https://sourceware.org/bugzilla/show_bug.cgi?id=33237

--- Comment #6 from H.J. Lu <hjl.tools at gmail dot com> ---
(In reply to Rainer Orth from comment #5)
> > --- Comment #4 from H.J. Lu <hjl.tools at gmail dot com> ---
> > (In reply to Rainer Orth from comment #3)
> >> Recently, the workaround (hack?) stopped working: the tests FAIL to run 
> >> with
> >> 
> >> Running: tmpdir/no-plt-1f > tmpdir/no-plt-1f.out
> >> ld.so.1: no-plt-1f: fatal: relocation error: R_AMD64_PC32: file
> >> tmpdir/no-plt-1f: symbol func_p: value 0xfffffffec86ffd1e does not fit
> >
> > On Linux, there is
> >
> > $ readelf -r tmpdir/no-plt-1f
> > Relocation section '.rela.dyn' at offset 0x12a8 contains 14 entries:
> >   Offset          Info           Type           Sym. Value    Sym. Name +
> > Addend
> > ...
> > 000000003008  000a00000005 R_X86_64_COPY     0000000000003008 func_p + 0
> >
> > Why isn't R_X86_64_COPY generated?
> 
> It is without -z nocopyreloc as originally reported, but then the test
> aborts.  Until recently, the -z nocopyreloc workaround let the test
> PASS, but not anymore.

Why does it abort on Solaris?

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

Reply via email to