starlingx/kernel
Code Issues Proposed changes

xfs: fix ioend batching log reservation deadlock

Browse Source 
Problem:
We received a report of a workload that causes an xfs task to be blocked
for more than 120 seconds on log reservation via iomap_ioend completion
batching.

 kernel: err [5636141.631454] INFO: task xfs-conv/dm-4:1788 blocked for
                                    more than 122 seconds.

 kernel: info [267022.728862] Workqueue: xfs-conv/dm-4 xfs_end_io [xfs]
 kernel: info [267022.728864] Call Trace:
 kernel: info [267022.728870] __schedule+0x340/0x810
 kernel: info [267022.728876] schedule+0x51/0xc0
 kernel: info [267022.728913] xlog_grant_head_wait+0xc7/0x200 [xfs]
 kernel: info [267022.728950] xlog_grant_head_check+0xd0/0x110 [xfs]
 kernel: info [267022.728985] xfs_log_reserve+0xc3/0x1e0 [xfs]
 kernel: info [267022.729023] xfs_trans_reserve+0x156/0x1b0 [xfs]
 kernel: info [267022.729184] xfs_trans_alloc+0xc6/0x190 [xfs]
 kernel: info [267022.729317] xfs_iomap_write_unwritten+0xaa/0x2c0 [xfs]
 kernel: info [267022.729333] ? stop_one_cpu+0x71/0xa0
 kernel: info [267022.729347] ? set_cpus_allowed_ptr+0x10/0x10
 kernel: info [267022.729396] xfs_end_ioend+0xc4/0x100 [xfs]
 kernel: info [267022.729444] ? xfs_setfilesize_ioend+0x60/0x60 [xfs]
 kernel: info [267022.729491] xfs_end_io+0xb9/0xe0 [xfs]
 kernel: info [267022.729505] process_one_work+0x1a1/0x370
 kernel: info [267022.729516] rescuer_thread+0x207/0x350
 kernel: info [267022.729528] ? worker_thread+0x370/0x370
 kernel: info [267022.729537] kthread+0x12e/0x150
 kernel: info [267022.729548] ? __kthread_cancel_work+0x40/0x40
 kernel: info [267022.729559] ret_from_fork+0x1f/0x30

After that, the connection via ssh to the controller is stuck,
Press Ctrl+C, it entered shell and the prompt displayed '-sh-4.2$'

Solution:
Removing the preallocated transaction from xfs append ioends to avoid
the ioend completion batching log reservation deadlock.
Now we continue to process the append ioend completions via the
workqueue, but let the wq task allocate the transaction similar to other
ioend types.

Backport the four patches from upstream(git://git.kernel.org/pub/scm/
linux/kernel/git/torvalds/linux.git) for debian-based StarlingX.
Only the 0034-xfs-use-current-journal_info-for-detecting-transacti.patch
for centos-based StarlingX is from stable tree(git://git.kernel.org/pub/
scm/linux/kernel/git/stable/linux.git linux-5.10.y branch), because the
kernel has been upgraded to v5.10.152 for debian-based StarlingX which
includes this fix, so we just apply it for the centos-based one.

TestPlan:
Pass: Execute bonnie++ test for xfs filesystem successfully without
      kernel panic and any xfs anomalies in the kernel logs.
      $mkfs.x /dev/sdc1
      $mount /dev/sdc1 ~/xfstests
      $sudo bonnie++ -u root:root -d ~/xfstests
Debian:
Pass: build-pkgs -c -a
Pass: build-image
Pass: boot successfully with std/rt.
CentOS:
Pass: build-pkgs
Pass: build-iso
Pass: boot successfully with std/rt.

Closes-Bug: 1996269

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
Change-Id: I1e5b85111b2b54cd249c116724b952042f9d781f
This commit is contained in:
Zhixiong Chi 2022-11-10 05:38:05 -08:00
parent db0c0bcfec
commit 5ab365dd41
22 changed files with 1984 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
kernel-rt/centos/kernel-rt.spec
View File

@ -809,6 +809,11 @@ Patch29: 0030-x86-Enumerate-AVX512-FP16-CPUID-feature-flag.patch
Patch30: 0031-KVM-x86-Expose-AVX512_FP16-for-supported-CPUID.patch
Patch31: 0032-tools-headers-cpufeatures-Sync-with-the-kernel-sourc.patch
Patch32: 0033-rcu-Avoid-running-boost-kthreads-on-isolated-CPUs.patch
Patch33: 0034-xfs-use-current-journal_info-for-detecting-transacti.patch
Patch34: 0035-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch
Patch35: 0036-xfs-open-code-ioend-needs-workqueue-helper.patch
Patch36: 0037-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch
Patch37: 0038-xfs-drop-unnecessary-setfilesize-helper.patch
# END OF PATCH DEFINITIONS
%endif

269
kernel-rt/centos/patches/0034-xfs-use-current-journal_info-for-detecting-transacti.patch Normal file
View File

@ -0,0 +1,269 @@
From aebe2063884ef5e8a969d73d75e1086ea67d8033 Mon Sep 17 00:00:00 2001
From: Dave Chinner <dchinner@redhat.com>
Date: Sun, 3 Jul 2022 08:04:50 +0300
Subject: [PATCH 1/5] xfs: use current->journal_info for detecting transaction
recursion
commit 756b1c343333a5aefcc26b0409f3fd16f72281bf upstream.
Because the iomap code using PF_MEMALLOC_NOFS to detect transaction
recursion in XFS is just wrong. Remove it from the iomap code and
replace it with XFS specific internal checks using
current->journal_info instead.
[djwong: This change also realigns the lifetime of NOFS flag changes to
match the incore transaction, instead of the inconsistent scheme we have
now.]
Fixes: 9070733b4efa ("xfs: abstract PF_FSTRANS to PF_MEMALLOC_NOFS")
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Acked-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[commit b261cd005ab980c4018634a849f77e036bfd4f80 from
git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
linux-5.10.y branch]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/iomap/buffered-io.c | 7 -------
fs/xfs/libxfs/xfs_btree.c | 12 ++++++++++--
fs/xfs/xfs_aops.c | 17 +++++++++++++++--
fs/xfs/xfs_trans.c | 20 +++++---------------
fs/xfs/xfs_trans.h | 30 ++++++++++++++++++++++++++++++
5 files changed, 60 insertions(+), 26 deletions(-)
diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
index cd9f7baa5..47279fe00 100644
--- a/fs/iomap/buffered-io.c
+++ b/fs/iomap/buffered-io.c
@@ -1459,13 +1459,6 @@ iomap_do_writepage(struct page *page, struct writeback_control *wbc, void *data)
PF_MEMALLOC))
goto redirty;
- /*
- * Given that we do not allow direct reclaim to call us, we should
- * never be called in a recursive filesystem reclaim context.
- */
- if (WARN_ON_ONCE(current->flags & PF_MEMALLOC_NOFS))
- goto redirty;
-
/*
* Is this page beyond the end of the file?
*
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
index 2d25bab68..fe87ecd76 100644
--- a/fs/xfs/libxfs/xfs_btree.c
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -2814,7 +2814,7 @@ xfs_btree_split_worker(
struct xfs_btree_split_args *args = container_of(work,
struct xfs_btree_split_args, work);
unsigned long pflags;
- unsigned long new_pflags = PF_MEMALLOC_NOFS;
+ unsigned long new_pflags = 0;
/*
* we are in a transaction context here, but may also be doing work
@@ -2826,12 +2826,20 @@ xfs_btree_split_worker(
new_pflags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
current_set_flags_nested(&pflags, new_pflags);
+ xfs_trans_set_context(args->cur->bc_tp);
args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
args->key, args->curp, args->stat);
- complete(args->done);
+ xfs_trans_clear_context(args->cur->bc_tp);
current_restore_flags_nested(&pflags, new_pflags);
+
+ /*
+ * Do not access args after complete() has run here. We don't own args
+ * and the owner may run and free args before we return here.
+ */
+ complete(args->done);
+
}
/*
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 4304c6416..b4186d666 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -62,7 +62,7 @@ xfs_setfilesize_trans_alloc(
* We hand off the transaction to the completion thread now, so
* clear the flag here.
*/
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ xfs_trans_clear_context(tp);
return 0;
}
@@ -125,7 +125,7 @@ xfs_setfilesize_ioend(
* thus we need to mark ourselves as being in a transaction manually.
* Similarly for freeze protection.
*/
- current_set_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ xfs_trans_set_context(tp);
__sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
/* we abort the update if there was an IO error */
@@ -568,6 +568,12 @@ xfs_vm_writepage(
{
struct xfs_writepage_ctx wpc = { };
+ if (WARN_ON_ONCE(current->journal_info)) {
+ redirty_page_for_writepage(wbc, page);
+ unlock_page(page);
+ return 0;
+ }
+
return iomap_writepage(page, wbc, &wpc.ctx, &xfs_writeback_ops);
}
@@ -578,6 +584,13 @@ xfs_vm_writepages(
{
struct xfs_writepage_ctx wpc = { };
+ /*
+ * Writing back data in a transaction context can result in recursive
+ * transactions. This is bad, so issue a warning and get out of here.
+ */
+ if (WARN_ON_ONCE(current->journal_info))
+ return 0;
+
xfs_iflags_clear(XFS_I(mapping->host), XFS_ITRUNCATED);
return iomap_writepages(mapping, wbc, &wpc.ctx, &xfs_writeback_ops);
}
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index c94e71f74..2d7deacea 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -68,6 +68,7 @@ xfs_trans_free(
xfs_extent_busy_clear(tp->t_mountp, &tp->t_busy, false);
trace_xfs_trans_free(tp, _RET_IP_);
+ xfs_trans_clear_context(tp);
if (!(tp->t_flags & XFS_TRANS_NO_WRITECOUNT))
sb_end_intwrite(tp->t_mountp->m_super);
xfs_trans_free_dqinfo(tp);
@@ -119,7 +120,8 @@ xfs_trans_dup(
ntp->t_rtx_res = tp->t_rtx_res - tp->t_rtx_res_used;
tp->t_rtx_res = tp->t_rtx_res_used;
- ntp->t_pflags = tp->t_pflags;
+
+ xfs_trans_switch_context(tp, ntp);
/* move deferred ops over to the new tp */
xfs_defer_move(ntp, tp);
@@ -153,9 +155,6 @@ xfs_trans_reserve(
int error = 0;
bool rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
- /* Mark this thread as being in a transaction */
- current_set_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
-
/*
* Attempt to reserve the needed disk blocks by decrementing
* the number needed from the number available. This will
@@ -163,10 +162,8 @@ xfs_trans_reserve(
*/
if (blocks > 0) {
error = xfs_mod_fdblocks(mp, -((int64_t)blocks), rsvd);
- if (error != 0) {
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ if (error != 0)
return -ENOSPC;
- }
tp->t_blk_res += blocks;
}
@@ -240,9 +237,6 @@ xfs_trans_reserve(
xfs_mod_fdblocks(mp, (int64_t)blocks, rsvd);
tp->t_blk_res = 0;
}
-
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
-
return error;
}
@@ -266,6 +260,7 @@ xfs_trans_alloc(
tp = kmem_cache_zalloc(xfs_trans_zone, GFP_KERNEL | __GFP_NOFAIL);
if (!(flags & XFS_TRANS_NO_WRITECOUNT))
sb_start_intwrite(mp->m_super);
+ xfs_trans_set_context(tp);
/*
* Zero-reservation ("empty") transactions can't modify anything, so
@@ -878,7 +873,6 @@ __xfs_trans_commit(
xfs_log_commit_cil(mp, tp, &commit_lsn, regrant);
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
xfs_trans_free(tp);
/*
@@ -910,7 +904,6 @@ __xfs_trans_commit(
xfs_log_ticket_ungrant(mp->m_log, tp->t_ticket);
tp->t_ticket = NULL;
}
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
xfs_trans_free_items(tp, !!error);
xfs_trans_free(tp);
@@ -970,9 +963,6 @@ xfs_trans_cancel(
tp->t_ticket = NULL;
}
- /* mark this thread as no longer being in a transaction */
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
-
xfs_trans_free_items(tp, dirty);
xfs_trans_free(tp);
}
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index 084658946..075eeade4 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -268,4 +268,34 @@ xfs_trans_item_relog(
return lip->li_ops->iop_relog(lip, tp);
}
+static inline void
+xfs_trans_set_context(
+ struct xfs_trans *tp)
+{
+ ASSERT(current->journal_info == NULL);
+ tp->t_pflags = memalloc_nofs_save();
+ current->journal_info = tp;
+}
+
+static inline void
+xfs_trans_clear_context(
+ struct xfs_trans *tp)
+{
+ if (current->journal_info == tp) {
+ memalloc_nofs_restore(tp->t_pflags);
+ current->journal_info = NULL;
+ }
+}
+
+static inline void
+xfs_trans_switch_context(
+ struct xfs_trans *old_tp,
+ struct xfs_trans *new_tp)
+{
+ ASSERT(current->journal_info == old_tp);
+ new_tp->t_pflags = old_tp->t_pflags;
+ old_tp->t_pflags = 0;
+ current->journal_info = new_tp;
+}
+
#endif /* __XFS_TRANS_H__ */
--
2.34.1

138
kernel-rt/centos/patches/0035-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch Normal file
View File

@ -0,0 +1,138 @@
From 8182ec00803085354761bbadf0287cad7eac0e2f Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:43 -0700
Subject: [PATCH 2/5] xfs: drop submit side trans alloc for append ioends
Per-inode ioend completion batching has a log reservation deadlock
vector between preallocated append transactions and transactions
that are acquired at completion time for other purposes (i.e.,
unwritten extent conversion or COW fork remaps). For example, if the
ioend completion workqueue task executes on a batch of ioends that
are sorted such that an append ioend sits at the tail, it's possible
for the outstanding append transaction reservation to block
allocation of transactions required to process preceding ioends in
the list.
Append ioend completion is historically the common path for on-disk
inode size updates. While file extending writes may have completed
sometime earlier, the on-disk inode size is only updated after
successful writeback completion. These transactions are preallocated
serially from writeback context to mitigate concurrency and
associated log reservation pressure across completions processed by
multi-threaded workqueue tasks.
However, now that delalloc blocks unconditionally map to unwritten
extents at physical block allocation time, size updates via append
ioends are relatively rare. This means that inode size updates most
commonly occur as part of the preexisting completion time
transaction to convert unwritten extents. As a result, there is no
longer a strong need to preallocate size update transactions.
Remove the preallocation of inode size update transactions to avoid
the ioend completion processing log reservation deadlock. Instead,
continue to send all potential size extending ioends to workqueue
context for completion and allocate the transaction from that
context. This ensures that no outstanding log reservation is owned
by the ioend completion worker task when it begins to process
ioends.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7cd3099f4925d7c15887d1940ebd65acd66100f5 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 45 +++------------------------------------------
1 file changed, 3 insertions(+), 42 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index b4186d666..60943b28f 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -39,33 +39,6 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
XFS_I(ioend->io_inode)->i_d.di_size;
}
-STATIC int
-xfs_setfilesize_trans_alloc(
- struct iomap_ioend *ioend)
-{
- struct xfs_mount *mp = XFS_I(ioend->io_inode)->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- ioend->io_private = tp;
-
- /*
- * We may pass freeze protection with a transaction. So tell lockdep
- * we released it.
- */
- __sb_writers_release(ioend->io_inode->i_sb, SB_FREEZE_FS);
- /*
- * We hand off the transaction to the completion thread now, so
- * clear the flag here.
- */
- xfs_trans_clear_context(tp);
- return 0;
-}
-
/*
* Update on-disk file size now that data has been written to disk.
*/
@@ -182,12 +155,10 @@ xfs_end_ioend(
error = xfs_reflink_end_cow(ip, offset, size);
else if (ioend->io_type == IOMAP_UNWRITTEN)
error = xfs_iomap_write_unwritten(ip, offset, size, false);
- else
- ASSERT(!xfs_ioend_is_append(ioend) || ioend->io_private);
+ if (!error && xfs_ioend_is_append(ioend))
+ error = xfs_setfilesize(ip, ioend->io_offset, ioend->io_size);
done:
- if (ioend->io_private)
- error = xfs_setfilesize_ioend(ioend, error);
iomap_finish_ioends(ioend, error);
memalloc_nofs_restore(nofs_flag);
}
@@ -237,7 +208,7 @@ xfs_end_io(
static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
{
- return ioend->io_private ||
+ return xfs_ioend_is_append(ioend) ||
ioend->io_type == IOMAP_UNWRITTEN ||
(ioend->io_flags & IOMAP_F_SHARED);
}
@@ -250,8 +221,6 @@ xfs_end_bio(
struct xfs_inode *ip = XFS_I(ioend->io_inode);
unsigned long flags;
- ASSERT(xfs_ioend_needs_workqueue(ioend));
-
spin_lock_irqsave(&ip->i_ioend_lock, flags);
if (list_empty(&ip->i_ioend_list))
WARN_ON_ONCE(!queue_work(ip->i_mount->m_unwritten_workqueue,
@@ -501,14 +470,6 @@ xfs_prepare_ioend(
ioend->io_offset, ioend->io_size);
}
- /* Reserve log space if we might write beyond the on-disk inode size. */
- if (!status &&
- ((ioend->io_flags & IOMAP_F_SHARED) ||
- ioend->io_type != IOMAP_UNWRITTEN) &&
- xfs_ioend_is_append(ioend) &&
- !ioend->io_private)
- status = xfs_setfilesize_trans_alloc(ioend);
-
memalloc_nofs_restore(nofs_flag);
if (xfs_ioend_needs_workqueue(ioend))
--
2.34.1

51
kernel-rt/centos/patches/0036-xfs-open-code-ioend-needs-workqueue-helper.patch Normal file
View File

@ -0,0 +1,51 @@
From edbf1eb9032b84631031d9b43570e262f3461c24 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 3/5] xfs: open code ioend needs workqueue helper
Open code xfs_ioend_needs_workqueue() into the only remaining
caller.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7adb8f14e134d5f885d47c4ccd620836235f0b7f upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 60943b28f..a74a9ead5 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -206,13 +206,6 @@ xfs_end_io(
}
}
-static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
-{
- return xfs_ioend_is_append(ioend) ||
- ioend->io_type == IOMAP_UNWRITTEN ||
- (ioend->io_flags & IOMAP_F_SHARED);
-}
-
STATIC void
xfs_end_bio(
struct bio *bio)
@@ -472,7 +465,9 @@ xfs_prepare_ioend(
memalloc_nofs_restore(nofs_flag);
- if (xfs_ioend_needs_workqueue(ioend))
+ /* send ioends that might require a transaction to the completion wq */
+ if (xfs_ioend_is_append(ioend) || ioend->io_type == IOMAP_UNWRITTEN ||
+ (ioend->io_flags & IOMAP_F_SHARED))
ioend->io_bio->bi_end_io = xfs_end_bio;
return status;
}
--
2.34.1

94
kernel-rt/centos/patches/0037-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch Normal file
View File

@ -0,0 +1,94 @@
From 170e31793806ce5e5a9647b6340954536244518e Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 4/5] xfs: drop unused ioend private merge and setfilesize code
XFS no longer attaches anthing to ioend->io_private. Remove the
unnecessary ->io_private merging code. This removes the only remaining
user of xfs_setfilesize_ioend() so remove that function as well.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 044c6449f18f174ba8d86640936add3fc7582e49 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 46 +---------------------------------------------
1 file changed, 1 insertion(+), 45 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index a74a9ead5..23a9bd9ff 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -85,31 +85,6 @@ xfs_setfilesize(
return __xfs_setfilesize(ip, tp, offset, size);
}
-STATIC int
-xfs_setfilesize_ioend(
- struct iomap_ioend *ioend,
- int error)
-{
- struct xfs_inode *ip = XFS_I(ioend->io_inode);
- struct xfs_trans *tp = ioend->io_private;
-
- /*
- * The transaction may have been allocated in the I/O submission thread,
- * thus we need to mark ourselves as being in a transaction manually.
- * Similarly for freeze protection.
- */
- xfs_trans_set_context(tp);
- __sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
-
- /* we abort the update if there was an IO error */
- if (error) {
- xfs_trans_cancel(tp);
- return error;
- }
-
- return __xfs_setfilesize(ip, tp, ioend->io_offset, ioend->io_size);
-}
-
/*
* IO write completion.
*/
@@ -163,25 +138,6 @@ xfs_end_ioend(
memalloc_nofs_restore(nofs_flag);
}
-/*
- * If the to be merged ioend has a preallocated transaction for file
- * size updates we need to ensure the ioend it is merged into also
- * has one. If it already has one we can simply cancel the transaction
- * as it is guaranteed to be clean.
- */
-static void
-xfs_ioend_merge_private(
- struct iomap_ioend *ioend,
- struct iomap_ioend *next)
-{
- if (!ioend->io_private) {
- ioend->io_private = next->io_private;
- next->io_private = NULL;
- } else {
- xfs_setfilesize_ioend(next, -ECANCELED);
- }
-}
-
/* Finish all pending io completions. */
void
xfs_end_io(
@@ -201,7 +157,7 @@ xfs_end_io(
while ((ioend = list_first_entry_or_null(&tmp, struct iomap_ioend,
io_list))) {
list_del_init(&ioend->io_list);
- iomap_ioend_try_merge(ioend, &tmp, xfs_ioend_merge_private);
+ iomap_ioend_try_merge(ioend, &tmp, NULL);
xfs_end_ioend(ioend);
}
}
--
2.34.1

74
kernel-rt/centos/patches/0038-xfs-drop-unnecessary-setfilesize-helper.patch Normal file
View File

@ -0,0 +1,74 @@
From 2fd609b6c90a88630a50fb317473b210759b3873 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:56 -0700
Subject: [PATCH 5/5] xfs: drop unnecessary setfilesize helper
xfs_setfilesize() is the only remaining caller of the internal
__xfs_setfilesize() helper. Fold them into a single function.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit e7a3d7e792a5ad50583a2e6c35e72bd2ca6096f4 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 29 +++++++++--------------------
1 file changed, 9 insertions(+), 20 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 23a9bd9ff..9e32d4135 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -42,14 +42,20 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
/*
* Update on-disk file size now that data has been written to disk.
*/
-STATIC int
-__xfs_setfilesize(
+int
+xfs_setfilesize(
struct xfs_inode *ip,
- struct xfs_trans *tp,
xfs_off_t offset,
size_t size)
{
+ struct xfs_mount *mp = ip->i_mount;
+ struct xfs_trans *tp;
xfs_fsize_t isize;
+ int error;
+
+ error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
+ if (error)
+ return error;
xfs_ilock(ip, XFS_ILOCK_EXCL);
isize = xfs_new_eof(ip, offset + size);
@@ -68,23 +74,6 @@ __xfs_setfilesize(
return xfs_trans_commit(tp);
}
-int
-xfs_setfilesize(
- struct xfs_inode *ip,
- xfs_off_t offset,
- size_t size)
-{
- struct xfs_mount *mp = ip->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- return __xfs_setfilesize(ip, tp, offset, size);
-}
-
/*
* IO write completion.
*/
--
2.34.1

138
kernel-rt/debian/patches/0035-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch Normal file
View File

@ -0,0 +1,138 @@
From 8182ec00803085354761bbadf0287cad7eac0e2f Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:43 -0700
Subject: [PATCH 1/4] xfs: drop submit side trans alloc for append ioends
Per-inode ioend completion batching has a log reservation deadlock
vector between preallocated append transactions and transactions
that are acquired at completion time for other purposes (i.e.,
unwritten extent conversion or COW fork remaps). For example, if the
ioend completion workqueue task executes on a batch of ioends that
are sorted such that an append ioend sits at the tail, it's possible
for the outstanding append transaction reservation to block
allocation of transactions required to process preceding ioends in
the list.
Append ioend completion is historically the common path for on-disk
inode size updates. While file extending writes may have completed
sometime earlier, the on-disk inode size is only updated after
successful writeback completion. These transactions are preallocated
serially from writeback context to mitigate concurrency and
associated log reservation pressure across completions processed by
multi-threaded workqueue tasks.
However, now that delalloc blocks unconditionally map to unwritten
extents at physical block allocation time, size updates via append
ioends are relatively rare. This means that inode size updates most
commonly occur as part of the preexisting completion time
transaction to convert unwritten extents. As a result, there is no
longer a strong need to preallocate size update transactions.
Remove the preallocation of inode size update transactions to avoid
the ioend completion processing log reservation deadlock. Instead,
continue to send all potential size extending ioends to workqueue
context for completion and allocate the transaction from that
context. This ensures that no outstanding log reservation is owned
by the ioend completion worker task when it begins to process
ioends.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7cd3099f4925d7c15887d1940ebd65acd66100f5 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 45 +++------------------------------------------
1 file changed, 3 insertions(+), 42 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index b4186d666..60943b28f 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -39,33 +39,6 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
XFS_I(ioend->io_inode)->i_d.di_size;
}
-STATIC int
-xfs_setfilesize_trans_alloc(
- struct iomap_ioend *ioend)
-{
- struct xfs_mount *mp = XFS_I(ioend->io_inode)->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- ioend->io_private = tp;
-
- /*
- * We may pass freeze protection with a transaction. So tell lockdep
- * we released it.
- */
- __sb_writers_release(ioend->io_inode->i_sb, SB_FREEZE_FS);
- /*
- * We hand off the transaction to the completion thread now, so
- * clear the flag here.
- */
- xfs_trans_clear_context(tp);
- return 0;
-}
-
/*
* Update on-disk file size now that data has been written to disk.
*/
@@ -182,12 +155,10 @@ xfs_end_ioend(
error = xfs_reflink_end_cow(ip, offset, size);
else if (ioend->io_type == IOMAP_UNWRITTEN)
error = xfs_iomap_write_unwritten(ip, offset, size, false);
- else
- ASSERT(!xfs_ioend_is_append(ioend) || ioend->io_private);
+ if (!error && xfs_ioend_is_append(ioend))
+ error = xfs_setfilesize(ip, ioend->io_offset, ioend->io_size);
done:
- if (ioend->io_private)
- error = xfs_setfilesize_ioend(ioend, error);
iomap_finish_ioends(ioend, error);
memalloc_nofs_restore(nofs_flag);
}
@@ -237,7 +208,7 @@ xfs_end_io(
static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
{
- return ioend->io_private ||
+ return xfs_ioend_is_append(ioend) ||
ioend->io_type == IOMAP_UNWRITTEN ||
(ioend->io_flags & IOMAP_F_SHARED);
}
@@ -250,8 +221,6 @@ xfs_end_bio(
struct xfs_inode *ip = XFS_I(ioend->io_inode);
unsigned long flags;
- ASSERT(xfs_ioend_needs_workqueue(ioend));
-
spin_lock_irqsave(&ip->i_ioend_lock, flags);
if (list_empty(&ip->i_ioend_list))
WARN_ON_ONCE(!queue_work(ip->i_mount->m_unwritten_workqueue,
@@ -501,14 +470,6 @@ xfs_prepare_ioend(
ioend->io_offset, ioend->io_size);
}
- /* Reserve log space if we might write beyond the on-disk inode size. */
- if (!status &&
- ((ioend->io_flags & IOMAP_F_SHARED) ||
- ioend->io_type != IOMAP_UNWRITTEN) &&
- xfs_ioend_is_append(ioend) &&
- !ioend->io_private)
- status = xfs_setfilesize_trans_alloc(ioend);
-
memalloc_nofs_restore(nofs_flag);
if (xfs_ioend_needs_workqueue(ioend))
--
2.34.1

51
kernel-rt/debian/patches/0036-xfs-open-code-ioend-needs-workqueue-helper.patch Normal file
View File

@ -0,0 +1,51 @@
From edbf1eb9032b84631031d9b43570e262f3461c24 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 2/4] xfs: open code ioend needs workqueue helper
Open code xfs_ioend_needs_workqueue() into the only remaining
caller.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7adb8f14e134d5f885d47c4ccd620836235f0b7f upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 60943b28f..a74a9ead5 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -206,13 +206,6 @@ xfs_end_io(
}
}
-static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
-{
- return xfs_ioend_is_append(ioend) ||
- ioend->io_type == IOMAP_UNWRITTEN ||
- (ioend->io_flags & IOMAP_F_SHARED);
-}
-
STATIC void
xfs_end_bio(
struct bio *bio)
@@ -472,7 +465,9 @@ xfs_prepare_ioend(
memalloc_nofs_restore(nofs_flag);
- if (xfs_ioend_needs_workqueue(ioend))
+ /* send ioends that might require a transaction to the completion wq */
+ if (xfs_ioend_is_append(ioend) || ioend->io_type == IOMAP_UNWRITTEN ||
+ (ioend->io_flags & IOMAP_F_SHARED))
ioend->io_bio->bi_end_io = xfs_end_bio;
return status;
}
--
2.34.1

94
kernel-rt/debian/patches/0037-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch Normal file
View File

@ -0,0 +1,94 @@
From 170e31793806ce5e5a9647b6340954536244518e Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 3/4] xfs: drop unused ioend private merge and setfilesize code
XFS no longer attaches anthing to ioend->io_private. Remove the
unnecessary ->io_private merging code. This removes the only remaining
user of xfs_setfilesize_ioend() so remove that function as well.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 044c6449f18f174ba8d86640936add3fc7582e49 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 46 +---------------------------------------------
1 file changed, 1 insertion(+), 45 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index a74a9ead5..23a9bd9ff 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -85,31 +85,6 @@ xfs_setfilesize(
return __xfs_setfilesize(ip, tp, offset, size);
}
-STATIC int
-xfs_setfilesize_ioend(
- struct iomap_ioend *ioend,
- int error)
-{
- struct xfs_inode *ip = XFS_I(ioend->io_inode);
- struct xfs_trans *tp = ioend->io_private;
-
- /*
- * The transaction may have been allocated in the I/O submission thread,
- * thus we need to mark ourselves as being in a transaction manually.
- * Similarly for freeze protection.
- */
- xfs_trans_set_context(tp);
- __sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
-
- /* we abort the update if there was an IO error */
- if (error) {
- xfs_trans_cancel(tp);
- return error;
- }
-
- return __xfs_setfilesize(ip, tp, ioend->io_offset, ioend->io_size);
-}
-
/*
* IO write completion.
*/
@@ -163,25 +138,6 @@ xfs_end_ioend(
memalloc_nofs_restore(nofs_flag);
}
-/*
- * If the to be merged ioend has a preallocated transaction for file
- * size updates we need to ensure the ioend it is merged into also
- * has one. If it already has one we can simply cancel the transaction
- * as it is guaranteed to be clean.
- */
-static void
-xfs_ioend_merge_private(
- struct iomap_ioend *ioend,
- struct iomap_ioend *next)
-{
- if (!ioend->io_private) {
- ioend->io_private = next->io_private;
- next->io_private = NULL;
- } else {
- xfs_setfilesize_ioend(next, -ECANCELED);
- }
-}
-
/* Finish all pending io completions. */
void
xfs_end_io(
@@ -201,7 +157,7 @@ xfs_end_io(
while ((ioend = list_first_entry_or_null(&tmp, struct iomap_ioend,
io_list))) {
list_del_init(&ioend->io_list);
- iomap_ioend_try_merge(ioend, &tmp, xfs_ioend_merge_private);
+ iomap_ioend_try_merge(ioend, &tmp, NULL);
xfs_end_ioend(ioend);
}
}
--
2.34.1

74
kernel-rt/debian/patches/0038-xfs-drop-unnecessary-setfilesize-helper.patch Normal file
View File

@ -0,0 +1,74 @@
From 2fd609b6c90a88630a50fb317473b210759b3873 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:56 -0700
Subject: [PATCH 4/4] xfs: drop unnecessary setfilesize helper
xfs_setfilesize() is the only remaining caller of the internal
__xfs_setfilesize() helper. Fold them into a single function.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit e7a3d7e792a5ad50583a2e6c35e72bd2ca6096f4 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 29 +++++++++--------------------
1 file changed, 9 insertions(+), 20 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 23a9bd9ff..9e32d4135 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -42,14 +42,20 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
/*
* Update on-disk file size now that data has been written to disk.
*/
-STATIC int
-__xfs_setfilesize(
+int
+xfs_setfilesize(
struct xfs_inode *ip,
- struct xfs_trans *tp,
xfs_off_t offset,
size_t size)
{
+ struct xfs_mount *mp = ip->i_mount;
+ struct xfs_trans *tp;
xfs_fsize_t isize;
+ int error;
+
+ error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
+ if (error)
+ return error;
xfs_ilock(ip, XFS_ILOCK_EXCL);
isize = xfs_new_eof(ip, offset + size);
@@ -68,23 +74,6 @@ __xfs_setfilesize(
return xfs_trans_commit(tp);
}
-int
-xfs_setfilesize(
- struct xfs_inode *ip,
- xfs_off_t offset,
- size_t size)
-{
- struct xfs_mount *mp = ip->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- return __xfs_setfilesize(ip, tp, offset, size);
-}
-
/*
* IO write completion.
*/
--
2.34.1

4
kernel-rt/debian/patches/series
View File

@ -30,3 +30,7 @@
0032-KVM-x86-Expose-AVX512_FP16-for-supported-CPUID.patch
0033-tools-headers-cpufeatures-Sync-with-the-kernel-sourc.patch
0034-rcu-Avoid-running-boost-kthreads-on-isolated-CPUs.patch
0035-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch
0036-xfs-open-code-ioend-needs-workqueue-helper.patch
0037-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch
0038-xfs-drop-unnecessary-setfilesize-helper.patch

5
kernel-std/centos/kernel.spec
View File

@ -840,6 +840,11 @@ Patch29: 0030-x86-Enumerate-AVX512-FP16-CPUID-feature-flag.patch
Patch30: 0031-KVM-x86-Expose-AVX512_FP16-for-supported-CPUID.patch
Patch31: 0032-tools-headers-cpufeatures-Sync-with-the-kernel-sourc.patch
Patch32: 0033-rcu-Avoid-running-boost-kthreads-on-isolated-CPUs.patch
Patch33: 0034-xfs-use-current-journal_info-for-detecting-transacti.patch
Patch34: 0035-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch
Patch35: 0036-xfs-open-code-ioend-needs-workqueue-helper.patch
Patch36: 0037-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch
Patch37: 0038-xfs-drop-unnecessary-setfilesize-helper.patch
# END OF PATCH DEFINITIONS
%endif

269
kernel-std/centos/patches/0034-xfs-use-current-journal_info-for-detecting-transacti.patch Normal file
View File

@ -0,0 +1,269 @@
From aebe2063884ef5e8a969d73d75e1086ea67d8033 Mon Sep 17 00:00:00 2001
From: Dave Chinner <dchinner@redhat.com>
Date: Sun, 3 Jul 2022 08:04:50 +0300
Subject: [PATCH 1/5] xfs: use current->journal_info for detecting transaction
recursion
commit 756b1c343333a5aefcc26b0409f3fd16f72281bf upstream.
Because the iomap code using PF_MEMALLOC_NOFS to detect transaction
recursion in XFS is just wrong. Remove it from the iomap code and
replace it with XFS specific internal checks using
current->journal_info instead.
[djwong: This change also realigns the lifetime of NOFS flag changes to
match the incore transaction, instead of the inconsistent scheme we have
now.]
Fixes: 9070733b4efa ("xfs: abstract PF_FSTRANS to PF_MEMALLOC_NOFS")
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Acked-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[commit b261cd005ab980c4018634a849f77e036bfd4f80 from
git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
linux-5.10.y branch]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/iomap/buffered-io.c | 7 -------
fs/xfs/libxfs/xfs_btree.c | 12 ++++++++++--
fs/xfs/xfs_aops.c | 17 +++++++++++++++--
fs/xfs/xfs_trans.c | 20 +++++---------------
fs/xfs/xfs_trans.h | 30 ++++++++++++++++++++++++++++++
5 files changed, 60 insertions(+), 26 deletions(-)
diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
index cd9f7baa5..47279fe00 100644
--- a/fs/iomap/buffered-io.c
+++ b/fs/iomap/buffered-io.c
@@ -1459,13 +1459,6 @@ iomap_do_writepage(struct page *page, struct writeback_control *wbc, void *data)
PF_MEMALLOC))
goto redirty;
- /*
- * Given that we do not allow direct reclaim to call us, we should
- * never be called in a recursive filesystem reclaim context.
- */
- if (WARN_ON_ONCE(current->flags & PF_MEMALLOC_NOFS))
- goto redirty;
-
/*
* Is this page beyond the end of the file?
*
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
index 2d25bab68..fe87ecd76 100644
--- a/fs/xfs/libxfs/xfs_btree.c
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -2814,7 +2814,7 @@ xfs_btree_split_worker(
struct xfs_btree_split_args *args = container_of(work,
struct xfs_btree_split_args, work);
unsigned long pflags;
- unsigned long new_pflags = PF_MEMALLOC_NOFS;
+ unsigned long new_pflags = 0;
/*
* we are in a transaction context here, but may also be doing work
@@ -2826,12 +2826,20 @@ xfs_btree_split_worker(
new_pflags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
current_set_flags_nested(&pflags, new_pflags);
+ xfs_trans_set_context(args->cur->bc_tp);
args->result = __xfs_btree_split(args->cur, args->level, args->ptrp,
args->key, args->curp, args->stat);
- complete(args->done);
+ xfs_trans_clear_context(args->cur->bc_tp);
current_restore_flags_nested(&pflags, new_pflags);
+
+ /*
+ * Do not access args after complete() has run here. We don't own args
+ * and the owner may run and free args before we return here.
+ */
+ complete(args->done);
+
}
/*
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 4304c6416..b4186d666 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -62,7 +62,7 @@ xfs_setfilesize_trans_alloc(
* We hand off the transaction to the completion thread now, so
* clear the flag here.
*/
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ xfs_trans_clear_context(tp);
return 0;
}
@@ -125,7 +125,7 @@ xfs_setfilesize_ioend(
* thus we need to mark ourselves as being in a transaction manually.
* Similarly for freeze protection.
*/
- current_set_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ xfs_trans_set_context(tp);
__sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
/* we abort the update if there was an IO error */
@@ -568,6 +568,12 @@ xfs_vm_writepage(
{
struct xfs_writepage_ctx wpc = { };
+ if (WARN_ON_ONCE(current->journal_info)) {
+ redirty_page_for_writepage(wbc, page);
+ unlock_page(page);
+ return 0;
+ }
+
return iomap_writepage(page, wbc, &wpc.ctx, &xfs_writeback_ops);
}
@@ -578,6 +584,13 @@ xfs_vm_writepages(
{
struct xfs_writepage_ctx wpc = { };
+ /*
+ * Writing back data in a transaction context can result in recursive
+ * transactions. This is bad, so issue a warning and get out of here.
+ */
+ if (WARN_ON_ONCE(current->journal_info))
+ return 0;
+
xfs_iflags_clear(XFS_I(mapping->host), XFS_ITRUNCATED);
return iomap_writepages(mapping, wbc, &wpc.ctx, &xfs_writeback_ops);
}
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index c94e71f74..2d7deacea 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -68,6 +68,7 @@ xfs_trans_free(
xfs_extent_busy_clear(tp->t_mountp, &tp->t_busy, false);
trace_xfs_trans_free(tp, _RET_IP_);
+ xfs_trans_clear_context(tp);
if (!(tp->t_flags & XFS_TRANS_NO_WRITECOUNT))
sb_end_intwrite(tp->t_mountp->m_super);
xfs_trans_free_dqinfo(tp);
@@ -119,7 +120,8 @@ xfs_trans_dup(
ntp->t_rtx_res = tp->t_rtx_res - tp->t_rtx_res_used;
tp->t_rtx_res = tp->t_rtx_res_used;
- ntp->t_pflags = tp->t_pflags;
+
+ xfs_trans_switch_context(tp, ntp);
/* move deferred ops over to the new tp */
xfs_defer_move(ntp, tp);
@@ -153,9 +155,6 @@ xfs_trans_reserve(
int error = 0;
bool rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
- /* Mark this thread as being in a transaction */
- current_set_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
-
/*
* Attempt to reserve the needed disk blocks by decrementing
* the number needed from the number available. This will
@@ -163,10 +162,8 @@ xfs_trans_reserve(
*/
if (blocks > 0) {
error = xfs_mod_fdblocks(mp, -((int64_t)blocks), rsvd);
- if (error != 0) {
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ if (error != 0)
return -ENOSPC;
- }
tp->t_blk_res += blocks;
}
@@ -240,9 +237,6 @@ xfs_trans_reserve(
xfs_mod_fdblocks(mp, (int64_t)blocks, rsvd);
tp->t_blk_res = 0;
}
-
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
-
return error;
}
@@ -266,6 +260,7 @@ xfs_trans_alloc(
tp = kmem_cache_zalloc(xfs_trans_zone, GFP_KERNEL | __GFP_NOFAIL);
if (!(flags & XFS_TRANS_NO_WRITECOUNT))
sb_start_intwrite(mp->m_super);
+ xfs_trans_set_context(tp);
/*
* Zero-reservation ("empty") transactions can't modify anything, so
@@ -878,7 +873,6 @@ __xfs_trans_commit(
xfs_log_commit_cil(mp, tp, &commit_lsn, regrant);
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
xfs_trans_free(tp);
/*
@@ -910,7 +904,6 @@ __xfs_trans_commit(
xfs_log_ticket_ungrant(mp->m_log, tp->t_ticket);
tp->t_ticket = NULL;
}
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
xfs_trans_free_items(tp, !!error);
xfs_trans_free(tp);
@@ -970,9 +963,6 @@ xfs_trans_cancel(
tp->t_ticket = NULL;
}
- /* mark this thread as no longer being in a transaction */
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
-
xfs_trans_free_items(tp, dirty);
xfs_trans_free(tp);
}
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index 084658946..075eeade4 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -268,4 +268,34 @@ xfs_trans_item_relog(
return lip->li_ops->iop_relog(lip, tp);
}
+static inline void
+xfs_trans_set_context(
+ struct xfs_trans *tp)
+{
+ ASSERT(current->journal_info == NULL);
+ tp->t_pflags = memalloc_nofs_save();
+ current->journal_info = tp;
+}
+
+static inline void
+xfs_trans_clear_context(
+ struct xfs_trans *tp)
+{
+ if (current->journal_info == tp) {
+ memalloc_nofs_restore(tp->t_pflags);
+ current->journal_info = NULL;
+ }
+}
+
+static inline void
+xfs_trans_switch_context(
+ struct xfs_trans *old_tp,
+ struct xfs_trans *new_tp)
+{
+ ASSERT(current->journal_info == old_tp);
+ new_tp->t_pflags = old_tp->t_pflags;
+ old_tp->t_pflags = 0;
+ current->journal_info = new_tp;
+}
+
#endif /* __XFS_TRANS_H__ */
--
2.34.1

138
kernel-std/centos/patches/0035-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch Normal file
View File

@ -0,0 +1,138 @@
From 8182ec00803085354761bbadf0287cad7eac0e2f Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:43 -0700
Subject: [PATCH 2/5] xfs: drop submit side trans alloc for append ioends
Per-inode ioend completion batching has a log reservation deadlock
vector between preallocated append transactions and transactions
that are acquired at completion time for other purposes (i.e.,
unwritten extent conversion or COW fork remaps). For example, if the
ioend completion workqueue task executes on a batch of ioends that
are sorted such that an append ioend sits at the tail, it's possible
for the outstanding append transaction reservation to block
allocation of transactions required to process preceding ioends in
the list.
Append ioend completion is historically the common path for on-disk
inode size updates. While file extending writes may have completed
sometime earlier, the on-disk inode size is only updated after
successful writeback completion. These transactions are preallocated
serially from writeback context to mitigate concurrency and
associated log reservation pressure across completions processed by
multi-threaded workqueue tasks.
However, now that delalloc blocks unconditionally map to unwritten
extents at physical block allocation time, size updates via append
ioends are relatively rare. This means that inode size updates most
commonly occur as part of the preexisting completion time
transaction to convert unwritten extents. As a result, there is no
longer a strong need to preallocate size update transactions.
Remove the preallocation of inode size update transactions to avoid
the ioend completion processing log reservation deadlock. Instead,
continue to send all potential size extending ioends to workqueue
context for completion and allocate the transaction from that
context. This ensures that no outstanding log reservation is owned
by the ioend completion worker task when it begins to process
ioends.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7cd3099f4925d7c15887d1940ebd65acd66100f5 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 45 +++------------------------------------------
1 file changed, 3 insertions(+), 42 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index b4186d666..60943b28f 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -39,33 +39,6 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
XFS_I(ioend->io_inode)->i_d.di_size;
}
-STATIC int
-xfs_setfilesize_trans_alloc(
- struct iomap_ioend *ioend)
-{
- struct xfs_mount *mp = XFS_I(ioend->io_inode)->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- ioend->io_private = tp;
-
- /*
- * We may pass freeze protection with a transaction. So tell lockdep
- * we released it.
- */
- __sb_writers_release(ioend->io_inode->i_sb, SB_FREEZE_FS);
- /*
- * We hand off the transaction to the completion thread now, so
- * clear the flag here.
- */
- xfs_trans_clear_context(tp);
- return 0;
-}
-
/*
* Update on-disk file size now that data has been written to disk.
*/
@@ -182,12 +155,10 @@ xfs_end_ioend(
error = xfs_reflink_end_cow(ip, offset, size);
else if (ioend->io_type == IOMAP_UNWRITTEN)
error = xfs_iomap_write_unwritten(ip, offset, size, false);
- else
- ASSERT(!xfs_ioend_is_append(ioend) || ioend->io_private);
+ if (!error && xfs_ioend_is_append(ioend))
+ error = xfs_setfilesize(ip, ioend->io_offset, ioend->io_size);
done:
- if (ioend->io_private)
- error = xfs_setfilesize_ioend(ioend, error);
iomap_finish_ioends(ioend, error);
memalloc_nofs_restore(nofs_flag);
}
@@ -237,7 +208,7 @@ xfs_end_io(
static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
{
- return ioend->io_private ||
+ return xfs_ioend_is_append(ioend) ||
ioend->io_type == IOMAP_UNWRITTEN ||
(ioend->io_flags & IOMAP_F_SHARED);
}
@@ -250,8 +221,6 @@ xfs_end_bio(
struct xfs_inode *ip = XFS_I(ioend->io_inode);
unsigned long flags;
- ASSERT(xfs_ioend_needs_workqueue(ioend));
-
spin_lock_irqsave(&ip->i_ioend_lock, flags);
if (list_empty(&ip->i_ioend_list))
WARN_ON_ONCE(!queue_work(ip->i_mount->m_unwritten_workqueue,
@@ -501,14 +470,6 @@ xfs_prepare_ioend(
ioend->io_offset, ioend->io_size);
}
- /* Reserve log space if we might write beyond the on-disk inode size. */
- if (!status &&
- ((ioend->io_flags & IOMAP_F_SHARED) ||
- ioend->io_type != IOMAP_UNWRITTEN) &&
- xfs_ioend_is_append(ioend) &&
- !ioend->io_private)
- status = xfs_setfilesize_trans_alloc(ioend);
-
memalloc_nofs_restore(nofs_flag);
if (xfs_ioend_needs_workqueue(ioend))
--
2.34.1

51
kernel-std/centos/patches/0036-xfs-open-code-ioend-needs-workqueue-helper.patch Normal file
View File

@ -0,0 +1,51 @@
From edbf1eb9032b84631031d9b43570e262f3461c24 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 3/5] xfs: open code ioend needs workqueue helper
Open code xfs_ioend_needs_workqueue() into the only remaining
caller.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7adb8f14e134d5f885d47c4ccd620836235f0b7f upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 60943b28f..a74a9ead5 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -206,13 +206,6 @@ xfs_end_io(
}
}
-static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
-{
- return xfs_ioend_is_append(ioend) ||
- ioend->io_type == IOMAP_UNWRITTEN ||
- (ioend->io_flags & IOMAP_F_SHARED);
-}
-
STATIC void
xfs_end_bio(
struct bio *bio)
@@ -472,7 +465,9 @@ xfs_prepare_ioend(
memalloc_nofs_restore(nofs_flag);
- if (xfs_ioend_needs_workqueue(ioend))
+ /* send ioends that might require a transaction to the completion wq */
+ if (xfs_ioend_is_append(ioend) || ioend->io_type == IOMAP_UNWRITTEN ||
+ (ioend->io_flags & IOMAP_F_SHARED))
ioend->io_bio->bi_end_io = xfs_end_bio;
return status;
}
--
2.34.1

94
kernel-std/centos/patches/0037-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch Normal file
View File

@ -0,0 +1,94 @@
From 170e31793806ce5e5a9647b6340954536244518e Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 4/5] xfs: drop unused ioend private merge and setfilesize code
XFS no longer attaches anthing to ioend->io_private. Remove the
unnecessary ->io_private merging code. This removes the only remaining
user of xfs_setfilesize_ioend() so remove that function as well.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 044c6449f18f174ba8d86640936add3fc7582e49 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 46 +---------------------------------------------
1 file changed, 1 insertion(+), 45 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index a74a9ead5..23a9bd9ff 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -85,31 +85,6 @@ xfs_setfilesize(
return __xfs_setfilesize(ip, tp, offset, size);
}
-STATIC int
-xfs_setfilesize_ioend(
- struct iomap_ioend *ioend,
- int error)
-{
- struct xfs_inode *ip = XFS_I(ioend->io_inode);
- struct xfs_trans *tp = ioend->io_private;
-
- /*
- * The transaction may have been allocated in the I/O submission thread,
- * thus we need to mark ourselves as being in a transaction manually.
- * Similarly for freeze protection.
- */
- xfs_trans_set_context(tp);
- __sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
-
- /* we abort the update if there was an IO error */
- if (error) {
- xfs_trans_cancel(tp);
- return error;
- }
-
- return __xfs_setfilesize(ip, tp, ioend->io_offset, ioend->io_size);
-}
-
/*
* IO write completion.
*/
@@ -163,25 +138,6 @@ xfs_end_ioend(
memalloc_nofs_restore(nofs_flag);
}
-/*
- * If the to be merged ioend has a preallocated transaction for file
- * size updates we need to ensure the ioend it is merged into also
- * has one. If it already has one we can simply cancel the transaction
- * as it is guaranteed to be clean.
- */
-static void
-xfs_ioend_merge_private(
- struct iomap_ioend *ioend,
- struct iomap_ioend *next)
-{
- if (!ioend->io_private) {
- ioend->io_private = next->io_private;
- next->io_private = NULL;
- } else {
- xfs_setfilesize_ioend(next, -ECANCELED);
- }
-}
-
/* Finish all pending io completions. */
void
xfs_end_io(
@@ -201,7 +157,7 @@ xfs_end_io(
while ((ioend = list_first_entry_or_null(&tmp, struct iomap_ioend,
io_list))) {
list_del_init(&ioend->io_list);
- iomap_ioend_try_merge(ioend, &tmp, xfs_ioend_merge_private);
+ iomap_ioend_try_merge(ioend, &tmp, NULL);
xfs_end_ioend(ioend);
}
}
--
2.34.1

74
kernel-std/centos/patches/0038-xfs-drop-unnecessary-setfilesize-helper.patch Normal file
View File

@ -0,0 +1,74 @@
From 2fd609b6c90a88630a50fb317473b210759b3873 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:56 -0700
Subject: [PATCH 5/5] xfs: drop unnecessary setfilesize helper
xfs_setfilesize() is the only remaining caller of the internal
__xfs_setfilesize() helper. Fold them into a single function.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit e7a3d7e792a5ad50583a2e6c35e72bd2ca6096f4 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 29 +++++++++--------------------
1 file changed, 9 insertions(+), 20 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 23a9bd9ff..9e32d4135 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -42,14 +42,20 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
/*
* Update on-disk file size now that data has been written to disk.
*/
-STATIC int
-__xfs_setfilesize(
+int
+xfs_setfilesize(
struct xfs_inode *ip,
- struct xfs_trans *tp,
xfs_off_t offset,
size_t size)
{
+ struct xfs_mount *mp = ip->i_mount;
+ struct xfs_trans *tp;
xfs_fsize_t isize;
+ int error;
+
+ error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
+ if (error)
+ return error;
xfs_ilock(ip, XFS_ILOCK_EXCL);
isize = xfs_new_eof(ip, offset + size);
@@ -68,23 +74,6 @@ __xfs_setfilesize(
return xfs_trans_commit(tp);
}
-int
-xfs_setfilesize(
- struct xfs_inode *ip,
- xfs_off_t offset,
- size_t size)
-{
- struct xfs_mount *mp = ip->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- return __xfs_setfilesize(ip, tp, offset, size);
-}
-
/*
* IO write completion.
*/
--
2.34.1

138
kernel-std/debian/patches/0034-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch Normal file
View File

@ -0,0 +1,138 @@
From 8182ec00803085354761bbadf0287cad7eac0e2f Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:43 -0700
Subject: [PATCH 1/4] xfs: drop submit side trans alloc for append ioends
Per-inode ioend completion batching has a log reservation deadlock
vector between preallocated append transactions and transactions
that are acquired at completion time for other purposes (i.e.,
unwritten extent conversion or COW fork remaps). For example, if the
ioend completion workqueue task executes on a batch of ioends that
are sorted such that an append ioend sits at the tail, it's possible
for the outstanding append transaction reservation to block
allocation of transactions required to process preceding ioends in
the list.
Append ioend completion is historically the common path for on-disk
inode size updates. While file extending writes may have completed
sometime earlier, the on-disk inode size is only updated after
successful writeback completion. These transactions are preallocated
serially from writeback context to mitigate concurrency and
associated log reservation pressure across completions processed by
multi-threaded workqueue tasks.
However, now that delalloc blocks unconditionally map to unwritten
extents at physical block allocation time, size updates via append
ioends are relatively rare. This means that inode size updates most
commonly occur as part of the preexisting completion time
transaction to convert unwritten extents. As a result, there is no
longer a strong need to preallocate size update transactions.
Remove the preallocation of inode size update transactions to avoid
the ioend completion processing log reservation deadlock. Instead,
continue to send all potential size extending ioends to workqueue
context for completion and allocate the transaction from that
context. This ensures that no outstanding log reservation is owned
by the ioend completion worker task when it begins to process
ioends.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7cd3099f4925d7c15887d1940ebd65acd66100f5 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 45 +++------------------------------------------
1 file changed, 3 insertions(+), 42 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index b4186d666..60943b28f 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -39,33 +39,6 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
XFS_I(ioend->io_inode)->i_d.di_size;
}
-STATIC int
-xfs_setfilesize_trans_alloc(
- struct iomap_ioend *ioend)
-{
- struct xfs_mount *mp = XFS_I(ioend->io_inode)->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- ioend->io_private = tp;
-
- /*
- * We may pass freeze protection with a transaction. So tell lockdep
- * we released it.
- */
- __sb_writers_release(ioend->io_inode->i_sb, SB_FREEZE_FS);
- /*
- * We hand off the transaction to the completion thread now, so
- * clear the flag here.
- */
- xfs_trans_clear_context(tp);
- return 0;
-}
-
/*
* Update on-disk file size now that data has been written to disk.
*/
@@ -182,12 +155,10 @@ xfs_end_ioend(
error = xfs_reflink_end_cow(ip, offset, size);
else if (ioend->io_type == IOMAP_UNWRITTEN)
error = xfs_iomap_write_unwritten(ip, offset, size, false);
- else
- ASSERT(!xfs_ioend_is_append(ioend) || ioend->io_private);
+ if (!error && xfs_ioend_is_append(ioend))
+ error = xfs_setfilesize(ip, ioend->io_offset, ioend->io_size);
done:
- if (ioend->io_private)
- error = xfs_setfilesize_ioend(ioend, error);
iomap_finish_ioends(ioend, error);
memalloc_nofs_restore(nofs_flag);
}
@@ -237,7 +208,7 @@ xfs_end_io(
static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
{
- return ioend->io_private ||
+ return xfs_ioend_is_append(ioend) ||
ioend->io_type == IOMAP_UNWRITTEN ||
(ioend->io_flags & IOMAP_F_SHARED);
}
@@ -250,8 +221,6 @@ xfs_end_bio(
struct xfs_inode *ip = XFS_I(ioend->io_inode);
unsigned long flags;
- ASSERT(xfs_ioend_needs_workqueue(ioend));
-
spin_lock_irqsave(&ip->i_ioend_lock, flags);
if (list_empty(&ip->i_ioend_list))
WARN_ON_ONCE(!queue_work(ip->i_mount->m_unwritten_workqueue,
@@ -501,14 +470,6 @@ xfs_prepare_ioend(
ioend->io_offset, ioend->io_size);
}
- /* Reserve log space if we might write beyond the on-disk inode size. */
- if (!status &&
- ((ioend->io_flags & IOMAP_F_SHARED) ||
- ioend->io_type != IOMAP_UNWRITTEN) &&
- xfs_ioend_is_append(ioend) &&
- !ioend->io_private)
- status = xfs_setfilesize_trans_alloc(ioend);
-
memalloc_nofs_restore(nofs_flag);
if (xfs_ioend_needs_workqueue(ioend))
--
2.34.1

51
kernel-std/debian/patches/0035-xfs-open-code-ioend-needs-workqueue-helper.patch Normal file
View File

@ -0,0 +1,51 @@
From edbf1eb9032b84631031d9b43570e262f3461c24 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 2/4] xfs: open code ioend needs workqueue helper
Open code xfs_ioend_needs_workqueue() into the only remaining
caller.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 7adb8f14e134d5f885d47c4ccd620836235f0b7f upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 11 +++--------
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 60943b28f..a74a9ead5 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -206,13 +206,6 @@ xfs_end_io(
}
}
-static inline bool xfs_ioend_needs_workqueue(struct iomap_ioend *ioend)
-{
- return xfs_ioend_is_append(ioend) ||
- ioend->io_type == IOMAP_UNWRITTEN ||
- (ioend->io_flags & IOMAP_F_SHARED);
-}
-
STATIC void
xfs_end_bio(
struct bio *bio)
@@ -472,7 +465,9 @@ xfs_prepare_ioend(
memalloc_nofs_restore(nofs_flag);
- if (xfs_ioend_needs_workqueue(ioend))
+ /* send ioends that might require a transaction to the completion wq */
+ if (xfs_ioend_is_append(ioend) || ioend->io_type == IOMAP_UNWRITTEN ||
+ (ioend->io_flags & IOMAP_F_SHARED))
ioend->io_bio->bi_end_io = xfs_end_bio;
return status;
}
--
2.34.1

94
kernel-std/debian/patches/0036-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch Normal file
View File

@ -0,0 +1,94 @@
From 170e31793806ce5e5a9647b6340954536244518e Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:55 -0700
Subject: [PATCH 3/4] xfs: drop unused ioend private merge and setfilesize code
XFS no longer attaches anthing to ioend->io_private. Remove the
unnecessary ->io_private merging code. This removes the only remaining
user of xfs_setfilesize_ioend() so remove that function as well.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit 044c6449f18f174ba8d86640936add3fc7582e49 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 46 +---------------------------------------------
1 file changed, 1 insertion(+), 45 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index a74a9ead5..23a9bd9ff 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -85,31 +85,6 @@ xfs_setfilesize(
return __xfs_setfilesize(ip, tp, offset, size);
}
-STATIC int
-xfs_setfilesize_ioend(
- struct iomap_ioend *ioend,
- int error)
-{
- struct xfs_inode *ip = XFS_I(ioend->io_inode);
- struct xfs_trans *tp = ioend->io_private;
-
- /*
- * The transaction may have been allocated in the I/O submission thread,
- * thus we need to mark ourselves as being in a transaction manually.
- * Similarly for freeze protection.
- */
- xfs_trans_set_context(tp);
- __sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
-
- /* we abort the update if there was an IO error */
- if (error) {
- xfs_trans_cancel(tp);
- return error;
- }
-
- return __xfs_setfilesize(ip, tp, ioend->io_offset, ioend->io_size);
-}
-
/*
* IO write completion.
*/
@@ -163,25 +138,6 @@ xfs_end_ioend(
memalloc_nofs_restore(nofs_flag);
}
-/*
- * If the to be merged ioend has a preallocated transaction for file
- * size updates we need to ensure the ioend it is merged into also
- * has one. If it already has one we can simply cancel the transaction
- * as it is guaranteed to be clean.
- */
-static void
-xfs_ioend_merge_private(
- struct iomap_ioend *ioend,
- struct iomap_ioend *next)
-{
- if (!ioend->io_private) {
- ioend->io_private = next->io_private;
- next->io_private = NULL;
- } else {
- xfs_setfilesize_ioend(next, -ECANCELED);
- }
-}
-
/* Finish all pending io completions. */
void
xfs_end_io(
@@ -201,7 +157,7 @@ xfs_end_io(
while ((ioend = list_first_entry_or_null(&tmp, struct iomap_ioend,
io_list))) {
list_del_init(&ioend->io_list);
- iomap_ioend_try_merge(ioend, &tmp, xfs_ioend_merge_private);
+ iomap_ioend_try_merge(ioend, &tmp, NULL);
xfs_end_ioend(ioend);
}
}
--
2.34.1

74
kernel-std/debian/patches/0037-xfs-drop-unnecessary-setfilesize-helper.patch Normal file
View File

@ -0,0 +1,74 @@
From 2fd609b6c90a88630a50fb317473b210759b3873 Mon Sep 17 00:00:00 2001
From: Brian Foster <bfoster@redhat.com>
Date: Fri, 9 Apr 2021 10:27:56 -0700
Subject: [PATCH 4/4] xfs: drop unnecessary setfilesize helper
xfs_setfilesize() is the only remaining caller of the internal
__xfs_setfilesize() helper. Fold them into a single function.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
[commit e7a3d7e792a5ad50583a2e6c35e72bd2ca6096f4 upstream
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git]
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
fs/xfs/xfs_aops.c | 29 +++++++++--------------------
1 file changed, 9 insertions(+), 20 deletions(-)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 23a9bd9ff..9e32d4135 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -42,14 +42,20 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
/*
* Update on-disk file size now that data has been written to disk.
*/
-STATIC int
-__xfs_setfilesize(
+int
+xfs_setfilesize(
struct xfs_inode *ip,
- struct xfs_trans *tp,
xfs_off_t offset,
size_t size)
{
+ struct xfs_mount *mp = ip->i_mount;
+ struct xfs_trans *tp;
xfs_fsize_t isize;
+ int error;
+
+ error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
+ if (error)
+ return error;
xfs_ilock(ip, XFS_ILOCK_EXCL);
isize = xfs_new_eof(ip, offset + size);
@@ -68,23 +74,6 @@ __xfs_setfilesize(
return xfs_trans_commit(tp);
}
-int
-xfs_setfilesize(
- struct xfs_inode *ip,
- xfs_off_t offset,
- size_t size)
-{
- struct xfs_mount *mp = ip->i_mount;
- struct xfs_trans *tp;
- int error;
-
- error = xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp);
- if (error)
- return error;
-
- return __xfs_setfilesize(ip, tp, offset, size);
-}
-
/*
* IO write completion.
*/
--
2.34.1

4
kernel-std/debian/patches/series
View File

@ -29,3 +29,7 @@
0031-KVM-x86-Expose-AVX512_FP16-for-supported-CPUID.patch
0032-tools-headers-cpufeatures-Sync-with-the-kernel-sourc.patch
0033-rcu-Avoid-running-boost-kthreads-on-isolated-CPUs.patch
0034-xfs-drop-submit-side-trans-alloc-for-append-ioends.patch
0035-xfs-open-code-ioend-needs-workqueue-helper.patch
0036-xfs-drop-unused-ioend-private-merge-and-setfilesize-.patch
0037-xfs-drop-unnecessary-setfilesize-helper.patch