https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114415
Jeffrey A. Law <law at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|[13/14 Regression] wrong |[13 Regression] wrong code |code with -Oz -fno-dce |with -Oz -fno-dce |-fno-forward-propagate |-fno-forward-propagate |-flive-range-shrinkage |-flive-range-shrinkage |-fweb since r13-1826 |-fweb since r13-1826 --- Comment #7 from Jeffrey A. Law <law at gcc dot gnu.org> --- Given the problems we've generally had in this space, it's probably quite reasonable to treat modifications to the stack pointer as memory accesses. It probably would have been enough to avoid the RISC-V bug I just fixed a week or so ago.