Re: [RFC] Cpu-hotplug: Using the Process Freezer (try2)

[Gautham R Shenoy <ego@in.ibm.com>]

On Mon, Apr 02, 2007 at 08:16:12AM +0200, Ingo Molnar wrote:
> i'm wondering about how TASK_UNINTERRUPTIBLE tasks are handled by the 
> freezer: are they assumed frozen immediately, or do we wait until they 
> notice their PF_FREEZING and go into try_to_freeze()? I'd expect 
> TASK_UNINTERRUPTIBLE to be the largest source of latency. (and hence be 
> the primary source for freezing 'failures')

Ok, are some luck. I panic()ed on freezer fail and checked
the stacktrace of the unfrozen tasks. The stacktrace of each one looks
like:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
PID: 7697   TASK: cc354a70  CPU: 7   COMMAND: "make"
#0 [cc37fe50] schedule at c0431752
#1 [cc37fec4] wait_for_completion at c04318d0
#2 [cc37ff24] do_fork at c01249a6
#3 [cc37ff94] sys_vfork at c0103c1f
#4 [cc37ffb4] system_call at c0104d8d
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Rafael had sent out a patch to fix the vfork race, which can be found at
http://lkml.org/lkml/2007/3/1/212

However, the following hunk seems to be missing from the latest -mm.

@@ -1393,7 +1394,9 @@ long do_fork(unsigned long clone_flags,
		tracehook_report_clone_complete(clone_flags, nr, p);
		if (clone_flags & CLONE_VFORK) {
+			freezer_do_not_count();
			wait_for_completion(&vfork);
+			freezer_count();
			tracehook_report_vfork_done(p, nr);
		}
		} else {



Rafael / Andrew,
        Any reasons for leaving this hunk out?

I reran my tests with this hunk applied, and it work just fine.
Even 'make -j' the maximum time taken to hotplug a cpu was 0m2.16s.

> 
> 	Ingo

Thanks and Regards
gautham.
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"