Re: allocation failures with CBQ bandwidth limiting & high net use(was Re: Filesystem kernel hangup, 2.6.3 (bad: scheduling while atomic!))

[Andrew Morton]

Brad Allen <Ulmo-Usenet@xxxxxxxxxxxx> wrote:
>
>  My MTU for GbE (e1000) is 9000, and NFS block size 8192 bytes.
>  That GbE is a consumer grade Intel model.
> ....
>  swapper: page allocation failure. order:3, mode:0x20

The kernel simply doesn't have a hope of being able to find eight
physically-contiguous free pages at interrupt time.

You'll get some benefit from increasing /proc/sys/vm/min_free_kbytes

But it's beginning to look like we need separately-managed higher-order
page pools for this.  At least.

-
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/