linux-kernel - Re: [tip: x86/asm] x86/asm/ftrace: Mark function

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20191018171354.GB20368@zn.tnic>
Date:   Fri, 18 Oct 2019 19:13:54 +0200
From:   Borislav Petkov <bp@...e.de>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     tip-bot2 for Jiri Slaby <tip-bot2@...utronix.de>,
        linux-kernel@...r.kernel.org, linux-tip-commits@...r.kernel.org,
        Jiri Slaby <jslaby@...e.cz>, "H. Peter Anvin" <hpa@...or.com>,
        Ingo Molnar <mingo@...hat.com>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        linux-arch@...r.kernel.org, Masami Hiramatsu <mhiramat@...nel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>, x86-ml <x86@...nel.org>,
        Ingo Molnar <mingo@...nel.org>, Borislav Petkov <bp@...en8.de>
Subject: Re: [tip: x86/asm] x86/asm/ftrace: Mark function_hook as function

On Fri, Oct 18, 2019 at 12:49:56PM -0400, Steven Rostedt wrote:
> On Fri, 18 Oct 2019 12:48:00 -0400
> Steven Rostedt <rostedt@...dmis.org> wrote:
> 
> > > Relabel function_hook to be marked really as a function. It is called
> > > from C and has the same expectations towards the stack etc.  
> > 
> 
> And to go even further, it *does not* have the same expectations
> towards the stack.
> 
> I think this patch should not be applied.

There are a couple more markings like that now:

$ git grep function_hook
Documentation/asm-annotations.rst:120:    SYM_FUNC_START(function_hook)
Documentation/asm-annotations.rst:122:    SYM_FUNC_END(function_hook)
arch/x86/kernel/ftrace_32.S:15:# define function_hook   __fentry__
arch/x86/kernel/ftrace_32.S:24:SYM_FUNC_START(function_hook)
arch/x86/kernel/ftrace_32.S:26:SYM_FUNC_END(function_hook)
arch/x86/kernel/ftrace_64.S:17:# define function_hook   __fentry__
arch/x86/kernel/ftrace_64.S:135:SYM_FUNC_START(function_hook)
arch/x86/kernel/ftrace_64.S:137:SYM_FUNC_END(function_hook)
arch/x86/kernel/ftrace_64.S:251:SYM_FUNC_START(function_hook)
arch/x86/kernel/ftrace_64.S:282:SYM_FUNC_END(function_hook)

Frankly, I wouldn't mark this function at all as it is special and I see
a little sense to have it in stack traces but maybe Jiri has another
angle here. I'll let him comment.

I guess with the new nomenclature that can be SYM_CODE_* now...

Then, this magic "function" or a global symbol with an address or
whatever that is (oh, there's #define trickery too) definitely deserves
a comment above it to explain what it is. I even have to build the .s
file to see what it turns into:

.globl __fentry__ ; .p2align 4, 0x90 ; __fentry__:
 retq
.type __fentry__, @function ; .size __fentry__, .-__fentry__

Yeah, it is called on every function entry:

callq  ffffffff81a01760 <__fentry__>

but can we please explain with a comment above it what it is?

Thx.

-- 
Regards/Gruss,
    Boris.

SUSE Software Solutions Germany GmbH, GF: Felix Imendörffer, HRB 36809, AG Nürnberg