Re: [PATCH] make setpriority POSIX compliant; introduce PRIO_THREAD extension

["Michael Kerrisk" <mtk.manpages@googlemail.com>]

Chris,

On Tue, Sep 9, 2008 at 6:42 PM, Chris Friesen <cfriesen@nortel.com> wrote:
> Michael Kerrisk wrote:
>
>> On 9/1/08, Denys Vlasenko <dvlasenk@redhat.com> wrote:
>>>
>>> Since there may be programs which use the fact that
>>>
>>> setpriority(PRIO_PROCESS, tid, value)
>>>
>>> prior to this patch was setting priority for selected thread,
>>> this behavior is retained in case when tid != pid.
>>>
>>> IOW: with PRIO_PROCESS, if pid specifies a thread group leader,
>>> all threads' prios are set. Otherwise, only selected thread's priority
>>> is set. (Alternative can be to just fail with ESRCH).
>
>
>> I would expect
>> setpriority(PRIO_PROCESS, getpid())
>> and
>> setpriority(PRIO_PROCESS, 0)
>> to have the same affect (because: which == PRIO_PRCESS, who == 0
>> conventionally means "the calling process").
>>
>> But they do not: the latter call only changes the priority of the
>> calling thread.  Is this intended?
>
> The patch interprets 0 as the current pid rather than the current tgid.
> It's up for discussion whether we should preserve old behaviour when
> specifying 0, or use a new and arguably more logical behaviour but possibly
> break old apps.

AFAICS, interpreting 0 as the TGID would be more consistent with those
parts of the interface that are specified by POSIX.1.


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html