Re: Logging Behavior under arch/x86/kernel

From: Borislav Petkov
Date: Fri Aug 19 2022 - 14:14:16 EST


Hi,

On Thu, Aug 18, 2022 at 03:08:44PM +0000, Zhihao Lin wrote:
> 2. At the end of wakeup_secondary_cpu_via_init, send_status is
> checked and given an error-level log message. However, according
> to the control flow analysis, it is unclear that which of
> safe_apic_wait_icr_idle (line 870 or line 918) is checked because
> the loop (which contains the third call) may or may not execute.

Well, does it matter which is checked?

AFAICT, it always checks the right one. :-)

> In this case, my tool alerts that this log statement doesn’t provide
> sufficient information for error diagnostic.

The pr_err() calls at the end of the function do catch the case when
either send_status or accept_status are != 0.

> pci_write_config_dword
>
> * arch/x86/kernel/amd_nb.c

That has been probably overlooked during review. If we had to be
pedantic, we should check all those retvals. I guess not that
important...

HTH.

--
Regards/Gruss,
Boris.

https://people.kernel.org/tglx/notes-about-netiquette