On Sun, Jul 04, 2021 at 07:04:10PM -0400, Sasha Levin wrote:
From: Qu Wenruo wqu@suse.com
[ Upstream commit 87b4d86baae219a9a79f6b0a1434b2a42fd40d09 ]
Currently we use page Private2 bit to indicate that we have ordered extent for the page range.
But the lifespan of it is not consistent, during regular writeback path, there are two locations to clear the same PagePrivate2:
T ----- Page marked Dirty | + ----- Page marked Private2, through btrfs_run_dealloc_range() | + ----- Page cleared Private2, through btrfs_writepage_cow_fixup() | in __extent_writepage_io() | ^^^ Private2 cleared for the first time | + ----- Page marked Writeback, through btrfs_set_range_writeback() | in __extent_writepage_io(). | + ----- Page cleared Private2, through | btrfs_writepage_endio_finish_ordered() | ^^^ Private2 cleared for the second time. | + ----- Page cleared Writeback, through btrfs_writepage_endio_finish_ordered()Currently PagePrivate2 is mostly to prevent ordered extent accounting being executed for both endio and invalidatepage. Thus only the one who cleared page Private2 is responsible for ordered extent accounting.
But the fact is, in btrfs_writepage_endio_finish_ordered(), page Private2 is cleared and ordered extent accounting is executed unconditionally.
The race prevention only happens through btrfs_invalidatepage(), where we wait for the page writeback first, before checking the Private2 bit.
This means, Private2 is also protected by Writeback bit, and there is no need for btrfs_writepage_cow_fixup() to clear Priavte2.
This patch will change btrfs_writepage_cow_fixup() to just check PagePrivate2, not to clear it. The clearing will happen in either btrfs_invalidatepage() or btrfs_writepage_endio_finish_ordered().
This makes the Private2 bit easier to understand, just meaning the page has unfinished ordered extent attached to it.
And this patch is a hard requirement for the incoming refactoring for how we finished ordered IO for endio context, as the coming patch will check Private2 to determine if we need to do the ordered extent accounting. Thus this patch is definitely needed or we will hang due to unfinished ordered extent.
Reviewed-by: Josef Bacik josef@toxicpanda.com Signed-off-by: Qu Wenruo wqu@suse.com Signed-off-by: David Sterba dsterba@suse.com Signed-off-by: Sasha Levin sashal@kernel.org
Please drop this patch from all autosel stable backports. This is not a standalone fix and the CC: stable@ is not there intentionally.
All patches that go through my tree are evaluated for stable backports up to 4.4 so it's unlikely the machinery you're using can find something I've overlooked.
The patches that autosel picks and do not get a complain^Wreply for me are below the bar I'd consider it for stable but if after another review the patch "does no harm", I let it pass because you're obviously handling the backports. I would not tag it myself to avoid increasing load of Greg with patches that don't matter much.