RFC: mips dwarf2 unwind info
Daniel Jacobowitz
drow@false.org
Fri Mar 11 02:21:00 GMT 2005
On Thu, Mar 10, 2005 at 06:08:36PM -0800, Eric Christopher wrote:
> So, I went ahead and implemented it for mips, but since we don't have
> DIFF_EXPR_OK and no way of doing pc relocs I had to change a couple of
> testcases to allow for either value. I also noticed that in gcc we have
> the debug offset slightly wrong. We have DWARF_CIE_DATA_ALIGNMENT as 4
> there, not -4 even though our stack grows downward. I can't think of any
> reason why I shouldn't check this in here, and correct it there, but I
> thought asking for a bit of review and comment wouldn't hurt either.
>
> I tested on mips-linux-gnu for gas and mips-elf for gcc and gas (along
> with the gcc change), but no more. Before I went through the effort of
> building an entire mips-linux build I thought I'd ask for comments here.
FWIW, it looks right to me. Do you need a tc_regname_to_dw2regnum?
I thought one was generally necessary - but it looks like the fallback
case is better than I thought. Not sure why any port needs one
actually.
And why don't you have any tests? :-)
--
Daniel Jacobowitz
CodeSourcery, LLC
More information about the Binutils
mailing list