ICE in reload_cse_simplify_operands, at postreload.c:402 with neon optimized code

[Luke Kim]

While building neon optimized code

$ cat insn-unsat-constr.c
#include <arm_neon.h>

void test(signed int *output)
{
 int16_t * out = (int16_t *) output;
 {
  int16x4_t a = {0, };
  int16x4_t b = {1, };
  int16x4x4_t z;
  z.val[0] = vadd_s16(a, b);
  z.val[1] = vadd_s16(a, b);
  z.val[2] = vsub_s16(a, b);
  z.val[3] = vadd_s16(a, b);

  vst4_lane_s16(&out[0]+0*4, z, 0);
  vst4_lane_s16(&out[8]+0*4, z, 1);

 }
}

$ arm-linux-gnueabi-gcc -mfloat-abi=softfp -mfpu=neon -O1 -Wall -o insn-unsat-constr.o -c insn-unsat-constr.c
insn-unsat-constr.c: In function ‘test’:
insn-unsat-constr.c:19:1: error: insn does not satisfy its constraints:
(insn 40 38 26 2 /usr/lib/gcc/arm-linux-gnueabi/4.5.2/include/arm_neon.h:10277 (set (reg:OI 95 d16 [orig:152 __b ] [152])
        (mem/s/c:OI (pre_dec:SI (reg/f:SI 3 r3 [151])) [0 __b+0 S32 A64])) 786 {*neon_movoi} (expr_list:REG_INC (reg/f:SI 3 r3 [151])
        (nil)))
insn-unsat-constr.c:19:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:402
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-4.5/README.Bugs> for instructions.

$ arm-linux-gnueabi-gcc -v
Using built-in specs.
COLLECT_GCC=arm-linux-gnueabi-gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/arm-linux-gnueabi/4.5.2/lto-wrapper
Target: arm-linux-gnueabi
Configured with: ../src/configure -v --with-pkgversion='Ubuntu/Linaro 4.5.2-5ubuntu2~ppa1' --with-bugurl=file:///usr/share/doc/gcc-4.5/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr --program-suffix=-4.5 --enable-shared --enable-multiarch --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/arm-linux-gnueabi/include/c++/4.5.2 --libdir=/usr/lib --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-plugin --enable-gold --enable-ld=default --with-plugin-ld=ld.gold --enable-objc-gc --disable-sjlj-exceptions --with-arch=armv7-a --with-float=softfp --with-fpu=vfpv3-d16 --with-mode=thumb --disable-werror --enable-checking=release --program-prefix=arm-linux-gnueabi- --includedir=/usr/arm-linux-gnueabi/include --build=i486-linux-gnu --host=i486-linux-gnu --target=arm-linux-gnueabi --with-headers=/usr/arm-linux-gnueabi/include --with-libs=/usr/arm-linux-gnueabi/lib
Thread model: posix
gcc version 4.5.2 (Ubuntu/Linaro 4.5.2-5ubuntu2~ppa1)

[Luke Kim (nereusuj)]

Created attachment 23794
preprocessed file

I have following error while building neon optimized code.

$ cat insn-unsat-constr.c
#include <arm_neon.h>

void test(signed int *output)
{
 int16_t * out = (int16_t *) output;
 {
  int16x4_t a = {0, };
  int16x4_t b = {1, };
  int16x4x4_t z;
  z.val[0] = vadd_s16(a, b);
  z.val[1] = vadd_s16(a, b);
  z.val[2] = vsub_s16(a, b);
  z.val[3] = vadd_s16(a, b);

  vst4_lane_s16(&out[0]+0*4, z, 0);
  vst4_lane_s16(&out[8]+0*4, z, 1);

 }
}

$ arm-none-linux-gnueabi-gcc -mfloat-abi=softfp -mfpu=neon -O1 -o insn-unsat-constr.o -c insn-unsat-constr.c
insn-unsat-constr.c: In function 'test':
insn-unsat-constr.c:19:1: error: insn does not satisfy its constraints:
(insn 40 38 26 2 /scratchbox/compilers/arm-linux-gnueabi-gcc4.5.1-2010.09-50/bin/../lib/gcc/arm-none-linux-gnueabi/4.5.1/include/arm_neon.h:10277 (set (reg:OI 95 d16 [orig:152 __b ] [152])
        (mem/s/c:OI (pre_dec:SI (reg/f:SI 3 r3 [151])) [0 __b+0 S32 A64])) 740 {*neon_movoi} (expr_list:REG_INC (reg/f:SI 3 r3 [151])
        (nil)))
insn-unsat-constr.c:19:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:403
Please submit a full bug report,
with preprocessed source if appropriate.

$ arm-none-linux-gnueabi-gcc -v
Using built-in specs.
COLLECT_GCC=/scratchbox/compilers/arm-linux-gnueabi-gcc4.5.1-2010.09-50/bin/arm-none-linux-gnueabi-gcc
COLLECT_LTO_WRAPPER=/scratchbox/compilers/arm-linux-gnueabi-gcc4.5.1-2010.09-50/bin/../libexec/gcc/arm-none-linux-gnueabi/4.5.1/lto-wrapper
Target: arm-none-linux-gnueabi
Configured with: /scratch/nathan/arm-lite/src/gcc-4.5-2010.09/configure --build=i686-pc-linux-gnu --host=i686-pc-linux-gnu --target=arm-none-linux-gnueabi --enable-threads --disable-libmudflap --disable-libssp --disable-libstdcxx-pch --enable-extra-sgxxlite-multilibs --with-arch=armv5te --with-gnu-as --with-gnu-ld --with-specs='%{save-temps: -fverbose-asm} %{funwind-tables|fno-unwind-tables|mabi=*|ffreestanding|nostdlib:;:-funwind-tables} -D__CS_SOURCERYGXX_MAJ__=2010 -D__CS_SOURCERYGXX_MIN__=9 -D__CS_SOURCERYGXX_REV__=50 %{O2:%{!fno-remove-local-statics: -fremove-local-statics}} %{O*:%{O|O0|O1|O2|Os:;:%{!fno-remove-local-statics: -fremove-local-statics}}}' --enable-languages=c,c++ --enable-shared --enable-lto --enable-symvers=gnu --enable-__cxa_atexit --with-pkgversion='Sourcery G++ Lite 2010.09-50' --with-bugurl=https://support.codesourcery.com/GNUToolchain/ --disable-nls --prefix=/opt/codesourcery --with-sysroot=/opt/codesourcery/arm-none-linux-gnueabi/libc --with-build-sysroot=/scratch/nathan/arm-lite/install/arm-none-linux-gnueabi/libc --with-gmp=/scratch/nathan/arm-lite/obj/host-libs-2010.09-50-arm-none-linux-gnueabi-i686-pc-linux-gnu/usr --with-mpfr=/scratch/nathan/arm-lite/obj/host-libs-2010.09-50-arm-none-linux-gnueabi-i686-pc-linux-gnu/usr --with-mpc=/scratch/nathan/arm-lite/obj/host-libs-2010.09-50-arm-none-linux-gnueabi-i686-pc-linux-gnu/usr --with-ppl=/scratch/nathan/arm-lite/obj/host-libs-2010.09-50-arm-none-linux-gnueabi-i686-pc-linux-gnu/usr --with-host-libstdcxx='-static-libgcc -Wl,-Bstatic,-lstdc++,-Bdynamic -lm' --with-cloog=/scratch/nathan/arm-lite/obj/host-libs-2010.09-50-arm-none-linu...

[Mikpe (mikpe)]

That's a heavily modified compiler by CodeSourcery. Please reproduce with a vanilla FSF GCC, or report the problem to CodeSourcery as their compiler clearly directs you to do (see the --with-bugurl= setting).

[Ibolton (ibolton)]

I get the same thing when I use r171282 of FSF 4.5 branch.

arm-none-linux-gnueabi-gcc pr48325.c -mfloat-abi=softfp -mfpu=neon -O1
pr48325.c: In function 'test':
pr48325.c:19:1: error: insn does not satisfy its constraints:
(insn 40 38 26 2 /work/ianbol01/cross-build/gcc45-r171282-thumb/arm-none-linux-gnueabi/tools/lib/gcc/arm-none-linux-gnueabi/4.5.3/include/arm_neon.h:10277 (set (reg:OI 95 d16 [orig:152 __b ] [152])
        (mem/s/c:OI (pre_dec:SI (reg/f:SI 3 r3 [151])) [0 __b+0 S32 A64])) 731 {*neon_movoi} (expr_list:REG_INC (reg/f:SI 3 r3 [151])
        (nil)))
pr48325.c:19:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:396

Here is the command-line just for cc1:

cc1 -quiet pr48325.c -mfloat-abi=softfp -mfpu=neon -marm -mcpu=cortex-a9 -O1

Doesn't work for thumb either.

It also fails on trunk.

There are two other bugs in flight that manifest in reload_cse_simplify_operands: PR48250 (broke on trunk for EABI, works on 4.5 for EABI) and PR42949 (works on EABI for trunk and gcc4.5, broke for OABI).

I do not know if they are duplicates of each other, or if there are two or more separate bugs causing this.

[Michael Hope (michaelh1)]

Confirmed in gcc-linaro-4.5-2011.03-0:

michaelh@ursa3:~/linaro/bugs$ /tools/toolchains/gcc-linaro-4.5-2011.03-0-armv7l-maverick-cbuild71-carina7-cortexa8r1/bin/gcc -mfpu=neon -O1 -c insn-unsat-constr.i
insn-unsat-constr.c: In function 'test':
insn-unsat-constr.c:19:1: error: insn does not satisfy its constraints:
(insn 40 38 26 2 /usr/lib/gcc/arm-linux-gnueabi/4.5.2/include/arm_neon.h:10277 (set (reg:OI 95 d16 [orig:152 __b ] [152])
        (mem/s/c:OI (pre_dec:SI (reg/f:SI 3 r3 [151])) [0 __b+0 S32 A64])) 786 {*neon_movoi} (expr_list:REG_INC (reg/f:SI 3 r3 [151])
        (nil)))
insn-unsat-constr.c:19:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:402

Does not occur in gcc-linaro-4.4-2010.10-0. Does not occur at -O0, but does at -O1, -O2, -O3.

The work around is to supply -fno-auto-inc-dec.

[Michael Hope (michaelh1)]

Also occurs in trunk in gcc 4.5.1 and gcc-4.5.2+svn171194.

Could you also report this in the GCC bugzilla please?

[Luke Kim (nereusuj)]

Upstream bug submitted http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48325

[Chung-Lin Tang (cltang)]

I also saw several other testsuite regressions on upstream trunk with a similar ICE too, will look at this.

[Chung-Lin Tang (cltang)]

Correct comment #6, the regressions I saw on upstream trunk should be different from this; this is another bug.

[Chung-Lin Tang (cltang)]

Upstream patch posted: http://gcc.gnu.org/ml/gcc-patches/2011-03/msg02204.html

[Michael Hope (michaelh1)]

Chung-Lin, could you ping this patch please?

[Ira Rosen (irar)]

I am out of the office until 17/04/2011.

Note: This is an automated response to your message "[Bug 744754] Re: ICE
in reload_cse_simplify_operands, at postreload.c:402 with neon optimized
code" sent on 15/4/2011 4:47:17.

This is the only notification you will receive while this person is away.

[Michael Hope (michaelh1)]

No longer occurs in gcc-linaro-4.5-2011.05 and gcc-linaro-4.6-2011.05 due to the changes Richard made on general NEON accesses.

Richard is looking into pre/post inc/dec in general so we'll mark this as fixed.

[Michael K. Edwards (m-k-edwards)]

I encountered this bug again with 4.6-2011.06-0 + patches. (See comment 3 on https://bugs.launchpad.net/gcc-linaro/+bug/803232 .) I will test with Chung-Lin's patch.

[Michael K. Edwards (m-k-edwards)]

Before Chung-Lin's patch, compiling jsimd_arm-min12.i fails:

build@ctbu-bld5:~/stuff$ ./compile.sh jsimd_arm-min12.i
jsimd_arm-min12.i: In function 'jsimd_idct_ifast':
jsimd_arm-min12.i:73:1: error: insn does not satisfy its constraints:
(insn 113 21 22 2 (set (mem/s/c:XI (post_inc:SI (reg/f:SI 4 r4 [184])) [0 MEM[(struct int8x8x4_t *)&coef].__o+0 S64 A128])
        (reg:XI 75 s12)) jsimd_arm-min12.i:40 757 {*neon_movxi}
     (expr_list:REG_INC (reg/f:SI 4 r4 [184])
        (nil)))
jsimd_arm-min12.i:73:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:403

After patch, compile succeeds.