Thread (4 messages) 4 messages, 2 authors, 29d ago

Re: [PATCH] net: wwan: mhi_wwan_mbim: validate datagram bounds before copy

flat view

From: Aamir Ahmed <hidden>
Date: 2026-09-08 00:37:40
Also in: stable

On Mon, Sep 07, 2026 at 09:57:09AM +0200, Loic Poulain wrote:
This seems redundant with the existing check in skb_copy_bits().
What would make more sense, however, is to check the return value of
skb_copy_bits().
Yes, this makes more sense than having to create a new guard - the
check already exists, the driver just ignores the return value.
Nothing is copied on failure, so the skb is delivered entirely
uninitialised, not partially as I first thought.

I will prepare v2 to check the return, free skbn, count an rx error
and break, as the copies above and the unknown-protocol path already
do. Will also drop the cdc_mbim rationale, which does not apply here.
That hash doesn't exist.
Tag will be corrected in v2, thank you:

  Fixes: aa730a9905b7 ("net: wwan: Add MHI MBIM network driver")
Please review AI-generated fixes carefully before submitting them, it
will save reviewers time...
Yes, many apologies for this! This was a clear mistake on my part. For
the future I will be more selective and thorough in my patch
submission process.

Kind Regards

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