https://sourceware.org/bugzilla/show_bug.cgi?id=34423
--- Comment #6 from Jan Beulich <jbeulich at suse dot com> --- (In reply to H.J. Lu from comment #4) > > Imo only if the source is saying so. And once it does, whether the symbol is > > a function one won't matter anymore (apart from possibly controlling As alluded to already here, ... > > emission of a warning). I.e. first thing is to make gas properly accept > > > > .long bar0@plt - .L4 > > > > (without it accepting other, bogus constructs). > > With > > .long bar0 - .L4 > .type bar0, @function > > assembler knows bar0 is a function symbol. ... I don't think function-ness should affect other behavior in the assembler, and only rarely in the linker. There's exactly one mention of the latter in the spec, afaics. As also said, one use case beyond what the spec says may perhaps be issuing of diagnostics for questionable constructs / uses. Therefore I also disagree with the suggested patch: It would unduly affect existing code, which may not want the new behavior. I continue to think that the new behavior should come into play only upon explicit request by the source, i.e. via use of the @plt modifier. -- You are receiving this mail because: You are on the CC list for the bug.
