[Sorry for the delay]

> My understanding is that Eric fixed the bug and all that needs to be done
> is simply adjust the testcase expected results (after verifying linker
> output is correct) for the added .got STT_SECTION symbol etc.

Right, tlssunpic32.rd only needs to account for .got in .dynsym again, which 
essentially boiled down to shifting the dynamic indexes.

> The test that was failing before Eric's patch was not even adjusted for
> my .dynsym changes.

Right, tlssunnopic32.rd was out of sync for the section order, the contents 
of .dynsym and the type of TLS symbols (it was NOTYPE).

Fixed thusly.


	* ld-sparc/tlssunnopic32.rd: Adjust for .dynsym changes.
	* ld-sparc/tlssunpic32.rd: Likewise.


-- 
Eric Botcazou