[Bug c/50928] New: ICE building RTEMS

[Bug c/50928] New: ICE building RTEMS

[joel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

             Bug #: 50928
           Summary: ICE building RTEMS
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: joel@gcc.gnu.org


gcc (GCC) 4.7.0 20111029 (experimental)
Newlib cvs head as of same date

m32c-rtems4.11-gcc --pipe  -mcpu=m32cm --pipe -DHAVE_CONFIG_H   -I..
-I../../../lib/include -D__RTEMS_INSIDE__   -g -O2 -Wall
-Wimplicit-function-declaration -Wstrict-prototypes -Wnested-externs 
-mcpu=m32cm -MT src/libscore_a-heap.o -MD -MP -MF src/.deps/libscore_a-heap.Tpo
-c -o src/libscore_a-heap.o `test -f 'src/heap.c' || echo
'/users/joel/test-gcc/rtems/cpukit/score/'`src/heap.c
In file included from ../../../lib/include/rtems/score/basedefs.h:25:0,
                 from ../../../lib/include/rtems/score/types.h:22,
                 from ../../../lib/include/rtems/score/cpu.h:40,
                 from ../../../lib/include/rtems/score/percpu.h:22,
                 from ../../../lib/include/rtems/system.h:23,
                 from /users/joel/test-gcc/rtems/cpukit/score/src/heap.c:28:
../../../lib/include/rtems/score/cpuopts.h:56:0: warning:
"__RTEMS_SIZEOF_VOID_P__" redefined [enabled by default]
In file included from /users/joel/test-gcc/rtems/cpukit/score/src/heap.c:23:0:
../config.h:465:0: note: this is the location of the previous definition
/users/joel/test-gcc/rtems/cpukit/score/src/heap.c: In function
'_Heap_Initialize':
/users/joel/test-gcc/rtems/cpukit/score/src/heap.c:300:1: error: unable to find
a register to spill in class 'A_REGS'
/users/joel/test-gcc/rtems/cpukit/score/src/heap.c:300:1: error: this is the
insn:
(insn 119 118 120 13 (set (mem/s:SI (subreg:PSI (reg/f:SI 30 [ first_block.3 ])
0) [3 first_block.3_22->prev_size+0 S4 A8])
        (reg/v:SI 28 [ heap_area_end ]))
/users/joel/test-gcc/rtems/cpukit/score/src/heap.c:260 98 {movsi_24}
     (nil))
/users/joel/test-gcc/rtems/cpukit/score/src/heap.c:300:1: internal compiler
error: in spill_failure, at reload1.c:2118
Please submit a full bug report,

[Bug c/50928] ICE building RTEMS

[joel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #1 from Joel Sherrill <joel at gcc dot gnu.org> 2011-10-30 21:09:56 UTC ---
Created attachment 25670
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25670
Preprocessed source code

[Bug c/50928] m32c ICE building RTEMS

[joel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #2 from Joel Sherrill <joel at gcc dot gnu.org> 2012-11-03 17:22:30 UTC ---
xgcc (GCC) 4.8.0 20121103 (experimental) [trunk revision 193124]

Now fails just building libgcc with what appears to be the same error.

Short command line for preprocessed source I will attach

 /home/joel/v850/tools/b-gcc-svn/./gcc/xgcc
-B/home/joel/v850/tools/b-gcc-svn/./gcc/ -c m32c_bug.c -O2 -mcpu=m32cm

At -O0, it compiles

Full command line:

/home/joel/v850/tools/b-gcc-svn/./gcc/xgcc
-B/home/joel/v850/tools/b-gcc-svn/./gcc/ -nostdinc
-B/home/joel/v850/tools/b-gcc-svn/m32c-rtems4.11/newlib/ -isystem
/home/joel/v850/tools/b-gcc-svn/m32c-rtems4.11/newlib/targ-include -isystem
/home/joel/v850/tools/gcc-svn/newlib/libc/include
-B/home/joel/v850/install/m32c-rtems4.11/bin/
-B/home/joel/v850/install/m32c-rtems4.11/lib/ -isystem
/home/joel/v850/install/m32c-rtems4.11/include -isystem
/home/joel/v850/install/m32c-rtems4.11/sys-include    -g -O2 -mcpu=m32cm -O2
-I../../../../gcc-svn/libgcc/../newlib/libc/sys/rtems/include -g -O2 -DIN_GCC
-DCROSS_DIRECTORY_STRUCTURE  -W -Wall -Wwrite-strings -Wcast-qual
-Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition  -isystem
./include   -g -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector
-Dinhibit_libc  -I. -I. -I../../.././gcc -I../../../../gcc-svn/libgcc
-I../../../../gcc-svn/libgcc/. -I../../../../gcc-svn/libgcc/../gcc
-I../../../../gcc-svn/libgcc/../include  -DHAVE_CC_TLS -DUSE_EMUTLS -o
_ffssi2.o -MT _ffssi2.o -MD -MP -MF _ffssi2.dep -DL_ffssi2 -c
../../../../gcc-svn/libgcc/libgcc2.c -fvisibility=hidden -DHIDE_EXPORTS
../../../../gcc-svn/libgcc/libgcc2.c: In function '__ffssi2':
../../../../gcc-svn/libgcc/libgcc2.c:524:1: error: unable to find a register to
spill in class 'A_REGS'
 }
 ^
../../../../gcc-svn/libgcc/libgcc2.c:524:1: error: this is the insn:
(insn 58 56 59 10 (set (reg:HI 0 r0 [orig:26 D.2772 ] [26])
        (zero_extend:HI (mem/u/j:QI (plus:PSI (subreg:PSI (reg:SI 68 [ D.2773
]) 0)
                    (symbol_ref:PSI ("__clz_tab") [flags 0x40] <var_decl
0xb77dc5c0 __clz_tab>)) [0 __clz_tab S1 A8])))
../../../../gcc-svn/libgcc/libgcc2.c:522 115 {zero_extendqihi2}
     (expr_list:REG_DEAD (reg:SI 68 [ D.2773 ])
        (nil)))
../../../../gcc-svn/libgcc/libgcc2.c:524:1: internal compiler error: in
spill_failure, at reload1.c:2124
0x849f565 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
    ../../gcc-svn/gcc/rtl-error.c:110
0x849f2c3 spill_failure
    ../../gcc-svn/gcc/reload1.c:2124
0x849f2c3 find_reload_regs
    ../../gcc-svn/gcc/reload1.c:2050
0x849f2c3 select_reload_regs
    ../../gcc-svn/gcc/reload1.c:2070
0x849f2c3 reload(rtx_def*, int)
    ../../gcc-svn/gcc/reload1.c:991
0x83cb8ea do_reload
    ../../gcc-svn/gcc/ira.c:4636
0x83cb8ea rest_of_handle_reload
    ../../gcc-svn/gcc/ira.c:4737

[Bug c/50928] m32c ICE building RTEMS

[joel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #3 from Joel Sherrill <joel at gcc dot gnu.org> 2012-11-03 17:23:38 UTC ---
Created attachment 28604
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28604
libgcc preprocessed file which trips bug

Description on how to reproduce given messages.

[Bug c/50928] m32c ICE building RTEMS

[corsepiu at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

Ralf Corsepius <corsepiu at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |corsepiu at gcc dot gnu.org
      Known to work|                            |4.7.2
      Known to fail|                            |4.8.0

--- Comment #4 from Ralf Corsepius <corsepiu at gcc dot gnu.org> 2013-03-26 08:30:02 UTC ---
This bug had vanished with gcc-4.7.2, but has returned with gcc-4.8.0:

/builddir/build/BUILD/rtems-4.11-m32c-rtems4.11-gcc-4.8.0/build/./gcc/xgcc
-B/builddir/build/BUILD/rtems-4.11-m32c-rtems4.11-gcc-4.8.0/build/./gcc/
-nostdinc
-B/builddir/build/BUILD/rtems-4.11-m32c-rtems4.11-gcc-4.8.0/build/m32c-rtems4.11/newlib/
-isystem
/builddir/build/BUILD/rtems-4.11-m32c-rtems4.11-gcc-4.8.0/build/m32c-rtems4.11/newlib/targ-include
-isystem
/builddir/build/BUILD/rtems-4.11-m32c-rtems4.11-gcc-4.8.0/gcc-4.8.0/newlib/libc/include
-B/opt/rtems-4.11/m32c-rtems4.11/bin/ -B/opt/rtems-4.11/m32c-rtems4.11/lib/
-isystem /opt/rtems-4.11/m32c-rtems4.11/include -isystem
/opt/rtems-4.11/m32c-rtems4.11/sys-include    -g -O2 -mcpu=m32cm -O2
-I../../../../gcc-4.8.0/libgcc/../newlib/libc/sys/rtems/include -g -O2 -DIN_GCC
-DCROSS_DIRECTORY_STRUCTURE  -W -Wall -Wno-narrowing -Wwrite-strings
-Wcast-qual -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition 
-isystem ./include   -g -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector
-Dinhibit_libc  -I. -I. -I../../.././gcc -I../../../../gcc-4.8.0/libgcc
-I../../../../gcc-4.8.0/libgcc/. -I../../../../gcc-4.8.0/libgcc/../gcc
-I../../../../gcc-4.8.0/libgcc/../include  -DHAVE_CC_TLS -DUSE_EMUTLS -o
_ffsdi2.o -MT _ffsdi2.o -MD -MP -MF _ffsdi2.dep -DL_ffsdi2 -c
../../../../gcc-4.8.0/libgcc/libgcc2.c -fvisibility=hidden -DHIDE_EXPORTS
../../../../gcc-4.8.0/libgcc/libgcc2.c: In function '__ffssi2':
../../../../gcc-4.8.0/libgcc/libgcc2.c:522:1: error: unable to find a register
to spill in class 'A_REGS'
 }
 ^
../../../../gcc-4.8.0/libgcc/libgcc2.c:522:1: error: this is the insn:
(insn 58 56 59 10 (set (reg:HI 0 r0 [orig:26 D.2817 ] [26])
        (zero_extend:HI (mem/u/j:QI (plus:PSI (subreg:PSI (reg:SI 44 [ D.2818
]) 0)
                    (symbol_ref:PSI ("__clz_tab") [flags 0x40] <var_decl
0x7f55b9a14ab0 __clz_tab>)) [0 __clz_tab S1 A8])))
../../../../gcc-4.8.0/libgcc/libgcc2.c:520 115 {zero_extendqihi2}
     (expr_list:REG_DEAD (reg:SI 44 [ D.2818 ])
        (nil)))
../../../../gcc-4.8.0/libgcc/libgcc2.c:522: confused by earlier errors, bailing
out

[Bug target/50928] m32c ICE building RTEMS

[joel at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #7 from Joel Sherrill <joel at gcc dot gnu.org> ---
DJ.. do you think the patch from Bernd can be applied to the 4.9 branch? and
maybe the head?

[Bug target/50928] m32c ICE building RTEMS

[dj at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #8 from DJ Delorie <dj at redhat dot com> ---
There are a few regressions in the testsuite (pr26255.c with -mcpu=m32c for
example) and libstdc++ still doesn't build, but ieee/920810-1 now passes and
newlib builds.  I suppose that's "better".

[Bug target/50928] m32c ICE building RTEMS

[joel at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #9 from Joel Sherrill <joel at gcc dot gnu.org> ---
I updated binutils, newlib and gcc. The build fails very early for me:

/users/joel/test-gcc/b-m32c-rtems4.11-gcc/./gcc/xgcc
-B/users/joel/test-gcc/b-m32c-rtems4.11-gcc/./gcc/ -nostdinc
-B/users/joel/test-gcc/b-m32c-rtems4.11-gcc/m32c-rtems4.11/newlib/ -isystem
/users/joel/test-gcc/b-m32c-rtems4.11-gcc/m32c-rtems4.11/newlib/targ-include
-isystem /users/joel/test-gcc/gcc/newlib/libc/include
-B/users/joel/test-gcc/install-head/m32c-rtems4.11/bin/
-B/users/joel/test-gcc/install-head/m32c-rtems4.11/lib/ -isystem
/users/joel/test-gcc/install-head/m32c-rtems4.11/include -isystem
/users/joel/test-gcc/install-head/m32c-rtems4.11/sys-include    -g -O2
-mcpu=m32cm -O2 -I../../../../gcc/libgcc/../newlib/libc/sys/rtems/include -g
-O2 -DIN_GCC  -DCROSS_DIRECTORY_STRUCTURE  -W -Wall -Wno-narrowing
-Wwrite-strings -Wcast-qual -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition  -isystem ./include   -g -DIN_LIBGCC2 -fbuilding-libgcc
-fno-stack-protector -Dinhibit_libc  -I. -I. -I../../.././gcc
-I../../../../gcc/libgcc -I../../../../gcc/libgcc/.
-I../../../../gcc/libgcc/../gcc -I../../../../gcc/libgcc/../include 
-DHAVE_CC_TLS -DUSE_EMUTLS -o _muldi3.o -MT _muldi3.o -MD -MP -MF _muldi3.dep
-DL_muldi3 -c ../../../../gcc/libgcc/libgcc2.c -fvisibility=hidden
-DHIDE_EXPORTS
/tmp/ccx0vJbF.s: Assembler messages:
/tmp/ccx0vJbF.s: Error: relocation out of range
make[4]: *** [_muldi3.o] Error 1

[Bug target/50928] m32c ICE building RTEMS

[bernd.edlinger at hotmail dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #10 from Bernd Edlinger <bernd.edlinger at hotmail dot de> ---
Hmm...

I use this configuration:

binutils-build:
../binutils-2.25/configure --prefix=/home/ed/gnu/m32c-rtems --target=m32c-rtems
--disable-werror

gcc-build:
../gcc-trunk/configure --prefix=/home/ed/gnu/m32c-rtems --target=m32c-rtems
--enable-languages=c --disable-libssp

Revision: 219900 with above patch.

=> build succeeds.

But pr26255.c does not work. I have no idea how to fix that. Something
wrong with addpsi3 constraints?

I think there is another relatively simple issue with exception handling.

What are your configure options?

[Bug target/50928] m32c ICE building RTEMS

[dj at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #11 from DJ Delorie <dj at redhat dot com> ---
I see the "relocation out of range" error too.
I'm configuring for m32c-elf

[Bug target/50928] m32c ICE building RTEMS

[dj at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #12 from DJ Delorie <dj at redhat dot com> ---
The reloc bug is caused when gcc puts a JMP.A at the very end of .text and also
adds debug info; this puts a 3-byte reloc right at the end of a section
(m32c-as pads the last section if it's a code section, the debug info makes
.text not last) but BFD has no option for a non-power-of-two reloc in the HOWTO
table, so it thinks it's 4 bytes and thus extends past the end of the section.

I suspect the way to fix this is to handle that one reloc specially (the rest
are handled by generic reloc handlers), unless someone has an alternate idea?

[Bug target/50928] m32c ICE building RTEMS

[bernd.edlinger at hotmail dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #13 from Bernd Edlinger <bernd.edlinger at hotmail dot de> ---
ok, now I see.
the binutils-2.25.tar.gz works,
but if we build directly from the git it does not.
is this a new check that makes this problem?


regarding my patch on pr26255.c, this happens:

m32c-elf-gcc -O1 -mcpu=m32c pr26255.c
pr26255.c: In function 'foo':
pr26255.c:31:1: error: insn does not satisfy its constraints:
 }
 ^
(insn 62 61 63 2 (set (reg:PSI 4 a0)
        (plus:PSI (reg:PSI 4 a0)
            (reg/v/f:PSI 28 [ w ]))) pr26255.c:29 5 {addpsi3}
     (expr_list:REG_EQUIV (plus:PSI (reg/v/f:PSI 28 [ w ])
            (const_int 128 [0x80]))
        (nil)))
pr26255.c:31:1: internal compiler error: in extract_constrain_insn, at
recog.c:2246
0x9e9478 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
    ../../gcc-trunk/gcc/rtl-error.c:110
0x9e949f _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
    ../../gcc-trunk/gcc/rtl-error.c:121
0x9c0505 extract_constrain_insn(rtx_insn*)
    ../../gcc-trunk/gcc/recog.c:2246
0x9a1a5d reload_cse_simplify_operands
    ../../gcc-trunk/gcc/postreload.c:430
0x9a4485 reload_cse_simplify
    ../../gcc-trunk/gcc/postreload.c:207
0x9a4485 reload_cse_regs_1
    ../../gcc-trunk/gcc/postreload.c:246
0x9a459b reload_cse_regs
    ../../gcc-trunk/gcc/postreload.c:94
0x9a459b execute
    ../../gcc-trunk/gcc/postreload.c:2367


I had similar traps and could avoid them by adding an explicit
mode to some patterns.  But here I have no idea at the moment.

DJ do you see anything obvious in the addpsi3 pattern?

[Bug target/50928] m32c ICE building RTEMS

[edlinger at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #14 from Bernd Edlinger <edlinger at gcc dot gnu.org> ---
Author: edlinger
Date: Fri Jan 23 16:32:34 2015
New Revision: 220048

URL: https://gcc.gnu.org/viewcvs?rev=220048&root=gcc&view=rev
Log:
2015-01-23  Bernd Edlinger  <bernd.edlinger@hotmail.de>

        PR target/50928
        * config/m32c/m32c.c (encode_pattern_1): Removed gcc_unreachable here.
        (DEBUG_RELOAD): Removed define.
        (m32c_limit_reload_class): Enable traces with if DEBUG0.
        (m32c_function_arg): Added a type cast.
        (m32c_legitimize_reload_address): Push A_REGS reload with PSImode.
        * config/m32c/addsub.md (addsi3_1): Specify the mode of all arguments.
        * config/m32c/bitops.md (andqi3_16): Likewise.
        * config/m32c/mov.md (m32c_immd_dbl_mov): Likewise.
        (push_a01_l): Likewise.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/m32c/addsub.md
    trunk/gcc/config/m32c/bitops.md
    trunk/gcc/config/m32c/m32c.c
    trunk/gcc/config/m32c/mov.md

[Bug target/50928] m32c ICE building RTEMS

[dj at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50928

--- Comment #15 from DJ Delorie <dj at redhat dot com> ---
The binutils team has been working a lot on patching vulnerabilities in the
binutils tools.  The m32c, however, has a 3-byte reloc that might occur at the
end of a section, and was implemented as three bytes of a four-byte "word",
which would then be outside the bounds of the section.  Recent patches check
for such bounds crossings, hence the breakage.  I've checked in a patch to
binutils head to manually process the R_M32C_24 relocations so that the range
checking is more appropriate to the three-byte relocation.