linux-kernel - Re: [PATCH v2 -tip] x86/percpu: Use C for arch_raw_cpu

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]

Date:   Mon, 16 Oct 2023 13:59:44 -0700
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Nadav Amit <namit@...are.com>
Cc:     Uros Bizjak <ubizjak@...il.com>,
        "the arch/x86 maintainers" <x86@...nel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Andy Lutomirski <luto@...nel.org>,
        Brian Gerst <brgerst@...il.com>,
        Denys Vlasenko <dvlasenk@...hat.com>,
        "H . Peter Anvin" <hpa@...or.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        Nick Desaulniers <ndesaulniers@...gle.com>
Subject: Re: [PATCH v2 -tip] x86/percpu: Use C for arch_raw_cpu_ptr()

On Mon, 16 Oct 2023 at 13:35, Nadav Amit <namit@...are.com> wrote:
>
> I don’t think it means that it the aliasing does not work; I think it all means that
> it actually works *too well*.

Hmm. The *only* case I can think of is __switch_to() itself when doing that

        raw_cpu_write(pcpu_hot.current_task, next_p);

and arguably that part should never have been done in C at all, but here we are.

I do think that from a sanity standpoint, it would be good to split
"__switch_to()" into two: a "before stack switch" and "after stack
switch", and change 'current' from within the asm section (ie do that
part in __switch_to_asm").

Or maybe we should just keep the single "__switch_to()" function, but
make it clear that it happens *after* current has been changed (and
maybe rename it to 'funish_switch_to()" or something like that to make
it clearer).

Because the situation with __switch_to() right now is outright
confusing, where it has 'prevp' and 'nextp', but then implicitly uses
'current' in two different ways.

Ugh.

               Linus