Re: [RFA] New bitflags type and eflags on i386/x86-64

[Andrew Cagney <ac131313@ges.redhat.com>]

> On Thu, Aug 29, 2002 at 04:57:42PM +0200, Michal Ludvig wrote:
> 
>> Daniel Jacobowitz wrote:
> 
>> >Well, actually, I like it. 
> 
>> 
>> Wow! That's a surprise! :-)))
>> 
> 
>> >Some textual changes and comments:
> 
>> 
>> Thanks, I'm not a native speaker ;-)
>> 
> 
>> >>+  builtin_type_i386_eflags = 
>> >>+    init_flags_type (32 /* EFLAGS_LENGTH */, 
>> >>+    	       "__i386_eflags", (struct objfile *) NULL);
>> >>+  add_flag_name (builtin_type_i386_eflags, 0, "CF");
>> >>+  builtin_type_simd_mxcsr =
>> >>+    init_flags_type (32 /* EFLAGS_LENGTH */,
>> >>+    	       "__simd_mxcsr", (struct objfile *) NULL);
>> >>+  add_flag_name (builtin_type_simd_mxcsr, 0, "IE");
> 
>> > Do these really need to be in common code?  That's gross.  Yes, I know
>> > a whole lot of others are, but those are all floatformats or standard
>> > vectors.
>> > This should be in i386-tdep.c.
> 
>> 
>> I don't compile i386-tdep.c for x86-64. Should I duplicate the code for 
>> x86-64 or better leave it here to have it only once?
> 
> 
> Hmm, there's already some common files between the ports; there should
> be another, I think.  I don't want something as i386-specific as this
> anywhere near gdbtypes.c.  How about i386-common-tdep.c or something
> like that?

It shouldn't be in gdbtypes.c.

MarkK, I believe, is planning on merging much of the i386 and x86-64 
stuff so that a single GDB would support both -- that would remove the 
issue.

Is there any immediate technical problem stopping x86-64 linking in 
i386-tdep.c?  (Yes scary, multi-arch).

Andrew