[Bug middle-end/78411] [7 Regression] FAIL: gcc.target/i386/pr45685.c scan-assembler-times cmov 6

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78411

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|                            |x86_64-*-*, i?86-*-*
             Status|UNCONFIRMED                 |NEW
           Keywords|                            |missed-optimization
   Last reconfirmed|                            |2016-11-18
                 CC|                            |rguenth at gcc dot gnu.org,
                   |                            |wschmidt at gcc dot gnu.org
             Blocks|                            |77848
     Ever confirmed|0                           |1
            Summary|FAIL:                       |[7 Regression] FAIL:
                   |gcc.target/i386/pr45685.c   |gcc.target/i386/pr45685.c
                   |scan-assembler-times cmov 6 |scan-assembler-times cmov 6
   Target Milestone|---                         |7.0

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ok, so this relied on tree loop if-conversion.  Not sure if it really should
though (tree loop if-conversion has no cost model).  I'd rather have RTL
if-conversion do the if-conversion if profitable.

Otherwise the regression can be easily fixed by adding -ftree-loop-if-convert
(though I'd recommend against this).


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77848
[Bug 77848] Gimple if-conversion results in redundant comparisons