linux-hardening - Re: [RFC PATCH 14/21] treewide: static_call: Pass call arguments to the macro

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 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, 2 May 2022 09:46:33 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Sami Tolvanen <samitolvanen@...gle.com>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Kees Cook <keescook@...omium.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>, X86 ML <x86@...nel.org>,
        Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Nathan Chancellor <nathan@...nel.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Joao Moreira <joao@...rdrivepizza.com>,
        Sedat Dilek <sedat.dilek@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        linux-hardening@...r.kernel.org,
        linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
        llvm@...ts.linux.dev
Subject: Re: [RFC PATCH 14/21] treewide: static_call: Pass call arguments to
 the macro

On Fri, Apr 29, 2022 at 05:49:21PM -0700, Sami Tolvanen wrote:
> On Fri, Apr 29, 2022 at 4:21 PM Peter Zijlstra <peterz@...radead.org> wrote:
> > Can you explain why this is needed? I don't think there are any indirect
> > calls to get confused about. That is, if you have STATIC_CALL_INLINE
> > then the compiler should be emitting direct calls to the trampoline.
> 
> Clang emits an indirect call for ({ &f; })(), which is optimized into
> a direct call when possible. Come to think of it, the recent
> InstCombine change to the compiler patch should solve this issue. Let
> me double check, I'd be more than happy to drop these two patches.

Oooh, but this must not require any magic. That is, we have a *ton* of
code that relies on constant propagation of function pointers to not
emit indirect calls.

Please make sure that 'just-works'.

Look at all the __always_inline functions in rbtree*.h for instance,
some like latch and augment rely on quite complicated const propagation
for the actual function pointer is in a const struct.

I've verified all that actually generates direct calls when we did that
code (on GCC, clang wasn't really a thing back then).