Folk periodically talk about evolving the "struct completion"
support ... here's one:
int wait_event (struct completion *x, signed long timeout_jiffies);
Returns zero if the event happened, nonzero if it timed out.
An example timeout action might be to cancel a request then
use wait_for_completion(x) to synchronize with that.
It's been behaving for me so far. Questions from me:
- Is this appropriate to merge as-is? The tradeoff I'm
thinking of is code duplication; this patch avoids it
("smaller==better" rule-of-thumb) but maybe someone has
real tuning knowledge, or a "standard" lk policy applies.
Or there might be bugs or api transgressions.
- One routine was unavailable to modules; now exported.
That seemed to just be an oversight.
This doesn't add an "interruptible wait" api mode, which
I'd expect some others might find a use for. This patch
will let us clean up some dubious code in usbcore which
is more or less trying to do a wait_event().
Thanks in advance for any comments.
- Dave
--- ./include/linux-dist/completion.h Sun Dec 8 10:57:47 2002
+++ ./include/linux/completion.h Mon Dec 9 15:11:51 2002
@@ -28,6 +28,7 @@
}
extern void FASTCALL(wait_for_completion(struct completion *));
+extern int FASTCALL(wait_timeout(struct completion *, signed long jiffies));
extern void FASTCALL(complete(struct completion *));
extern void FASTCALL(complete_all(struct completion *));
--- ./kernel-dist/ksyms.c Sun Dec 8 10:57:48 2002
+++ ./kernel/ksyms.c Mon Dec 9 15:13:47 2002
@@ -404,7 +404,9 @@ EXPORT_SYMBOL(autoremove_wake_function);
/* completion handling */
EXPORT_SYMBOL(wait_for_completion);
+EXPORT_SYMBOL(wait_timeout);
EXPORT_SYMBOL(complete);
+EXPORT_SYMBOL(complete_all);
/* The notion of irq probe/assignment is foreign to S/390 */
--- ./kernel-dist/sched.c Sun Dec 8 10:57:48 2002
+++ ./kernel/sched.c Mon Dec 9 15:34:36 2002
@@ -1204,6 +1204,11 @@ void complete_all(struct completion *x)
void wait_for_completion(struct completion *x)
{
+ wait_timeout (x, MAX_SCHEDULE_TIMEOUT);
+}
+
+int wait_timeout(struct completion *x, signed long timeout_jiffies)
+{
might_sleep();
spin_lock_irq(&x->wait.lock);
if (!x->done) {
@@ -1214,13 +1219,19 @@ void wait_for_completion(struct completi
do {
__set_current_state(TASK_UNINTERRUPTIBLE);
spin_unlock_irq(&x->wait.lock);
- schedule();
+ timeout_jiffies = schedule_timeout(timeout_jiffies);
+ if (timeout_jiffies == 0) {
+ __remove_wait_queue(&x->wait, &wait);
+ /* caller should wait again */
+ return 1;
+ }
spin_lock_irq(&x->wait.lock);
} while (!x->done);
__remove_wait_queue(&x->wait, &wait);
}
x->done--;
spin_unlock_irq(&x->wait.lock);
+ return 0;
}
#define SLEEP_ON_VAR \
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Dec 15 2002 - 22:00:24 EST