* [PATCH v2] xfs: reintroduce PF_FSTRANS for transaction reservation recursion protection
@ 2020-06-24 15:05 Yafang Shao
2020-06-26 9:02 ` Christoph Hellwig
0 siblings, 1 reply; 3+ messages in thread
From: Yafang Shao @ 2020-06-24 15:05 UTC (permalink / raw)
To: david, mhocko, darrick.wong, hch, akpm, bfoster, vbabka, holger
Cc: linux-xfs, linux-fsdevel, linux-mm, Yafang Shao
PF_FSTRANS which is used to avoid transaction reservation recursion, is
dropped since commit 9070733b4efa ("xfs: abstract PF_FSTRANS to
PF_MEMALLOC_NOFS") and commit 7dea19f9ee63 ("mm: introduce
memalloc_nofs_{save,restore} API") and replaced by PF_MEMALLOC_NOFS which
means to avoid filesystem reclaim recursion. That change is subtle.
Let's take the exmple of the check of WARN_ON_ONCE(current->flags &
PF_MEMALLOC_NOFS)) to explain why this abstraction from PF_FSTRANS to
PF_MEMALLOC_NOFS is not proper.
Bellow comment is quoted from Dave,
> It wasn't for memory allocation recursion protection in XFS - it was for
> transaction reservation recursion protection by something trying to flush
> data pages while holding a transaction reservation. Doing
> this could deadlock the journal because the existing reservation
> could prevent the nested reservation for being able to reserve space
> in the journal and that is a self-deadlock vector.
> IOWs, this check is not protecting against memory reclaim recursion
> bugs at all (that's the previous check [1]). This check is
> protecting against the filesystem calling writepages directly from a
> context where it can self-deadlock.
> So what we are seeing here is that the PF_FSTRANS ->
> PF_MEMALLOC_NOFS abstraction lost all the actual useful information
> about what type of error this check was protecting against.
[1]. Bellow check is to avoid memory reclaim recursion.
if (WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD)) ==
PF_MEMALLOC))
goto redirty;
Suggested-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
---
fs/iomap/buffered-io.c | 4 ++--
fs/xfs/libxfs/xfs_btree.c | 2 +-
fs/xfs/xfs_aops.c | 4 ++--
fs/xfs/xfs_trans.c | 12 ++++++------
include/linux/sched.h | 1 +
5 files changed, 12 insertions(+), 11 deletions(-)
diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
index bcfc288..0f1945c 100644
--- a/fs/iomap/buffered-io.c
+++ b/fs/iomap/buffered-io.c
@@ -1500,9 +1500,9 @@ static void iomap_writepage_end_bio(struct bio *bio)
/*
* Given that we do not allow direct reclaim to call us, we should
- * never be called in a recursive filesystem reclaim context.
+ * never be called while in a filesystem transaction.
*/
- if (WARN_ON_ONCE(current->flags & PF_MEMALLOC_NOFS))
+ if (WARN_ON_ONCE(current->flags & PF_FSTRANS))
goto redirty;
/*
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
index 2d25bab..4a7c8b7 100644
--- a/fs/xfs/libxfs/xfs_btree.c
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -2814,7 +2814,7 @@ struct xfs_btree_split_args {
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 = PF_MEMALLOC_NOFS | PF_FSTRANS;
/*
* we are in a transaction context here, but may also be doing work
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index b356118..79da028 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -62,7 +62,7 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *ioend)
* 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);
+ current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
return 0;
}
@@ -125,7 +125,7 @@ static inline bool xfs_ioend_is_append(struct iomap_ioend *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);
+ current_set_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
__sb_writers_acquired(VFS_I(ip)->i_sb, SB_FREEZE_FS);
/* we abort the update if there was an IO error */
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 3c94e5f..011f52f 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -153,7 +153,7 @@
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);
+ current_set_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
/*
* Attempt to reserve the needed disk blocks by decrementing
@@ -163,7 +163,7 @@
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);
+ current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
return -ENOSPC;
}
tp->t_blk_res += blocks;
@@ -240,7 +240,7 @@
tp->t_blk_res = 0;
}
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
return error;
}
@@ -861,7 +861,7 @@
xfs_log_commit_cil(mp, tp, &commit_lsn, regrant);
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
xfs_trans_free(tp);
/*
@@ -893,7 +893,7 @@
xfs_log_ticket_ungrant(mp->m_log, tp->t_ticket);
tp->t_ticket = NULL;
}
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
xfs_trans_free_items(tp, !!error);
xfs_trans_free(tp);
@@ -954,7 +954,7 @@
}
/* mark this thread as no longer being in a transaction */
- current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS);
+ current_restore_flags_nested(&tp->t_pflags, PF_MEMALLOC_NOFS | PF_FSTRANS);
xfs_trans_free_items(tp, dirty);
xfs_trans_free(tp);
diff --git a/include/linux/sched.h b/include/linux/sched.h
index b62e6aa..02045e8 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1511,6 +1511,7 @@ static inline int is_global_init(struct task_struct *tsk)
#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
#define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
#define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
+#define PF_FSTRANS 0x01000000 /* Inside a filesystem transaction */
#define PF_UMH 0x02000000 /* I'm an Usermodehelper process */
#define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */
#define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
--
1.8.3.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH v2] xfs: reintroduce PF_FSTRANS for transaction reservation recursion protection
2020-06-24 15:05 [PATCH v2] xfs: reintroduce PF_FSTRANS for transaction reservation recursion protection Yafang Shao
@ 2020-06-26 9:02 ` Christoph Hellwig
2020-06-27 9:32 ` Yafang Shao
0 siblings, 1 reply; 3+ messages in thread
From: Christoph Hellwig @ 2020-06-26 9:02 UTC (permalink / raw)
To: Yafang Shao
Cc: david, mhocko, darrick.wong, hch, akpm, bfoster, vbabka, holger,
linux-xfs, linux-fsdevel, linux-mm
On Wed, Jun 24, 2020 at 11:05:42AM -0400, Yafang Shao wrote:
> PF_FSTRANS which is used to avoid transaction reservation recursion, is
> dropped since commit 9070733b4efa ("xfs: abstract PF_FSTRANS to
> PF_MEMALLOC_NOFS") and commit 7dea19f9ee63 ("mm: introduce
> memalloc_nofs_{save,restore} API") and replaced by PF_MEMALLOC_NOFS which
> means to avoid filesystem reclaim recursion. That change is subtle.
> Let's take the exmple of the check of WARN_ON_ONCE(current->flags &
> PF_MEMALLOC_NOFS)) to explain why this abstraction from PF_FSTRANS to
> PF_MEMALLOC_NOFS is not proper.
>
> Bellow comment is quoted from Dave,
> > It wasn't for memory allocation recursion protection in XFS - it was for
> > transaction reservation recursion protection by something trying to flush
> > data pages while holding a transaction reservation. Doing
> > this could deadlock the journal because the existing reservation
> > could prevent the nested reservation for being able to reserve space
> > in the journal and that is a self-deadlock vector.
> > IOWs, this check is not protecting against memory reclaim recursion
> > bugs at all (that's the previous check [1]). This check is
> > protecting against the filesystem calling writepages directly from a
> > context where it can self-deadlock.
> > So what we are seeing here is that the PF_FSTRANS ->
> > PF_MEMALLOC_NOFS abstraction lost all the actual useful information
> > about what type of error this check was protecting against.
>
> [1]. Bellow check is to avoid memory reclaim recursion.
> if (WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD)) ==
> PF_MEMALLOC))
> goto redirty;
>
> Suggested-by: Dave Chinner <david@fromorbit.com>
> Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> Cc: Michal Hocko <mhocko@kernel.org>
This generally looks sane, but:
- adds a bunch of overly long lines for no good reason
- doesn't really hide this behind a useful informatin, e.g. a
xfs_trans_context_start/end helpers for the normal case, plus
an extra helper with kswapd in the name for that case.
The latter should also help to isolate a bit against the mm-area
changes to the memalloc flags proposed.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v2] xfs: reintroduce PF_FSTRANS for transaction reservation recursion protection
2020-06-26 9:02 ` Christoph Hellwig
@ 2020-06-27 9:32 ` Yafang Shao
0 siblings, 0 replies; 3+ messages in thread
From: Yafang Shao @ 2020-06-27 9:32 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Dave Chinner, Michal Hocko, Darrick J. Wong, Andrew Morton,
Brian Foster, Vlastimil Babka, Holger Hoffstätte, linux-xfs,
linux-fsdevel, Linux MM, Matthew Wilcox
On Fri, Jun 26, 2020 at 5:02 PM Christoph Hellwig <hch@infradead.org> wrote:
>
> On Wed, Jun 24, 2020 at 11:05:42AM -0400, Yafang Shao wrote:
> > PF_FSTRANS which is used to avoid transaction reservation recursion, is
> > dropped since commit 9070733b4efa ("xfs: abstract PF_FSTRANS to
> > PF_MEMALLOC_NOFS") and commit 7dea19f9ee63 ("mm: introduce
> > memalloc_nofs_{save,restore} API") and replaced by PF_MEMALLOC_NOFS which
> > means to avoid filesystem reclaim recursion. That change is subtle.
> > Let's take the exmple of the check of WARN_ON_ONCE(current->flags &
> > PF_MEMALLOC_NOFS)) to explain why this abstraction from PF_FSTRANS to
> > PF_MEMALLOC_NOFS is not proper.
> >
> > Bellow comment is quoted from Dave,
> > > It wasn't for memory allocation recursion protection in XFS - it was for
> > > transaction reservation recursion protection by something trying to flush
> > > data pages while holding a transaction reservation. Doing
> > > this could deadlock the journal because the existing reservation
> > > could prevent the nested reservation for being able to reserve space
> > > in the journal and that is a self-deadlock vector.
> > > IOWs, this check is not protecting against memory reclaim recursion
> > > bugs at all (that's the previous check [1]). This check is
> > > protecting against the filesystem calling writepages directly from a
> > > context where it can self-deadlock.
> > > So what we are seeing here is that the PF_FSTRANS ->
> > > PF_MEMALLOC_NOFS abstraction lost all the actual useful information
> > > about what type of error this check was protecting against.
> >
> > [1]. Bellow check is to avoid memory reclaim recursion.
> > if (WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD)) ==
> > PF_MEMALLOC))
> > goto redirty;
> >
> > Suggested-by: Dave Chinner <david@fromorbit.com>
> > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > Cc: Michal Hocko <mhocko@kernel.org>
>
> This generally looks sane, but:
>
> - adds a bunch of overly long lines for no good reason
> - doesn't really hide this behind a useful informatin, e.g. a
> xfs_trans_context_start/end helpers for the normal case, plus
> an extra helper with kswapd in the name for that case.
>
Good point. I will try to think about it.
> The latter should also help to isolate a bit against the mm-area
> changes to the memalloc flags proposed.
I have read the patchset from Matthew. Agree with you that we should
do it the same way.
[adding Matthew to cc]
--
Thanks
Yafang
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2020-06-27 9:33 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-06-24 15:05 [PATCH v2] xfs: reintroduce PF_FSTRANS for transaction reservation recursion protection Yafang Shao
2020-06-26 9:02 ` Christoph Hellwig
2020-06-27 9:32 ` Yafang Shao
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).