how close to binutils-2.12 previews

[Thiemo Seufer]

Eric Christopher wrote:
> Macej,
> 	So, I'm going to go with Thiemo's comments on this and ask that most of
> the patch be delayed while we wait for him to finish 64-bit support for
> mips. As Thiemo said the checking for la and dla based on
> register/address size causes (I believe valid) testsuite failures.

To be clear about this:
The warning output introduced by the patch causes the testsuite
failures, which means the testsuite should be fixed. Such code went
through unchecked before, even if it was harmful.

> It
> could warn erroneously as well (I'm thinking of a 64-bit arch with
> -mgp32, as well as, the n32 case or even a 64-bit arch with a 32-bit
> pointer range).

The code snippet I posted in reply to Maciej should solve this at
the price of dropping the warning for n32 and alike.


Thiemo