acpi: Narrow workaround for broken interrupt settings on x86

[jhb]

Summary

Commit 9a7bf07ccdc1 from 2016 introduced a workaround for some broken
BIOSes that specified active-lo instead of active-hi polarity for ISA
IRQs for UARTs. The workaround assumed that edge-sensitive ISA IRQs
on x86 should always be active-hi. However, some recent AMD systems
actually use active-lo edge-sensitive ISA IRQs (and not just for
UARTs, but also for the keyboard and PS/2 mouse devices) and the
override causes interrupts to be dropped resulting in boot time hangs,
non-working keyboards, etc.

Add a hw.acpi.override_isa_irq_polarity tunable (readable as a sysctl
post-boot) to control this quirk. It can be set to 1 to force enable
the override and 0 to disable it. The log of original message
mentions an Intel motherboard as the sample case, so default the
tunable to 1 on systems with an Intel CPU and 0 otherwise.

Special thanks to Matthias Lanter <[e-mail]> for tracking
down boot time issues on recent AMD systems to mismatched interrupt
polarity.

PR: 270707
Reported by: [e-mail], Michael Dexter
Reported by: [e-mail], Hannes Hauswedell <[e-mail]>
Reported by: Matthias Lanter <[e-mail]>
Reported by: William Bulley <[e-mail]>
MFC after: 1 week

[imp]

It's like OKish as is, but a tuneable would let our future selves have an easy-to-deploy diagnostic workaround should this guess become incorrect in the future.

sys/dev/acpica/acpi_resource.c

163

maybe add a 'chicken bit' here so one can turn it on/off via sysctl / tuneable?

[jhb]

sys/dev/acpica/acpi_resource.c

163

I think having chicken bits per IRQ is probably a bit much to manage, but having a single hw.acpi.isa_irq_override_polarity that is -1 (default, only override for Intel), 0 (never override), 1 (always override) seems sensible to me?

[imp]

sys/dev/acpica/acpi_resource.c

163

Yea. Do it or do not globally. If there's a big problem, then we can tailor the solution to what we find.

[jhb]

Add a tunable to control the override

[imp]

Is there a nan page that needs to be updated for tye new tuneable?
Otherwise looks good

[emaste]

What about printing a message when the workaround is applied? It might be useful if non-broken Intel systems show up in the future?

[jhb]

In D45554#1048064, @emaste wrote:

What about printing a message when the workaround is applied? It might be useful if non-broken Intel systems show up in the future?

That is hard to do usefully. We have no device name to print when that happens. I guess we could print the ACPI handle though.

[emaste]

That is hard to do usefully. We have no device name to print when that happens. I guess we could print the ACPI handle though.

Even if someone can report on a mailing list that the last thing they see before the hang is applying ACPI ISA IRQ polarity workaround there's a breadcrumb to follow; if the ACPI handle is the most detail we could provide that's fine. Anyhow I'd be happy for this to land soon with or without the additional log.

[jhb]

Add printf and tunable to man page

[imp]

Liked it before, love it now...