[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-block] [PATCH 10/18] async: optimize aio_bh_poll
From: |
Paolo Bonzini |
Subject: |
[Qemu-block] [PATCH 10/18] async: optimize aio_bh_poll |
Date: |
Thu, 6 Aug 2015 15:36:08 +0200 |
Avoid entering the slow path of qemu_lockcnt_dec_and_lock if
no bottom half has to be deleted. If a bottom half deletes itself,
it will be picked up on the next visit of the list, or when the
AioContext itself is finalized.
Signed-off-by: Paolo Bonzini <address@hidden>
---
async.c | 13 +++++++++++--
1 file changed, 11 insertions(+), 2 deletions(-)
diff --git a/async.c b/async.c
index 198c53d..1fce3e4 100644
--- a/async.c
+++ b/async.c
@@ -64,19 +64,24 @@ int aio_bh_poll(AioContext *ctx)
{
QEMUBH *bh, **bhp, *next;
int ret;
+ bool deleted = false;
qemu_lockcnt_inc(&ctx->list_lock);
ret = 0;
for (bh = atomic_rcu_read(&ctx->first_bh); bh; bh = next) {
next = atomic_rcu_read(&bh->next);
+ if (bh->deleted) {
+ deleted = true;
+ continue;
+ }
/* The atomic_xchg is paired with the one in qemu_bh_schedule. The
* implicit memory barrier ensures that the callback sees all writes
* done by the scheduling thread. It also ensures that the scheduling
* thread sees the zero before bh->cb has run, and thus will call
* aio_notify again if necessary.
*/
- if (!bh->deleted && atomic_xchg(&bh->scheduled, 0)) {
+ if (atomic_xchg(&bh->scheduled, 0)) {
/* Idle BHs don't count as progress */
if (!bh->idle) {
ret = 1;
@@ -89,6 +94,11 @@ int aio_bh_poll(AioContext *ctx)
}
/* remove deleted bhs */
+ if (!deleted) {
+ qemu_lockcnt_dec(&ctx->list_lock);
+ return ret;
+ }
+
if (qemu_lockcnt_dec_and_lock(&ctx->list_lock)) {
bhp = &ctx->first_bh;
while (*bhp) {
@@ -102,7 +112,6 @@ int aio_bh_poll(AioContext *ctx)
}
qemu_lockcnt_unlock(&ctx->list_lock);
}
-
return ret;
}
--
2.4.3
- [Qemu-block] [RFC PATCH 00/18] Fine-grained AioContext critical sections, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 02/18] aio: rename bh_lock to list_lock, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 03/18] qemu-thread: introduce QemuLockCnt, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 04/18] aio: make ctx->list_lock a QemuLockCnt, subsuming ctx->walking_bh, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 06/18] aio-posix: remove walking_handlers, protecting AioHandler list with list_lock, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 09/18] aio: push aio_context_acquire/release down to dispatching, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 10/18] async: optimize aio_bh_poll,
Paolo Bonzini <=
- [Qemu-block] [PATCH 16/18] quorum: split quorum_fifo_aio_cb from quorum_aio_cb, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 05/18] aio: tweak walking in dispatch phase, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 08/18] aio: document locking, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 14/18] block: explicitly acquire aiocontext in timers that need it, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 01/18] iothread: release iothread around aio_poll, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 15/18] quorum: use atomics for rewrite_count, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 07/18] aio-win32: remove walking_handlers, protecting AioHandler list with list_lock, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 18/18] aio: update locking documentation, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 12/18] block: explicitly acquire aiocontext in callbacks that need it, Paolo Bonzini, 2015/08/06
- [Qemu-block] [PATCH 17/18] block: explicitly acquire aiocontext in aio callbacks that need it, Paolo Bonzini, 2015/08/06