Re: [PATCH v1] perf tests: Fix data symbol test with LTO builds

[James Clark]

On 27/02/2025 5:42 am, Ian Rogers wrote:
> On Wed, Feb 26, 2025 at 3:01 PM Ian Rogers <irogers@xxxxxxxxxx> wrote:
> > With LTO builds, although regular builds could also see this as
> > all the code is in one file, the datasym workload can realize the
> > buf1.reserved data is never accessed. The compiler moves the
> > variable to bss and only keeps the data1 and data2 parts as
> > separate variables. This causes the symbol check to fail in the
> > test. Make the variable volatile to disable the more aggressive
> > optimization. Rename the variable to make which buf1 in perf is
> > being referred to.
> >
> > Before:
> > ```
> > $ perf test -vv "data symbol"
> > 126: Test data symbol:
> > --- start ---
> > test child forked, pid 299808
> > perf does not have symbol 'buf1'
> > perf is missing symbols - skipping test
> > ---- end(-2) ----
> > 126: Test data symbol                                                : Skip
> > $ nm perf|grep buf1
> > 0000000000a5fa40 b buf1.0
> > 0000000000a5fa48 b buf1.1
> > ```
> >
> > After:
> > ```
> > $ nm perf|grep buf1
> > 0000000000a53a00 d buf1
> > $ perf test -vv "data symbol"126: Test data symbol:
> > --- start ---
> > test child forked, pid 302166
> >   a53a00-a53a39 l buf1
> > perf does have symbol 'buf1'
> > Recording workload...
> > Waiting for "perf record has started" message
> > OK
> > Cleaning up files...
> > ---- end(0) ----
> > 126: Test data symbol                                                : Ok
> > ```
> >
> > Fixes: 3dfc01fe9d12 ("perf test: Add 'datasym' test workload")
> > Signed-off-by: Ian Rogers <irogers@xxxxxxxxxx>
>
> Ah, I see we're trying to force -O0 and -fno-inline in the Makefile
> (git.kernel.org is giving 403s):
> https://github.com/torvalds/linux/blob/master/tools/perf/tests/workloads/Build#L11
> Which LTO later undoes. I'm not seeing LTO breakages for brstack.c and
> the shell test "Check branch stack sampling". I think LTO is able to
> optimize this case as it is a variable/struct being optimized, so the
> "-O0" and "-fno-inline" mustn't be being made to apply. Not a wholly
> satisfactory reason to add the volatile, but I'm short on
> alternatives.
>
> Thanks,
> Ian

Adding -fno-lto to that file works for me (llvm-15):

  CFLAGS_datasym.o          = -fno-lto -g -O0 -fno-inline ...

In fact it looks like we should add this to everywhere we're already 
doing -g and -O0. If lto is just another form of optimisation it should 
be disabled as well.