nikic wrote: RFC: https://discourse.llvm.org/t/rfc-allocator-provenance-model/91106
Though seeing the last comments on this thread, one thing I forgot to account for is the realloc case. We could make `dealloc` not poison the memory, so that realloc can be a `dealloc` + `alloc` combination. It would require a change to `allockind` to specify whether poisoning happens, as DSE assumes it does. Alternatively, could have a dedicated `realloc` intrinsic. https://github.com/llvm/llvm-project/pull/198202 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
