[Bug ld/27441] Small inconsistency in between gold and bfd

[rguenth at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27441

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

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

--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> ---
Note with -flto it is difficult to argue about the order of "source" objects
and library references with -l, say you'd have

gcc t.c -lpthread t2.c -flto

gcc can then, via -flto, happily produce only a single "real" ltrans object.
Of course via the plugin it should get the original refs of t.c and t2.c
in the correct order but my understanding is that BFD re-scans objects
(because GCC is also prone to introduce new refs, for example to memcpy,
libm functions or libgcc provided functions).  But "placement" of the
generated ltrans object on the linker command-line is not well-defined since
the plugin merely provides LD with extra objects it should link.

-- 
You are receiving this mail because:
You are on the CC list for the bug.