linux-kernel - Re: [PATCH v2] x86/debug: Fix DR6 handling

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>] [day] [month] [year] [list]

Message-ID: <50670b90-6d53-8c65-b3c8-e76fc9e2a1d3@suse.de>
Date:   Fri, 29 Jan 2021 17:59:54 +0100
From:   Tom de Vries <tdevries@...e.de>
To:     Borislav Petkov <bp@...en8.de>,
        Peter Zijlstra <peterz@...radead.org>
Cc:     x86@...nel.org, linux-kernel@...r.kernel.org,
        andrew.cooper3@...rix.com, Frederic Weisbecker <fweisbec@...il.com>
Subject: Re: [PATCH v2] x86/debug: Fix DR6 handling

On 1/29/21 3:48 PM, Borislav Petkov wrote:
> On Thu, Jan 28, 2021 at 10:16:27PM +0100, Peter Zijlstra wrote:
>>
>> Tom reported that one of the GDB test-cases failed, and Boris bisected
>> it to commit:
>>
>>   d53d9bc0cf78 ("x86/debug: Change thread.debugreg6 to thread.virtual_dr6")
>>
>> The debugging session led us to commit:
>>
>>   6c0aca288e72 ("x86: Ignore trap bits on single step exceptions")
>>
>> It turns out that TF and data breakpoints are both traps and will be
>> merged, while instruction breakpoints are faults and will not be
>> merged. This means 6c0aca288e72 is wrong, we only need to exclude TF
>> and instruction breakpoints while we can merge TF and data
>> breakpoints.
>>
>> Fixes: d53d9bc0cf78 ("x86/debug: Change thread.debugreg6 to thread.virtual_dr6")
>> Fixes: 6c0aca288e72 ("x86: Ignore trap bits on single step exceptions")
>> Reported-by: Tom de Vries <tdevries@...e.de>
>> Bisected-by: Borislav Petkov <bp@...en8.de>
>> Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
> 
> I guess
> 
> Cc: <stable@...r.kernel.org>
> 
> Also,
> 
> Reviewed-by: Borislav Petkov <bp@...e.de>
> 
> And gdb testsuite is a bit happier:
> 
> --- before
> +++ after
>                  === gdb Summary ===
>  
> -# of expected passes            70822
> -# of unexpected failures        899
> +# of expected passes            70852
> +# of unexpected failures        869
>  # of expected failures          74
>  # of known failures             99
>  # of untested testcases         114
> 
> You just fixed 30(!) testcases.
> 
> :-)
> 

Hi Boris,

thanks for testing this, and just to confirm: the total number of
regressions I see in the gdb testsuite related to watchpoints is indeed 30.

Thanks,
- Tom