Re: [PATCH v2 3/2] sched/deadline: Check bandwidth overflow earlier for hotplug

[Juri Lelli]

On 17/02/25 16:10, Jon Hunter wrote:
> 
> On 17/02/2025 16:08, Juri Lelli wrote:
> > On 14/02/25 10:05, Jon Hunter wrote:
> > 
> > ...
> > 
> > > Sorry for the delay, the day got away from me. However, it is still not
> > > working :-(
> > 
> > Ouch.
> > 
> > > Console log is attached.
> > 
> > Thanks. Did you happen to also collect a corresponding trace?
> 
> Sorry, but I am not sure exactly what trace do you want?

Ah, sorry, I think I mentioned it somewhere else in this long thread.

The idea would be to boot with something like "ftrace=nop
trace_buf_size=50K" added to kernel cmdline.

I would then try to offline CPUs 'manually' in the order suspend seems
to be doing (starting from CPU5). Offlining CPU1 should still fail. At
that point collect the trace with

# cat /sys/kernel/debug/tracing/trace > trace.out

and share it together with dmesg output as you have been doing so far.

Thanks!
Juri