| Message ID | 20250415231635.83960-1-dave@stgolabs.net (mailing list archive) |
|---|---|
| Headers | show |
| Series | fs/buffer: split pagecache lookups into atomic or blocking | expand |
On Tue, Apr 15, 2025 at 04:16:28PM -0700, Davidlohr Bueso wrote: > Hello, > > This is a respin of the series[0] to address the sleep in atomic scenarios for > noref migration with large folios, introduced in: > > 3c20917120ce61 ("block/bdev: enable large folio support for large logical block sizes") > > The main difference is that it removes the first patch and moves the fix (reducing > the i_private_lock critical region in the migration path) to the final patch, which > also introduces the new BH_Migrate flag. It also simplifies the locking scheme in > patch 1 to avoid folio trylocking in the atomic lookup cases. So essentially blocking > users will take the folio lock and hence wait for migration, and otherwise nonblocking > callers will bail the lookup if a noref migration is on-going. Blocking callers > will also benefit from potential performance gains by reducing contention on the > spinlock for bdev mappings. > > It is noteworthy that this series is probably too big for Linus' tree, so there are > two options: > > 1. Revert 3c20917120ce61, add this series + 3c20917120ce61 for next. Or, Reverting due to a fix series is odd, I'd advocate this series as a set of fixes to Linus' tree because clearly folio migration was not complete for buffer_migrate_folio_norefs() and this is part of the loose bits to help it for large folios. This issue was just hard to reproduce. The enabler of large folios on the block device cache is actually commit 47dd67532303 ("block/bdev: lift block size restrictions to 64k") which goes later after 3c20917120ce61. Jan Kara, since you've already added your Reviewed-by for all patches do you have any preference how this trickles to Linus? > 2. Cherry pick patch 7 as a fix for Linus' tree, and leave the rest for next. > But that could break lookup callers that have been deemed unfit to bail. > > Patch 1: carves a path for callers that can block to take the folio lock. > Patch 2: adds sleeping flavors to pagecache lookups, no users. > Patches 3-6: converts to the new call, where possible. > Patch 7: does the actual sleep in atomic fix. > > Thanks! kdevops has tested this patch series and compared it to the baseline [0] and has found no regressions on ext4. Tested-by: kdevops@lists.linux.dev Detailed test results below: Comparing commits: Baseline: a74831cc4300 | linux-ext4-kpd: Linux 6.15-rc2 Test: 6b337686249b | 6.15-rc2 + these patches Baseline Kernel:6.15.0-rc2-g8ffd015db85f Test Kernel: 6.15.0-rc2-00006-g89e084d709fc Verbose Test Results Comparison: ================================================================================ Profile: ext4_1k | BASELINE | TEST -------------------|--------------|-------------- ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/082 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/235 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_2k | BASELINE | TEST -------------------|--------------|-------------- ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/082 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/235 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_4k | BASELINE | TEST -------------------|--------------|-------------- ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/082 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/235 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_advanced_features | BASELINE | TEST -------------------|--------------|-------------- ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/082 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/235 | [fail] | [fail] generic/270 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/477 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_bigalloc1024k_4k | BASELINE | TEST -------------------|--------------|-------------- ext4/033 | [fail] | [fail] ext4/034 | [fail] | [fail] ext4/045 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/075 | [fail] | [fail] generic/082 | [fail] | [fail] generic/091 | [fail] | [fail] generic/112 | [fail] | [fail] generic/127 | [fail] | [fail] generic/219 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/234 | [fail] | [fail] generic/235 | [fail] | [fail] generic/251 | [fail] | [fail] generic/263 | [fail] | [fail] generic/280 | [fail] | [fail] generic/365 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/435 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/614 | [fail] | [fail] generic/629 | [fail] | [fail] generic/634 | [fail] | [fail] generic/635 | [fail] | [fail] generic/643 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/698 | [fail] | [fail] generic/732 | [fail] | [fail] generic/738 | [fail] | [fail] generic/741 | [fail] | [fail] generic/754 | [fail] | [fail] Profile: ext4_bigalloc16k_4k | BASELINE | TEST -------------------|--------------|-------------- ext4/033 | [fail] | [fail] ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/075 | [fail] | [fail] generic/082 | [fail] | [fail] generic/091 | [fail] | [fail] generic/112 | [fail] | [fail] generic/127 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/234 | [fail] | [fail] generic/235 | [fail] | [fail] generic/263 | [fail] | [fail] generic/280 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_bigalloc2048k_4k | BASELINE | TEST -------------------|--------------|-------------- ext4/033 | [fail] | [fail] ext4/034 | [fail] | [fail] ext4/045 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/075 | [fail] | [fail] generic/082 | [fail] | [fail] generic/091 | [fail] | [fail] generic/112 | [fail] | [fail] generic/127 | [fail] | [fail] generic/219 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/234 | [fail] | [fail] generic/235 | [fail] | [fail] generic/251 | [fail] | [fail] generic/263 | [fail] | [fail] generic/280 | [fail] | [fail] generic/365 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/435 | [fail] | [fail] generic/471 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/614 | [fail] | [fail] generic/629 | [fail] | [fail] generic/634 | [fail] | [fail] generic/635 | [fail] | [fail] generic/643 | [fail] | [fail] generic/645 | [fail] | [fail] generic/676 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/698 | [fail] | [fail] generic/732 | [fail] | [fail] generic/736 | [fail] | [fail] generic/738 | [fail] | [fail] generic/741 | [fail] | [fail] generic/754 | [fail] | [fail] Profile: ext4_bigalloc32k_4k | BASELINE | TEST -------------------|--------------|-------------- ext4/033 | [fail] | [fail] ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/075 | [fail] | [fail] generic/082 | [fail] | [fail] generic/091 | [fail] | [fail] generic/112 | [fail] | [fail] generic/127 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/234 | [fail] | [fail] generic/235 | [fail] | [fail] generic/263 | [fail] | [fail] generic/280 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_bigalloc64k_4k | BASELINE | TEST -------------------|--------------|-------------- ext4/033 | [fail] | [fail] ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/075 | [fail] | [fail] generic/082 | [fail] | [fail] generic/091 | [fail] | [fail] generic/112 | [fail] | [fail] generic/127 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/234 | [fail] | [fail] generic/235 | [fail] | [fail] generic/263 | [fail] | [fail] generic/280 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Profile: ext4_defaults | BASELINE | TEST -------------------|--------------|-------------- ext4/034 | [fail] | [fail] ext4/055 | [fail] | [fail] generic/082 | [fail] | [fail] generic/219 | [fail] | [fail] generic/223 | [fail] | [fail] generic/230 | [fail] | [fail] generic/231 | [fail] | [fail] generic/232 | [fail] | [fail] generic/233 | [fail] | [fail] generic/235 | [fail] | [fail] generic/270 | [fail] | [fail] generic/381 | [fail] | [fail] generic/382 | [fail] | [fail] generic/566 | [fail] | [fail] generic/587 | [fail] | [fail] generic/600 | [fail] | [fail] generic/601 | [fail] | [fail] generic/681 | [fail] | [fail] generic/682 | [fail] | [fail] generic/741 | [fail] | [fail] Summary: - Total regressions: 0 - Total fixes: 0 - Unchanged failures: 261 [0] https://lore.kernel.org/all/Z__vQcCF9xovbwtT@bombadil.infradead.org/ Luis
On Wed, Apr 16, 2025 at 12:27:57PM -0700, Luis Chamberlain wrote: > On Tue, Apr 15, 2025 at 04:16:28PM -0700, Davidlohr Bueso wrote: > > Hello, > > > > This is a respin of the series[0] to address the sleep in atomic scenarios for > > noref migration with large folios, introduced in: > > > > 3c20917120ce61 ("block/bdev: enable large folio support for large logical block sizes") > > > > The main difference is that it removes the first patch and moves the fix (reducing > > the i_private_lock critical region in the migration path) to the final patch, which > > also introduces the new BH_Migrate flag. It also simplifies the locking scheme in > > patch 1 to avoid folio trylocking in the atomic lookup cases. So essentially blocking > > users will take the folio lock and hence wait for migration, and otherwise nonblocking > > callers will bail the lookup if a noref migration is on-going. Blocking callers > > will also benefit from potential performance gains by reducing contention on the > > spinlock for bdev mappings. > > > > It is noteworthy that this series is probably too big for Linus' tree, so there are > > two options: > > > > 1. Revert 3c20917120ce61, add this series + 3c20917120ce61 for next. Or, > > Reverting due to a fix series is odd, I'd advocate this series as a set > of fixes to Linus' tree because clearly folio migration was not complete I agree.
On Tue, Apr 15, 2025 at 04:16:28PM -0700, Davidlohr Bueso wrote: > Hello, > > This is a respin of the series[0] to address the sleep in atomic scenarios for > noref migration with large folios, introduced in: > > 3c20917120ce61 ("block/bdev: enable large folio support for large logical block sizes") Please resend based on vfs.fixes. > The main difference is that it removes the first patch and moves the fix (reducing > the i_private_lock critical region in the migration path) to the final patch, which > also introduces the new BH_Migrate flag. It also simplifies the locking scheme in > patch 1 to avoid folio trylocking in the atomic lookup cases. So essentially blocking > users will take the folio lock and hence wait for migration, and otherwise nonblocking > callers will bail the lookup if a noref migration is on-going. Blocking callers > will also benefit from potential performance gains by reducing contention on the > spinlock for bdev mappings. > > It is noteworthy that this series is probably too big for Linus' tree, so there are > two options: > > 1. Revert 3c20917120ce61, add this series + 3c20917120ce61 for next. Or, > 2. Cherry pick patch 7 as a fix for Linus' tree, and leave the rest for next. > But that could break lookup callers that have been deemed unfit to bail. > > Patch 1: carves a path for callers that can block to take the folio lock. > Patch 2: adds sleeping flavors to pagecache lookups, no users. > Patches 3-6: converts to the new call, where possible. > Patch 7: does the actual sleep in atomic fix. > > Thanks! > > [0] https://lore.kernel.org/all/20250410014945.2140781-1-mcgrof@kernel.org/ > > Davidlohr Bueso (7): > fs/buffer: split locking for pagecache lookups > fs/buffer: introduce sleeping flavors for pagecache lookups > fs/buffer: use sleeping version of __find_get_block() > fs/ocfs2: use sleeping version of __find_get_block() > fs/jbd2: use sleeping version of __find_get_block() > fs/ext4: use sleeping version of sb_find_get_block() > mm/migrate: fix sleep in atomic for large folios and buffer heads > > fs/buffer.c | 73 +++++++++++++++++++++++++++---------- > fs/ext4/ialloc.c | 3 +- > fs/ext4/mballoc.c | 3 +- > fs/jbd2/revoke.c | 15 +++++--- > fs/ocfs2/journal.c | 2 +- > include/linux/buffer_head.h | 9 +++++ > mm/migrate.c | 8 ++-- > 7 files changed, 82 insertions(+), 31 deletions(-) > > -- > 2.39.5 >
Hello, This is a respin of the series[0] to address the sleep in atomic scenarios for noref migration with large folios, introduced in: 3c20917120ce61 ("block/bdev: enable large folio support for large logical block sizes") The main difference is that it removes the first patch and moves the fix (reducing the i_private_lock critical region in the migration path) to the final patch, which also introduces the new BH_Migrate flag. It also simplifies the locking scheme in patch 1 to avoid folio trylocking in the atomic lookup cases. So essentially blocking users will take the folio lock and hence wait for migration, and otherwise nonblocking callers will bail the lookup if a noref migration is on-going. Blocking callers will also benefit from potential performance gains by reducing contention on the spinlock for bdev mappings. It is noteworthy that this series is probably too big for Linus' tree, so there are two options: 1. Revert 3c20917120ce61, add this series + 3c20917120ce61 for next. Or, 2. Cherry pick patch 7 as a fix for Linus' tree, and leave the rest for next. But that could break lookup callers that have been deemed unfit to bail. Patch 1: carves a path for callers that can block to take the folio lock. Patch 2: adds sleeping flavors to pagecache lookups, no users. Patches 3-6: converts to the new call, where possible. Patch 7: does the actual sleep in atomic fix. Thanks! [0] https://lore.kernel.org/all/20250410014945.2140781-1-mcgrof@kernel.org/ Davidlohr Bueso (7): fs/buffer: split locking for pagecache lookups fs/buffer: introduce sleeping flavors for pagecache lookups fs/buffer: use sleeping version of __find_get_block() fs/ocfs2: use sleeping version of __find_get_block() fs/jbd2: use sleeping version of __find_get_block() fs/ext4: use sleeping version of sb_find_get_block() mm/migrate: fix sleep in atomic for large folios and buffer heads fs/buffer.c | 73 +++++++++++++++++++++++++++---------- fs/ext4/ialloc.c | 3 +- fs/ext4/mballoc.c | 3 +- fs/jbd2/revoke.c | 15 +++++--- fs/ocfs2/journal.c | 2 +- include/linux/buffer_head.h | 9 +++++ mm/migrate.c | 8 ++-- 7 files changed, 82 insertions(+), 31 deletions(-) -- 2.39.5