Re: [Scst-devel] Changes for RHEL6?

[Vladislav B. <vs...@vl...>]

Hi Guys,

Richard Sharpe, on 04/26/2011 11:23 AM wrote:
> On Tue, Apr 26, 2011 at 11:07 AM, Bart Van Assche
> <bar...@gm...> wrote:
>> On Tue, Apr 26, 2011 at 5:01 PM, Richard Sharpe
>> <rea...@gm...> wrote:
>>>
>>> On Tue, Apr 26, 2011 at 1:58 AM, ido benda <id...@gm...> wrote:
>>>> On Tue, Apr 26, 2011 at 12:00 AM, Richard Sharpe
>>>> <rea...@gm...> wrote:
>>>>> Hi,
>>>>>
>>>>> We are seeing some changes required for RHEL6 around the use of
>>>>> set_cpus_allowed ... Has anyone generated any patches as yet?
>>>>>
>>>>
>>>> Hi Richard,
>>>>
>>>> I sent a patch for RHEL6 on this mailing list some time ago.
>>>> See attached
>>>
>>> Was that with the final released version? We are currently seeing a
>>> problem where code like this:
>>>
>>> #ifdef RHEL_MAJOR
>>>                        rc = set_cpus_allowed(thr->cmd_thread,
>>>                                tgt_dev->acg_dev->acg->acg_cpu_mask);
>>> #else
>>>                        rc = set_cpus_allowed_ptr(thr->cmd_thread,
>>>                                &tgt_dev->acg_dev->acg->acg_cpu_mask);
>>> #endif
>>>
>>> must have the first #ifdef changes to:
>>>
>>> #if defined(RHEL_MAJOR) && RHEL_MAJOR -0 <= 5
>>>
>>> It seems that in the version of 2.6.32 that we are running with,
>>> things have changed.
>>
>> Hello Richard,
>>
>> In "#if defined(RHEL_MAJOR) && RHEL_MAJOR -0 <= 5" the "defined(RHEL_MAJOR)
>> &&" part is superfluous. If the C preprocessor encounters an undefined
>> symbol it gets replaced by whitespace. So the test "RHEL_MAJOR -0 <= 5" will
>> be expanded to "-0 <= 5" if RHEL_MAJOR is not defined.
> 
> That's intriguing because there are three instances of:
> 
>     || defined(RHEL_MAJOR) && RHEL_MAJOR -0 <= 5
> 
> in the code I just scanned, which is pristine from SVN.
> 
> Secondly, before we changed things, we were getting errors during
> build about set_cpus_allowed being undefined, and after the change, it
> built OK.
> 
> Perhaps RHEL_MAJOR was not being passed into our build, but then I
> would expect it to have build without the change, unless #ifdef
> __NOTHING__ is true ...

I'm somehow lost in the proposed patches. So, when you have a fully
working one, please (re)send it here.

Thanks,
Vlad