[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[andrew.goodbody at linaro dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

andrew.goodbody at linaro dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |andrew.goodbody at linaro dot org

--- Comment #1 from andrew.goodbody at linaro dot org ---
gcc 10.2.0 built from tarball running native on aarch64 shows the same
behaviour. I will attach the pre-processed source if I can although its over
9MB.

In file included from
external/eigen_archive/unsupported/Eigen/CXX11/Tensor:144,
                 from ./third_party/eigen3/unsupported/Eigen/CXX11/Tensor:1,
                 from ./tensorflow/core/framework/numeric_types.h:20,
                 from ./tensorflow/core/framework/allocator.h:26,
                 from ./tensorflow/core/framework/op_kernel.h:25,
                 from ./tensorflow/core/kernels/histogram_op.h:19,
                 from tensorflow/core/kernels/histogram_op.cc:20:
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h: In
static member function 'static void Eigen::internal::EvalRange<Evaluator,
StorageIndex, true>::run(Evaluator*, StorageIndex, StorageIndex) [with
Evaluator = Eigen::TensorEvaluator<const
Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<int, 1, 1, long int>, 16,
Eigen::MakePointer>, const Eigen::TensorConversionOp<int, const
Eigen::TensorCwiseBinaryOp<Eigen::internal::scalar_min_op<double, double>,
const
Eigen::TensorCwiseUnaryOp<Eigen::internal::bind2nd_op<Eigen::internal::scalar_quotient_op<double,
double> >, const Eigen::TensorConversionOp<double, const
Eigen::TensorCwiseBinaryOp<Eigen::internal::scalar_difference_op<Eigen::half,
Eigen::half>, const
Eigen::TensorCwiseBinaryOp<Eigen::internal::scalar_max_op<const Eigen::half,
const Eigen::half>, const Eigen::TensorMap<Eigen::Tensor<const Eigen::half, 1,
1, long int>, 16, Eigen::MakePointer>, const
Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<const
Eigen::half>, const Eigen::TensorMap<Eigen::Tensor<const Eigen::half, 1, 1,
long int>, 16, Eigen::MakePointer> > >, const
Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<const
Eigen::half>, const Eigen::TensorMap<Eigen::Tensor<const Eigen::half, 1, 1,
long int>, 16, Eigen::MakePointer> > > > >, const
Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<double>, const
Eigen::TensorCwiseUnaryOp<Eigen::internal::bind2nd_op<Eigen::internal::scalar_quotient_op<double,
double> >, const Eigen::TensorConversionOp<double, const
Eigen::TensorCwiseBinaryOp<Eigen::internal::scalar_difference_op<Eigen::half,
Eigen::half>, const
Eigen::TensorCwiseBinaryOp<Eigen::internal::scalar_max_op<const Eigen::half,
const Eigen::half>, const Eigen::TensorMap<Eigen::Tensor<const Eigen::half, 1,
1, long int>, 16, Eigen::MakePointer>, const
Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<const
Eigen::half>, const Eigen::TensorMap<Eigen::Tensor<const Eigen::half, 1, 1,
long int>, 16, Eigen::MakePointer> > >, const
Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<const
Eigen::half>, const Eigen::TensorMap<Eigen::Tensor<const Eigen::half, 1, 1,
long int>, 16, Eigen::MakePointer> > > > > > > > >, Eigen::ThreadPoolDevice>;
StorageIndex = long int]':
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h:349:3:
error: could not split insn
  349 |   }
      |   ^
(insn 1615 2102 1433 (set (reg:VNx4SF 34 v2 [orig:554 _1597 ] [554])
        (unspec:VNx4SF [
                (reg:VNx4BI 72 p4 [orig:376 vec_mask_and_1413 ] [376])
                (unspec:VNx4SF [
                        (reg:VNx4BI 78 p10 [orig:376 vec_mask_and_1413 ] [376])
                        (const_int 1 [0x1])
                        (reg:VNx4SF 49 v17 [2017])
                        (reg:VNx4SF 32 v0 [2049])
                    ] UNSPEC_COND_FADD)
                (reg:VNx4SF 39 v7 [orig:392 vect__ifc__887.8029 ] [392])
            ] UNSPEC_SEL)) 5628 {*cond_addvnx4sf_any}
     (expr_list:REG_DEAD (reg:VNx4BI 78 p10 [orig:376 vec_mask_and_1413 ]
[376])
        (expr_list:REG_DEAD (reg:VNx4BI 72 p4 [orig:376 vec_mask_and_1413 ]
[376])
            (expr_list:REG_DEAD (reg:VNx4SF 39 v7 [orig:392 vect__ifc__887.8029
] [392])
                (expr_list:REG_EQUAL (unspec:VNx4SF [
                            (reg:VNx4BI 78 p10 [orig:376 vec_mask_and_1413 ]
[376])
                            (unspec:VNx4SF [
                                    (reg:VNx4BI 78 p10 [orig:376
vec_mask_and_1413 ] [376])
                                    (const_int 1 [0x1])
                                    (reg:VNx4SF 49 v17 [2017])
                                    (const_vector:VNx4SF [
                                            (const_double:SF -6.103515625e-5
[-0x0.8p-13])
                                        ])
                                ] UNSPEC_COND_FADD)
                            (subreg:VNx4SF (reg:VNx4SI 39 v7 [orig:392
vect__ifc__887.8029 ] [392]) 0)
                        ] UNSPEC_SEL)
                    (nil))))))
during RTL pass: final
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h:349:3:
internal compiler error: in final_scan_insn_1, at final.c:3073
0x69255f _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        /home/builder/src/gcc/10.2.0/gcc-10.2.0/gcc/rtl-error.c:108
0x9debcf final_scan_insn_1
        /home/builder/src/gcc/10.2.0/gcc-10.2.0/gcc/final.c:3073
0x9dede3 final_scan_insn(rtx_insn*, _IO_FILE*, int, int, int*)
        /home/builder/src/gcc/10.2.0/gcc-10.2.0/gcc/final.c:3152
0x9df0a7 final_1
        /home/builder/src/gcc/10.2.0/gcc-10.2.0/gcc/final.c:2020
0x9df823 rest_of_handle_final
        /home/builder/src/gcc/10.2.0/gcc-10.2.0/gcc/final.c:4658
0x9df823 execute
        /home/builder/src/gcc/10.2.0/gcc-10.2.0/gcc/final.c:4736
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[andrew.goodbody at linaro dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #2 from andrew.goodbody at linaro dot org ---
Created attachment 50260
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50260&action=edit
Pre-processed source showing bug

Sorry its an archive but its too big as plain text. Produced along with the
output from the previous comment.

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[andrew.goodbody at linaro dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #3 from andrew.goodbody at linaro dot org ---
Command line used was this

/usr/local/bin/gcc    -U_FORTIFY_SOURCE     -Wall    
-Wunused-but-set-parameter     -Wno-free-nonheap-object    
-fno-omit-frame-pointer     -g0     -O2     '-D_FORTIFY_SOURCE=1'     -DNDEBUG 
   -ffunction-sections     -fdata-sections     '-std=c++0x'    -MD     -MF    
bazel-out/aarch64-opt/bin/tensorflow/core/kernels/_objs/histogram_op/histogram_op.pic.d
   
'-frandom-seed=bazel-out/aarch64-opt/bin/tensorflow/core/kernels/_objs/histogram_op/histogram_op.pic.o'
    -fPIC     -D__CLANG_SUPPORT_DYN_ANNOTATION__     -DEIGEN_MPL2_ONLY    
'-DEIGEN_MAX_ALIGN_BYTES=64'     '-DEIGEN_HAS_TYPE_TRAITS=0'    
-DEIGEN_USE_BLAS     -DEIGEN_USE_LAPACK     -iquote     .     -iquote    
bazel-out/aarch64-opt/bin     -iquote     external/com_google_absl     -iquote 
   bazel-out/aarch64-opt/bin/external/com_google_absl    -iquote    
external/eigen_archive     -iquote    
bazel-out/aarch64-opt/bin/external/eigen_archive     -iquote    
external/local_config_sycl     -iquote    
bazel-out/aarch64-opt/bin/external/local_config_sycl     -iquote    
external/nsync     -iquote     bazel-out/aarch64-opt/bin/external/nsync    
-iquote     external/com_googlesource_code_re2     -iquote    
bazel-out/aarch64-opt/bin/external/com_googlesource_code_re2     -iquote    
external/gif     -iquote     bazel-out/aarch64-opt/bin/external/gif     -iquote
    external/libjpeg_turbo     -iquote    
bazel-out/aarch64-opt/bin/external/libjpeg_turbo     -iquote    
external/com_google_protobuf     -iquote    
bazel-out/aarch64-opt/bin/external/com_google_protobuf     -iquote    
external/farmhash_archive     -iquote    
bazel-out/aarch64-opt/bin/external/farmhash_archive     -iquote    
external/fft2d     -iquote     bazel-out/aarch64-opt/bin/external/fft2d    
-iquote     external/highwayhash     -iquote    
bazel-out/aarch64-opt/bin/external/highwayhash     -iquote    
external/zlib_archive     -iquote    
bazel-out/aarch64-opt/bin/external/zlib_archive     -isystem    
third_party/eigen3/mkl_include     -isystem    
bazel-out/aarch64-opt/bin/third_party/eigen3/mkl_include     -isystem    
external/eigen_archive    -isystem    
bazel-out/aarch64-opt/bin/external/eigen_archive     -isystem    
external/nsync/public     -isystem    
bazel-out/aarch64-opt/bin/external/nsync/public     -isystem     external/gif  
  -isystem     bazel-out/aarch64-opt/bin/external/gif     -isystem    
external/com_google_protobuf/src     -isystem    
bazel-out/aarch64-opt/bin/external/com_google_protobuf/src     -isystem    
external/farmhash_archive/src     -isystem    
bazel-out/aarch64-opt/bin/external/farmhash_archive/src     -isystem    
external/zlib_archive     -isystem    
bazel-out/aarch64-opt/bin/external/zlib_archive     '-march=armv8.2-a+sve'    
-O3     '-std=c++14'     '-march=armv8.2-a+sve'     -O3    
'-D_GLIBCXX_USE_CXX11_ABI=0'     -DEIGEN_AVOID_STL_ARRAY    
-Iexternal/gemmlowp     -Wno-sign-compare     '-ftemplate-depth=900'    
-fno-exceptions     '-DINTEL_MKL=1'     -DEIGEN_USE_VML    
-DINTEL_MKL_DNN_ONLY     -DENABLE_MKL     -pthread     '-DINTEL_MKL=1'    
-DINTEL_MKL_DNN_ONLY     -DENABLE_MKL     -fno-canonical-system-headers    
-Wno-builtin-macro-redefined     '-D__DATE__="redacted"'    
'-D__TIMESTAMP__="redacted"'     '-D__TIME__="redacted"'     -c    
tensorflow/core/kernels/histogram_op.cc     -o    
bazel-out/aarch64-opt/bin/tensorflow/core/kernels/_objs/histogram_op/histogram_op.pic.o

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #4 from Segher Boessenkool <segher at gcc dot gnu.org> ---
I cannot get the reduced testcase to fail.  Are any special options needed?

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[asolokha at gmx dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #5 from Arseny Solokha <asolokha at gmx dot com> ---
(In reply to Segher Boessenkool from comment #4)
> I cannot get the reduced testcase to fail.  Are any special options needed?

If you've been asking me: no, the compiler invocation posted in comment 0 is
explicit, but maybe you need -m32 -f{,no-}PIC -f{,no-}stack-protector which one
often does for reproducing my PRs because of the configuration I use. I'll
certainly test the current snapshot, but I won't be able to do so at least one
more week.

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I can reproduce #c0 with an older trunk and -m32 -mcpu=e300c3 in a cross, but
it has been fixed with
r11-3606-g4c69e61f4307865b95151006e480ae2022b30454
So maybe that fix needs backporting?
As for #c3-#c5, please file a separate bug, a rs6000 backend bug certainly has
nothing to do with completely unrelated aarch64 ICE.

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
And to explain what was happening,
rs6000_legitimize_address has been called on x
(plus:SI (reg/f:SI 117)
    (const_int 2147483647 [0x7fffffff]))
and mode QImode, oldx == x.
((0x7fffffff & 0xffff) ^ 0x8000) - 0x8000
is low_int -1 and
      if (low_int >= 0x8000 - extra)
is not true and 0x7fffffff - -1 is 0x80000000 (with UB on the compiler side).
So maybe the above mentioned commit wasn't sufficient and
we should
-      high_int = INTVAL (XEXP (x, 1)) - low_int;
+      high_int = UINTVAL (XEXP (x, 1)) - low_int;
But also
      && ((unsigned HOST_WIDE_INT) (INTVAL (XEXP (x, 1)) + 0x8000)
can invoke UB in the compiler, shouldn't it be just
      && ((UINTVAL (XEXP (x, 1)) + 0x8000)
?

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #8 from Segher Boessenkool <segher at gcc dot gnu.org> ---
(In reply to Arseny Solokha from comment #5)
> (In reply to Segher Boessenkool from comment #4)
> > I cannot get the reduced testcase to fail.  Are any special options needed?
> 
> If you've been asking me:

Well you reported this, so probably?  :-)

> no, the compiler invocation posted in comment 0 is
> explicit, but maybe you need -m32 -f{,no-}PIC -f{,no-}stack-protector which
> one often does for reproducing my PRs because of the configuration I use.

Maybe that is why I so often cannot reproduce your PRs?  Please always state
the
exact compiler configuration / invocation needed.

> I'll certainly test the current snapshot, but I won't be able to do so at
> least one more week.

Jakub in comment 7 seems to have found the problem.  I cc:ed Alan, who did
4c69e61f4307, which seems to have fixed it on trunk (there is no PR for that
so far?)

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

--- Comment #9 from Segher Boessenkool <segher at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #7)
>       if (low_int >= 0x8000 - extra)
> is not true and 0x7fffffff - -1 is 0x80000000 (with UB on the compiler side).

These are HWIs, so there is no UB.

> But also
>       && ((unsigned HOST_WIDE_INT) (INTVAL (XEXP (x, 1)) + 0x8000)
> can invoke UB in the compiler, shouldn't it be just
>       && ((UINTVAL (XEXP (x, 1)) + 0x8000)
> ?

That sounds right, yes.

[Bug target/93353] ICE: in final_scan_insn_1, at final.c:3073 (error: could not split insn)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93353

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
   Target Milestone|---                         |11.0
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #10 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed so closing as such as GCC 10 no longer being updated.


Note the aarch64 bug which was reported (incorrectly in this bug report) is a
dup of PR 96357.