Alison Schofield wrote:
> On Wed, Jun 18, 2025 at 03:21:28PM -0700, Dan Williams wrote:
> > With current kernel+tracecmd combinations stdout is no longer purely trace
> > records and column "21" is no longer the vmfault_t result.
> >
> > Drop, if present, the diagnostic print of how many CPUs are in the trace
> > and use the more universally compatible assumption that the fault result is
> > the last column rather than a specific column.
> >
> > Signed-off-by: Dan Williams <dan.j.willi...@intel.com>
> > ---
> > test/dax.sh | 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/test/dax.sh b/test/dax.sh
> > index 3ffbc8079eba..98faaf0eb9b2 100755
> > --- a/test/dax.sh
> > +++ b/test/dax.sh
> > @@ -37,13 +37,14 @@ run_test() {
> > rc=1
> > while read -r p; do
> > [[ $p ]] || continue
> > + [[ $p == cpus=* ]] && continue
> remove above line
> > if [ "$count" -lt 10 ]; then
> > if [ "$p" != "0x100" ] && [ "$p" != "NOPAGE" ]; then
> > cleanup "$1"
> > fi
> > fi
> > count=$((count + 1))
> > - done < <(trace-cmd report | awk '{ print $21 }')
> > + done < <(trace-cmd report | awk '{ print $NF }')
> replace above line w
> done < <(trace-cmd report | grep dax_pmd_fault_done | awk '{ print $NF
> }')
>
>
> Thanks for all of these Dan!
>
> For this one, I ran into more metadata in the trace file, other than
> 'cpu=' causing the test to fail. I've tested what I'm showing above,
> which makes it immune to other things in the trace file.
>
> Tell me you are OK w this and I'll apply this set to pending.
Yup, looks good to me and is even more robust. Thanks for the fixups!
