httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dean Gaudet <>
Subject Re: mutex problems on Irix 6.2
Date Tue, 23 Jun 1998 07:06:57 GMT
As documented somewhere... perf-tuning probably... uslock is an order of
magnitude faster than sysvsem on multiprocessor irix boxes.  It is an
order of magnitude slower than sysvsem on single cpu irix boxes.  Also,
fcntl (or flock, forget which it used in 1.2) is an order of magnitude
slower than sysvsem on multiprocessor boxes... and about half as fast as
sysvsem on single cpu boxes. 

If anything I'd say revert it to the 1.2 default. 

Don't ask me why uslocks are so lame. 

That isn't the whole story... irix has another set of primitives that
nobody has bothered to compare... search the bugdb for a patch and
everything... the fellow never responded to my questions.

It'd be cool if you could investigate why the below stuff is failing.  It
never failed on the IRIX boxes I had access to (I no longer have access
to)... and they were 5.3 and 6.2.  It could just be the need to turn on


On 23 Jun 1998, Randy Terbush wrote:

> I've got an Irix 6.2 system failing to run Apache because of an EINVAL 
> return from the following piece of code. Anyone else have 1.3 running
> on Irix 6.2? Setting #define USE_USLOCK_SERIALIZED_ACCEPT seems to
> work fine. I'm wondering if this shouldn't be the default for this
> version of the OS?
> -Randy
> if (!getuid()) {
> /* restrict it to use only by the appropriate user_id ... not that this 
>  * stops CGIs from acquiring it and dinking around with it.
>  */
>     buf.sem_perm.uid = ap_user_id;
>     buf.sem_perm.gid = ap_group_id;
>     buf.sem_perm.mode = 0600;
>     ick.buf = &buf;
>     if (semctl(sem_id, 0, IPC_SET, ick) < 0) {
>         perror("semctl(IPC_SET)");
>         exit(APEXIT_INIT);
>     }
> }

View raw message