[Bug tree-optimization/24997] New: [4.1/4.2 regression] ICE with -ftree-vectorize

[Bug tree-optimization/24997] New: [4.1/4.2 regression] ICE with -ftree-vectorize

[phython at gcc dot gnu dot org]

wmadec.c ICE's with:
/home/jim/cvs/mplayer/ffmpeg/libavcodec/wmadec.c:1193: error: unrecognizable
insn:
(insn 4809 3267 2762 272 /home/jim/cvs/mplayer/ffmpeg/libavcodec/wmadec.c:1117
(set (reg:SI 9 9)
        (and:SI (plus:SI (reg:SI 27 27)
                (reg:SI 11 11 [orig:284 base_off.641 ] [284]))
            (const_int -16 [0xfffffffffffffff0]))) -1 (nil)
    (nil))
/home/jim/cvs/mplayer/ffmpeg/libavcodec/wmadec.c:1193: internal compiler error:
in extract_insn, at recog.c:2084

the command line is:
 /usr/lib/gcc-snapshot/bin/gcc -O3 -g -Wall -Wno-switch  -maltivec
-mabi=altivec  -DHAVE_AV_CONFIG_H -I.. -I/home/jim/cvs/mplayer/ffmpeg/libavutil
-D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_GNU_SOURCE  -c -o wmadec.o
/home/jim/cvs/mplayer/ffmpeg/libavcodec/wmadec.c
The gcc-snapshot is from the debian package version 20051122-1, the ice also
occurs in earlier versions of the package.


-- 
           Summary: [4.1/4.2 regression] ICE with -ftree-vectorize
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: phython at gcc dot gnu dot org
 GCC build triplet: powerpc-linux
  GCC host triplet: powerpc-linux
GCC target triplet: powerpc-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug tree-optimization/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #1 from amodra at bigpond dot net dot au  2005-11-23 12:20 -------
Please attach preprocessed source.


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amodra at bigpond dot net
                   |                            |dot au


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2005-11-23 13:15 -------
Hmm, r9, wtf.  Any ways, we need the .i file :).


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |WAITING
          Component|tree-optimization           |middle-end
           Keywords|                            |ice-on-valid-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[phython at gcc dot gnu dot org]

------- Comment #3 from phython at gcc dot gnu dot org  2005-11-23 16:05 -------
Created an attachment (id=10327)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10327&action=view)
preprocessed source

Here is the preprocessed source.  Sorry, I haven't had a change to run delta on
it yet, nor can I reproduce this ICE without using -O3 -ftree-vectorize.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[phython at gcc dot gnu dot org]

-- 

phython at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-11-23 16:06:14
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[pinskia at gcc dot gnu dot org]

------- Comment #4 from pinskia at gcc dot gnu dot org  2005-11-23 16:14 -------
Reducing.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[pinskia at gcc dot gnu dot org]

------- Comment #5 from pinskia at gcc dot gnu dot org  2005-11-23 18:54 -------
I will finish reducing this after class (I stopped and started reducing a
different testcase anyways).


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2005-11-23 21:38 -------
Actually this bug is too funny to reduce, reload is too funny for the life of
me.

Compiling with -da with a semi reduced testcases makes it pass.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug tree-optimization/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #7 from amodra at bigpond dot net dot au  2005-11-24 06:51 -------
I'm not surprised that reducing this testcase isn't easy.  Register pressure is
needed to trigger the bug.  The problem is that in
(insn:HI 2761 3601 2769 279 wmadec.c:1117 (set (reg:V4SF 280 [ vect_var_.655 ])
        (mem:V4SF (and:SI (plus:SI (reg/f:SI 1824)
                    (reg:SI 284 [ base_off.641 ]))
                (const_int -16 [0xfffffffffffffff0])) [4 S16 A128])) 628
{altivec_lvx_v4sf} (insn_list:REG_DEP_TRUE 2757 (nil))
    (expr_list:REG_DEAD (reg:SI 284 [ base_off.641 ])
        (nil)))

reg 1824 doesn't get a hard register, so is replaced with a stack slot

(insn:HI 2761 3601 2769 279 wmadec.c:1117 (set (reg:V4SF 89 12 [orig:280
vect_var_.655 ] [280])
        (mem:V4SF (and:SI (plus:SI (plus:SI (reg/f:SI 1 1)
                        (reg:SI 11 11 [orig:284 base_off.641 ] [284]))
                    (const_int 16512 [0x4080]))
                (const_int -16 [0xfffffffffffffff0])) [4 S16 A128])) 628
{altivec_lvx_v4sf} (insn_list:REG_DEP_TRUE 2757 (nil))
    (expr_list:REG_DEAD (reg:SI 11 11 [orig:284 base_off.641 ] [284])
        (nil)))

reload just tries to put the mem address, including the AND, into a reg.  It
needs to be taught that the AND should stay with the insn.  The real address is
the part inside the AND.


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |tree-optimization
   Target Milestone|4.1.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug tree-optimization/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #8 from amodra at bigpond dot net dot au  2005-11-24 07:04 -------
Oops, no, that's not a stack slot.  That's a stack slot gone wrong.  Where's
the MEM for the slot?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug tree-optimization/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #9 from amodra at bigpond dot net dot au  2005-11-24 08:59 -------
Indeed it's not a stack slot.  Instead, reg 1824 is an invariant, equal to
sfp+16512.  Code in reload1.c:eliminate_regs_1 substitutes r1 for sfp, then
reorganises the order of the additions, so we get (r1+r11)+const not
(r1+const)+r11.  This of course isn't a valid address on powerpc, where indexed
addresses can't take an offset.

In case it isn't obvious, I'm a little lost.  Hopefully my analysis will let
someone more familiar with reload sort this out.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug tree-optimization/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #10 from amodra at bigpond dot net dot au  2005-11-24 10:21 -------
I think I've sorted this out after all..  Testing a fix.


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |amodra at bigpond dot net
                   |dot org                     |dot au
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2005-11-23 16:06:14         |2005-11-24 10:21:53
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug middle-end/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|tree-optimization           |middle-end
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug target/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2005-
                   |                            |11/msg01916.html
          Component|middle-end                  |target
   Target Milestone|4.1.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug target/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[amodra at gcc dot gnu dot org]

------- Comment #11 from amodra at gcc dot gnu dot org  2005-11-28 03:52 -------
Subject: Bug 24997

Author: amodra
Date: Mon Nov 28 03:52:01 2005
New Revision: 107591

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=107591
Log:
        PR target/24997
        * config/rs6000/rs6000.c (legitimate_indexed_address_p): Allow pattern
        generated by reload.
        * config/rs6000/predicates.md (indexed_or_indirect_operand): Use
        indexed_or_indirect_address.
        (indexed_or_indirect_address): Don't test for base reg.  Call
        address_operand last.  Make it a special predicate.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/predicates.md
    trunk/gcc/config/rs6000/rs6000.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug target/24997] [4.1/4.2 regression] ICE with -ftree-vectorize

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug target/24997] [4.1 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #12 from amodra at bigpond dot net dot au  2005-12-07 05:50 -------
pr25212 has a more complete fix that I propose applying to the 4.1 branch.


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  BugsThisDependsOn|                            |25212


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997

[Bug target/24997] [4.1 regression] ICE with -ftree-vectorize

[amodra at bigpond dot net dot au]

------- Comment #13 from amodra at bigpond dot net dot au  2005-12-12 10:05 -------
Fixed 4.1 by pr25212 commit.


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to fail|4.1.0                       |
      Known to work|4.0.3 4.2.0                 |4.0.3 4.1.0 4.2.0
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24997