Re: 5.8: LTO exposes some new issues

[Daniel Shahaf <d.s@xxxxxxxxxxxxxxxxxx>]

Roman Perepelitsa wrote on Mon, 27 Jul 2020 16:19 +0200:
> FYI: I won't be doing anything about the warning in gettempname (which
> I'm not getting with my toolchain).

I'm not sure there's anything we can do about that warning.  When I last
looked, the issue was that:

- The linker emits the warning whenever mktemp(3) is used;

- At least one callsite uses mktemp(3) unavoidably: it follows the call
  not by open(2) but by mknod(2), so the warning's rationale (that
  mkstemp(3) should be used in lieu of mktemp(3)) does not apply;

- The linker gives no way to disable the warning.

Unless there's an alternative to mktemp(3)+mknod(2) that I'm unaware
of, I'd triage this as a linker bug, in that it should be possible to
silence the warning when it's a false positive.

Cheers,

Daniel