g:9b7cad5884f21cc5783075be0043777448db3fab, r15-512-g9b7cad5884f21c: 1122 failures I saw these on powerpc64 both BE and LE FAIL: g++.dg/tree-ssa/pr83215.C -std=gnu++14 scan-tree-dump-times fre1 "\\*i" 1 FAIL: g++.dg/tree-ssa/pr83215.C -std=gnu++17 scan-tree-dump-times fre1 "\\*i" 1 FAIL: g++.dg/tree-ssa/pr83215.C -std=gnu++20 scan-tree-dump-times fre1 "\\*i" 1 FAIL: g++.dg/tree-ssa/pr83215.C -std=gnu++98 scan-tree-dump-times fre1 "\\*i" 1 FAIL: gcc.dg/tree-ssa/ssa-lim-15.c scan-tree-dump lim2 "Executing store motion" FAIL: gfortran.dg/vect/vect-do-concurrent-1.f90 -O at line 14 (test for warnings, line )
commit 9b7cad5884f21cc5783075be0043777448db3fab (HEAD) Author: Jan Hubicka <jh@suse.cz> Date: Wed May 15 14:14:27 2024 +0200 Avoid pointer compares on TYPE_MAIN_VARIANT in TBAA
I see those on x86_64-linux and i686-linux too.
I hate to pile on, but riscv64-linux-gnu: FAIL: gcc.dg/tree-ssa/ssa-lim-15.c scan-tree-dump lim2 "Executing store motion" I'm certainly willing to believe that perhaps we can't do this store motion without using TBAA to disambiguate certain objects. I haven't actually dug into the regression to determine if that's the case or not.
Confirmed.
There's also PR ada/115106 where this patch broke 32-bit Solaris/x86 Ada bootstrap.
I'm seeing a bootstrap failure on aarch64-none-linux-gnu as well, bisected to this commit. The error, in Stage 2, is: In file included from ../../src/gcc/system.h:726, from ../../src/gcc/tree-affine.cc:21: In member function 'long int* widest_int_storage<N>::write_val(unsigned int) [with int N = 131072]', inlined from 'typename wi::binary_traits<T, T>::result_type wi::sext(const T&, unsigned int) [with T = generic_wide_int<widest_int_storage<131072> >]' at ../../src/gcc/wide-int.h:2582:28, inlined from 'poly_int<N, typename poly_result<Ca, Ca>::type> wi::sext(const poly_int<N, C>&, unsigned int) [with unsigned int N = 2; Ca = generic_wide_int<widest_int_storage<131072> >]' at ../../src/gcc/poly-int.h:746:5, inlined from 'poly_widest_int wide_int_ext_for_comb(const poly_widest_int&, tree)' at ../../src/gcc/tree-affine.cc:49:46: ../../src/gcc/../include/libiberty.h:370:38: error: 'void free(void*)' called on a pointer to an unallocated object '13450750900837400303' [-Werror=free-nonheap-object] 370 | #define XDELETEVEC(P) free ((void*) (P)) | ~~~~~^~~~~~~~~~~~~ ../../src/gcc/wide-int.h:1630:5: note: in expansion of macro 'XDELETEVEC' 1630 | XDELETEVEC (u.valp); | ^~~~~~~~~~ ../../src/gcc/../include/libiberty.h:370:38: error: 'void free(void*)' called on a pointer to an unallocated object '13450750900837400303' [-Werror=free-nonheap-object] 370 | #define XDELETEVEC(P) free ((void*) (P)) | ~~~~~^~~~~~~~~~~~~ ../../src/gcc/wide-int.h:1630:5: note: in expansion of macro 'XDELETEVEC' 1630 | XDELETEVEC (u.valp); | ^~~~~~~~~~ Our regression testing also shows the test failures reported in the first comment for powerpc64.
I have a patch.
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>: https://gcc.gnu.org/g:a5b3721c06646bf5b9b50a22964e8e2bd4d03f5f commit r15-626-ga5b3721c06646bf5b9b50a22964e8e2bd4d03f5f Author: Richard Biener <rguenther@suse.de> Date: Fri May 17 11:02:29 2024 +0200 middle-end/115110 - Fix view_converted_memref_p view_converted_memref_p was checking the reference type against the pointer type of the offset operand rather than its pointed-to type which leads to all refs being subject to view-convert treatment in get_alias_set causing numerous testsuite fails but with its new uses from r15-512-g9b7cad5884f21c is also a wrong-code issue. PR middle-end/115110 * tree-ssa-alias.cc (view_converted_memref_p): Fix.
Fixed.
*** Bug 115141 has been marked as a duplicate of this bug. ***