Skip to content

ext4: scoped NOFS for the handle in nojournal mode

Theodore Ts'o linux-ext4 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest2 patches 2 repliesIn review

Changes in v3 from the cover letter

* Export ext4_handle_cache to fix a build break when ext4 is compiled as module
* Add an optimization in how h_type and h_line_no is laid out in the jbd2_journal_handle structure.
Link to v2: https://lore.kernel.org/r/[email protected]

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/2ext4: enable scoped NOFS when starting a handle in nojournal modefs/ext4/ext4_jbd2.c fs/ext4/ext4_jbd2.h fs/jbd2/journal.c +1 more+25 -192 replies
2/2jbd2: align h_type and h_line_no in the handle structure on byte boundariesinclude/linux/jbd2.h+6 -60 replies

Reviews trailers per patch

1 of 2 patches has a Reviewed-by or Acked-by.

PatchAndreas Dilger
1/2R-b
2/2

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/2 -v3] ext4: scoped NOFS for the handle in nojournal mode

This patch series makes it safe to remove the GFP_NOFS flag when an active is handle and slightly optimized the data layout of the jbd2 handle.

Theodore Ts'o (2):
ext4: enable scoped NOFS when starting a handle in nojournal mode
jbd2: align h_type and h_line_no in the handle structure on byte
boundaries

 fs/ext4/ext4_jbd2.c  | 36 ++++++++++++++++++++++--------------
 fs/ext4/ext4_jbd2.h  |  6 +-----
 fs/jbd2/journal.c    |  1 +
 include/linux/jbd2.h | 11 ++++++-----
 4 files changed, 30 insertions(+), 24 deletions(-)

Changes in v3:
* Export ext4_handle_cache to fix a build break when ext4 is compiled as module
* Add an optimization in how h_type and h_line_no is laid out in the jbd2_journal_handle structure.
Link to v2: https://lore.kernel.org/r/[email protected]

Changes in v2:
* Remove an unused variable in ext4_get_nojournal()
* Add a safety check in case ext4_handle_valid() is passed a NULL pointer, even though the fucntion is documented that the caller must not be called with a NULL handle.
Link to v1: https://lore.kernel.org/r/[email protected]

Signature · 1 line
2.53.0