problems with g10 0.2.19 on sun-sparc-solaris2.5.1

[Tom Spindler]

Well, and with egcs as well. The first problem is a misspelling, I believe:

diff -u -r -x .o ../gnupg-0.2.19/mpi/sparc32/mpih-rshift.S ./gnupg-0.2.19/mpi/sparc32/mpih-rshift.S
--- ../gnupg-0.2.19/mpi/sparc32/mpih-rshift.S	Mon May 18 07:33:20 1998
+++ ./gnupg-0.2.19/mpi/sparc32/mpih-rshift.S	Tue Jun  9 11:26:22 1998
@@ -29,7 +29,7 @@
 
 	.text
 	.align	4
-	.global C_SYMBOL_NAME(mpohelp_rshift)
+	.global C_SYMBOL_NAME(mpihelp_rshift)
 C_SYMBOL_NAME(mpihelp_rshift):
 	ld	[%o1],%g2	! load first limb
 	sub	%g0,%o3,%o5	! negate shift count

The second is that egcs-1.0.3 doesn't like the definition of variables
in blocks; I believe it only wants them at the top of the function,
before any instructions are generated.

in particular, rand-dummy.c has code that looks like

  #if HAVE_GETHRTIME
    {
        hrtime_t tv;
        tv = gethrtime(void);
        add_randomness( &tv, sizeof(tv), 1 );
    }

And egcs complains mightily about this. If you either put the hrtime_t tv;
definition at the top of the function, or remove the braces, the compiler
doesn't complain.