Re: [PATCH v9 7/7] arm64: kgdb: Roundup cpus using the debug IPI

[Mark Rutland]

On Mon, Aug 07, 2023 at 04:24:44PM +0100, Daniel Thompson wrote:
> On Mon, Aug 07, 2023 at 11:28:52AM +0100, Mark Rutland wrote:
> > On Thu, Jun 01, 2023 at 02:31:51PM -0700, Douglas Anderson wrote:
> > > From: Sumit Garg <sumit.garg@xxxxxxxxxx>
> > >
> > > Let's use the debug IPI for rounding up CPUs in kgdb. When the debug
> > > IPI is backed by an NMI (or pseudo NMI) then this will let us debug
> > > even hard locked CPUs. When the debug IPI isn't backed by an NMI then
> > > this won't really have any huge benefit but it will still work.
> > >
> > > Signed-off-by: Sumit Garg <sumit.garg@xxxxxxxxxx>
> > > Signed-off-by: Douglas Anderson <dianders@xxxxxxxxxxxx>
> > > ---
> > >
> > > Changes in v9:
> > > - Remove fallback for when debug IPI isn't available.
> > > - Renamed "NMI IPI" to "debug IPI" since it might not be backed by NMI.
> > >
> > >  arch/arm64/kernel/ipi_debug.c |  5 +++++
> > >  arch/arm64/kernel/kgdb.c      | 14 ++++++++++++++
> > >  2 files changed, 19 insertions(+)
> >
> > This looks fine to me, but I'd feel a bit happier if we had separate SGIs for
> > the backtrace and the KGDB callback as they're logically unrelated.
> 
> I've no objections to seperate SGIs (if one can be found) but I'm curious
> what benefit emerges from giving them seperate IPIs.

Mostly an "I'd feel happier"; they're logically unrelated and having them
separate avoids the risk of them unintentionally getting in the way of the
other.

> Both interfaces are already designed to share and NMI-like IPI nicely
> (and IIUC they must share one on x86), neither is performance
> critical[1] and the content of /proc/interrupts for the IPI is seldom
> going to be of much interest.

Sure; I understand that. The flip side of "neither is performance critical" is
that they're seldom tested in terms of interaction with one another, and hence
for robustness I'd prefer they're separate.

I agree it's not strictly necessary, but given we can easily free up an SGI
slot, I think it'd be worthwhile. We can always decide to fold them together in
future if we have to.

I realise a similar argument could be applied to IPI_WAKEUP and IPI_RESCHEDULE,
but given that IPI_RESCHEDULE happens *all the time* and the wakeup handler
does literally nothing, I think the risk there is substantially lower.

Thanks,
Mark.

> As mentioned it is perfectly OK to separate them if there is space in
> the SGI allocations. However if any two functions are good candidates to
> share a scarce resource such as an SGI then it is these!
> 
> 
> Daniel.
> 
> 
> [1] In both cases their results are only required at human-scale
>     and the work of the both handlers is hugely more expensive than
>     either up front quick-check.