OSDN Git Service

RDMA/cxgb4: Use non-atomic bitmap functions when possible
authorChristophe JAILLET <christophe.jaillet@wanadoo.fr>
Wed, 24 Nov 2021 21:40:26 +0000 (22:40 +0100)
committerJason Gunthorpe <jgg@nvidia.com>
Thu, 25 Nov 2021 17:29:06 +0000 (13:29 -0400)
commit223b4d5c8702a92fb18d14e3f32953e474313e5b
tree4d1051dc79a19d2265bfa90b7b7375b615a634bc
parent967a578af0c6d0c6c50ee5f9c48216cd2eebe4ef
RDMA/cxgb4: Use non-atomic bitmap functions when possible

The accesses to the 'alloc->table' bitmap are protected by the
'alloc->lock' spinlock, so no concurrent accesses can happen.

So prefer the non-atomic '__[set|clear]_bit()' functions to save a few
cycles.

Link: https://lore.kernel.org/r/0c1c4505ca32f5ba4126e3e324041da191513ef2.1637789139.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
drivers/infiniband/hw/cxgb4/id_table.c