Skip to content

Better sysfs lockdep

Eric W. Biederman lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest4 patches 9 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/4sysfs: Remove sysfs_get/put_active_twofs/sysfs/bin.c fs/sysfs/dir.c fs/sysfs/file.c +1 more+38 -771 reply
1/4sysfs: Remove sysfs_get/put_active_twofs/sysfs/bin.c fs/sysfs/dir.c fs/sysfs/file.c +1 more+38 -770 replies
2/4sysfs: Only take active references on attributes.fs/sysfs/dir.c fs/sysfs/file.c fs/sysfs/sysfs.h+6 -14 replies
3/4sysfs: Use one lockdep class per sysfs attribute.fs/sysfs/sysfs.h include/linux/sysfs.h+23 -21 reply
4/4sysfs: Use sysfs_attr_init and sysfs_bin_attr_init on dynamic attributesarch/x86/kernel/cpu/mcheck/mce.c drivers/acpi/system.c drivers/pci/pci-sysfs.c +1 more+9 -05 replies

Reviews trailers per patch

All 5 patches have a Reviewed-by or Acked-by.

PatchWANG Cong
1/4A-b
1/4A-b
2/4A-b
3/4A-b
4/4A-b

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 0/4] Better sysfs lockdep

It turns out that it is pretty simple to rework the lockdep support to give each sysfs attribute it's own lockdep class, which should kill the annoying false positives we have been seeing lately.

The one caveat is that it requires dynamically allocated sysfs attributes to be explicitly initialized. Luckily it is brainless trivial whack-a-mole and took maybe 15 minutes to find all of the dynamically allocates sysfs attributes on my test machine, because the errors show up when the attributes are registered with sysfs, and all of the fixes are one-liners.

Hopefully with this we can head back to fixing the real issues.

Eric