Created attachment 57098 [details] reduced testcase This has been reduced from gcc.target/riscv/rvv/vsetvl/vlmax_back_prop-27.c ; it fails the same way with -ftree-parallelize-loops=2. Compiler output: $ riscv64-unknown-linux-gnu-gcc -O -march=rv64gcv -ftree-parallelize-loops=2 testcase.c during GIMPLE pass: parloops testcase.c: In function 'f': testcase.c:4:1: internal compiler error: in force_constant_size, at gimplify.cc:743 4 | f(long out, int n) | ^ 0xa18136 force_constant_size /repo/gcc-trunk/gcc/gimplify.cc:743 0x134e297 gimple_add_tmp_var(tree_node*) /repo/gcc-trunk/gcc/gimplify.cc:781 0x1304676 create_tmp_var(tree_node*, char const*) /repo/gcc-trunk/gcc/gimple-expr.cc:491 0x1780c4d separate_decls_in_region /repo/gcc-trunk/gcc/tree-parloops.cc:2125 0x1782dd1 gen_parallel_loop /repo/gcc-trunk/gcc/tree-parloops.cc:3162 0x1784a48 parallelize_loops /repo/gcc-trunk/gcc/tree-parloops.cc:4140 0x1785dc1 execute /repo/gcc-trunk/gcc/tree-parloops.cc:4228 0x1785dc1 execute /repo/gcc-trunk/gcc/tree-parloops.cc:4207 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. $ riscv64-unknown-linux-gnu-gcc -O -march=rv32gcv -mabi=ilp32 -ftree-parallelize-loops=2 testcase.c during RTL pass: expand testcase.c: In function 'f._loopfn.0': testcase.c:6:17: internal compiler error: in convert_move, at expr.cc:223 6 | for (int i; i < n; i++) { | ^ 0x9f3dd1 convert_move(rtx_def*, rtx_def*, int) /repo/gcc-trunk/gcc/expr.cc:223 0x127807a store_expr(tree_node*, rtx_def*, int, bool, bool) /repo/gcc-trunk/gcc/expr.cc:6810 0x1279909 expand_assignment(tree_node*, tree_node*, bool) /repo/gcc-trunk/gcc/expr.cc:6443 0x112715c expand_gimple_stmt_1 /repo/gcc-trunk/gcc/cfgexpand.cc:3966 0x112715c expand_gimple_stmt /repo/gcc-trunk/gcc/cfgexpand.cc:4064 0x112e2be expand_gimple_basic_block /repo/gcc-trunk/gcc/cfgexpand.cc:6120 0x112fec7 execute /repo/gcc-trunk/gcc/cfgexpand.cc:6855 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. $ riscv64-unknown-linux-gnu-gcc -v Using built-in specs. COLLECT_GCC=/repo/gcc-trunk/binary-latest-riscv64/bin/riscv64-unknown-linux-gnu-gcc COLLECT_LTO_WRAPPER=/repo/gcc-trunk/binary-trunk-r14-7258-20240115151353-g731444b3c39-checking-yes-rtl-df-extra-riscv64/bin/../libexec/gcc/riscv64-unknown-linux-gnu/14.0.1/lto-wrapper Target: riscv64-unknown-linux-gnu Configured with: /repo/gcc-trunk//configure --enable-languages=c,c++ --enable-valgrind-annotations --disable-nls --enable-checking=yes,rtl,df,extra --with-cloog --with-ppl --with-isl --with-isa-spec=2.2 --with-sysroot=/usr/riscv64-unknown-linux-gnu --build=x86_64-pc-linux-gnu --host=x86_64-pc-linux-gnu --target=riscv64-unknown-linux-gnu --with-ld=/usr/bin/riscv64-unknown-linux-gnu-ld --with-as=/usr/bin/riscv64-unknown-linux-gnu-as --disable-multilib --disable-libstdcxx-pch --prefix=/repo/gcc-trunk//binary-trunk-r14-7258-20240115151353-g731444b3c39-checking-yes-rtl-df-extra-riscv64 Thread model: posix Supported LTO compression algorithms: zlib zstd gcc version 14.0.1 20240115 (experimental) (GCC)
I thought I saw something similar before for aarch64 and SVE. Maybe OpenMP .
Confirmed. Also fails a similar way with ARMv9-a's SVE: ``` #include <arm_sve.h> void f(char* out, int n, svint8_t v2) { for (int i = 0; i < n; i++) { *(svint8_t *)(out + i) = v2; } } ``` Compile with ` -O2 -g0 -march=armv9-a -ftree-parallelize-loops=2`.