Re: [Patch v4 1/2] freezer: check OOM kill while being frozen

[Cong Wang]

On Tue, Sep 9, 2014 at 1:48 PM, Rafael J. Wysocki <rjw@xxxxxxxxxxxxx> wrote:
> On Wednesday, September 10, 2014 12:23:36 AM Tejun Heo wrote:
>> On Tue, Sep 09, 2014 at 05:16:55PM +0200, Michal Hocko wrote:
>> > But OOM killer doesn't kill kernel threads as they do not own any
>> > memory. So the check should be safe, no?
>>
>> Even for userland tasks, try_to_freeze() can currently be anywhere in
>> the kernel.  The frequently used ones are few but there are some odd
>> ones out, and, again, there's nothing enforcing any structure on
>> try_to_freeze() usage.  The other thing is that we may do quite a bit
>> during exiting including allocating memory.  Are those safe for system
>> PM?  Rafael, what exactly are the rules for PM?  What shouldn't
>> change?
>
> We can't make any assumptions regarding the availability of any devices.
> That is, whatever can end up in device access may potentially fail.
>

So we still have to rule out PM freeze, right?

I am thinking about adding a new task flag or adding a new parameter
to __refrigerator() to distinguish PM freeze with cgroup freeze. Will try
to code tomorrow.

Thanks!
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/