[Bug target/54093] New: ICE in in extract_insn, at recog.c:2129

[Bug target/54093] New: ICE in in extract_insn, at recog.c:2129

[rmansfield at qnx dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

             Bug #: 54093
           Summary: ICE in in extract_insn, at recog.c:2129
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: rmansfield@qnx.com
                CC: amodra@gmail.com
              Host: x86_64-unknown-linux-gnu
            Target: powerpc-e500v2-linux-gnuspe
             Build: x86_64-unknown-linux-gnu


Created attachment 27871
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27871
preprocessed src

Appears to be introduced by rev189801

$ ./xgcc -v
Using built-in specs.
COLLECT_GCC=./xgcc
Target: powerpc-e500v2-linux-gnuspe
Configured with: ../configure --target=powerpc-e500v2-linux-gnuspe
--prefix=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe
--with-sysroot=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe/powerpc-e500v2-linux-gnuspe/sys-root
--enable-languages=c,c++ --disable-multilib --with-cpu=8548 --with-tune=8548
--with-gmp=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe
--with-mpfr=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe
--enable-__cxa_atexit
--with-local-prefix=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe/powerpc-e500v2-linux-gnuspe/sys-root
--disable-nls --enable-threads=posix --enable-symvers=gnu --enable-c99
--enable-long-long --enable-target-optspace --enable-e500_double
--with-long-double-128 : (reconfigured) ../configure
--target=powerpc-e500v2-linux-gnuspe
--prefix=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe
--with-sysroot=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe/powerpc-e500v2-linux-gnuspe/sys-root
--enable-languages=c,c++ --disable-multilib --with-cpu=8548 --with-tune=8548
--with-gmp=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe
--with-mpfr=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe
--enable-__cxa_atexit
--with-local-prefix=/home/ryan/x-tools/powerpc-e500v2-linux-gnuspe/powerpc-e500v2-linux-gnuspe/sys-root
--disable-nls --enable-threads=posix --enable-symvers=gnu --enable-c99
--enable-long-long --enable-target-optspace --enable-e500_double
--with-long-double-128
Thread model: posix
gcc version 4.8.0 20120724 (experimental) [trunk revision 189843] (GCC) 
$ ./xgcc -B. ~/ice.i -c
/home/ryan/ice.i: In function 'cbq_stat_loop':
/home/ryan/ice.i:67:8: warning: incompatible implicit declaration of built-in
function 'printf' [enabled by default]
        printf (",\tMeasured: %s [bps]\n",
        ^
/home/ryan/ice.i:74:1: error: unrecognizable insn:
 }
 ^
(insn 79 78 80 5 (set (reg:SI 11 11)
        (plus:SI (reg/f:SI 31 31)
            (const_int 32824 [0x8038]))) /home/ryan/ice.i:67 -1
     (nil))
/home/ryan/ice.i:74:1: internal compiler error: in extract_insn, at
recog.c:2129
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


I've also seen another couple ICEs with the spe configuration which look
related. I'll reduce those testcases shortly, but the errors are:

 error: unrecognizable insn:
 "memory reqd: %20d bytes\n" , ( int )( 4 * _53 + _72 + 100 )); return 0 ; }
 ^
(insn 938 937 939 34 (set (reg:SI 9 9)
        (plus:SI (reg/f:SI 1 1)
            (const_int 70800 [0x11490])))
/home/ryan/latest/svn/lib/sqlite3/cerod_build/cerod_build.c:92 -1
     (nil))
/home/ryan/latest/svn/lib/sqlite3/cerod_build/cerod_build.c:98:1: internal
compiler error: in extract_insn, at recog.c:2129


and 

: error: unrecognizable insn:
 }
 ^
(insn 4114 4113 4115 137 (set (reg:SI 8 8)
        (plus:SI (reg/f:SI 1 1)
            (const_int 82016 [0x14060])))
/home/ryan/latest/svn/lib/io-pkt/crypto/external/bsd/openssh/dist/sftp-server.c:283
-1
     (nil))
/home/ryan/latest/svn/lib/io-pkt/crypto/external/bsd/openssh/dist/sftp-server.c:1521:1:
internal compiler error: in extract_insn, at recog.c:2129

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[rmansfield at qnx dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

--- Comment #1 from Ryan Mansfield <rmansfield at qnx dot com> 2012-07-25 19:27:00 UTC ---
Created attachment 27873
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27873
2nd preprocessed src example

Happens at -O2

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[rmansfield at qnx dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

--- Comment #2 from Ryan Mansfield <rmansfield at qnx dot com> 2012-07-25 19:28:40 UTC ---
Created attachment 27874
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27874
3rd preprocessed src example

Happens at -O1

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[amodra at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

Alan Modra <amodra at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2012-07-26
         AssignedTo|unassigned at gcc dot       |amodra at gmail dot com
                   |gnu.org                     |
   Target Milestone|---                         |4.8.0
     Ever Confirmed|0                           |1

--- Comment #3 from Alan Modra <amodra at gmail dot com> 2012-07-26 03:58:05 UTC ---
This comment of mine has come back to bite me, I think.
> - I haven't actually seen the 32-bit gpr secondary reload patterns
> trigger (it's hard to make a testcase), so that code is largely
> untested.  Fortunately the code is very similar to the 64-bit gpr
> secondary reload code.

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[amodra at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

--- Comment #4 from Alan Modra <amodra at gmail dot com> 2012-07-26 13:16:23 UTC ---
Created attachment 27877
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27877
proposed fix

Please try out this patch.

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[rmansfield at qnx dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

--- Comment #5 from Ryan Mansfield <rmansfield at qnx dot com> 2012-07-27 02:56:45 UTC ---
(In reply to comment #4)
> Created attachment 27877 [details]
> proposed fix
> 
> Please try out this patch.

The patch resolves the ICEs, and I didn't see any others problems building a
fairly large code. I didn't have a chance to run the gcc regression testsuite
though.

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[amodra at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

--- Comment #6 from Alan Modra <amodra at gcc dot gnu.org> 2012-07-28 00:04:15 UTC ---
Author: amodra
Date: Sat Jul 28 00:04:10 2012
New Revision: 189921

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=189921
Log:
    PR target/54093
    * config/rs6000/rs6000.c (rs6000_secondary_reload): Limit 32-bit
    multi-gpr reload to cases where predicate passes.  Do the same for
    64-bit multi-gpr reload.


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

[Bug target/54093] ICE in in extract_insn, at recog.c:2129

[amodra at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54093

Alan Modra <amodra at gmail dot com> changed:

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

--- Comment #7 from Alan Modra <amodra at gmail dot com> 2012-07-28 00:06:42 UTC ---
Fixed