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] [day] [month] [year] [list]
Message-ID: <9e0a8efd-9af8-49eb-a9b9-950ee582921b@roeck-us.net>
Date: Tue, 5 Mar 2024 06:20:48 -0800
From: Guenter Roeck <linux@...ck-us.net>
To: Huacai Chen <chenhuacai@...nel.org>, Marco Elver <elver@...gle.com>
Cc: Peter Zijlstra <peterz@...radead.org>, Dmitry Vyukov
 <dvyukov@...gle.com>, linux-perf-users@...r.kernel.org,
 linux-kernel@...r.kernel.org,
 Linux ARM <linux-arm-kernel@...ts.infradead.org>,
 Catalin Marinas <catalin.marinas@....com>, Will Deacon <will@...nel.org>,
 Mark Rutland <mark.rutland@....com>, loongarch@...ts.linux.dev
Subject: Re: hw_breakpoint unit test failures with various
 architectures/platforms in qemu

On 3/5/24 01:05, Huacai Chen wrote:
> Hi, Marco,
> 
> On Tue, Mar 5, 2024 at 4:34 AM Marco Elver <elver@...gle.com> wrote:
>>
>> On Mon, 4 Mar 2024 at 19:12, Guenter Roeck <linux@...ck-us.net> wrote:
>>>
>>> Hi,
>>>
>>> I see a number of failures and tracebacks when running the hw_breakpoint
>>> unit tests on various architectures in qemu. Some examples are below.
>>>
>>> Is this a potential problem with the unit tests, with the hardware,
>>> or with the qemu emulation of that hardware ?
>>
>> The test is testing the accounting logic (allocating/deallocating HW
>> breakpoints), not actually using the breakpoints. That requires that
>> the architecture reports the correct number of HW breakpoints
>> available. It looks like it's not doing that.
>>
>>> In other words, is it worthwhile to look into this further, or would
>>> it make more sense to just disable those tests if they fail on a
>>> given hardware/platform ?
>>
>> It's an arch bug:
>> https://lore.kernel.org/lkml/Ytl9L0Zn1PVuL1cB@FVFF77S0Q05N.cambridge.arm.com/
>>
>> Back then we decided to leave the test as-is, given it's reporting an
>> actual issue. However, since then nothing has changed and the test
>> continues to exist as a reminder the arch code needs fixing. I suspect
>> none of us have the time to get to that soon, so we can either leave
>> things as is, or skip the test on all the broken architectures.
>>
>> No idea what's going on with loongarch.
> LoongArch's problem comes from QEMU, I have tested on real machine.
> 
> [    7.222500] KTAP version 1
> [    7.226067] 1..1
> [    7.228795]     KTAP version 1
> [    7.232737]     # Subtest: hw_breakpoint
> [    7.237512]     # module: hw_breakpoint_test
> [    7.237520]     1..9
> [    7.442899]     ok 1 test_one_cpu
> [    7.443912]     ok 2 test_many_cpus
> [    7.451298]     ok 3 test_one_task_on_all_cpus
> [    7.456486]     ok 4 test_two_tasks_on_all_cpus
> [    7.462291]     ok 5 test_one_task_on_one_cpu
> [    7.468293]     ok 6 test_one_task_mixed
> [    7.474405]     ok 7 test_two_tasks_on_one_cpu
> [    7.479884]     ok 8 test_two_tasks_on_one_all_cpus
> [    7.485699]     ok 9 test_task_on_all_and_one_cpu
> [    7.491488] # hw_breakpoint: pass:9 fail:0 skip:0 total:9
> [    7.496985] # Totals: pass:9 fail:0 skip:0 total:9
> [    7.503179] ok 1 hw_breakpoint
> 
> Huacai
> 

That is good to know. Thanks a lot for testing!

Guenter


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ