Re: [PATCH v3 5/5] alpha: fill in pc for software event samples

[Magnus Lindholm]

Hi Matt,

On Sun, Oct 18, 2026 at 4:22 PM Matt Turner <mattst88@xxxxxxxxx> wrote:
>
> Context switch and tracepoint events have no trap frame, so perf builds
> one with perf_fetch_caller_regs(). Alpha does not define
> perf_arch_fetch_caller_regs(), which leaves that pt_regs zeroed, and
> these samples report pc as 0.
>
> Record the caller's address as pc. The zero ps already marks the frame
> as kernel mode, and pt_regs has nowhere to keep sp, so pc is all that
> can be filled in.
>
> perf_fetch_caller_regs() gets that address from CALLER_ADDR0, which
> alpha defines as 0 when CONFIG_FRAME_POINTER is disabled. Finding the
> level zero return address does not need a frame pointer, so make
> alpha_ftrace_return_address0() available in both configurations. Only
> ftrace_return_address(n) still depends on frame pointers.
>
> Reported-by: Magnus Lindholm <linmag7@xxxxxxxxx>
> Suggested-by: Magnus Lindholm <linmag7@xxxxxxxxx>
> Closes: https://lore.kernel.org/all/CA%2B%3DFv5S8-Zk8-LUpZnWiBfTGSQXthepmRc_-a9O%2BpEPmVW0mRA@xxxxxxxxxxxxxx/
> Assisted-by: Claude:claude-opus-5-5
> Reviewed-by: Magnus Lindholm <linmag7@xxxxxxxxx>
> Acked-by: Ian Rogers <irogers@xxxxxxxxxx>
> Signed-off-by: Matt Turner <mattst88@xxxxxxxxx>
> ---
>  arch/alpha/include/asm/ftrace.h     | 5 ++---
>  arch/alpha/include/asm/perf_event.h | 7 +++++++
>  2 files changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/arch/alpha/include/asm/ftrace.h b/arch/alpha/include/asm/ftrace.h
> index 7ec44134c804..5b75a294a7c9 100644
> --- a/arch/alpha/include/asm/ftrace.h
> +++ b/arch/alpha/include/asm/ftrace.h
> @@ -2,8 +2,6 @@
>  #ifndef _ASM_ALPHA_FTRACE_H
>  #define _ASM_ALPHA_FTRACE_H
>
> -#ifdef CONFIG_FRAME_POINTER
> -
>  static void *alpha_ftrace_return_address0(void)
>         noinline notrace;
>  static void *alpha_ftrace_return_address0(void)
> @@ -13,6 +11,8 @@ static void *alpha_ftrace_return_address0(void)
>
>  #define ftrace_return_address0 alpha_ftrace_return_address0()
>
> +#ifdef CONFIG_FRAME_POINTER
> +
>  /*
>   * __builtin_return_address() requires a constant integer argument.
>   * Keep this as a macro so the value is seen at the callsite.
> @@ -21,7 +21,6 @@ static void *alpha_ftrace_return_address0(void)
>
>  #else  /* !CONFIG_FRAME_POINTER */
>
> -#define ftrace_return_address0 0UL
>  #define ftrace_return_address(n) ((void)(n), 0UL)
>
>  #endif /* CONFIG_FRAME_POINTER */
> diff --git a/arch/alpha/include/asm/perf_event.h b/arch/alpha/include/asm/perf_event.h
> index 5996e7a6757e..72536566014a 100644
> --- a/arch/alpha/include/asm/perf_event.h
> +++ b/arch/alpha/include/asm/perf_event.h
> @@ -1,4 +1,11 @@
>  #ifndef __ASM_ALPHA_PERF_EVENT_H
>  #define __ASM_ALPHA_PERF_EVENT_H
>
> +/*
> + * Software events have no trap frame, so perf_fetch_caller_regs() fills
> + * in a zeroed pt_regs. A zero ps already means kernel mode; only pc is
> + * missing.
> + */
> +#define perf_arch_fetch_caller_regs(regs, __ip)        ((regs)->pc = (__ip))
> +
>  #endif /* __ASM_ALPHA_PERF_EVENT_H */
>
> --
> 2.55.0
>

Tested on my AlphaServer ES40 with CONFIG_FRAME_POINTER disabled.
Context-switch and sched_switch samples report valid PCs, with
synthetic SP correctly remaining zero.

Tested-by: Magnus Lindholm <linmag7@xxxxxxxxx>

Thanks,
Magnus