Thread (8 messages) flat view 8 messages, 3 authors, 2012-03-24

Re: [PATCH v2] iwlwifi: fix skb truesize underestimation

From: Guy, Wey-Yi <hidden>
Date: 2012-03-24 15:36:30

Hi Eric,

On Fri, 2012-03-23 at 17:29 -0700, Eric Dumazet wrote:
By default, iwlwifi uses order-1 pages (8 KB) to store incoming frames,
but doesnt say so in skb->truesize.

This makes very possible to exhaust kernel memory since these skb evade
normal socket memory accounting.

As struct ieee80211_hdr is going to be pulled before calling IP stack,
there is no need to use dev_alloc_skb() to reserve NET_SKB_PAD bytes.
alloc_skb() is ok in this driver, allowing more tailroom.

Pull beginning of frame in skb header, in the hope we can reuse order-1
pages in the driver immediately for small frames and reduce their
truesize to the minimum (linear skbs)

Signed-off-by: Eric Dumazet <redacted>
Cc: Wey-Yi Guy <redacted>
Cc: "John W. Linville" <redacted>
Cc: Neal Cardwell <ncardwell@google.com>
---
Is it ok I pull your patch into our internal tree first for regression
testing, once the test is done, I will push your patch along with all
our other patches together to wireless-next(John Linville).

By doing so, first, we know the patch is not breaking any functionality
for all the devices(legacy/new), second, it is easier for us to sync-up
with all the other changes we are making now.

Thanks
Wey
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help