Re: [PATCH] ipc/sem.c: fix update sem_otime when calling sem_op insemaphore initialization

[Jia He]

On Mon, 23 Sep 2013 03:08:36 +0200 from bitbucket@xxxxxxxxx wrote:
> On Sun, 2013-09-22 at 12:42 +0200, Manfred Spraul wrote:
>
>> Mike: no, your patch makes it worse:
>> - wait-for-zero semops still don't update sem_otime
>> - sem_otime is initialized to sem_ctime. That's not mentioned in the 
>> sysv standard.
> So sem_otime = 0 is a specified semaphore state?  I thought the proggy
> was busted for spinning on a (busted and) irrelevant stamp.
Please refer to the words from Unix Network Programming - Volume 2 2nd
Edition Chapter 11
"Fortunately, there is a way around this race condition. We are guaranteed
that thesem-otime member of the semid-ds structure is set to 0 when a
new semaphore set iscreated. (The System V manuals have stated this
fact for a long time, as do the XPG3and Unix 98 standards.) This member
is set to the current time only by a successful callto semop."
>
> Man lernt nie aus.
>
> -Mike
>
>

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/