This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch][loop-invariant.c] Fix a couple of bugs regarding loop invariant motion discovered by spec2k6 on aarch64


On 05/18/2015 02:16 AM, David Sherwood wrote:
Hi Jeff,

Thanks for the suggestion. I did a bootstrap x86_64 build before and after my
patch and looked for differences in the last stage object files and there were
plenty of them. I chose a nice simple function (check_callers) from
ipa-inline-analysis.c and reduced it to a small test case. Hope this is ok.

Testing done:

  *  aarch64 built, "make check" no regressions
  *  aarch64_be built, "make check" no regressions
  *  x86_64 built, "make check" no regressions

ChangeLog:
     2015-05-15  David Sherwood  <david.sherwood@arm.com>

         * loop-invariant.c (create_new_invariant): Don't calculate address cost
         if mode is not a scalar integer.
          (get_inv_cost): Increase computational cost for unused invariants.
         * testsuite/gcc.dg/loop-invariant.c: New testcase.
Thanks. Installed on the trunk after fixing a couple trivial whitespace nits.

Jeff


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]