Re: [Regression v4.2 ?] 32-bit seccomp-BPF returned errno values wrong in VM?

From: Andy Lutomirski <luto@amacapital.net>
To: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Kees Cook <keescook@chromium.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	David Drysdale <drysdale@google.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Will Drewry <wad@chromium.org>, Ingo Molnar <mingo@kernel.org>,
	Alok Kataria <akataria@vmware.com>,
	Borislav Petkov <bp@alien8.de>,
	Alexei Starovoitov <ast@plumgrid.com>,
	Frederic Weisbecker <fweisbec@gmail.com>,
	"H. Peter Anvin" <hpa@zytor.com>, Oleg Nesterov <oleg@redhat.com>,
	Steven Rostedt <rostedt@goodmis.org>, X86 ML <x86@kernel.org>
Subject: Re: [Regression v4.2 ?] 32-bit seccomp-BPF returned errno values wrong in VM?
Date: Fri, 14 Aug 2015 07:27:24 -0700	[thread overview]
Message-ID: <CALCETrUM9z=txEmxe0KFP_9uzpvaQfefWtu7ffTsrLPFfy0o8g@mail.gmail.com> (raw)
In-Reply-To: <55CDD7CE.5070503@redhat.com>

On Fri, Aug 14, 2015 at 4:58 AM, Denys Vlasenko <dvlasenk@redhat.com> wrote:
> On 08/14/2015 12:59 AM, Andy Lutomirski wrote:
>> On Thu, Aug 13, 2015 at 3:56 PM, Kees Cook <keescook@chromium.org> wrote:
>>> On Thu, Aug 13, 2015 at 3:54 PM, Linus Torvalds
>>> <torvalds@linux-foundation.org> wrote:
>>>> On Thu, Aug 13, 2015 at 3:49 PM, Linus Torvalds
>>>> <torvalds@linux-foundation.org> wrote:
>>>>>
>>>>> Does the attached patch make sense and work?
>>>>
>>>> Btw, I'm not all that happy with it anyway.
>>>>
>>>> I still think Denys' patch also potentially changed what audit and
>>>> strace see for %rax in the pt_regs to -ENOSYS, which I'm not convinced
>>>> is a good change.
>>>>
>>>> But maybe that three-liner patch fixes the immediate problem that
>>>> David sees. David?
>>>
>>> Your patch fixes it for me. The seccomp compat selftests pass again
>>> with audit enabled.
>>
>> Kees, would it be straightforward to rig up the seccomp tests to
>> automatically test compat?  The x86 selftests automatically test both
>> native and compat, and that might be usable as a model.  I did that
>> because it's extremely easy to regress one and not the other.
>
> BTW, why 64-bt code doesn't need this RAX read-back?
>

It's hiding inside of RESTORE_C_REGS_EXCEPT_RCX_R11.

--Andy