linux-kernel - Re: [PATCH] objtool: Improve UACCESS coverage

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:   Wed, 24 Jul 2019 09:54:46 -0700
From:   Nathan Chancellor <natechancellor@...il.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Josh Poimboeuf <jpoimboe@...hat.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        LKML <linux-kernel@...r.kernel.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        clang-built-linux <clang-built-linux@...glegroups.com>,
        x86@...nel.org, Arnd Bergmann <arnd@...db.de>,
        Sedat Dilek <sedat.dilek@...il.com>,
        Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH] objtool: Improve UACCESS coverage

On Wed, Jul 24, 2019 at 06:48:21PM +0200, Peter Zijlstra wrote:
> On Wed, Jul 24, 2019 at 04:10:40PM +0200, Peter Zijlstra wrote:
> > And that most certainly should trigger...
> > 
> > Let me gdb that objtool thing.
> 
> ---
> Subject: objtool: Improve UACCESS coverage
> 
> A clang build reported an (obvious) double CLAC while a GCC build did
> not; it turns out we only re-visit instructions if the first visit was
> with AC=0. If OTOH the first visit was with AC=1, we completely ignore
> any subsequent visit, even when it has AC=0.
> 
> Fix this by using a visited mask, instead of boolean and (explicitly)
> mark the AC state.
> 
> $ ./objtool check -b --no-fp --retpoline --uaccess ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool: .altinstr_replacement+0x22: redundant UACCESS disable
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool:   eb_copy_relocations.isra.34()+0xea: (alt)
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool:   .altinstr_replacement+0xffffffffffffffff: (branch)
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool:   eb_copy_relocations.isra.34()+0xd9: (alt)
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool:   eb_copy_relocations.isra.34()+0xb2: (branch)
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool:   eb_copy_relocations.isra.34()+0x39: (branch)
> ../../defconfig-build/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool:   eb_copy_relocations.isra.34()+0x0: <=== (func)
> 
> Reported-by: Josh Poimboeuf <jpoimboe@...hat.com>
> Reported-by: Thomas Gleixner <tglx@...utronix.de>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>

This doesn't regress clang and I see the warning on GCC now too.

Tested-by: Nathan Chancellor <natechancellor@...il.com>