linux-kernel - Re: [RESEND][PATCH v3 14/17] static_call: Add static_cond

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: <CAHk-=wiPrOd_yM5NGHQDE3AFtNWqV=MDFno2xgZOYV3jwGQU4w@mail.gmail.com>
Date:   Wed, 25 Mar 2020 11:26:38 -0700
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Andy Lutomirski <luto@...capital.net>,
        "the arch/x86 maintainers" <x86@...nel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Steven Rostedt <rostedt@...dmis.org>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Daniel Bristot de Oliveira <bristot@...hat.com>,
        Jason Baron <jbaron@...mai.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...nel.org>, Nadav Amit <namit@...are.com>,
        Peter Anvin <hpa@...or.com>,
        Andrew Lutomirski <luto@...nel.org>,
        Ard Biesheuvel <ard.biesheuvel@...aro.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>
Subject: Re: [RESEND][PATCH v3 14/17] static_call: Add static_cond_call()

On Wed, Mar 25, 2020 at 11:13 AM Peter Zijlstra <peterz@...radead.org> wrote:
>
> To clarify; the problem is a task getting preempted with its RIP at the
> RET. Then when we rewrite the text to be a CALL/JMP.d32 it will read
> garbage (1 byte into the displacement of the instruction) instead of a
> RET when it resumes.

Yeah, I realized it when you mentioned it. I was trying to come up
with some clever way to avoid it, but can't see anything.

I can come up with insane models - you could replace the xor;ret
sequence with a jump to a trampoline where you've aligned the target
trampoline so that the third byte in the "jmp xxx" remains a 'ret'
instruction. Then replace _that_ one with a "call_rcu()" callback.
Wild handwaving of "I'm sure this could be made to work".

But nothing remotely sane comes to mind.

              Linus