[Bug lto/95224] New: -flto -save-temps uses very unusual name for resolution file, looks arbitrary

[Bug lto/95224] New: -flto -save-temps uses very unusual name for resolution file, looks arbitrary

[slyfox at inbox dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95224

            Bug ID: 95224
           Summary: -flto -save-temps uses very unusual name for
                    resolution file, looks arbitrary
           Product: gcc
           Version: 10.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: lto
          Assignee: unassigned at gcc dot gnu.org
          Reporter: slyfox at inbox dot ru
                CC: hubicka at gcc dot gnu.org, marxin at gcc dot gnu.org,
                    rguenth at gcc dot gnu.org
  Target Milestone: ---

Encountered when was debugging lto/95194.

Reproducer:

  # good
  $ rm -- *; echo 'int main(){}' | gcc -c -x c -flto - -o main.o; gcc main.o -o
main -flto -save-temps; ls -- *.res
  main.res

  # unexpected
  $ rm -- *; echo 'int main(){}' | gcc -c -x c -flto - -o main.o; gcc main.o -o
main -flto -save-temps -lm; ls -- *.res
  -lm.res

Expected behaviour: always get 'main.res' resolution file.

In real scenario I was building 'genmatch' and got '-lm.res'. It confused me
quite a bit: I explored very strange possibilities: driver did not recognize
-lm as a library, '-lm.res' is not the only response file out there, etc.

Would it be reasonable for response file to always use the same basename as
'-o' paramemter?

[Bug lto/95224] -flto -save-temps uses very unusual name for resolution file, looks arbitrary

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95224

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |aoliva at gcc dot gnu.org

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
ISTR seeing dups about this.  We seem to be happy to pick anything remotely
resembling a linker output as basename (-lm here).

Alex, do your patches change anything here?

[Bug lto/95224] -flto -save-temps uses very unusual name for resolution file, looks arbitrary

[aoliva at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95224

--- Comment #2 from Alexandre Oliva <aoliva at gcc dot gnu.org> ---
That's not expected, but it seems highly suspicious indeed.  I'll have a look.

[Bug lto/95224] -flto -save-temps uses very unusual name for resolution file, looks arbitrary

[aoliva at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95224

Alexandre Oliva <aoliva at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from Alexandre Oliva <aoliva at gcc dot gnu.org> ---
The changes to the handling of '%u' under -save-temps have probably fixed this.
 Previously, we'd use input_basename unconditionally.