linux-kernel - Re: [PATCH] x86/xsave: Robustify and merge macros

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: <20150403173306.GG14902@chrystal.uk.oracle.com>
Date:	Fri, 3 Apr 2015 19:33:06 +0200
From:	Quentin Casasnovas <quentin.casasnovas@...cle.com>
To:	Borislav Petkov <bp@...en8.de>
Cc:	Quentin Casasnovas <quentin.casasnovas@...cle.com>,
	X86 ML <x86@...nel.org>, LKML <linux-kernel@...r.kernel.org>,
	"H. Peter Anvin" <hpa@...or.com>, Ingo Molnar <mingo@...nel.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Oleg Nesterov <oleg@...hat.com>,
	Andy Lutomirski <luto@...capital.net>
Subject: Re: [PATCH] x86/xsave: Robustify and merge macros

On Fri, Apr 03, 2015 at 07:06:25PM +0200, Borislav Petkov wrote:
> On Fri, Apr 03, 2015 at 05:40:55PM +0200, Quentin Casasnovas wrote:
> > So yeah I still think we're not properly padding, if you take my earlier
> > example where repl2 = 5 bytes, repl1 = 4 bytes and orin_insn = 3.
> > 
> > I'll let you re-read my original mail and come back to me to tell me what'd
> > I really miss! :)
> 
> Dammit, dammit, dammit!
> 
> And I thought this aspect was taken care of. I went into the old
> branches where I had done this and there I have:
> 
> +#define OLDINSTR_2(oldinstr, num1, num2)                               \
> +       __OLDINSTR(oldinstr, num1)                                      \
> +       ".skip -(((" alt_rlen(num2) ")-(" alt_rlen(num1) ")) > 0) * "   \
> +               "((" alt_rlen(num2) ")-(" alt_rlen(num1) ")),0x90\n"    \
> +       alt_end_marker ":\n"
> +
> 
> without the size of the orig_insn factored in into the padding.
> 
> And that would work for your example because it would add 1+1 bytes
> padding.
> 
> Basically, the idea was:
> 
> .skip len(repl1) - len(orig), 0x90
> .skip len(repl2) - len(repl1), 0x90
> 
> BUT!, for some reason I changed it to what's there now and I can't
> remember why anymore.

I think it would not work in the case where repl1 is smaller or equal than
orig_insn (i.e. no padding in the first .skip) but orig_insn is strictly
smaller than repl2 (since we're never comparing repl2 with insn in this
new-old code).

Anything wrong with the two different approaches I've suggested in my
original mail?  One is using a one-liner .skip directive inspired by yours,
and the other is using .if directives.  FWIW I think exploding the logic
using conditionnals '.if' is way more readable and less error-prone.

Quentin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/