Re: [PATCH v8 1/4] nmi_backtrace: add more trigger_*_cpu_backtrace() methods

From: Petr Mladek
Date: Thu Aug 18 2016 - 10:38:29 EST


On Tue 2016-08-16 15:50:21, Chris Metcalf wrote:
> Currently you can only request a backtrace of either all cpus, or
> all cpus but yourself. It can also be helpful to request a remote
> backtrace of a single cpu, and since we want that, the logical
> extension is to support a cpumask as the underlying primitive.
>
> This change modifies the existing lib/nmi_backtrace.c code to take
> a cpumask as its basic primitive, and modifies the linux/nmi.h code
> to use the new "cpumask" method instead.
>
> The mips code ignored the "include_self" boolean but with this change
> it will now also dump a local backtrace if requested.
>
> diff --git a/arch/mips/kernel/process.c b/arch/mips/kernel/process.c
> index 7429ad09fbe3..fea1fa7726e3 100644
> --- a/arch/mips/kernel/process.c
> +++ b/arch/mips/kernel/process.c
> @@ -569,9 +569,16 @@ static void arch_dump_stack(void *info)
> dump_stack();
> }
>
> -void arch_trigger_all_cpu_backtrace(bool include_self)
> +void arch_trigger_cpumask_backtrace(const cpumask_t *mask, bool exclude_self)
> {
> - smp_call_function(arch_dump_stack, NULL, 1);
> + long this_cpu = get_cpu();
> +
> + if (cpumask_test_cpu(this_cpu, mask) && !exclude_self)
> + dump_stack();

The bit is not cleared in the mask. Therefore arch_dump_stack
will get called for this CPU as well.

We could either use similar tricks as in lib/nmi_backtrace().
I mean to copy the mask into a global variable and prevent
parallel call with a backtrace_flag.

Or we could ignore the exclude_self flag as it was done
before. It is a separate problem after all and it need
not be solved in this patch(set).

> +
> + smp_call_function_many(mask, arch_dump_stack, NULL, 1);
> +
> + put_cpu();
> }

Otherwise the patch patch looks good to me.

Best Regards,
Petr