From: Bart Van Assche <[email protected]>
To: Jens Axboe <[email protected]>
Cc: [email protected], Bart Van Assche <[email protected]>
Subject: [PATCH liburing 3/3] Convert __io_uring_get_sqe() from a macro into an inline function
Date: Sun, 21 Jun 2020 13:36:46 -0700 [thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>
This patch makes gcc 10 insert one additional assembly instruction in
io_uring_get_sqe(), namely xor %r8d,%r8d. That shouldn't cause any slowdown
since that instruction does not access memory:
0x0000000000000360 <+0>: mov (%rdi),%rax
0x0000000000000363 <+3>: xor %r8d,%r8d
0x0000000000000366 <+6>: mov (%rax),%ecx
0x0000000000000368 <+8>: mov 0x44(%rdi),%eax
0x000000000000036b <+11>: lea 0x1(%rax),%edx
0x000000000000036e <+14>: mov %edx,%esi
0x0000000000000370 <+16>: sub %ecx,%esi
0x0000000000000372 <+18>: mov 0x18(%rdi),%rcx
0x0000000000000376 <+22>: cmp (%rcx),%esi
0x0000000000000378 <+24>: ja 0x38e <io_uring_get_sqe+46>
0x000000000000037a <+26>: mov 0x10(%rdi),%rcx
0x000000000000037e <+30>: and (%rcx),%eax
0x0000000000000380 <+32>: mov %edx,0x44(%rdi)
0x0000000000000383 <+35>: shl $0x6,%rax
0x0000000000000387 <+39>: add 0x38(%rdi),%rax
0x000000000000038b <+43>: mov %rax,%r8
0x000000000000038e <+46>: mov %r8,%rax
0x0000000000000391 <+49>: retq
Signed-off-by: Bart Van Assche <[email protected]>
---
src/queue.c | 22 ++++++++++++----------
1 file changed, 12 insertions(+), 10 deletions(-)
diff --git a/src/queue.c b/src/queue.c
index 3db52bd17b86..88e0294c19fb 100644
--- a/src/queue.c
+++ b/src/queue.c
@@ -273,16 +273,18 @@ int io_uring_submit_and_wait(struct io_uring *ring, unsigned wait_nr)
return __io_uring_submit_and_wait(ring, wait_nr);
}
-#define __io_uring_get_sqe(sq, __head) ({ \
- unsigned __next = (sq)->sqe_tail + 1; \
- struct io_uring_sqe *__sqe = NULL; \
- \
- if (__next - __head <= *(sq)->kring_entries) { \
- __sqe = &(sq)->sqes[(sq)->sqe_tail & *(sq)->kring_mask];\
- (sq)->sqe_tail = __next; \
- } \
- __sqe; \
-})
+static inline struct io_uring_sqe *
+__io_uring_get_sqe(struct io_uring_sq *sq, unsigned int __head)
+{
+ unsigned int __next = (sq)->sqe_tail + 1;
+ struct io_uring_sqe *__sqe = NULL;
+
+ if (__next - __head <= *(sq)->kring_entries) {
+ __sqe = &(sq)->sqes[(sq)->sqe_tail & *(sq)->kring_mask];
+ (sq)->sqe_tail = __next;
+ }
+ return __sqe;
+}
/*
* Return an sqe to fill. Application must later call io_uring_submit()
next prev parent reply other threads:[~2020-06-21 20:36 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-21 20:36 [PATCH liburing 0/3] Three small liburing patches Bart Van Assche
2020-06-21 20:36 ` [PATCH liburing 1/3] Makefiles: Enable -Wextra Bart Van Assche
2020-06-21 20:36 ` [PATCH liburing 2/3] src/include/liburing/barrier.h: Use C11 atomics Bart Van Assche
2020-06-21 20:36 ` Bart Van Assche [this message]
2020-06-22 2:50 ` [PATCH liburing 0/3] Three small liburing patches Jens Axboe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox