afs: Fix writeback
Subsystems (2)
David Howells
maintainercover afs: Fix writeback
Hi Linus,
Could you apply these fixes to AFS writeback code from Vishal?
(1) Release the acquired batch before returning if we got >=5 skips.
(2) Retry a page we had to wait for rather than skipping over it after the wait.
The patches can be found here:
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=afs-fixes
Thanks,
David
---
%(shortlog)s
%(diffstat)s
Vishal Moola (Oracle) (2):
afs: Fix dangling folio ref counts in writeback
afs: Fix waiting for writeback then skipping folio
fs/afs/write.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
David Howells
maintainerPATCH 1/2 afs: Fix dangling folio ref counts in writeback
From: "Vishal Moola (Oracle)" <[email protected]>
Commit acc8d8588cb7 converted afs_writepages_region() to write back a folio batch. If writeback needs rescheduling, the function exits without dropping the references to the folios in fbatch. This patch fixes that.
[DH: Moved the added line before the _leave()]
--- fs/afs/write.c | 1 + 1 file changed, 1 insertion(+)
fs/afs/write.c +1 -0
@@ -763,6 +763,7 @@ static int afs_writepages_region(struct address_space *mapping,if (wbc->sync_mode ==WB_SYNC_NONE ) {if (skips >=5 ||need_resched ()) { *_next = start; +folio_batch_release (&fbatch);_leave (" = 0 [%llx]" , *_next);return 0 ; }
David Howells
maintainerPATCH 2/2 afs: Fix waiting for writeback then skipping folio
In reply to David HowellsFrom: "Vishal Moola (Oracle)" <[email protected]>
Commit acc8d8588cb7 converted afs_writepages_region() to write back a folio batch. The function waits for writeback to a folio, but then proceeds to the rest of the batch without trying to write that folio again. This patch fixes has it attempt to write the folio again.
[DH: Also remove an 'else' that adding a goto makes redundant]
--- fs/afs/write.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
fs/afs/write.c +4 -2
@@ -731,6 +731,7 @@ static int afs_writepages_region(struct address_space *mapping,* (changing page->mapping to NULL), or even swizzled * back from swapper_space to tmpfs file mapping */ +try_again:if (wbc->sync_mode !=WB_SYNC_NONE ) { ret =folio_lock_killable (folio);if (ret <0 ) {
@@ -757,9 +758,10 @@ static int afs_writepages_region(struct address_space *mapping,#ifdef CONFIG_AFS_FSCACHE folio_wait_fscache (folio);#endif - }else { - start +=folio_size (folio); + goto try_again; } + + start +=folio_size (folio);if (wbc->sync_mode ==WB_SYNC_NONE ) {if (skips >=5 ||need_resched ()) { *_next = start;
David Howells
maintainerQuoted 4 lines · David Howells
Hi Linus,
Could you apply these fixes to AFS writeback code from Vishal?
Quoted 8 lines
(1) Release the acquired batch before returning if we got >=5 skips.
(2) Retry a page we had to wait for rather than skipping over it after the wait.
The patches can be found here:
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=afs-fixes
Let me do that with a signed tag.
David