This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Remove mips aout, coff, and pe support


Alan, Nick --

> >  I'm fine with letting `mips-ecoff' and `mipsel-ecoff' targets go, but I 
> > thought we'd keep the `ecoff-bigmips' and `ecoff-littlemips' BFDs as 
> > secondaries for 'mips*-linux*' targets so that `objcopy' and `objdump' 
> > remain functional with these BFDs.  Especially the latter tool seems 
> > useful to examine various kinds of bootable images that are still around 
> > (people have been used `elf2ecoff' rather than `objcopy' to create such 
> > images).  I think having to keep an old binary version around just for 
> > this purpose would be unnecessary pain.
> 
> I was hoping we could kill off ecoff.  Oh well, if you think we still
> need mips_ecoff_be_vec and mips_ecoff_le_vec, please revert those
> parts of my patch that you think are wrong.  (I guess that means
> bringing mips-coff.c back to life and reverting some config.bfd and
> Makefile.am changes.)

 I won't be able to do it tonight, and I'm rather loaded with other stuff 
the following days, but I'll see what I can do.

 I think we want to bring coff-mips.c back as it was, and then adjust 
anything required by `mips*-linux*' targets for the `ecoff-bigmips' and 
`ecoff-littlemips' BFDs to work.
  
 Ideally we'd rebase the tree with:

$ git rebase -i 3596d8ceb2cd^

(editing 3596d8ceb2cd appropriately) so that things look like the bits 
have never been removed in the first place and `git blame', etc. works as 
expected, but I fear that would cause some people trouble.  And for that 
reason I'm not even sure if our policy allows for rebases as we've been 
using our repository pretty linearly so far, as with older revision 
control systems (Nick, would you please clarify that?).

 If it does, then I'll try to do the rebase in the next couple of days, 
hopefully tomorrow.  The sooner the better, as the risk of the need to 
resolve conflicts is going to grow with time.  Otherwise it'll have to be 
a pretty straightforward revert via an ordinary commit.

 Once that part is complete, then we can look into removing those parts of 
coff-mips.c that are not needed anymore (if any are still left, i.e. all 
the linker bits should have gone with `mips-ecoff' and `mipsel-ecoff' 
target support in LD).

  Maciej


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]