io_uring/rsrc: optimise single entry advance
authorPavel Begunkov <asml.silence@gmail.com>
Wed, 22 Feb 2023 14:36:50 +0000 (14:36 +0000)
committerJens Axboe <axboe@kernel.dk>
Wed, 22 Feb 2023 16:57:23 +0000 (09:57 -0700)
commitb000ae0ec2d709046ac1a3c5722fea417f8a067e
tree91a0bf402c0a9b2eb5f5f9ba4471f3af3de956c6
parentedd478269640b360c6f301f2baa04abdda563ef3
io_uring/rsrc: optimise single entry advance

Iterating within the first bvec entry should be essentially free, but we
use iov_iter_advance() for that, which shows up in benchmark profiles
taking up to 0.5% of CPU. Replace it with a hand coded version.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
io_uring/rsrc.c