linux-kernel - Re: [PATCH 5/6] uaccess: add force_uaccess

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200714071211.GC776@lst.de>
Date:   Tue, 14 Jul 2020 09:12:11 +0200
From:   Christoph Hellwig <hch@....de>
To:     Geert Uytterhoeven <geert@...ux-m68k.org>
Cc:     Mark Rutland <mark.rutland@....com>,
        Christoph Hellwig <hch@....de>, Nick Hu <nickhu@...estech.com>,
        Greentime Hu <green.hu@...il.com>,
        Vincent Chen <deanbo422@...il.com>,
        Paul Walmsley <paul.walmsley@...ive.com>,
        Palmer Dabbelt <palmer@...belt.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        linux-riscv <linux-riscv@...ts.infradead.org>,
        Linux-Arch <linux-arch@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 5/6] uaccess: add force_uaccess_{begin,end} helpers

On Mon, Jul 13, 2020 at 03:19:42PM +0200, Geert Uytterhoeven wrote:
> > This used to set KERNEL_DS, and now it sets USER_DS, which looks wrong
> > superficially.
> 
> Thanks for noticing, and sorry for missing that myself.
> 
> The same issue is present for SuperH:
> 
>     -               set_fs(KERNEL_DS);
>     +               oldfs = force_uaccess_begin();
> 
> So the patch description should be:
> 
>     "Add helpers to wraper the get_fs/set_fs magic for undoing any damage
>      done by set_fs(USER_DS)."
> 
> and leave alone users setting KERNEL_DS?

Yes, this was broken.  Fixed for the next version.

> > If the new behaviour is fine it suggests that the old behaviour was
> > wrong, or that this is superfluous and could go entirely.
> >
> > Geert?
> 
> Nope, on m68k, TLB cache operations operate on the current address space.
> Hence to flush a kernel TLB entry, you have to switch to KERNEL_DS first.
> 
> If we're guaranteed to be already using KERNEL_DS, I guess the
> address space handling can be removed.  But can we be sure?

We can't be sure yet.  But with a lot of my pending work we should be
able to get there in the not too far future.