diff options
author | Michael Ellerman <mpe@ellerman.id.au> | 2019-12-04 13:20:28 +0300 |
---|---|---|
committer | Michael Ellerman <mpe@ellerman.id.au> | 2019-12-04 13:20:28 +0300 |
commit | 4f4afc2c9599520300b3f2b3666d2034fca03df3 (patch) | |
tree | 8c102fbde90f353422669b0e235c498ab523a952 /.gitignore | |
parent | 5bece3d66153d78f1fd62108a1553c3f15e71412 (diff) | |
download | linux-4f4afc2c9599520300b3f2b3666d2034fca03df3.tar.xz |
docs/core-api: Remove possibly confusing sub-headings from Bit Operations
The recent commit 81d2c6f81996 ("kasan: support instrumented bitops
combined with generic bitops"), split the KASAN instrumented bitops
into separate headers for atomic, non-atomic and locking operations.
This was done to allow arches to include just the instrumented bitops
they need, while also using some of the generic bitops in
asm-generic/bitops (which are automatically instrumented). The generic
bitops are already split into atomic, non-atomic and locking headers.
This split required an update to kernel-api.rst because it included
include/asm-generic/bitops-instrumented.h, which no longer exists. So
now kernel-api.rst includes all three instrumented headers to get the
definitions for all the bitops.
When adding the three headers it seemed sensible to add sub-headings
for each, ie. "Atomic", "Non-atomic" and "Locking".
The confusion is that test_bit() is (and always has been) in
non-atomic.h, but is documented elsewhere (atomic_bitops.txt) as being
atomic. So having it appear under the "Non-atomic" heading is possibly
confusing.
Probably test_bit() should move from bitops/non-atomic.h to atomic.h,
but that has flow on effects. For now just remove the newly added
sub-headings in the documentation, so we at least aren't adding to the
confusion about whether test_bit() is atomic or not.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Diffstat (limited to '.gitignore')
0 files changed, 0 insertions, 0 deletions