Re: [PATCH v2 0/6] Add port for Imagination Technologies Meta

[Will Newton <will dot newton at gmail dot com>]

On Thu, Jan 10, 2013 at 9:46 AM, nick clifton <nickc@redhat.com> wrote:

Hi nick,

> Hi Will,
>
>> Ping?
>
>
> Many apologies for dropping the ball on this submission. :-(

Many thanks for committing the changes!

> I have now been through the patches, and although I found a few problems, I
> did not find any that warranted refusing the submission. Thus I have now
> committed your patches to the repository.
>
> There was one small addition that I made to your patch - I added a line to
> gas/NEWS and ld/NEWS mentioning the support for the new architecture.
>
> Are you volunteering to act as the maintainer for this port ?  If so, please
> add your details at the appropriate location in the binutils/MAINTAINERS
> file.

I may be the maintainer in the short term but I'm not sure who the
longer term maintainer will be. Either way I will submit a patch to
add a MAINTAINERS entry.

> Cheers
>   Nick
>
> PS. The changelog for gas/testsuite mentioned changes to gas/all/gas.exp and
> gas/ieee-fp/x930509a.exp.  But these files were not actually changed.  Thus
> I have omitted these lines from the changelog entry.

Thanks, yes those were a couple of tests I re-enabled late in the day
and forgot to remove the ChangeLog entries for.

> PPS. For the record the compilation problems were all local variables that
> were set but never used:
>
> bfd/elf32-metag.c: In function 'elf_metag_relocate_section':
> bfd/elf32-metag.c:1503:8: error: variable 'dynobj' set but not used
> bfd/elf32-metag.c:3610:25: error: variable 'info'
> bfd/elf32-metag.c:3686:37: error: variable 'htab'
> opcodes/metag-dis.c:2437:15: error: variable 'template_reg'
> opcodes/metag-dis.c:3367:7: error: variable 'status'
> opcodes/metag-dis.c:2437:16: error: variable 'tidx'
> gas/config/tc-metag.c:646:8: error: variable 'exp_seg'
> gas/config/tc-metag.c:1596:8: error: variable 'exp_seg'
> gas/config/tc-metag.c:1802:8: error: variable 'exp_seg'
> gas/config/tc-metag.c:1842:8: error: variable 'exp_seg'
> gas/config/tc-metag.c:2090:8: error: variable 'exp_seg'

Thanks for fixing these.

> PPPS. When I ran the gas testsuite I encountered an unexpected failure:
>
> regexp_diff match failure
> regexp "^.*:    c1 3f b0 ca     DB        MMOVD
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,\[D1Re0\+\+\]$"
> line   "   19950:       c1 3f b0 ca     DB        MMOVD
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RA,[D1Re0++]"
> regexp_diff match failure
> regexp "^.*:    a1 ff b1 ca     DB        MMOVD
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,\[D0\.7\+\+\]$"
> line   "   19964:       a1 ff b1 ca     DB        MMOVD
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RA,[D0.7++]"
> regexp_diff match failure
> regexp "^.*:    e1 3f b2 ca     DB        MMOVD
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,\[A0\.8\+\+\]$"
> line   "   1996c:       e1 3f b2 ca     DB        MMOVD
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RA,[A0.8++]"
> regexp_diff match failure
> regexp "^.*:    c1 3f b0 cb     DB        MMOVL
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,\[D1Re0\+\+\]$"
> line   "   199f8:       c1 3f b0 cb     DB        MMOVL
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RA,[D1Re0++]"
> regexp_diff match failure
> regexp "^.*:    a1 ff b1 cb     DB        MMOVL
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,\[D0\.7\+\+\]$"
> line   "   19a0c:       a1 ff b1 cb     DB        MMOVL
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RA,[D0.7++]"
> regexp_diff match failure
> regexp "^.*:    e1 3f b2 cb     DB        MMOVL
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,\[A0\.8\+\+\]$"
> line   "   19a14:       e1 3f b2 cb     DB        MMOVL
> RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RAM8X32,RA,[A0.8++]"
> FAIL: metadsp21
>
> I did not find out where the truncation of the disassembler's output is
> occurring, I just assumed that it was correct and adjusted the metagdsp21.d
> file accordingly.  Possibly this is connected to the fact that I was running
> the tests on a 32-bit host ?

It *should* be ok on 32 or 64 bit, I'll investigate. It looks like the
gas testsuite files are not visible in CVS, although sourceware.org
seems to be having some trouble at the moment, I'll give it a little
while.