[Bug rtl-optimization/61608] New: [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[Bug rtl-optimization/61608] New: [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

            Bug ID: 61608
           Summary: [4.10 regression] FAIL: gcc.target/arm/epilog-1.c
                    scan-assembler tests
           Product: gcc
           Version: 4.10.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jgreenhalgh at gcc dot gnu.org

Revision 211919 changes the position of the peephole2 pass causing a code-size
regression for ARM targets.

My compiler is configured with:

Target: arm-none-eabi
Configured with: /work/gcc-clean/src/gcc/configure --target=arm-none-eabi
--prefix=/work/gcc-clean/build-arm-none-eabi/install
--with-gmp=/work/gcc-clean/build-arm-none-eabi/host-tools
--with-mpfr=/work/gcc-clean/build-arm-none-eabi/host-tools
--with-mpc=/work/gcc-clean/build-arm-none-eabi/host-tools
--with-pkgversion=unknown --disable-shared --disable-nls --disable-threads
--disable-tls --enable-checking=yes --enable-languages=c,c++,fortran
--with-newlib --with-arch=armv8-a --with-float=hard
--with-fpu=crypto-neon-fp-armv8
Thread model: single
gcc version 4.10.0 20140624

---
Code generation Before:

foo:
    @ args = 0, pretend = 0, frame = 0
    @ frame_needed = 0, uses_anonymous_args = 0
    push    {r4, lr}
    ldr    r4, .L6
    ldr    r4, [r4]
    lsls    r4, r4, #29
    it    mi
    addmi    r2, r2, #1
    bl    bar
    movs    r0, #0
    pop    {r4, pc}
.
---
Code generation After:

foo:
    @ args = 0, pretend = 0, frame = 0
    @ frame_needed = 0, uses_anonymous_args = 0
    push    {r4, lr}
    ldr    r4, .L6
    ldr    r4, [r4]
    tst    r4, #4
    it    ne
    addne    r2, r2, #1
    bl    bar
    movs    r0, #0
    pop    {r4, pc}
---

The tst (immediate) instruction is a 4-byte thumb instruction, while lsls is a
2-byte thumb instruction.

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

jgreenhalgh at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|                            |arm*-none-*
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-06-25
                 CC|                            |jgreenhalgh at gcc dot gnu.org,
                   |                            |ktietz at gcc dot gnu.org
     Ever confirmed|0                           |1

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[ktietz at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

--- Comment #1 from Kai Tietz <ktietz at gcc dot gnu.org> ---
Does this issue get fixed by adding the peephole2 also at old place too?

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

--- Comment #2 from jgreenhalgh at gcc dot gnu.org ---
Yes, it does.

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[ktietz at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

--- Comment #3 from Kai Tietz <ktietz at gcc dot gnu.org> ---
Thanks for testing.  I will sent a patch for it.
It seems after all that we need to run peephole2 pass twice.

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Kai Tietz from comment #3)
> Thanks for testing.  I will sent a patch for it.
> It seems after all that we need to run peephole2 pass twice.

Bad for compile-time.

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.10.0

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[rth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

--- Comment #5 from Richard Henderson <rth at gcc dot gnu.org> ---
Author: rth
Date: Mon Jun 30 20:14:42 2014
New Revision: 212172

URL: https://gcc.gnu.org/viewcvs?rev=212172&root=gcc&view=rev
Log:
PR rtl-opt/61608

        PR target/39284
        * bb-reorder.c (pass_duplicate_computed_gotos::execute): Cleanup
        the cfg if there were any changes.
        * passes.def: Revert move of peephole2 after reorder_blocks;
        move duplicate_computed_gotos before peephole2.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/bb-reorder.c
    trunk/gcc/passes.def

[Bug rtl-optimization/61608] [4.10 regression] FAIL: gcc.target/arm/epilog-1.c scan-assembler tests

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61608

jgreenhalgh at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #6 from jgreenhalgh at gcc dot gnu.org ---
This looks resolved to me.