[ELF] - exit on --version call

[grimar]

Summary

Sorry for posting such a minor one
, but just found I am not sure this is
expected behaviour though it seems so according to
https://llvm.org/bugs/show_bug.cgi?id=31057.

So do we want that one right ?

[davide]

The patch is correct, but please add a testcase.

[grimar]

• Updated test.

[grimar]

In D26865#600142, @davide wrote:

The patch is correct, but please add a testcase.

I missed I need to update test as well, thanks !

[ruiu]

This is the patch I mentioned. http://llvm.org/viewvc/llvm-project?view=revision&revision=278797

[emaste]

Seems it's a difference in behaviour between -v and --version. AFAICT for -v both ld.bfd and ld.gold print a one-line version number and carry on with the link if there are args (or exit with 0 if not). For --version they print a multi-line version & This program is free software info, and do not link.

[ruiu]

So we probably should stop handling --version as an alias to -v and make --verbose behave differently than -v.

[grimar]

In D26865#600168, @ruiu wrote:

This is the patch I mentioned. http://llvm.org/viewvc/llvm-project?view=revision&revision=278797

Hmm, I really forgot about this one. Confusing.

[grimar]

In D26865#600186, @ruiu wrote:

So we probably should stop handling --version as an alias to -v and make --verbose behave differently than -v.

I can do that tomorrow. (if nobody needs to fix it faster).

[grimar]

• Updated.

Behavior of ld.bfd and ld.gold is inconsistent. I checked the next combinations:
ld -v + <normal files>. Result: link OK.
gold -v + <normal files>. Result: link OK.
ld -v <no files>. Result: exit with code 0.
gold -v <no files>. Result: exit with code 0.
ld -version <no files>. Result: exit with code 0.
gold -version <no files>. Result: exit with code 0.

Until here behavior was equal. But the last case has different result:
ld -version <normal files>. Result: no link happens, exit with 0.
gold -version <normal file>. Result: link OK.

That is what Rui mentioned on bug page, I did not realized it at first.

So looks we can live with just a minor change like this patch do to fix
PR31057, thoughts ?

[ruiu]

LGTM

ELF/Driver.cpp

 
 void LinkerDriver::main(ArrayRef<const char *> ArgsArr, bool CanExitEarly) {
   ELFOptTable Parser;
   opt::InputArgList Args = Parser.parse(ArgsArr.slice(1));
   if (Args.hasArg(OPT_help)) {
     printHelp(ArgsArr[0]);
     return;
   }
-  if (Args.hasArg(OPT_version))
+  if (Args.hasArg(OPT_version)) {
     outs() << getLLDVersion() << "\n";
+    exitLld(0);
+  }
   Config->ExitEarly = CanExitEarly && !Args.hasArg(OPT_full_shutdown);
 
   if (const char *Path = getReproduceOption(Args)) {
     // Note that --reproduce is a debug option so you can ignore it
     // if you are trying to understand the whole picture of the code.
     ErrorOr<CpioFile *> F = CpioFile::create(Path);
     if (F) {
       Cpio.reset(*F);