mirror of
https://github.com/torvalds/linux.git
synced 2024-11-24 13:11:40 +00:00
f75d48644c
__clear_bit_unlock() is a special little snowflake. While it carries the
non-atomic '__' prefix, it is specifically documented to pair with
test_and_set_bit() and therefore should be 'somewhat' atomic.
Therefore the generic implementation of __clear_bit_unlock() cannot use
the fully non-atomic __clear_bit() as a default.
If an arch is able to do better; is must provide an implementation of
__clear_bit_unlock() itself.
Specifically, this came up as a result of hackbench livelock'ing in
slab_lock() on ARC with SMP + SLUB + !LLSC.
The issue was incorrect pairing of atomic ops.
slab_lock() -> bit_spin_lock() -> test_and_set_bit()
slab_unlock() -> __bit_spin_unlock() -> __clear_bit()
The non serializing __clear_bit() was getting "lost"
80543b8e: ld_s r2,[r13,0] <--- (A) Finds PG_locked is set
80543b90: or r3,r2,1 <--- (B) other core unlocks right here
|
||
---|---|---|
.. | ||
__ffs.h | ||
__fls.h | ||
arch_hweight.h | ||
atomic.h | ||
builtin-__ffs.h | ||
builtin-__fls.h | ||
builtin-ffs.h | ||
builtin-fls.h | ||
const_hweight.h | ||
ext2-atomic-setbit.h | ||
ext2-atomic.h | ||
ffs.h | ||
ffz.h | ||
find.h | ||
fls64.h | ||
fls.h | ||
hweight.h | ||
le.h | ||
lock.h | ||
non-atomic.h | ||
sched.h |