[Bug rtl-optimization/67462] New: [6 Regression] FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump ce1 "3 true changes made"

[Bug rtl-optimization/67462] New: [6 Regression] FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump ce1 "3 true changes made"

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67462

            Bug ID: 67462
           Summary: [6 Regression] FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump
                    ce1 "3 true changes made"
           Product: gcc
           Version: 6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hjl.tools at gmail dot com
                CC: ktkachov at gcc dot gnu.org
  Target Milestone: ---

On Linux/x86-64, r227368 caused:

spawn -ignore SIGHUP /export/build/gnu/gcc-x32/build-x86_64-linux/gcc/xgcc
-B/export/build/gnu/gcc-x32/build-x86_64-linux/gcc/
/export/gnu/import/git/sources/gcc/gcc/testsuite/gcc.dg/ifcvt-3.c -m32
-fno-diagnostics-show-caret -fdiagnostics-color=never -fdump-rtl-ce1 -O2
-ffat-lto-objects -S -o ifcvt-3.s
PASS: gcc.dg/ifcvt-3.c (test for excess errors)
FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump ce1 "3 true changes made"

[Bug rtl-optimization/67462] [6 Regression] FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump ce1 "3 true changes made"

[ktkachov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67462

ktkachov at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-09-08
     Ever confirmed|0                           |1

--- Comment #1 from ktkachov at gcc dot gnu.org ---
This is just a failure to if-convert with -m32 due to costs.
The costing logic deems it to not be profitable, which may well be a valid
decision to make. In this case, if-converting would actually always be
profitable
as the code:
 s64 d = a - b;

  if (d == 0)
    return a + c;
  else
    return b + d + c;

can be simplified to just "a + c" but that's not something we can see in ifcvt
(shouldn't the tree-level optimizers catch this earlier?).

I propose to skip the testcase for -m32, what is the standard way of specifying
that in the testsuite?

[Bug target/67462] [6 Regression] FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump ce1 "3 true changes made"

[ktkachov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67462

ktkachov at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|rtl-optimization            |target

--- Comment #2 from ktkachov at gcc dot gnu.org ---
IMO this is a target issue.
If you think if-conversion should happen for -m32 then the backend costs should
be fixed.
If not, then this test should be skipped.

[Bug target/67462] [6 Regression] FAIL: gcc.dg/ifcvt-3.c scan-rtl-dump ce1 "3 true changes made"

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67462

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |6.0

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
The tree level should indeed do a better job here but it gets "confused" by
narrowing the return expressions to int before it gets a chance to do
that optimization.  It's

 s64 d = a - b;

  if (d == 0)
    return (unsigned)a + (unsigned)c;
  else
    return (unsigned)b + (unsigned)d + (unsigned)c;

to them and 'd' is not handled the same way because the shortening happens
in the frontend.

You might want to file a separate PR about this missed optimization.