Re: [PATCH] selftests/ftrace: Fix unique symbol check in kprobe_non_uniq_symbol.tc

[Masami Hiramatsu (Google) <mhiramat@kernel.org>]

On Wed, 9 Sep 2026 13:41:56 -0400
Steven Rostedt <rostedt@goodmis.org> wrote:

> On Wed,  9 Sep 2026 11:29:53 +0200
> Sven Schnelle <svens@linux.ibm.com> wrote:
> 
> > The current regex also matches symbols in modules, which makes the
> > test fail on s390 where name_show is present only once in the kernel,
> > but also multiple times in modules:
> > 
> > 000001b1401cdc20 t name_show
> > 000001b0c05e6c40 t name_show    [mdev]
> > 000001b0c0495f30 t name_show    [i2c_core]
> > 
> > Fix this by changing the regular expression to only match the function
> > name.
> > 
> > Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
> > ---
> >  .../selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc    | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc
> > index bc9514428dba..07b1177c1634 100644
> > --- a/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc
> > +++ b/tools/testing/selftests/ftrace/test.d/kprobe/kprobe_non_uniq_symbol.tc
> > @@ -6,7 +6,7 @@
> >  SYMBOL='name_show'
> >  
> >  # We skip this test on kernel where SYMBOL is unique or does not exist.
> > -if [ "$(grep -c -E "[[:alnum:]]+ t ${SYMBOL}" /proc/kallsyms)" -le '1' ]; then
> > +if [ "$(grep -c -E "[[:alnum:]]+ t ${SYMBOL}$" /proc/kallsyms)" -le '1' ]; then
> >  	exit_unsupported
> >  fi
> >  
> 
> Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
> 
> But I wonder if "type_show" or "init_once" is a better option than
> "name_show"?

Yeah, we have many other candidates :)

$ grep -E '[[:alnum:]]+ t [[:alpha:]][[:alnum:]_]+$' /proc/kallsyms | cut -d " " -f 3  | sort | uniq -cd | sort -nr | head 
     27 p4d_offset
     20 type_show
     20 dev_attr_name
     17 dev_attr_modalias
     16 pfn_to_mfn
     15 name_show
     14 modalias_show
     14 dev_attr_type
     13 init_once
     13 format_attr_event

Maybe we can just fallback to this logic for finding available symbols?

Thanks,

-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>