Copied to assorted architecture maintainers and mailing lists, please
trim cc: list back to lkml plus arch if you reply.
The generic version of read_trylock() never tests if the lock is in
use, it always spins waiting for the lock to be free. IOW, it behaves
like read_lock(). Given the different implementations of rwlock_t, it
is hard for generic__raw_read_trylock() to do anything else.
I strongly recommend that the architectures below define their own
version of __raw_read_trylock() that really test the lock first, then
we can ditch generic__raw_read_trylock(). I have already sent an ia64
patch to that mailing list.
include/asm-arm/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)
include/asm-ia64/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)
include/asm-m32r/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)
include/asm-mips/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)
include/asm-parisc/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)
include/asm-sh/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)
include/asm-sparc64/spinlock.h:#define __raw_read_trylock(lock) generic__raw_read_trylock(lock)