55e858c848
- Remove the BKL from agp_open - Perform a few clean-ups. Analysis: --------- int minor is local to the function. The following are protected by agp_fe.agp_mutex struct agp_file_private *priv; struct agp_client *client; Call-outs: kzalloc should be safe to call under the mutex_lock agp_find_client_by_pid: - agp_mmap calls that under agp_fe.agp_mutex which we hold in agp_open - agpioc_reserve_wrap calls it without any locking what-so-ever. - Is that an error? Or is that okay because it has pid that is a unique handle? agp_insert_file_private: - This function only manipulates struct agp_file_private, once again while agp_fe.agp_mutex is held Signed-off-by: John Kacur <jkacur@redhat.com> Acked-by: David Airlie <airlied@linux.ie> LKML-Reference: <alpine.LFD.2.00.0910112216060.12574@localhost.localdomain> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> |
||
---|---|---|
.. | ||
agp.h | ||
ali-agp.c | ||
alpha-agp.c | ||
amd64-agp.c | ||
amd-k7-agp.c | ||
ati-agp.c | ||
backend.c | ||
compat_ioctl.c | ||
compat_ioctl.h | ||
efficeon-agp.c | ||
frontend.c | ||
generic.c | ||
hp-agp.c | ||
i460-agp.c | ||
intel-agp.c | ||
isoch.c | ||
Kconfig | ||
Makefile | ||
nvidia-agp.c | ||
parisc-agp.c | ||
sgi-agp.c | ||
sis-agp.c | ||
sworks-agp.c | ||
uninorth-agp.c | ||
via-agp.c |