martong added a comment. In D122150#3408156 <https://reviews.llvm.org/D122150#3408156>, @balazske wrote:
> This errno check will work if a state split is added at every standard > function that may set `errno` at failure. (The success and failure branches > have different //errno check state//, and probably different return value for > the function.) Probably this is too many state splits. But a correct program > should have anyway checks for the return value that make the same state > splits. I think this is a good initiative and valuable work. You might be right about the too many state splits, however, we should measure this empirically. We should have a measurement on opensource projects to highlight peak memory usage and runtime discrepancies compared to the baseline. It might turn out that we indeed have to create a list of functions on which we are allowed to do the state split. Also, would be useful to evaluate the new reports. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D122150/new/ https://reviews.llvm.org/D122150 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
