Subject: pid_max madness

Let's do:
#echo "-1" >/proc/sys/kernel/pid_max
#cat /proc/sys/kernel/pid_max
-1
#

Madness, isn't ?

I guess that isn't what author ment it to behave like.
Anyway, does it mean that after max unsigned value is reached pids are
going to be negative in value ??

--
GJ


2004-11-17 20:50:42

by Valdis Klētnieks

[permalink] [raw]
Subject: Re: pid_max madness

On Wed, 17 Nov 2004 21:12:07 +0100, Grzegorz Piotr Jaskiewicz said:

> Anyway, does it mean that after max unsigned value is reached pids are
> going to be negative in value ??

No, because something in the /proc drivers will die of indigestion *much*
sooner (I think it's some very low value like 64K or 128K on i386?)


Attachments:
(No filename) (226.00 B)

2004-11-17 20:53:14

by William Lee Irwin III

[permalink] [raw]
Subject: Re: pid_max madness

On Wed, Nov 17, 2004 at 09:12:07PM +0100, Grzegorz Piotr Jaskiewicz wrote:
> Let's do:
> #echo "-1" >/proc/sys/kernel/pid_max
> #cat /proc/sys/kernel/pid_max
> -1
> #
> Madness, isn't ?
> I guess that isn't what author ment it to behave like.
> Anyway, does it mean that after max unsigned value is reached pids are
> going to be negative in value ??

Kernel version please?


-- wli

Subject: Re: pid_max madness

William Lee Irwin III wrote:
> On Wed, Nov 17, 2004 at 09:12:07PM +0100, Grzegorz Piotr Jaskiewicz wrote:
>
>>Let's do:
>>#echo "-1" >/proc/sys/kernel/pid_max
>>#cat /proc/sys/kernel/pid_max
>>-1
>>#
>>Madness, isn't ?
>>I guess that isn't what author ment it to behave like.
>>Anyway, does it mean that after max unsigned value is reached pids are
>>going to be negative in value ??
>
>
> Kernel version please?
all 2.6.x on intel86, this exact one is 2.6.9.
on 2.6.10-rc1-mm4 it doesn't change it anymore.

--
GJ

-rwxr-xr-x 1 root 5.89824e37 Oct 22 1990 /usr/bin/emacs

2004-11-17 21:14:54

by William Lee Irwin III

[permalink] [raw]
Subject: Re: pid_max madness

William Lee Irwin III wrote:
>> Kernel version please?

On Wed, Nov 17, 2004 at 09:59:37PM +0100, Grzegorz Piotr Jaskiewicz wrote:
> all 2.6.x on intel86, this exact one is 2.6.9.
> on 2.6.10-rc1-mm4 it doesn't change it anymore.

Okay, good to hear I didn't miss this piece of the fixes.

Thanks.


-- wli

2004-11-18 02:01:57

by linux-os

[permalink] [raw]
Subject: Re: pid_max madness

On Wed, 17 Nov 2004, Grzegorz Piotr Jaskiewicz wrote:

> Let's do:
> #echo "-1" >/proc/sys/kernel/pid_max
> #cat /proc/sys/kernel/pid_max
> -1
> #
>
> Madness, isn't ?
>
> I guess that isn't what author ment it to behave like.
> Anyway, does it mean that after max unsigned value is reached pids are going
> to be negative in value ??
>
> --
> GJ
> -

Seems to set them back to 300 on this system Linux-2.6.9. That's
an interesting side-effect. I like it! They never get much over
300 either!


Cheers,
Dick Johnson
Penguin : Linux version 2.6.9 on an i686 machine (5537.79 BogoMips).
Notice : All mail here is now cached for review by John Ashcroft.
98.36% of all statistics are fiction.