Re: ieee80211.h virtual_map splat

[Kees Cook <kees@kernel.org>]

On Fri, Jun 21, 2024 at 07:25:08AM -0700, Jeff Johnson wrote:
> On 6/21/2024 3:31 AM, Koen Vandeputte wrote:
> > On Fri, Jun 21, 2024 at 11:30 AM Johannes Berg
> > <johannes@sipsolutions.net> wrote:
> >>
> >>
> >>> will this one get backported also?
> >>
> >> Why? It's not even a bug.
> >>
> >> johannes
> > 
> > Because without this patch, it produces a splat on kernel 6.6 (which
> > is an LTS) at least ? :-)
> 
> @Kees, have you been backporting all your flexible array changes?

I haven't done anything explicit for them. This is especially true for
netdev where maintainers have asked that contributors not include "Cc:
stable" tags, as they want to evaluate for themselves whether patches
should go to stable or not.

> Or are you suggesting folks disable FORTIFY_SOURCE (is that the controlling
> config?)

I do not want people turning off FORTIFY_SOURCE. By design, this is
a warning only -- the memcpy() still works like it did before. The
goal was to catch any of these stragglers going forward, not to break
existing code. I expect that in a few more years it can be flipped to
warn-and-block for these kinds of detected memcpy()s, but for now there
should not be any behavioral changes seen besides the WARN appearing.

-Kees

-- 
Kees Cook