Hi Peter
On 10/23/20 8:39 AM, Peter Zijlstra wrote:
On Thu, Oct 22, 2020 at 03:20:33PM -0600, Mathieu Poirier wrote:
Suzuki's depiction of the usecase is accurate. Using the pid of the process
that created the events comes out of a discussion you and I had in the common
area by the Intel booth at ELC in Edinburgh in the fall of 2018. At the time I
exposed the problem of having multiple events sharing the same HW resources and
you advised to proceed this way.
Bah, I was afraid of that. I desperately tried to find correspondence on
it, but alas, verbal crap doesn't end up in the Sent folder :-/
That being said it is plausible that I did not expressed myself clearly enough
for you to understand the full extend of the problem. If that is the case we
are more than willing to revisit that solution. Do you see a better option than
what has currently been implemented?
Moo... that really could've done with a comment I suppose.
So then I don't understand the !->owner issue, that only happens when
the task dies, which cannot be concurrent with event creation. Are you
Part of the patch from Sai, fixes this by avoiding the dereferencing
after event creation (by caching it). But the kernel events needs
fixing.
One follow up question on the !->owner issue. Given the ->owner is
dying, does it prevent events from being scheduled ? Or is there a delay
between that and eventually stopping the events. In this case, we hit
the issue when :
A A or B ?
event_start()
... event->owner = NULL
READ_ONCE(event->owner);
Is this expected ?
somehow accessing ->owner later?
As for the kernel events.. why do you care about the actual task_struct
* in there? I see you're using it to grab the task-pid, but how is that
useful?
Correct, kernel events are something that the driver didn't account for.
May be we could handle this case with a "special pid" and simply
disallow sharing (which is fine I believe, given there are not grouping
for the kernel created events).
Suzuki
