From 3beec1d446fba335f07787636920892dd3b2c658 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Thu, 14 Oct 2010 13:31:48 +0200 Subject: drbd: tag a few error messages with "assert failed" If those messages ever get logged, clearly state that they are actually failed ASSERTS, so our regression tests can pick them up from the logs more easily. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'drivers/block/drbd/drbd_int.h') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index c07c370c4c82..e0e0bf6f16a1 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 */ -- cgit v1.2.3 From 82f59cc6353889b426cf13b6596d5a3d100fa09e Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Sat, 16 Oct 2010 12:13:47 +0200 Subject: drbd: fix potential deadlock on detach If we have contention in drbd_al_begin_iod (heavy randon IO), an administrative request to detach the disk may deadlock for similar reasons as the recently fixed deadlock if detaching because of IO-error. The approach taken here is to either go through the intermediate cleanup state D_FAILED, or first lock out application io, don't just go directly to D_DISKLESS. We need an additional state bit (WAS_IO_ERROR) to distinguish the -> D_FAILED because of IO-error from other failures. Sanitize D_ATTACHING -> D_FAILED to D_ATTACHING -> D_DISKLESS. If only attaching, ldev may be missing still, but would be referenced from within the after_state_ch for -> D_FAILED, potentially dereferencing a NULL pointer. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-) (limited to 'drivers/block/drbd/drbd_int.h') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index e0e0bf6f16a1..03c15e317c37 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -852,7 +852,8 @@ enum { BITMAP_IO, /* suspend application io; once no more io in flight, start bitmap io */ BITMAP_IO_QUEUED, /* Started bitmap IO */ - GO_DISKLESS, /* Disk failed, local_cnt reached zero, we are going diskless */ + 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 */ @@ -1281,6 +1282,7 @@ 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 @@ -1798,17 +1800,17 @@ 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; } @@ -2127,7 +2129,11 @@ static inline void put_ldev(struct drbd_conf *mdev) __release(local); 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); } @@ -2138,6 +2144,10 @@ 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) -- cgit v1.2.3 From 2451fc3b2bd3a7205270da75a21dde0d5d7c96a2 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Tue, 24 Aug 2010 13:43:11 +0200 Subject: drbd: Removed the BIO_RW_BARRIER support form the receiver/epoch code Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 12 ------------ 1 file changed, 12 deletions(-) (limited to 'drivers/block/drbd/drbd_int.h') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 03c15e317c37..1b915fd9278f 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -749,17 +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 */ }; @@ -801,11 +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, @@ -820,7 +810,6 @@ enum { }; #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) @@ -948,7 +937,6 @@ enum write_ordering_e { WO_none, WO_drain_io, WO_bdev_flush, - WO_bio_barrier }; struct fifo_buffer { -- cgit v1.2.3 From a8a4e51e6965db84d2af041370ea2ab6232aa4f1 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Wed, 25 Aug 2010 10:21:04 +0200 Subject: drbd: REQ_HARDBARRIER -> REQ_FUA transition for meta data accesses Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'drivers/block/drbd/drbd_int.h') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 1b915fd9278f..575bfba1b0da 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -835,8 +835,7 @@ enum { 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 */ @@ -2404,13 +2403,13 @@ 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); 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); } } -- cgit v1.2.3 From 650789c87f16dcdf1dd0a67ac7461b7537534855 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Wed, 25 Aug 2010 10:47:17 +0200 Subject: drbd: Removed checks for REQ_HARDBARRIER on incomming BIOs Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 1 - 1 file changed, 1 deletion(-) (limited to 'drivers/block/drbd/drbd_int.h') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 575bfba1b0da..0c527d281622 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -832,7 +832,6 @@ 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_FUA, /* Users wants us to not use FUA/FLUSH on meta data dev */ -- cgit v1.2.3 From 77304d2abac6101f7249754ffdd4421258877ab0 Mon Sep 17 00:00:00 2001 From: Mike Snitzer Date: Mon, 8 Nov 2010 14:39:12 +0100 Subject: block: read i_size with i_size_read() Convert direct reads of an inode's i_size to using i_size_read(). i_size_{read,write} use a seqcount to protect reads from accessing incomple writes. Concurrent i_size_write()s require mutual exclussion to protect the seqcount that is used by i_size_{read,write}. But i_size_read() callers do not need to use additional locking. Signed-off-by: Mike Snitzer Acked-by: NeilBrown Acked-by: Lars Ellenberg Signed-off-by: Jens Axboe --- drivers/block/drbd/drbd_int.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/block/drbd/drbd_int.h') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 9bdcf4393c0a..2637f499f77f 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1874,7 +1874,7 @@ 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; } /** -- cgit v1.2.3