old binutils-2.13-msp.diff and binutils 2.19

old binutils-2.13-msp.diff and binutils 2.19

[Andrew Randrianasulu]

Hello. Sorry for asking this question - but while trying to move forward this patch, i faced with duplicated opcode in opcodes/mips-ops.c One "vmulu" was defined for Octeon, and one - for SGI O2 VICE coprocessor. After commenting out one from Octeon - my [patched] binutils finally was able to pass gcc-3.4.6 compilation. I saw some duplicates in this file, apart from my case,  but i'm really unsure what to do in this case? Move patch parts around? But moving them in random order will break assembler, already learned this ....

Also I'm lost in gas/config/tc-mips.c Original patch was designed for (K, m, n) [i don't know what they mean .. some form of internal markers?] and i changed it for (+K, +m,+n). But i'm really lost in those big switches there. Right now my new code disabled, looking at old patch i must add some logic before yet another switch. Where is the best place for discussing this - here or on gcc mail list?


      

Re: old binutils-2.13-msp.diff and binutils 2.19

[Thiemo Seufer]

Andrew Randrianasulu wrote:
> Hello. Sorry for asking this question - but while trying to move 
> forward this patch, i faced with duplicated opcode in
> opcodes/mips-ops.c One "vmulu" was defined for Octeon, and one -
> for SGI O2 VICE coprocessor. After commenting out one from Octeon
> - my [patched] binutils finally was able to pass gcc-3.4.6 compilation.
> I saw some duplicates in this file, apart from my case,  but i'm
> really unsure what to do in this case? Move patch parts around?
> But moving them in random order will break assembler, already learned
> this ....
> 
> Also I'm lost in gas/config/tc-mips.c Original patch was designed for
> (K, m, n) [i don't know what they mean .. some form of internal
> markers?]

Exactly that, their meaning is documented in include/opcode/mips.h

> and i changed it for (+K, +m,+n). But i'm really lost in those big
> switches there. Right now my new code disabled, looking at old patch
> i must add some logic before yet another switch. Where is the best
> place for discussing this - here or on gcc mail list?

Use binutils@sourceware.org


Thiemo