The read event loop in the blocking path is also inefficient. For every
event it reap (if not blocking), it does the following in a loop:
while (i < nr) {
prepare_to_wait_exclusive
aio_read_evt
finish_wait
...
}
Given the previous patch "aio: add per task aio wait event condition"
that we properly wake up event waiting process knowing that we have
enough events to reap, it's just plain waste of time to insert itself
into a wait queue, and then immediately remove itself from the wait
queue for *every* event reap iteration.
This patch factors out the wait queue insertion/deletion out of the event
reap loop, streamlines the event reaping after the process wakes up.
Signed-off-by: Ken Chen <[email protected]>
--- ./fs/aio.c.orig 2006-12-24 17:04:52.000000000 -0800
+++ ./fs/aio.c 2006-12-24 17:05:10.000000000 -0800
@@ -1174,42 +1174,40 @@ retry:
}
aio_init_wait(&wait);
+wait:
+ prepare_to_wait_exclusive(&ctx->wait, &wait.wait, TASK_INTERRUPTIBLE);
+ ret = aio_read_evt(ctx, &ent);
+ if (!ret) {
+ wait.nr_wait = min_nr - i;
+ schedule();
+ if (signal_pending(tsk))
+ ret = -EINTR;
+ }
+ finish_wait(&ctx->wait, &wait.wait);
+
+ if (ret < 0)
+ goto out_cleanup;
+
while (likely(i < nr)) {
- do {
- prepare_to_wait_exclusive(&ctx->wait, &wait.wait,
- TASK_INTERRUPTIBLE);
- ret = aio_read_evt(ctx, &ent);
- if (ret)
- break;
- if (min_nr <= i)
- break;
- ret = 0;
- if (to.timed_out) /* Only check after read evt */
- break;
- wait.nr_wait = min_nr - i;
- schedule();
- if (signal_pending(tsk)) {
- ret = -EINTR;
+ if (ret) {
+ if (unlikely(copy_to_user(event, &ent, sizeof(ent)))) {
+ dprintk("aio: lost an event due to EFAULT.\n");
+ ret = -EFAULT;
break;
}
- /*ret = aio_read_evt(ctx, &ent);*/
- } while (1) ;
- finish_wait(&ctx->wait, &wait.wait);
-
- if (unlikely(ret <= 0))
- break;
+ event++;
+ i++;
+ }
- ret = -EFAULT;
- if (unlikely(copy_to_user(event, &ent, sizeof(ent)))) {
- dprintk("aio: lost an event due to EFAULT.\n");
+ ret = aio_read_evt(ctx, &ent);
+ if (unlikely(!ret)) {
+ if (i < min_nr && !to.timed_out)
+ goto wait;
break;
}
-
- /* Good, event copied to userland, update counts. */
- event ++;
- i ++;
}
+out_cleanup:
if (timeout)
clear_timeout(&to);
out:
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
