Re: page allocation failure

[Andrew Morton]

Dominik Karall <dominik.karall@xxxxxxx> wrote:
>
> is this a kernel bug, or smbd failure? I think it could be caused by kernel 
>  and less memory. Cause the machine is running with 56MB ram. But IMHO I think 
>  the kernel shouldn't handle it this way. Running 2.6.8-rc4-mm1.

It's networking trying to allocate eight physically-contiguous pages with
GFP_ATOMIC.  Can you say "snowball's chance in hell"?

Probably we should kill off those noisy printk's, or make them dependent on
some debugging option.  But we keep on finding quite serious cases, such as
this one.

Sure, networking will recover from memory allocation failures - presumably
by dropping a packet.  But if it's doing frequent atomic 3-order allocations
then it will end up dropping a *lot* of packets, and performance will suffer.
-
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/