RFC: weakref GCC attribute and .weakref assembly directive

[Nick Clifton]

Hi Alex,

> Ok to install?

Sorry no.  The code itself is fine, but the new weakref1.[sd] test is 
not.  It causes new failures for lots of different toolchains, 
including, but not limited to:

  arc-elf
  h8300-elf
  i386-coff
  i686-pc-cygwin
  ip2k-elf
  m32r-elf
  m68k-elf
  maxq-coff
  mcore-pe
  mingw32-pe
  mips-elf
  mn10300-elf
  powerpc-ibm-aix5.2.0
  sh-nto

I have only looked at a few logs so far, but I suspect that the problem 
is going to be the same with all of them - your weakref1.d file is too 
restrictive in the output it is checking for, and it does not make 
allowances for extra symbols that particular ports may insert into an 
object file.

If you can fix this testcase so that it does not introduce new failures, 
(or if you can show for specific toolchains that the failures are due to 
real, target-specific bugs) then you have my approval to check your 
patch in.

Cheers
   Nick