialize rq->cmd in prepare_flush_fn hook block/blk-barrier.c:blk_ordered_cur_seq() mustn't be inline block/elevator.c:elv_rq_merge_ok() mustn't be inline block: make queue flags non-atomic block: add dma alignment and padding support to blk_rq_map_kern unexport blk_max_pfn ps3disk: Remove superfluous cast block: make rq_init() do a full memset() relay: fix splice problem
| * block: no need to initialize rq->cmd with blk_get_requestFUJITA Tomonori2008-04-293-4/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | blk_get_request initializes rq->cmd (rq_init does) so the users don't need to do that. The purpose of this patch is to remove sizeof(rq->cmd) and &rq->cmd, as a preparation for large command support, which changes rq->cmd from the static array to a pointer. sizeof(rq->cmd) will not make sense and &rq->cmd won't work. Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Alasdair G Kergon <agk@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
| * block: make queue flags non-atomicNick Piggin2008-04-292-3/+7
| | | | | | | | | | | | | | We can save some atomic ops in the IO path, if we clearly define the rules of how to modify the queue flags. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
* | drivers: use non-racy method for proc entries creation (2)Denis V. Lunev2008-04-291-5/+1
|/ | | | | | | | | | | | | | | | | | | Use proc_create()/proc_create_data() to make sure that ->proc_fops and ->data be setup before gluing PDE to main tree. Signed-off-by: Denis V. Lunev <den@openvz.org> Cc: Greg Kroah-Hartman <gregkh@suse.de> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Peter Osterlund <petero2@telia.com> Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> Cc: Dmitry Torokhov <dtor@mail.ru> Cc: Neil Brown <neilb@suse.de> Cc: Mauro Carvalho Chehab <mchehab@infradead.org> Cc: Bjorn Helgaas <bjorn.helgaas@hp.com> Cc: Alessandro Zummo <a.zummo@towertech.it> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* drivers/md: use time_before, time_before_eq, etcJulia Lawall2008-04-281-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The functions time_before, time_before_eq, time_after, and time_after_eq are more robust for comparing jiffies against other values. A simplified version of the semantic patch making this change is as follows: (http://www.emn.fr/x-info/coccinelle/) // <smpl> @ change_compare_np @ expression E; @@ ( - jiffies <= E + time_before_eq(jiffies,E) | - jiffies >= E + time_after_eq(jiffies,E) | - jiffies < E + time_before(jiffies,E) | - jiffies > E + time_after(jiffies,E) ) @ include depends on change_compare_np @ @@ #include <linux/jiffies.h> @ no_include depends on !include && change_compare_np @ @@ #include <linux/...> + #include <linux/jiffies.h> // </smpl> [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Julia Lawall <julia@diku.dk> Cc: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* raid: remove leading TAB on printk messagesNick Andrew2008-04-284-7/+8
| | | | | | | | | | | | | | | | MD drivers use one printk() call to print 2 log messages and the second line may be prefixed by a TAB character. It may also output a trailing space before newline. klogd (I think) turns the TAB character into the 2 characters '^I' when logging to a file. This looks ugly. Instead of a leading TAB to indicate continuation, prefix both output lines with 'raid:' or similar. Also remove any trailing space in the vicinity of the affected code and consistently end the sentences with a period. Signed-off-by: Nick Andrew <nick@nick-andrew.net> Cc: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: raid5.c convert simple_strtoul to strict_strtoulDan Williams2008-04-281-9/+5
| | | | | | | | | | strict_strtoul handles the open-coded sanity checks in raid5_store_stripe_cache_size and raid5_store_preread_threshold Acked-by: NeilBrown <neilb@suse.de> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: introduce get_priority_stripe() to improve raid456 write performanceDan Williams2008-04-281-10/+112
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Improve write performance by preventing the delayed_list from dumping all its stripes onto the handle_list in one shot. Delayed stripes are now further delayed by being held on the 'hold_list'. The 'hold_list' is bypassed when: * a STRIPE_IO_STARTED stripe is found at the head of 'handle_list' * 'handle_list' is empty and i/o is being done to satisfy full stripe-width write requests * 'bypass_count' is less than 'bypass_threshold'. By default the threshold is 1, i.e. every other stripe handled is a preread stripe provided the top two conditions are false. Benchmark data: System: 2x Xeon 5150, 4x SATA, mem=1GB Baseline: 2.6.24-rc7 Configuration: mdadm --create /dev/md0 /dev/sd[b-e] -n 4 -l 5 --assume-clean Test1: dd if=/dev/zero of=/dev/md0 bs=1024k count=2048 * patched: +33% (stripe_cache_size = 256), +25% (stripe_cache_size = 512) Test2: tiobench --size 2048 --numruns 5 --block 4096 --block 131072 (XFS) * patched: +13% * patched + preread_bypass_threshold = 0: +37% Changes since v1: * reduce bypass_threshold from (chunk_size / sectors_per_chunk) to (1) and make it configurable. This defaults to fairness and modest performance gains out of the box. Changes since v2: * [neilb@suse.de]: kill STRIPE_PRIO_HI and preread_needed as they are not necessary, the important change was clearing STRIPE_DELAYED in add_stripe_bio and this has been moved out to make_request for the hang fix. * [neilb@suse.de]: simplify get_priority_stripe * [dan.j.williams@intel.com]: reset the bypass_count when ->hold_list is sampled empty (+11%) * [dan.j.williams@intel.com]: decrement the bypass_count at the detection of stripes being naturally promoted off of hold_list +2%. Note, resetting bypass_count instead of decrementing on these events yields +4% but that is probably too aggressive. Changes since v3: * cosmetic fixups Tested-by: James W. Laferriere <babydr@baby-dragons.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: replace remaining __FUNCTION__ occurrencesHarvey Harrison2008-04-282-25/+25
| | | | | | | | | __FUNCTION__ is gcc-specific, use __func__ Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Cc: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* md: fix integer as NULL pointer warnings in md.cHarvey Harrison2008-04-281-4/+4
| | | | | | | | | | | | drivers/md/md.c:734:16: warning: Using plain integer as NULL pointer drivers/md/md.c:1115:16: warning: Using plain integer as NULL pointer Add some braces to match the else-block as well. Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Cc: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* dm: remove md argument from specific_minorFrederik Deweerdt2008-04-251-8/+6
| | | | | | | | | | | The small patch below: - Removes the unused md argument from both specific_minor() and next_free_minor() - Folds kmalloc + memset(0) into a single kzalloc call in alloc_dev() This has been compile tested on x86. Signed-off-by: Frederik Deweerdt <frederik.deweerdt@gmail.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm table: remove unused dm_create_error_tableAdrian Bunk2008-04-251-38/+0
| | | | | | | dm_create_error_table() was added in kernel 2.6.18 and never used... Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm table: drop void suspend_targets returnAdrian Bunk2008-04-251-2/+2
| | | | | | | | | | void returning functions returned the return value of another void returning function... Spotted by sparse. Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm: unplug queues in threadsMikulas Patocka2008-04-253-4/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Remove an avoidable 3ms delay on some dm-raid1 and kcopyd I/O. It is specified that any submitted bio without BIO_RW_SYNC flag may plug the queue (i.e. block the requests from being dispatched to the physical device). The queue is unplugged when the caller calls blk_unplug() function. Usually, the sequence is that someone calls submit_bh to submit IO on a buffer. The IO plugs the queue and waits (to be possibly joined with other adjacent bios). Then, when the caller calls wait_on_buffer(), it unplugs the queue and submits the IOs to the disk. This was happenning: When doing O_SYNC writes, function fsync_buffers_list() submits a list of bios to dm_raid1, the bios are added to dm_raid1 write queue and kmirrord is woken up. fsync_buffers_list() calls wait_on_buffer(). That unplugs the queue, but there are no bios on the device queue as they are still in the dm_raid1 queue. wait_on_buffer() starts waiting until the IO is finished. kmirrord is scheduled, kmirrord takes bios and submits them to the devices. The submitted bio plugs the harddisk queue but there is no one to unplug it. (The process that called wait_on_buffer() is already sleeping.) So there is a 3ms timeout, after which the queues on the harddisks are unplugged and requests are processed. This 3ms timeout meant that in certain workloads (e.g. O_SYNC, 8kb writes), dm-raid1 is 10 times slower than md raid1. Every time we submit something asynchronously via dm_io, we must unplug the queue actually to send the request to the device. This patch adds an unplug call to kmirrord - while processing requests, it keeps the queue plugged (so that adjacent bios can be merged); when it finishes processing all the bios, it unplugs the queue to submit the bios. It also fixes kcopyd which has the same potential problem. All kcopyd requests are submitted with BIO_RW_SYNC. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com> Acked-by: Jens Axboe <jens.axboe@oracle.com>
* dm raid1: use timerMikulas Patocka2008-04-251-20/+28
| | | | | | | | | | | | This patch replaces the schedule() in the main kmirrord thread with a timer. The schedule() could introduce an unwanted delay when work is ready to be processed. The code instead calls wake() when there's work to be done immediately, and delayed_wake() after a failure to give a short delay before retrying. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm: move include filesAlasdair G Kergon2008-04-2510-274/+11
| | | | | | Publish the dm-io, dm-log and dm-kcopyd headers in include/linux. Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm kcopyd: renameAlasdair G Kergon2008-04-252-0/+0
| | | | | | Rename kcopyd.[ch] to dm-kcopyd.[ch]. Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm: expose macrosAlasdair G Kergon2008-04-251-89/+0
| | | | | | Make dm.h macros and inlines available in include/linux/device-mapper.h Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm kcopyd: remove redundant client countingMikulas Patocka2008-04-253-98/+29
| | | | | | | | | | | | Remove client counting code that is no longer needed. Initialization and destruction is made globally from dm_init and dm_exit and is not based on client counts. Initialization allocates only one empty slab cache, so there is no negative impact from performing the initialization always, regardless of whether some client uses kcopyd or not. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm kcopyd: private mempoolMikulas Patocka2008-04-251-13/+19
| | | | | | | | Change the global mempool in kcopyd into a per-device mempool to avoid deadlock possibilities. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm kcopyd: per deviceMikulas Patocka2008-04-251-59/+73
| | | | | | | | Make one kcopyd thread per device. The original shared kcopyd could deadlock. Configuration:
* dm log: make module use tracking internalJonathan Brassow2008-04-252-31/+94
| | | | | | | Remove internal module reference fields from the interface. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm log: move register functionsAlasdair G Kergon2008-04-251-26/+26
| | | | | | Reorder a couple of functions in the file so the next patch is readable. Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm log: clean interfaceHeinz Mauelshagen2008-04-254-103/+112
| | | | | | | Clean up the dm-log interface to prepare for publishing it in include/linux. Signed-off-by: Heinz Mauelshagen <hjm@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
* dm kcopyd: clean interfaceHeinz Mauelshagen2008-04-255-58/+62
| | | | | | | Clean up the kcopyd interface to prepare for publishing it in include/linux. Signed-off-by: Heinz Mauelshagen <hjm@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
generated by cgit v1.1 at 2024-10-31 01:24:17 +0000