On Mon, May 18, 2026 at 6:45 AM Boris Brezillon
<[email protected]> wrote:
>
> On Wed, 13 May 2026 10:47:28 -0700
> Chia-I Wu <[email protected]> wrote:
>
> > > >
> > > > Multiple things happen in this commit. I try to identify things that
> > > > can be separate commits. If this does not make sense, feel free to
> > > > ignore.
> > > >
> > > > > /** @tiler_oom_work: Work used to process tiler OOM events
> > > > > happening on this group. */
> > > > > struct work_struct tiler_oom_work;
> > > > >
> > >
> > > [...]
> > >
> > > > > /**
> > > > > * panthor_sched_report_fw_events() - Report FW events to the
> > > > > scheduler.
> > > > > * @ptdev: Device.
> > > > > @@ -1902,8 +1953,19 @@ void panthor_sched_report_fw_events(struct
> > > > > panthor_device *ptdev, u32 events)
> > > > This can be renamed to panthor_sched_handle_fw_events.
> > >
> > > It's not quite handling events though. For most of them, it's really
> > > just deferring the processing to work items, SYNC_UPDATE is the
> > > exception.
> > panthor_sched_report_fw_events no longer just queues
> > process_fw_events_work. It processes fw events immediately. If
> > "handle" is not the right verb, perhaps we can go with "process".
>
> I guess "demux" would be more accurate, but do we need to rename this
> function in the first place? I mean, panthor_sched_report_fw_events()
> doesn't imply that events are processed/handled, it just reflects the
> fact FW events are reported to the scheduler. Up to the scheduler to
> do what it wants with this piece of information (process some of them
> immediately, defer the processing for others, etc).
process_fw_events_work gets inlined into
panthor_sched_report_fw_events. Either report/process/handle should be
fine. Let's drop this comment.
>
> >
> >
> > >
> > > >
> > > > > if (!ptdev->scheduler)
> > > > > return;
> > > > >
> > > > > - atomic_or(events, &ptdev->scheduler->fw_events);
> > > > > - sched_queue_work(ptdev->scheduler, fw_events);
> > > > > + guard(spinlock_irqsave)(&ptdev->scheduler->events_lock);
> > > > > +
> > > > > + if (events & JOB_INT_GLOBAL_IF) {
> > > > > + sched_process_global_irq_locked(ptdev);
> > > > > + events &= ~JOB_INT_GLOBAL_IF;
> > > > > + }
> > > > > +
> > > > > + while (events) {
> > > > > + u32 csg_id = ffs(events) - 1;
> > > > > +
> > > > > + sched_process_csg_irq_locked(ptdev, csg_id);
> > > > > + events &= ~BIT(csg_id);
> > > > > + }
> > > > This handles all fw events in the irq context. Are there concerns that
> > > > it may take too long? I might be wrong, but it seems possible to
> > > > handle only CSG_SYNC_UPDATE and defer the rest as before.
> > >
> > > I started with just the SYNC_UPDATE processing done in the hard-irq
> > > context, but after auditing the other stuff done in the handler, I
> > > realized it's basically just deferring all actual processing to work
> > > items. Yes, there's the overhead of demuxing the events from the
> > > ack/req regs, but part of this is already done to get to SYNC_UPDATE
> > > anyway, so at this point we're probably better off demuxing everything
> > > and scheduling works for all kind of events.
> > >
> > > I also compared the perfs between the two approaches (though I didn't
> > > do as much testing as I did with the new version, so I might have
> > > missed something), and it didn't seem to matter at all, because the
> > > interrupts we receive the most are SYNC_UPDATE and IDLE events, and
> > > those are at the same level.
> > Looking at ftrace irq events, when there is one active csg,
> > panthor-job takes 6us (median) / 17us (95%) / 27us (slowest).
> >
> > I don't have a good sense if that's considered normal in hardirq. But
> > if that is ever an issue, and if the majority of the time is spent in
> > CSG_SYNC_UPDATE anyway, we can always revert the last patch to move
> > processing to threaded handler.
>
> Actually, the threaded -> hard transition (patch 9) is where the perf
> gain is.
hardirq is even more timely for sure. For our use case, the threaded
handler is RT and is also good enough.
