[Bug rtl-optimization/43653] ICE: in reload, at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[Bug rtl-optimization/43653] ICE: in reload, at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[devurandom at gmx dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #3 from devurandom at gmx dot net 2010-12-25 23:43:13 UTC ---
4.5.1 shows the same issue when building Clang 2.8.

[Bug rtl-optimization/43653] ICE: in reload, at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[devurandom at gmx dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #4 from devurandom at gmx dot net 2011-01-03 18:24:15 UTC ---
(In reply to comment #3)
> 4.5.1 shows the same issue when building Clang 2.8.
Persists with 4.5.2.

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE: in reload, at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.3.6

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE: in reload, at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE: in reload, at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu.org

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-03 20:42:33 UTC ---
It is triggered by revision 124324:

http://gcc.gnu.org/ml/gcc-cvs/2007-05/msg00015.html

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at redhat dot com

--- Comment #6 from Jeffrey A. Law <law at redhat dot com> 2011-02-15 17:27:44 UTC ---
Working on trunk, still investigating if there was a patch designed to fix this
problem or if the bug has just gone latent.

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #7 from Uros Bizjak <ubizjak at gmail dot com> 2011-02-15 18:35:43 UTC ---
(In reply to comment #6)
> Working on trunk, still investigating if there was a patch designed to fix this
> problem or if the bug has just gone latent.

It still crashes for me (with the test from attachment 20315):

~/gcc-build/gcc/cc1 -O1 -ftree-vectorize -quiet pr43653.c 
pr43653.c: In function ‘foo’:
pr43653.c:10:1: internal compiler error: in reload, at reload1.c:1059
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


~/gcc-build/gcc/xgcc -v
Using built-in specs.
COLLECT_GCC=/home/uros/gcc-build/gcc/xgcc
Target: x86_64-unknown-linux-gnu
Configured with: ../gcc-svn/trunk/configure --enable-languages=c,c++,fortran
Thread model: posix
gcc version 4.6.0 20110214 (experimental) [trunk revision 170139] (GCC)

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #8 from Jeffrey A. Law <law at redhat dot com> 2011-02-15 19:01:20 UTC ---
ah.  Fails with -O1 -ftree-vectorize, works with -O2 -ftree-vectorize on trunk.

Appears that we allocate memory (via get_secondary_mem) after the point at
which we believe the size of frame should be fixed.  This would tend to
indicate that during find_reloads we thought no secondary memory was needed,
but then later decided it was needed.  That's clearly bad.

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #9 from Jeffrey A. Law <law at redhat dot com> 2011-02-15 21:06:23 UTC ---
I think an x86 maintainer is going to need to take a look at this.

The problem as I see it is we're trying to use the address of a stack slot as a
vector initializer.  

(insn 20 17 21 2 (set (reg:V2DI 74 [ vect_cst_.2 ])
        (vec_duplicate:V2DI (reg/f:DI 20 frame))) j.c:4 1788 {*vec_dupv2di}
     (expr_list:REG_EQUIV (vec_duplicate:V2DI (reg/f:DI 20 frame))
        (nil)))

There's nothing inherently wrong with that.  Eventually frame gets turned into
sp+offset like this:

(insn 20 17 21 2 (set (reg:V2DI 21 xmm0 [orig:74 vect_cst_.2 ] [74])
        (vec_duplicate:V2DI (plus:DI (reg/f:DI 7 sp)
                (const_int 392 [0x188])))) j.c:4 1788 {*vec_dupv2di}
     (expr_list:REG_EQUIV (vec_duplicate:V2DI (plus:DI (reg/f:DI 7 sp)
                (const_int 392 [0x188])))
        (nil)))

This is a natural result of FP elimination and IMHO, we're still OK. That later
gets turned into:

(insn 20 17 21 2 (set (reg:V2DI 21 xmm0 [orig:74 vect_cst_.2 ] [74])
        (vec_duplicate:V2DI (plus:DI (reg/f:DI 7 sp)
                (mem/u/c/i:DI (symbol_ref/u:DI ("*.LC0") [flags 0x2]) [0 S8
A64])))) j.c:4 1788 {*vec_dupv2di}
     (expr_list:REG_EQUIV (vec_duplicate:V2DI (plus:DI (reg/f:DI 7 sp)
                (const_int 392 [0x188])))
        (nil)))

Dubious, but I understand why the code did this.    We'll have the following
reload recorded:

Reload 0: reload_in (DI) = (plus:DI (reg/f:DI 7 sp)
                                                    (mem/u/c/i:DI
(symbol_ref/u:DI ("*.LC0") [flags 0x2]) [0 S8 A64]))
    reload_out (V2DI) = (reg:V2DI 21 xmm0 [orig:74 vect_cst_.2 ] [74])
    SSE_REGS, RELOAD_OTHER (opnum = 0), can't combine
    reload_in_reg: (plus:DI (reg/f:DI 7 sp)
                                                    (const_int 392 [0x188]))
    reload_out_reg: (reg:V2DI 21 xmm0 [orig:74 vect_cst_.2 ] [74])
    reload_reg_rtx: (reg:V2DI 22 xmm1)

Which simply isn't going to work and it's all downhill from there, including a
surprise secondary memory allocation which ultimately triggers the ICE.

I was able to hack things to work by enabling a secondary reload when copying
(plus (sp) (whatever)) into SSE_REGS, but that may not be the best solution. 
Someone with more x86 internal knowledge should look at this.

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #10 from Uros Bizjak <ubizjak at gmail dot com> 2011-02-15 22:41:40 UTC ---
(In reply to comment #9)
> I think an x86 maintainer is going to need to take a look at this.

> Which simply isn't going to work and it's all downhill from there, including a
> surprise secondary memory allocation which ultimately triggers the ICE.
> 
> I was able to hack things to work by enabling a secondary reload when copying
> (plus (sp) (whatever)) into SSE_REGS, but that may not be the best solution. 
> Someone with more x86 internal knowledge should look at this.

Hm, indeed. SSE can't handle non-vector PLUS insn, so reload needs a bit of
help.

Following patch fixes the testcase:

Index: i386.c
===================================================================
--- i386.c    (revision 170189)
+++ i386.c    (working copy)
@@ -28313,6 +28313,11 @@ ix86_secondary_reload (bool in_p, rtx x,
     return Q_REGS;
     }

+  if (in_p && GET_CODE (x) == PLUS
+      && !VECTOR_MODE_P (mode)
+      && rclass == SSE_REGS)
+    return GENERAL_REGS;
+
   return NO_REGS;
 }


And generates:

    movq    .LC1(%rip), %rax
    leaq    (%rsp,%rax), %rax
    movq    %rax, -112(%rsp)
    movq    -112(%rsp), %xmm1
    punpcklqdq    %xmm1, %xmm1

[Bug rtl-optimization/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #11 from Jeffrey A. Law <law at redhat dot com> 2011-02-16 04:03:41 UTC ---
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 02/15/11 15:42, ubizjak at gmail dot com wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653
> 
> --- Comment #10 from Uros Bizjak <ubizjak at gmail dot com> 2011-02-15 22:41:40 UTC ---
> (In reply to comment #9)
>> I think an x86 maintainer is going to need to take a look at this.
> 
>> Which simply isn't going to work and it's all downhill from there, including a
>> surprise secondary memory allocation which ultimately triggers the ICE.
>>
>> I was able to hack things to work by enabling a secondary reload when copying
>> (plus (sp) (whatever)) into SSE_REGS, but that may not be the best solution. 
>> Someone with more x86 internal knowledge should look at this.
> 
> Hm, indeed. SSE can't handle non-vector PLUS insn, so reload needs a bit of
> help.
It's unfortunate that reload decides to reload the constant offset, but
I suspect this kind of thing is exceedingly rare.

I'll let you own taking care of the PR since you've got a patch :-)

jeff
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJNW0yQAAoJEBRtltQi2kC7VJsH/isQ/CjWMiBfrNea5JEqF6Hw
rc955z1QyN4584IfL2Vcl1KTwvMr5zy7FJTiWbybl812a24V5b7lFwGlgDmag3/6
59XY1RdEH4j+JLYuT+bEu3IBL4orqNZSHMxDRAqzeYGbzLj16KVJeS1+cNQV4iV5
foXZGrZ+1jr4saqpBCkpA0ecid5a4eX8vtMv2EmGWnW7LW/fy2w8casgn24iGio+
0HdMn8gBQuTGknU/u7yLf0ArS9/UBlrk8iE3VM9ssgz4oD4w3BJt0J5VMlBaqCpQ
G8BgZkfrmVCAvYcEiHpa6Cj30OmR3VeWN9IDu+jUWiZhSOluo5EDIxWCU8S1FBU=
=7vqY
-----END PGP SIGNATURE-----

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
          Component|rtl-optimization            |target
         AssignedTo|unassigned at gcc dot       |ubizjak at gmail dot com
                   |gnu.org                     |

--- Comment #12 from Uros Bizjak <ubizjak at gmail dot com> 2011-02-16 08:27:50 UTC ---
Target bug. I have a patch in testing.

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

Uros Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2011-02/txt00085.txt

--- Comment #13 from Uros Bizjak <ubizjak at gmail dot com> 2011-02-16 11:21:03 UTC ---
Patch at [1].

[1] http://gcc.gnu.org/ml/gcc-patches/2011-02/txt00085.txt

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #14 from uros at gcc dot gnu.org 2011-02-17 08:29:55 UTC ---
Author: uros
Date: Thu Feb 17 08:29:50 2011
New Revision: 170241

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=170241
Log:
    PR target/43653
    * config/i386/i386.c (ix86_secondary_reload): Handle SSE
    input reload with PLUS RTX.

testsuite/ChangeLog:

    PR target/43653
    * gcc.target/i386/pr43653.c: New test.


Added:
    trunk/gcc/testsuite/gcc.target/i386/pr43653.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/i386/i386.c
    trunk/gcc/testsuite/ChangeLog

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|4.3.6                       |4.6.0

--- Comment #15 from Jeffrey A. Law <law at redhat dot com> 2011-02-17 19:46:39 UTC ---
Fixed by Uros's patch.

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #16 from Uros Bizjak <ubizjak at gmail dot com> 2011-02-17 21:05:11 UTC ---
The assembly from -O1 -ftree-vectorize -msse3 shows another opportunity for
enhancement PR19398 (secondary reloads don't consider "m" alternatives):

.LFB0:
    .cfi_startproc
    subq    $416, %rsp
    .cfi_def_cfa_offset 424
    movq    .LC1(%rip), %rax
    leaq    (%rsp,%rax), %rax
    movq    %rax, -112(%rsp)
(*)    movq    -112(%rsp), %xmm1
(*)    punpcklqdq    %xmm1, %xmm1
    movdqa    %xmm1, %xmm0
    leaq    -104(%rsp), %rax
    leaq    408(%rsp), %rdx
.L2:


Looking at the definition of

(define_insn "*vec_dupv2di_sse3"
  [(set (match_operand:V2DI 0 "register_operand"     "=x,x")
    (vec_duplicate:V2DI
      (match_operand:DI 1 "nonimmediate_operand" " 0,m")))]
  "TARGET_SSE3"
  "@
   punpcklqdq\t%0, %0
   movddup\t{%1, %0|%0, %1}"
  [(set_attr "type" "sselog1")
   (set_attr "mode" "TI,DF")])

the two insns marked with (*) can be substituted with the second alternative:

    movddup    -112(%rsp), %xmm1.

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #17 from uros at gcc dot gnu.org 2011-02-17 21:22:05 UTC ---
Author: uros
Date: Thu Feb 17 21:22:02 2011
New Revision: 170256

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=170256
Log:
    PR target/43653
    * config/i386/i386.c (ix86_secondary_reload): Handle SSE
    input reload with PLUS RTX.

testsuite/ChangeLog:

    PR target/43653
    * gcc.target/i386/pr43653.c: New test.


Added:
    branches/gcc-4_5-branch/gcc/testsuite/gcc.target/i386/pr43653.c
      - copied unchanged from r170241,
trunk/gcc/testsuite/gcc.target/i386/pr43653.c
Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/config/i386/i386.c
    branches/gcc-4_5-branch/gcc/testsuite/ChangeLog

[Bug target/43653] [4.3/4.4/4.5/4.6 Regression] ICE at reload1.c:1188 with -O1 -ftree-vectorize and empty struct

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43653

--- Comment #18 from uros at gcc dot gnu.org 2011-02-17 22:10:40 UTC ---
Author: uros
Date: Thu Feb 17 22:10:37 2011
New Revision: 170258

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=170258
Log:
    PR target/43653
    * config/i386/i386.c (ix86_secondary_reload): Handle SSE
    input reload with PLUS RTX.

testsuite/ChangeLog:

    PR target/43653
    * gcc.target/i386/pr43653.c: New test.


Added:
    branches/gcc-4_4-branch/gcc/testsuite/gcc.target/i386/pr43653.c
      - copied unchanged from r170241,
trunk/gcc/testsuite/gcc.target/i386/pr43653.c
Modified:
    branches/gcc-4_4-branch/gcc/ChangeLog
    branches/gcc-4_4-branch/gcc/config/i386/i386.c
    branches/gcc-4_4-branch/gcc/testsuite/ChangeLog