Duplicate FDEs in .eh_frame

[Rene Gollent]

Hi there,

I was wondering, is there a known/intended case where gcc + binutils
will produce .eh_frame sections with duplicate FDEs? I'm seeing this
with gcc 4.8.3 + binutils 2.24 on Haiku, and thought I'd see if this
is intentional before I try to come up with a reduced testcase on
another platform, as I don't believe all of our patches have been
upstreamed at this point. The FDEs in question have the exact same
lowpc/highpc, CIE and call frame instructions, the only discernable
difference is that they're located in completely different parts of
the section.

Unfortunately, I'm not seeing any obvious pattern to the FDEs that are
duplicated; some of them map to program counter ranges associated with
weak symbols in the target executable that came from a static library
that was linked in, while others don't match up with anything at all,
so it may take some experimentation to find a way to replicate this
behavior in a simplified manner suitable for testing/analysis, so I
thought I'd ask here first before I start to attempt that.

Regards,

Rene