Re: [PATCH v2] x86/acpi/boot: Do not register processors that cannot be onlined for x2apic

[Guy Durrieu]

Le 05/01/2023 à 05:10, Kishon Vijay Abraham I a écrit :

> Section 5.2.12.12 Processor Local x2APIC Structure in the ACPI v6.5
> spec mandates that both "enabled" and "online capable" Local APIC Flags
> should be used to determine if the processor is usable or not.
>
> However, Linux doesn't use the "online capable" flag for x2APIC to
> determine if the processor is usable. As a result, cpu_possible_mask has
> incorrect value and results in more memory getting allocated for per_cpu
> variables than it is going to be used.
>
> Make sure Linux parses both "enabled" and "online capable" flags for
> x2APIC to correctly determine if the processor is usable.
>
> Fixes: aa06e20f1be6 ("x86/ACPI: Don't add CPUs that are not online capable")
> Reviewed-by: Borislav Petkov (AMD) <bp@xxxxxxxxx>
> Reported-by: Leo Duran <leo.duran@xxxxxxx>
> Signed-off-by: Kishon Vijay Abraham I <kvijayab@xxxxxxx>

Hello everyone,

My system worked fine with kernel 6.1.15, but stopped booting after
upgrading to 6.1.20 and resulted in a kernel panic:

---
[ 0.117782] Kernel panic — not syncing: timer doesn’t work through Interrupt-remapped IO-APIC
[ 0.117848] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 6.1.0-7-and64 #1 Debian 6.1.20-1
[ 0.117913] Hardware name: Gigabyte Technology Co., Ltd. ABS50M-Gaming 3/AB350M-Gaming 3-CF, BIOS F50d 07/02/2020
[ 0.117982] Call Trace:
[ 0.118634] <TASK>
[ 0.118685] dump_stack_lvl+0x44/0x5c
[ 0.118143] panic+0x118/0x2ed
[ 0.118198] panic_if_irq_remap.cold+0x5/0x5
[ 0.118256] setup_I0_APIC+0x3db/0x64b
[ 0.118313] ? _raw_spin_unlock_irqrestore+0x23/0x40
[ 0.118372] ? clear_IO_APIC_pin+0x169/0x240
[ 0.118429] apic_intr_node_init+0x101/0x106
[ 0.118485] x86_late_time_init+0x20/0x34
[ 0.118542] start_kerne1+0x667/0x727
[ 0.118598] secondary_startup_64_no_verify+0xe5/0xeb
[ 0.118658] </TASK>
[ 0.118711] ---[ end Kernel panic - not syncing: timer doesn’t work through Interrupt-remapped IO-APIC J---
---
I tried an update of the BIOS up to F51h without any effect.

I sent a bug report to Debian Bug Tracking System. In reply Bjørn Mork 
identified ce7d894bed1a539a8d6cff42f6f78f9db0c9c26b from the linux-6.1.y 
branch as the likely culprit.

After building a 6.1.20 kernel with just that commit reverted, my system boots normally again.

I reported that all tohttps://bugs.debian.org/1033732  and were asked to report the issue 'upstream' (which is what that mail is).

Hope it will help!

Best regards.

-- Guy Durrieu