[PATCH v2 1/5] uaccess: Fix scoped_user_read_access() for 'pointer to const'

[David Laight]

On Mon, 2 Mar 2026 09:26:31 -0800
Linus Torvalds <torvalds at linux-foundation.org> wrote:

> On Mon, 2 Mar 2026 at 06:59, Christophe Leroy (CS GROUP)
> <chleroy at kernel.org> wrote:
> >
> > Can we get this fix merged in 7.0-rc3 so that we can start building 7.1
> > on top of it ?  
> 
> Applied this first patch. I'm not so convinced about the others in the
> series, although people can always try to argue for them..

Patches 2 and 3 seemed a reasonable idea to me.
Removes a lot of code that is only there to make the whole thing work.
The 'with' is a bit of a take on Pascal - but without the 'making the
code completely unreadable' side effect.

I don't do WARN=1 builds, never mind WARN=2 :-)
Although -Wshadow can find real bugs - so I would turn it on and
suffer the annoyances. 

Patch 5 was just what I was experimenting with.
Doing the equivalent change to the non-compat version (IIRC it
uses the (likely) much slower copy_to/from_user() because the
structures match) might even be more sensible.

	David


> 
>               Linus