Re: [PATCH v6 2/4] x86/traps: Print address on #GP
From: Andy Lutomirski
Date: Wed Dec 11 2019 - 12:22:36 EST
> On Dec 11, 2019, at 9:06 AM, Borislav Petkov <bp@xxxxxxxxx> wrote:
> ïOn Mon, Dec 09, 2019 at 03:31:18PM +0100, Jann Horn wrote:
>> I have already sent a patch to syzkaller that relaxes their parsing of GPF
>> messages (https://github.com/google/syzkaller/commit/432c7650) such that
>> changes like the one in this patch don't break it.
>> That patch has already made its way into syzbot's syzkaller instances
>> according to <https://syzkaller.appspot.com/upstream>.
> Ok, cool.
> I still think we should do the oops number marking, though, as it has
> more benefits than just syzkaller scanning for it. The first oops has always
> been of crucial importance so having the number in there:
> [ 2.542218]  general protection fault while derefing a non-canonical address 0xdfff000000000001: 0000 [#1] PREEMPT SMP
> would make eyeballing oopses even easier. Basically the same reason why
> you're doing this enhancement. :)
Could we spare a few extra bytes to make this more readable? I can never keep track of which number is the oops count, which is the cpu, and which is the error code. How about:
OOPS 1: general protection blah blah blah (CPU 0)
and put in the next couple lines â#GP(0)â.
> So let me know if you don't have time to do it or you don't care about
> it etc, and I'll have a look. Independent of those patches, of course -
> those look good so far.