Post #2279979
2026-04-15 06:02 UTC
@pinskia@hachyderm.io @karolherbst@chaos.social no, I don't have a problem with that as a fix. I do think (and know this is unpopular) that the compiler has some obligation to keep old code that previously had a deterministic behavior to have a consistent result. I know the compiler has to move forward, but I think it currently fails many use cases (including one important to me) of keeping code that works working.
I don't think we should stop the compiler moving forward in those cases, but I do think we should mark it and let the users ask for specific prior behavior. I think not doing so significantly adds a burden to many maintainers, and I can easily quantify that in my case.
Replies (1)
-
@pinskia@hachyderm.io 2026-04-15 06:06
@vathpela@infosec.exchange @karolherbst@chaos.social But in most cases it just happened "deterministic behavior" was due to stack being cleared or addresses being the same. Both changes in the kernel not the compiler. Or slightly different library writing to the stack gets that old behavior. E.g. reading after a free, in a secured system might read bad values. Also.