Re: Question about the ipi_raise filter usage and output

[richard clark]

On Tue, Feb 6, 2024 at 12:05 AM Valentin Schneider <vschneid@redhatcom> wrote:
>
> On 05/02/24 14:39, Mark Rutland wrote:
> > [adding Valentin]
> >
>
> Thanks!
>
> > On Mon, Feb 05, 2024 at 08:06:09AM -0500, Steven Rostedt wrote:
> >> On Mon, 5 Feb 2024 10:28:57 +0000
> >> Mark Rutland <mark.rutland@xxxxxxx> wrote:
> >>
> >> > > I try to write below:
> >> > > echo 'target_cpus == 11 && reason == "Function call interrupts"' >
> >> > > events/ipi/ipi_raise/filter
> >> >
> >> > The '=' checks if the target_cpus bitmap *only* contains CPU 11. If the cpumask
> >> > contains other CPUs, the filter will skip the call.
> >> >
> >> > I believe you can use '&' to check whether a cpumask contains a CPU, e.g.
> >> >
> >> >    'target_cpus & 11'
> >>
> >> 11 == 0xb = b1011
> >>
> >> So the above would only be true for CPUs 0,1 and 3 ;-)
> >
> > Sorry, I misunderstood the scalar logic and thought that we treated:
> >
> >       '$mask $OP $scalar', e.g. 'target_cpus & 11'
> >
> > .. as a special case meaning a cpumask with that scalar bit set, i.e.
> >
> >       '$mask $OP CPUS{$scalar}', e.g. 'target_cpus & CPUS{11}'
> >
> > .. but evidently I was wrong.
> >
> >> I think you meant: 'target_cpus & 0x800'
> >>
> >> I tried "1 << 11' but it appears to not allow shifts. I wonder if we should add that?
> >
> > Hmm... shouldn't we make 'CPUS{11}' work for that?
> >
>
> It /should/ already be the case, the user input with the curly braces is
> parsed as a cpulist. So CPUS{11} really does mean CPU11, not a hex value to
> be interpreted as a cpumask.
>
> However...
>
> > From a quick test (below), that doesn't seem to work, though I think it
> > probably should?
> > Have I completely misunderstood how this is supposed to work, or is that a bug?
> >
>
> The CPUS{} thingie only works with an event field that is either declared as a
> cpumask (__cpumask) or a scalar. That's not the case for ipi_raise, the
> target_cpus event field is saved as a "raw" bitmask.
>
> There /should/ have been a warning about the event filter though, but I
> think it's not happening because I'm allowing more than just FILTER_CPUMASK
> in parse_pred() to make it work for scalars. I'll go poke around some more.
>
> Generally for this sort of IPI investigation I'd recommend using the newer
> trace_ipi_send_cpu() and trace_ipi_send_cpumask() (for which CPUS{}
> filtering works).
> If it's only the function call interrupts you're interesting in, have a
> look at trace_csd_queue_cpu().

This should be supported by newer version kernels like v6.5, but I am
using v6.1 and this trace event has not been supported yet... so ipi
is more suitable for me. ipi_entry and ipi_exit is ok, but seems the
filter doesn't support a specific cpu, maybe we need to add this?
>
> > Mark.
>