Re: page allocation failure. order:3, mode:0x20

[Andrew Morton]

"Marco d'Itri" <md@xxxxxxxx> wrote:
>
> Linux attila 2.6.3 #1 Fri Feb 20 02:12:34 CET 2004 ppc GNU/Linux
> 
> (Linus tree.)
> 
> kernel: tinyproxy: page allocation failure. order:3, mode:0x20
> kernel: Call trace:
> kernel:  [c00097e8] dump_stack+0x18/0x28
> kernel:  [c0037fd8] __alloc_pages+0x31c/0x364
> kernel:  [c003804c] __get_free_pages+0x2c/0x74
> kernel:  [c003b14c] cache_grow+0x94/0x328
> kernel:  [c003b5a0] cache_alloc_refill+0x1c0/0x25c
> kernel:  [c003b9ac] __kmalloc+0xa8/0xb4
> kernel:  [c0167dc8] alloc_skb+0x4c/0xe0
> kernel:  [c0127004] loopback_xmit+0x100/0x110
> kernel:  [c016d0c0] dev_queue_xmit+0xcc/0x248
> kernel:  [c0184464] ip_finish_output+0x124/0x26c
> kernel:  [c018698c] dst_output+0x28/0x5c
> kernel:  [c01784c8] nf_hook_slow+0xf0/0x14c
> kernel:  [c0184c50] ip_queue_xmit+0x408/0x520
> kernel:  [c01960b4] tcp_transmit_skb+0x3e4/0x5e4
> kernel: tinyproxy: page allocation failure. order:4, mode:0x20

When the slab code creates a new slab it decides up-front on a nice
underlying allocation size which will pack a good number of the requested
objects.  So for the size-2048 slab it is currently using a 2-order (16k)
allocation.

Here, you're getting 3- and even 4-order allocation attempts.  Presumably
because loopback uses a 16k MTU and slab has sized that up to use 64k
allocations.

It's all crazy - these allocations are sure to fail all over the place. 
Manfred is working on fixing it up.

-
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/