linux-kernel - RE: [PATCH 1/5] selftests/resctrl: Clear unused initalization code in MBM tests

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]

Message-ID: <TYAPR01MB63302FA7919E8C95B9C56FC08B579@TYAPR01MB6330.jpnprd01.prod.outlook.com>
Date:   Thu, 29 Sep 2022 05:28:17 +0000
From:   "tan.shaopeng@...itsu.com" <tan.shaopeng@...itsu.com>
To:     'Reinette Chatre' <reinette.chatre@...el.com>,
        Fenghua Yu <fenghua.yu@...el.com>,
        Shuah Khan <shuah@...nel.org>
CC:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-kselftest@...r.kernel.org" <linux-kselftest@...r.kernel.org>
Subject: RE: [PATCH 1/5] selftests/resctrl: Clear unused initalization code in
 MBM tests

Hi Reinette,

> On 9/27/2022 2:01 AM, tan.shaopeng@...itsu.com wrote:
> >> On 9/13/2022 6:51 PM, Shaopeng Tan wrote:
> >>> There is a comment "Set up shemata with 100% allocation on the first run"
> >>> in function mbm_setup(), but the condition "num_of_runs == 0" will
> >>> never be met and write_schemata() will never be called to set
> >>> schemata to 100%.
> >>
> >> Thanks for catching this.
> >>
> >>>
> >>> Since umount/mount resctrl file system is run on each resctrl test,
> >>> at the same time the default schemata will also be set to 100%.
> >>
> >> This is the case when a test is run with struct
> >> resctrl_val_param->mum_resctrlfs == 1, but if the test is run with
> >> struct resctrl_val_param->mum_resctrlfs == 0 then resctrl filesystem
> >> will not be remounted.
> >>
> >> I do think that this setup function should support both cases.
> >
> > In mbm test(mbm_test.c), resctrl_val_param.mum_resctrlfs is set to 1
> > and never be changed, and umount/mount resctrl file system is always
> executed.
> > So it is not necessary to run "if (num_of_runs == 0)".
> 
> This is true for the current usage. You could also add a warning here ("running
> test with stale config") if a future test sets mum_resctrlfs - but with all the
> current output of the selftests a warning may be lost in the noise.
> 
> I think it would just be simpler to support both cases. Having the tests be more
> robust is good.

I understand that mum_resctrlfs should support both cases(0&1).

However, "num_of_runs++" is executed before "if (num_of_runs == 0)",
So write_schemata() is never executed regardless of mum_rectrlfs is 0 or 1.

97         if (num_of_runs++ >= NUM_OF_RUNS)
105         if (num_of_runs == 0)
106                 ret = write_schemata(p->ctrlgrp, "100", p->cpu_no,

I will fix this in the next version

Best Regards,
Shaopeng