From e3a746f5aab71f2dd0a83116772922fb37ae29d6 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Thu, 12 Jul 2012 07:40:42 +1000 Subject: xfs: really fix the cursor leak in xfs_alloc_ag_vextent_near The current cursor is reallocated when retrying the allocation, so the existing cursor needs to be destroyed in both the restart and the failure cases. Signed-off-by: Dave Chinner Tested-by: Mike Snitzer Signed-off-by: Ben Myers --- fs/xfs/xfs_alloc.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c index 9d1aeb7e2734..f654f51b0c67 100644 --- a/fs/xfs/xfs_alloc.c +++ b/fs/xfs/xfs_alloc.c @@ -1074,13 +1074,13 @@ restart: * If we couldn't get anything, give up. */ if (bno_cur_lt == NULL && bno_cur_gt == NULL) { + xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR); + if (!forced++) { trace_xfs_alloc_near_busy(args); xfs_log_force(args->mp, XFS_LOG_SYNC); goto restart; } - - xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR); trace_xfs_alloc_size_neither(args); args->agbno = NULLAGBLOCK; return 0; -- cgit v1.2.3 From aa292847b9fc6e187547110de833a7d3131bbddf Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Thu, 12 Jul 2012 07:40:43 +1000 Subject: xfs: don't defer metadata allocation to the workqueue Almost all metadata allocations come from shallow stack usage situations. Avoid the overhead of switching the allocation to a workqueue as we are not in danger of running out of stack when making these allocations. Metadata allocations are already marked through the args that are passed down, so this is trivial to do. Signed-off-by: Dave Chinner Reported-by: Mel Gorman Tested-by: Mel Gorman Signed-off-by: Ben Myers --- fs/xfs/xfs_alloc.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c index f654f51b0c67..4f33c32affe3 100644 --- a/fs/xfs/xfs_alloc.c +++ b/fs/xfs/xfs_alloc.c @@ -2434,13 +2434,22 @@ xfs_alloc_vextent_worker( current_restore_flags_nested(&pflags, PF_FSTRANS); } - -int /* error */ +/* + * Data allocation requests often come in with little stack to work on. Push + * them off to a worker thread so there is lots of stack to use. Metadata + * requests, OTOH, are generally from low stack usage paths, so avoid the + * context switch overhead here. + */ +int xfs_alloc_vextent( - xfs_alloc_arg_t *args) /* allocation argument structure */ + struct xfs_alloc_arg *args) { DECLARE_COMPLETION_ONSTACK(done); + if (!args->userdata) + return __xfs_alloc_vextent(args); + + args->done = &done; INIT_WORK_ONSTACK(&args->work, xfs_alloc_vextent_worker); queue_work(xfs_alloc_wq, &args->work); -- cgit v1.2.3 From 40a9b7963df32e743c45d79a5f41445fe2476f15 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 2 Jul 2012 06:00:04 -0400 Subject: xfs: prevent recursion in xfs_buf_iorequest If the b_iodone handler is run in calling context in xfs_buf_iorequest we can run into a recursion where xfs_buf_iodone_callbacks keeps calling back into xfs_buf_iorequest because an I/O error happened, which keeps calling back into xfs_buf_iorequest. This chain will usually not take long because the filesystem gets shut down because of log I/O errors, but even over a short time it can cause stack overflows if run on the same context. As a short term workaround make sure we always call the iodone handler in workqueue context. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Ben Myers --- fs/xfs/xfs_buf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c index a4beb421018a..687b275f165c 100644 --- a/fs/xfs/xfs_buf.c +++ b/fs/xfs/xfs_buf.c @@ -1243,7 +1243,7 @@ xfs_buf_iorequest( */ atomic_set(&bp->b_io_remaining, 1); _xfs_buf_ioapply(bp); - _xfs_buf_ioend(bp, 0); + _xfs_buf_ioend(bp, 1); xfs_buf_rele(bp); } -- cgit v1.2.3 From 1632dcc93f55f9ab407b373da1957a727b1a7fe3 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Fri, 13 Jul 2012 02:24:10 -0400 Subject: xfs: do not call xfs_bdstrat_cb in xfs_buf_iodone_callbacks xfs_bdstrat_cb only adds a check for a shutdown filesystem over xfs_buf_iorequest, but xfs_buf_iodone_callbacks just checked for a shut down filesystem a little earlier. In addition the shutdown handling in xfs_bdstrat_cb is not very suitable for this caller. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Ben Myers --- fs/xfs/xfs_buf.c | 51 ++++++++++++++++++++++----------------------------- fs/xfs/xfs_buf.h | 1 - fs/xfs/xfs_buf_item.c | 2 +- 3 files changed, 23 insertions(+), 31 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c index 687b275f165c..269b35c084da 100644 --- a/fs/xfs/xfs_buf.c +++ b/fs/xfs/xfs_buf.c @@ -989,27 +989,6 @@ xfs_buf_ioerror_alert( (__uint64_t)XFS_BUF_ADDR(bp), func, bp->b_error, bp->b_length); } -int -xfs_bwrite( - struct xfs_buf *bp) -{ - int error; - - ASSERT(xfs_buf_islocked(bp)); - - bp->b_flags |= XBF_WRITE; - bp->b_flags &= ~(XBF_ASYNC | XBF_READ | _XBF_DELWRI_Q); - - xfs_bdstrat_cb(bp); - - error = xfs_buf_iowait(bp); - if (error) { - xfs_force_shutdown(bp->b_target->bt_mount, - SHUTDOWN_META_IO_ERROR); - } - return error; -} - /* * Called when we want to stop a buffer from getting written or read. * We attach the EIO error, muck with its flags, and call xfs_buf_ioend @@ -1079,14 +1058,7 @@ xfs_bioerror_relse( return EIO; } - -/* - * All xfs metadata buffers except log state machine buffers - * get this attached as their b_bdstrat callback function. - * This is so that we can catch a buffer - * after prematurely unpinning it to forcibly shutdown the filesystem. - */ -int +STATIC int xfs_bdstrat_cb( struct xfs_buf *bp) { @@ -1107,6 +1079,27 @@ xfs_bdstrat_cb( return 0; } +int +xfs_bwrite( + struct xfs_buf *bp) +{ + int error; + + ASSERT(xfs_buf_islocked(bp)); + + bp->b_flags |= XBF_WRITE; + bp->b_flags &= ~(XBF_ASYNC | XBF_READ | _XBF_DELWRI_Q); + + xfs_bdstrat_cb(bp); + + error = xfs_buf_iowait(bp); + if (error) { + xfs_force_shutdown(bp->b_target->bt_mount, + SHUTDOWN_META_IO_ERROR); + } + return error; +} + /* * Wrapper around bdstrat so that we can stop data from going to disk in case * we are shutting down the filesystem. Typically user data goes thru this diff --git a/fs/xfs/xfs_buf.h b/fs/xfs/xfs_buf.h index 7f1d1392ce37..79344c48008e 100644 --- a/fs/xfs/xfs_buf.h +++ b/fs/xfs/xfs_buf.h @@ -180,7 +180,6 @@ extern void xfs_buf_unlock(xfs_buf_t *); extern int xfs_bwrite(struct xfs_buf *bp); extern void xfsbdstrat(struct xfs_mount *, struct xfs_buf *); -extern int xfs_bdstrat_cb(struct xfs_buf *); extern void xfs_buf_ioend(xfs_buf_t *, int); extern void xfs_buf_ioerror(xfs_buf_t *, int); diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c index 45df2b857d48..d9e451115f98 100644 --- a/fs/xfs/xfs_buf_item.c +++ b/fs/xfs/xfs_buf_item.c @@ -954,7 +954,7 @@ xfs_buf_iodone_callbacks( if (!XFS_BUF_ISSTALE(bp)) { bp->b_flags |= XBF_WRITE | XBF_ASYNC | XBF_DONE; - xfs_bdstrat_cb(bp); + xfs_buf_iorequest(bp); } else { xfs_buf_relse(bp); } -- cgit v1.2.3