[LTP] [PATCH v3 08/16] controllers: Expand cgroup_lib shell library

[Li Wang]

Hi Luke,

On Thu, Mar 17, 2022 at 5:47 AM Luke Nowakowski-Krijger <
luke.nowakowskikrijger@canonical.com> wrote:

> Hi Richard and Li,
>
> On Wed, Mar 16, 2022 at 2:46 AM Li Wang <liwang@redhat.com> wrote:
>
>> Richard Palethorpe <rpalethorpe@suse.de> wrote:
>>
>>
>>
>>> >> +
>>> >> +    _cgroup_state=$(tst_cgctl require "$ctrl" $$)
>>> >> +
>>> >> +    [ "$_cgroup_state" = "" ] && tst_brk TBROK "cgroup_require: No
>>> >state was set after call. Controller '$ctrl' maybe does not exist?"
>>>
>>> Perhaps we can just check the return status of tst_cgctl and exit the
>>> script if it is >0?
>>>
>>
>> That should be work, but we need to check the _cgrou_state
>> at other places because many shell APIs require getting Cgroup
>> info, I guess this does not make things more easily if check the
>> return value only one time.
>>
>>
>>
> An idea I had to make the failure of tst_cgctl more graceful when there is
> a tst_brk and when it exits is to check the return of tst_cgctl and
> basically just mirror the error with a better message. E.g.
>
> tst_cgroup.c:702: TCONF: IO controller found on V2 root, skipping blkio
> mount that would unmount IO controller
> cgroup_fj_function 1 TCONF: cgroup_require: tst_cgtl require exited
>
> Something that would look like this:
> _cgroup_state=$(tst_cgctl require "$ctrl" $$)
>
> _cgroup_check_return "$?" "cgroup_require: tst_cgctl require exited"
>
> [ "$_cgroup_state" = "" ] && tst_brk TBROK "cgroup_require: No state was
> set after call to tst_cgctl require?"
>
> Where in _cgroup_check_return we would do the appropriate tst_brk for the
> return and then return a nice message (not exactly sure what it should be
> but it should be obvious that one error is caused by the other). Because if
> there was an error in tst_cgroup we would definitely want to break the test
> because the _cgroup_state was most likely not set. Then it would fall
> through to checking the _cgroup_state if everything looked good per normal
> which now should never happen if tst_cgctl returned a non error.
>
> Let me know what you think :)
>

That sounds good to me.


-- 
Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20220317/ad13cbfb/attachment-0001.htm>