Re: [patch] ipv4: don't warn about skb ack allocation failures

[ebiederm@xmission.com (Eric W. Biederman)]

David Miller <davem@davemloft.net> writes:

> From: David Rientjes <rientjes@google.com>
> Date: Thu, 18 Jun 2009 12:23:28 -0700 (PDT)
>
>> On Thu, 18 Jun 2009, David Miller wrote:
>> 
>>> > I disagree, page allocation failure messages show vital information about 
>>> > the state of the VM so that we can find bugs and GFP_ATOMIC allocations 
>>> > are the most common trigger for these diagnostic messages since 
>>> > __GFP_WAIT allocations can trigger direct reclaim (and __GFP_FS 
>>> > allocations can trigger the oom killer) to free memory and will retry the 
>>> > allocation if ~__GFP_NORETRY.
>>> 
>>> It's COMPLETELY and ABSOLUTELY normal for GFP_ATOMIC allocations to
>>> fail in the networking.
>>> 
>> 
>> __GFP_NOWARN exists for that reason.
>
> You're going to have to put that into every driver, every part of
> the core networking, every protocol.
>
> That's dumb.
>
>> I understand what you're trying to avoid, but I disagree with the
>> approach of altering the default behavior of GFP_ATOMIC.
>
> The default got changed at some point because it never did
> crap like this before.

I started seeing this about when I upgraded to 2.6.28.

>> I may suggest that emitting the page allocation failures become a
>> compile time option; CONFIG_DEBUG_VM would be my suggestion.
>
> Use statistics gathering and tracing for this, not log spam.
>
> It serves all of your needs without spewing junk into the log.  It
> allows complete diagnosis and gathering of whatever information you
> may need.

I know my logs are overloaded with this noise, even on my laptop!

But Mr. Reintjes if you really want the traces I can set up 
an script to email them to you every time it happens.  Say about
1 a minute from my paltry little farm of machines.

Eric