Patch to prototype in insn-*.[ch]

Kaveh R. Ghazi ghazi@caip.rutgers.edu
Tue Aug 24 06:51:00 GMT 1999


 > From: Jeffrey A Law <law@cygnus.com>
 > 
 >   In message < 199908110133.VAA18890@caip.rutgers.edu >you write:
 >   > 	The following patch adds prototypes in gen*.c as well as in
 >   > their generated files insn-*.[ch].  There are three notable exceptions
 >   > which I didn't fix.
 >   > 
 >   > 1.  In insn.attr.h, I couldn't readily determine the right args for
 >   >     some of the functions since they are generated on the fly and I
 >   >     wasn't able to figure it out yet.  I put in placeholders for now.
 > 
 > I'd rather not have the placeholders.  They'll just make it more
 > difficult to track these down unless I'm missing something.


	Hmm, well I thought replacing () with PROTO(()) would be lots
easier to find when grepping for missing prototypes since () can
appear legitimately as a call to a function taking void arguments.

But I won't press this issue since I'd like to solve the underlying
problem instead. :-)



 > 
 >   > 2.  In insn-flags.h, there were a couple of prototypes wrapped by
 >   >     MD_CALL_PROTOTYPES.  See:
 >   >     http://egcs.cygnus.com/ml/gcc/1999-08/msg00316.html
 > 
 > Right.  Known issue.  Let's revisit after we get some of the other
 > backlogged stuff dealt with.


	Okay.



 >   > 3.  In insn-output.c, the `insn_gen_function' function ptr array
 >   >     defines itself with unknown args.  E.g. "()".  However the
 >   >     `insn_gen_function' prototype in expr.h uses PROTO ((rtx, ...)).
 >   >     This looked wrong since the functions assigned to the array are
 >   >     definitely not variable argument functions.  I seem to recall this
 >   >     might lose on system which pass args differently when using
 >   >     varargs like Irix6. (?)  Although I can't point to any specific
 >   >     failures, I left it alone for now.
 > 
 > Sounds reasonable.  We can revisit this too.
 > 
 > So, let's consider the patch without the placeholders approved, get it
 > installed and iterate on the remaining issues.
 > jeff


Done.  Thanks for reviewing the patch.

		--Kaveh
--
Kaveh R. Ghazi			Engagement Manager / Project Services
ghazi@caip.rutgers.edu		Qwest Internet Solutions


More information about the Gcc-patches mailing list