add patches to GCCcore 14.2.0 & 14.3.0 to fix ICE with SVE on aarch64

[Thyre]

(created using eb --new-pr)

Should avoid the following error when building PyTorch in #24926:

/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/aten/src/ATen/native/cpu/Unfold2d.cpp:225:1: error: unrecognizable insn:
  225 | }
      | ^
(insn 1302 1301 1303 97 (set (reg:VNx16BI 3167)
        (unspec:VNx16BI [
                (reg:VNx16BI 3164)
                (reg:VNx8BI 3166)
                (const_vector:VNx4BI [
                        (const_int 0 [0]) repeated x8
                    ])
            ] UNSPEC_TRN1_CONV)) "/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/torch/headeronly/util/bit_cast.h":40:14 -1
     (nil))
during RTL pass: vregs
/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/aten/src/ATen/native/cpu/Unfold2d.cpp:225:1: internal compiler error: in extract_insn, at recog.cc:2812
0x7d30df _fatal_insn(char const*, rtx_def const*, char const*, int, char const*)
        ../../gcc/rtl-error.cc:108
0x7d3113 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        ../../gcc/rtl-error.cc:116
0xec1d17 extract_insn(rtx_insn*)
        ../../gcc/recog.cc:2812
0xc2a28b instantiate_virtual_regs_in_insn
        ../../gcc/function.cc:1612
0xc2a28b instantiate_virtual_regs
        ../../gcc/function.cc:1995
0xc2a28b execute
        ../../gcc/function.cc:2042                                                                                                                                                                                                                             Please submit a full bug report, with preprocessed source (by using -freport-bug).                                                                                                                                                                             Please include the complete backtrace with any bug report.                                                                                                                                                                                                     See <https://gcc.gnu.org/bugs/> for instructions.

See also:

• Build error: unrecognizable insn with using gcc-14 on aarch64 pytorch/pytorch#157842
• AArch64 GCC-14 - internal compiler error: in extract_insn, at recog.cc:2812 pytorch/pytorch#172630
• https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121027
• https://gcc.gnu.org/cgit/gcc/commit/?id=86ecf3e5832a3fd1fc0ecfbb03e91ca8e6aa806e
• https://gcc.gnu.org/cgit/gcc/commit/?id=43f59fcd1e33fc98adc5ff70ac82cc16f77b9ab4

[Thyre]

I'll mark this as draft until I was able to check if I get further in PyTorch with a rebuilt GCCcore...

[Thyre]

Looking through pytorch/pytorch#172445, we might need a few more GCC patches to get forward.
See e.g. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123457, just raised a few days ago.

[Thyre]

Test report by @Thyre
SUCCESS
Build succeeded for 2 out of 2 (total: 1 hour 9 mins 0 secs) (2 easyconfigs in total)
jrc0900.jureca - Linux Rocky Linux 9.6, AArch64, ARM UNKNOWN (neoverse_v2), 1 x NVIDIA NVIDIA GH200 480GB, 580.95.05, Python 3.9.21
See https://gist.github.com/Thyre/9a756a8f83ce39803686cc1d5be2a5e3 for a full test report.

[Thyre]

Adding these patches unfortunately doesn't change anything:

/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/aten/src/ATen/native/cpu/Unfold2d.cpp: In function ‘void at::native::{anonymous}::unfolded2d_acc_kernel(c10::ScalarType, void*, void*, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, int64_t, bool)’:
/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/aten/src/ATen/native/cpu/Unfold2d.cpp:221:1: error: unrecognizable insn:
  221 | }
      | ^
(insn 1302 1301 1303 97 (set (reg:VNx16BI 3167)
        (unspec:VNx16BI [
                (reg:VNx16BI 3164)
                (reg:VNx8BI 3166)
                (const_vector:VNx4BI [
                        (const_int 0 [0]) repeated x8
                    ])
            ] UNSPEC_TRN1_CONV)) "/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/torch/headeronly/util/bit_cast.h":40:14 -1
     (nil))
during RTL pass: vregs
/dev/shm/reuter1/easybuild/build/PyTorch/2.9.1/foss-2025b-CUDA-12.9.1/pytorch-v2.9.1/aten/src/ATen/native/cpu/Unfold2d.cpp:221:1: internal compiler error: in extract_insn, at recog.cc:2812
0x7d30df _fatal_insn(char const*, rtx_def const*, char const*, int, char const*)
        ../../gcc/rtl-error.cc:108
0x7d3113 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        ../../gcc/rtl-error.cc:116
0xec1d17 extract_insn(rtx_insn*)
        ../../gcc/recog.cc:2812
0xc2a28b instantiate_virtual_regs_in_insn
        ../../gcc/function.cc:1612
0xc2a28b instantiate_virtual_regs
        ../../gcc/function.cc:1995
0xc2a28b execute
        ../../gcc/function.cc:2042
Please submit a full bug report, with preprocessed source (by using -freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

Maybe there's another patch for GCC 14 which fixes the issue. The test case in the bug report works with the patch.