On Tue, Jul 21, 2020 at 1:14 AM Sandra Loosemore <san...@codesourcery.com> wrote: > > On 7/20/20 2:15 AM, Richard Biener wrote: > > > I think at least parts of tree-prof.exp exercises sample-based profiling > > which might require more iterations. For example cold_partition_label.c > > was changed by > > > > commit f63ba78ce6d50bf627dd18018179eb03bf89716f > > Author: Andi Kleen <a...@linux.intel.com> > > Date: Thu Jul 14 02:14:56 2016 +0000 > > > > Some fixes for profile test cases for autofdo > > > > This fixes some basic issues with the profile test cases with autofdo. > > > > - Disable checking for value transformations that autofdo does not > > support. > > - Disable checking for fixed hit counts which autofdo does not support > > - Enable dumping of afdo log file and check right log file. > > - Increase run time of test cases to 1M iterations because autofdo > > needs > > a few samples to make sense of a program. The test case don't run > > noticeable slower with that. > > > > There are still failures unfortunately, especially the indirect call > > transformations do not trigger because autofdo thinks they are not hot. > > This can be addressed later. > > > > so the change to a larger number of iterations was intended. Maybe > > we can arrange to pass -DFOR_AUTOFDO_TESTING for the > > autofdo compiles and gate the larger number of iterations on that > > (most targets do not support autofdo and to not run that mode)? > > Something like the attached updated patch? Unfortunately I'm not set up > to test that this actually works on an autofdo target, maybe somebody > else could give it a try?
Looks good to me, thus OK without further tests. Thanks, Richard. > -Sandra
