Re: [PATCH bpf-next] bpf: Add kernel symbol for struct_ops trampoline

[Yonghong Song <yonghong.song@linux.dev>]

On 10/30/24 4:15 AM, Xu Kuohai wrote:
> From: Xu Kuohai <xukuohai@huawei.com>
>
> Without kernel symbols for struct_ops trampoline, the unwinder may
> produce unexpected stacktraces.
>
> For example, the x86 ORC and FP unwinders check if an IP is in kernel
> text by verifying the presence of the IP's kernel symbol. When a
> struct_ops trampoline address is encountered, the unwinder stops due
> to the absence of symbol, resulting in an incomplete stacktrace that
> consists only of direct and indirect child functions called from the
> trampoline.

Please give some concrete examples here, e.g. stack trace before and
after this patch, so it will be clear what is fixed.

>
> The arm64 unwinder is another example. While the arm64 unwinder can
> proceed across a struct_ops trampoline address, the corresponding
> symbol name is displayed as "unknown", which is confusing.
>
> Thus, add kernel symbol for struct_ops trampoline. The name is
> bpf_trampoline_<PROG_NAME>, where PROG_NAME is the name of the
> struct_ops prog linked to the trampoline.
>
> Fixes: 85d33df357b6 ("bpf: Introduce BPF_MAP_TYPE_STRUCT_OPS")
> Signed-off-by: Xu Kuohai <xukuohai@huawei.com>

There is a warning in kernel test bot, please fix it. Otherwise,
the patch LGTM. I also tried with one struct_ops example and it
does show full *good* stack with this patch, and without
this patch, the backtrace stops right before trampoline symbols.

Acked-by: Yonghong Song <yonghong.song@linux.dev>