DevHeads.net

/proc/sys/net/core/optmem_max on armv7l

Hello together,

I just noticed, there is a difference in the default value of
`/proc/sys/net/core/optmem_max` on armv7l:

On all arches it is 20480, but on armv7l it is 10240.

Is there any specific reason for limiting the maximum ancillary buffer
size allowed per socket on this arch?

Cheers,
Björn

Comments

Re: /proc/sys/net/core/optmem_max on armv7l

By Peter Robinson at 01/11/2018 - 22:17

No specific reason, I think that would be an upstream kernel default,
I'm not aware we tweak that setting on any of the architectures so it
should be the arch default.

Re: /proc/sys/net/core/optmem_max on armv7l

By Laura Abbott at 01/15/2018 - 12:42

On 01/11/2018 06:17 PM, Peter Robinson wrote:
The actual root cause is that the default is based off of
sizeof(unsigned long) which is different on 32-bit targets.

Thanks,
Laura

Re: /proc/sys/net/core/optmem_max on armv7l

By =?UTF-8?Q?Bj=c3... at 01/15/2018 - 13:26

Am Montag, den 15.01.2018, 08:42 -0800 schrieb Laura Abbott:

I just checked and found `unsigned long` to be 32 bits on %{arm},
only. All other arches (even %{ix86}) have it set to 64 bits.

unsigned long long seems to be at least 64-bit on all arches.

Well, yes… One cannot change it using sysctl (or using other ways)
without being root (and thus isn't an option for Koji builds).

The preset can be changed to be the same on all arches (based on
changing that in systemd) by adding `net.core.optmem_max = 20480` to
`/usr/lib/sysctl.d/50-default.conf`, tho.

Any ideas towards that?

Let's keep the further discussion in [1].

Cheers,
Björn

[1] <a href="https://bugzilla.redhat.com/show_bug.cgi?id=1534620" title="https://bugzilla.redhat.com/show_bug.cgi?id=1534620">https://bugzilla.redhat.com/show_bug.cgi?id=1534620</a>

Re: /proc/sys/net/core/optmem_max on armv7l

By Florian Weimer at 01/15/2018 - 15:42

* Björn 'besser82' Esser:

Huh? I'm not aware of *any* vaguely Linux-compatible 32-bit
architecture which has 64-bit longs (unsigned or signed).