https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124905
--- Comment #2 from Jonathan Wakely <redi at gcc dot gnu.org> --- Again, component=diagnostics is not for reporting false positive warnings: A bug relating to the "gcc/diagnostics" subdirectory of the source tree: the shared code implementing diagnostics (text output, SARIF output, source quoting, etc). Bugs relating to specific diagnostics should instead use the component emitting them, and have the "diagnostic" keyword set.' I agree that the warning seems correct here. There is no situation when reading -1zu bytes will be correct, so it's not a good argument to memchr. The content of the string is not relevant to the warning. If you really want to pass that value, you can suppress the warning with diagnostic pragmas.
