[PATCH] [RFC] Fix bitfield behaviour

[Neil Booth]

Joseph S. Myers wrote:-

> On Mon, 14 Apr 2003, Christian Ehrhardt wrote:
> 
> > bit-fields currently have several issues including some where wrong
> > code is generated. Neil proposed a patch for these some time ago but
> > it had to be reverted. This problem is the reason for at least the following
> > PRs: 9620, 7696, 6515, 6346, 3326, 3325 and 2511. There are also XFAILing
> > testcases for this problem in the testsuite.
> > 
> > This patch fixes these PRs. Bootstrapped and regtested on x86 with
> > c, c++, objc, fortran and java and on sparc-sun-solaris2.9 with c and c++.
> 
> Your patch needs to remove all the XFAILs from Neil's testcases - if any
> still fail, that means the patch is incomplete.

I'm not sure I put all the testcases in.  If I did, great.  If not, they
should probably go in too.

I'm not convinced creating separate types for bitfield widths is ideal,
but it's better than what we have now.  Thanks Christian!

Neil.