https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77691
Rainer Orth <ro at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |redi at gcc dot gnu.org --- Comment #22 from Rainer Orth <ro at gcc dot gnu.org> --- While debugging what could be memory corruption as reported in PR bootstrap/87134, I found another related issue: when I preload libumem.so (an alternative debugging allocator library), two new failures occur: +FAIL: experimental/memory_resource/new_delete_resource.cc execution test /vol/gcc/src/hg/trunk/local/libstdc++-v3/testsuite/experimental/memory_resource/new_delete_resource.cc:117: void test03(): Assertion 'aligned<max_align_t>(p)' failed. +FAIL: experimental/memory_resource/resource_adaptor.cc execution test /vol/gcc/src/hg/trunk/local/libstdc++-v3/testsuite/experimental/memory_resource/resource_adaptor.cc:136: void test05(): Assertion 'aligned<max_align_t>(p)' failed. this time for the 64-bit test only. I find that malloc(1) returns 8-byte aligned storage, while alignof(max_align_t) is 16. Reading https://github.com/jemalloc/jemalloc/issues/1072 suggests that this might be in accordance with C11.