| Message ID | 1488853256-10867-1-git-send-email-bo.li.liu@oracle.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
At 03/07/2017 10:20 AM, Liu Bo wrote: > The bug is a regression after commit > (da2c7009f6ca "btrfs: teach __process_pages_contig about PAGE_LOCK operation") > and commit > (76c0021db8fd "Btrfs: use helper to simplify lock/unlock pages"). > > So if the dirty pages which are under writeback got truncated partially > before we lock the dirty pages, we couldn't find all pages mapping to the > delalloc range, and the bug didn't return an error so it kept going on and > found that the delalloc range got truncated and got to unlock the dirty > pages, and then the ASSERT could caught the error, and showed > > ----------------------------------------------------------------------------- > assertion failed: page_ops & PAGE_LOCK, file: fs/btrfs/extent_io.c, line: 1716 > ----------------------------------------------------------------------------- We also triggered such bug when running xfstests btrfs/070, although possibility is somewhat low. However since it's more about possibility, any fsstress may trigger it though. Thanks, Qu > > This fixes the bug by returning the proper -EAGAIN. > > Cc: David Sterba <dsterba@suse.com> > Reported-by: Dave Jones <davej@codemonkey.org.uk> > Signed-off-by: Liu Bo <bo.li.liu@oracle.com> > --- > fs/btrfs/extent_io.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c > index 28e8192..8df7974 100644 > --- a/fs/btrfs/extent_io.c > +++ b/fs/btrfs/extent_io.c > @@ -1714,7 +1714,8 @@ static int __process_pages_contig(struct address_space *mapping, > * can we find nothing at @index. > */ > ASSERT(page_ops & PAGE_LOCK); > - return ret; > + err = -EAGAIN; > + goto out; > } > > for (i = 0; i < ret; i++) { > -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 07, 2017 at 01:54:31PM +0800, Qu Wenruo wrote: > > > At 03/07/2017 10:20 AM, Liu Bo wrote: > > The bug is a regression after commit > > (da2c7009f6ca "btrfs: teach __process_pages_contig about PAGE_LOCK operation") > > and commit > > (76c0021db8fd "Btrfs: use helper to simplify lock/unlock pages"). > > > > So if the dirty pages which are under writeback got truncated partially > > before we lock the dirty pages, we couldn't find all pages mapping to the > > delalloc range, and the bug didn't return an error so it kept going on and > > found that the delalloc range got truncated and got to unlock the dirty > > pages, and then the ASSERT could caught the error, and showed > > > > ----------------------------------------------------------------------------- > > assertion failed: page_ops & PAGE_LOCK, file: fs/btrfs/extent_io.c, line: 1716 > > ----------------------------------------------------------------------------- > > We also triggered such bug when running xfstests btrfs/070, although > possibility is somewhat low. > > However since it's more about possibility, any fsstress may trigger it > though. > Yes, I think buffer writes with O_SYNC and some concurrent truncate could get us to ASSERT. Thanks, -liubo > Thanks, > Qu > > > > > This fixes the bug by returning the proper -EAGAIN. > > > > Cc: David Sterba <dsterba@suse.com> > > Reported-by: Dave Jones <davej@codemonkey.org.uk> > > Signed-off-by: Liu Bo <bo.li.liu@oracle.com> > > --- > > fs/btrfs/extent_io.c | 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c > > index 28e8192..8df7974 100644 > > --- a/fs/btrfs/extent_io.c > > +++ b/fs/btrfs/extent_io.c > > @@ -1714,7 +1714,8 @@ static int __process_pages_contig(struct address_space *mapping, > > * can we find nothing at @index. > > */ > > ASSERT(page_ops & PAGE_LOCK); > > - return ret; > > + err = -EAGAIN; > > + goto out; > > } > > > > for (i = 0; i < ret; i++) { > > > > -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 28e8192..8df7974 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1714,7 +1714,8 @@ static int __process_pages_contig(struct address_space *mapping, * can we find nothing at @index. */ ASSERT(page_ops & PAGE_LOCK); - return ret; + err = -EAGAIN; + goto out; } for (i = 0; i < ret; i++) {
The bug is a regression after commit (da2c7009f6ca "btrfs: teach __process_pages_contig about PAGE_LOCK operation") and commit (76c0021db8fd "Btrfs: use helper to simplify lock/unlock pages"). So if the dirty pages which are under writeback got truncated partially before we lock the dirty pages, we couldn't find all pages mapping to the delalloc range, and the bug didn't return an error so it kept going on and found that the delalloc range got truncated and got to unlock the dirty pages, and then the ASSERT could caught the error, and showed ----------------------------------------------------------------------------- assertion failed: page_ops & PAGE_LOCK, file: fs/btrfs/extent_io.c, line: 1716 ----------------------------------------------------------------------------- This fixes the bug by returning the proper -EAGAIN. Cc: David Sterba <dsterba@suse.com> Reported-by: Dave Jones <davej@codemonkey.org.uk> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> --- fs/btrfs/extent_io.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)