Re: egcs on unixware 7 fixed...mostly.

Re: egcs on unixware 7 fixed...mostly.

[Kaveh R. Ghazi]

 > From: Robert Lipe <robertl@dgii.com>
 > 
 > I just committed the following patch which ALMOST allows a bootstrap using
 > the native compiler.   I haven't run the testsuite, nor am I likely to do
 > so in the immediate future.    Donations of shell accounts on systems that 
 > can build and test egcs in a reasonable time period are welcome. ;-) 
 > 
 > For reasons I don't yet understand, you must run configure, then edit 
 > gcc/auto-config.h.  Find the section that starts:
 > 
 > 	/* Define if your cpp understands the stringify operator.  */
 > and change the following line to:
 > 	#define HAVE_CPP_STRINGIFY 1


	It might be of use if you looked in gcc/config.log for the
place where configure ran the stringify test.  You should see in there
the code snippet configure tried to compile and the error message if
any which was produced by the compiler.

	This may provide a clue as to why the test failed when it
shouldn't have.

		--Kaveh
--
Kaveh R. Ghazi			Project Manager / Custom Development
ghazi@caip.rutgers.edu		Icon CMT Corp.

egcs on unixware 7 fixed...mostly.

[Robert Lipe]

UW7 (SVR5) changed slightly between the betas when I first added EGCS 
support and the final customer shipment version.   (The next compiler 
vendor to dork with alloca support needs to be smacked...) Now it's 
even more like SVR4.2 from egcs' view, but I left the separate cases 
in to ease future changes as necessary.

I just committed the following patch which ALMOST allows a bootstrap using
the native compiler.   I haven't run the testsuite, nor am I likely to do
so in the immediate future.    Donations of shell accounts on systems that 
can build and test egcs in a reasonable time period are welcome. ;-) 

For reasons I don't yet understand, you must run configure, then edit 
gcc/auto-config.h.  Find the section that starts:

	/* Define if your cpp understands the stringify operator.  */
and change the following line to:
	#define HAVE_CPP_STRINGIFY 1

If you don't do this, the stage 1 gengenrtl contains tokenization mush 
which makes all the functions end in *ENUM which makes things fail in 
really wierd ways.

Jack, perhaps you'd like to tear into why the configure script isn't 
playing nicely with  HAVE_CPP_STRINGIFY with UW7's compiler.   Look in
gcc/config.log for clues.   I've already dismantled my UW7 machine for
the weekend.

The symptoms were bizarre, but here was the fix:

        Make UnixWare 7 bootstrap support work with final shipping product.
        * configure.in: (i[34567]86-*-sysv5): append, not overwrite, xm_file.
        Pick up xm-siglist and xm-alloca.
        (xm_defines): Add USG so dbxout will build.
        * configure: Regenerate.

RCS file: /egcs/carton/cvsfiles/egcs/gcc/configure.in,v
retrieving revision 1.93
diff -u -p -r1.93 configure.in
--- configure.in        1998/04/28 05:57:03     1.93
+++ configure.in        1998/05/04 04:57:13
@@ -1013,7 +1013,8 @@ for machine in $build $host $target; do
                fi
                ;;
        i[[34567]]86-*-sysv5*)           # Intel x86 on System V Release 5
-               xm_file=i386/xm-sysv4.h
+               xm_file="xm-alloca.h xm-siglist.h ${xm_file}"
+               xm_defines="USG POSIX"
                tm_file=i386/sysv4.h
                if [[ x$stabs = xyes ]]
                then

-- 
Robert Lipe       http://www.dgii.com/people/robertl       robertl@dgii.com
              (WEB ADDRESS MAY BE TEMPORARILY UNAVAILABLE)