Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1481173 > unrolled thread

[PATCH 1/5] ipc/sem: do not call wake_sem_queue_do() prematurely

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH 1/5] ipc/sem: do not call wake_sem_queue_do() prematurely Davidlohr Bueso <dave@stgolabs.net> - 2016-09-12 14:00 +0200
    Re: [PATCH 1/5] ipc/sem: do not call wake_sem_queue_do() prematurely Manfred Spraul <manfred@colorfullife.com> - 2016-09-13 06:20 +0200
      Re: [PATCH 1/5] ipc/sem: do not call wake_sem_queue_do() prematurely Davidlohr Bueso <dave@stgolabs.net> - 2016-09-13 10:20 +0200

#1481173 — [PATCH 1/5] ipc/sem: do not call wake_sem_queue_do() prematurely

... as this call should obviously be paired with its _prepare()
counterpart. At least whenever possible, as there is no harm in
calling it bogusly as we do now in a few places. Immediate error
semop(2) paths that are far from ever having the task block can
be simplified and avoid a few unnecessary loads on their way out
of the call as it is not deeply nested.

Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
---
 ipc/sem.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/ipc/sem.c b/ipc/sem.c
index 5e318c5f749d..a4e8bb2fae38 100644
--- a/ipc/sem.c
+++ b/ipc/sem.c
@@ -1887,16 +1887,22 @@ SYSCALL_DEFINE4(semtimedop, int, semid, struct sembuf __user *, tsops,
 	}
 
 	error = -EFBIG;
-	if (max >= sma->sem_nsems)
-		goto out_rcu_wakeup;
+	if (max >= sma->sem_nsems) {
+		rcu_read_unlock();
+		goto out_free;
+	}
 
 	error = -EACCES;
-	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO))
-		goto out_rcu_wakeup;
+	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO)) {
+		rcu_read_unlock();
+		goto out_free;
+	}
 
 	error = security_sem_semop(sma, sops, nsops, alter);
-	if (error)
-		goto out_rcu_wakeup;
+	if (error) {
+		rcu_read_unlock();
+		goto out_free;
+	}
 
 	error = -EIDRM;
 	locknum = sem_lock(sma, sops, nsops);
@@ -2039,7 +2045,6 @@ sleep_again:
 
 out_unlock_free:
 	sem_unlock(sma, locknum);
-out_rcu_wakeup:
 	rcu_read_unlock();
 	wake_up_sem_queue_do(&tasks);
 out_free:
-- 
2.6.6

[toc] | [next] | [standalone]

#1482137

Hi Davidlohr,

On 09/12/2016 01:53 PM, Davidlohr Bueso wrote:
> ... as this call should obviously be paired with its _prepare()
> counterpart. At least whenever possible, as there is no harm in
> calling it bogusly as we do now in a few places.
I would define the interface differently:
WAKE_Q creates an initialized wake queue. There is no need to track if 
any tasks were added to the wake queue, it is safe to call wake_up_q().
So especially for error paths, there is no need to optimize out calls to 
wake_up_q()
>   Immediate error
> semop(2) paths that are far from ever having the task block can
> be simplified and avoid a few unnecessary loads on their way out
> of the call as it is not deeply nested.
> Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
> ---
>   ipc/sem.c | 19 ++++++++++++-------
>   1 file changed, 12 insertions(+), 7 deletions(-)
>
> diff --git a/ipc/sem.c b/ipc/sem.c
> index 5e318c5f749d..a4e8bb2fae38 100644
> --- a/ipc/sem.c
> +++ b/ipc/sem.c
> @@ -1887,16 +1887,22 @@ SYSCALL_DEFINE4(semtimedop, int, semid, struct sembuf __user *, tsops,
>   	}
>   
>   	error = -EFBIG;
> -	if (max >= sma->sem_nsems)
> -		goto out_rcu_wakeup;
> +	if (max >= sma->sem_nsems) {
> +		rcu_read_unlock();
> +		goto out_free;
> +	}
>   
>   	error = -EACCES;
> -	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO))
> -		goto out_rcu_wakeup;
> +	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO)) {
> +		rcu_read_unlock();
> +		goto out_free;
> +	}
>   
Is this really better/simpler?
You replace "if (error) goto cleanup" with "if (error) {cleanup_1(); 
goto cleanup_2()}".

 From my point of view, this just increases the risks that some cleanup 
steps are forgotten.

--
     Manfred

[toc] | [prev] | [next] | [standalone]

#1482246

On Tue, 13 Sep 2016, Manfred Spraul wrote:

>>-	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO))
>>-		goto out_rcu_wakeup;
>>+	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO)) {
>>+		rcu_read_unlock();
>>+		goto out_free;
>>+	}
>Is this really better/simpler?
>You replace "if (error) goto cleanup" with "if (error) {cleanup_1(); 
>goto cleanup_2()}".

I believe it is better as it clearly separates blocking from non-blocking
exit paths. Overhead of course is irrelevant in both in-house wake_up_sem_queue_do
and wake_up_q.

Thanks,
Davidlohr

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web