GAS vs. GCC

[Mark Mitchell]

I'm catching up on the last few days of email.

I see (by my own experiment) that GCC now causes GAS to crash on the
mainline, although, thankfully, not on the release branch.

Would someone mind explaining what happenned here?

In particular,

  - What's going wrong?

  - Was this a latent bug in GAS, or is GCC now generating some
    opcode or other construct that it did not before?

In addition, we must either:

  - Cause `configure' to abort in the event that the version of GAS
    is not up to date.

  - Or, configure away the use of the new feature if GAS is not
    up to date.

It's not kind to developers to make them get all the way to a V3 build
before the assembler aborts.  

Would whoever checked in the change that causes GAS to abort please
implement one of the configury changes above?

Thanks in advance,

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com