ARM interpreter bug fixes

[Edward Nevill]

Hi folks,

Two bug fixes here,

1) irem of negative number returns positive.

This was introduced in my last checkin. The new dispatch code corrupted the
flags inc. the N flags so irem didn't handle -ve nos.

2) icedtea-debug build fails.

The ARM assembler still does not build for icedtea-debug, however at least
with this patch it will not fall over in the test_gamma, instead it builds
a C only vsn.

Changelog to follow

Regards,
Ed.

--- CUT HERE -------------------------------------------------------------
diff -ruNE orig/icedtea6/ports/hotspot/src/cpu/zero/vm/cppInterpreter_arm.S new/icedtea6/ports/hotspot/src/cpu/zero/vm/cppInterpreter_arm.S
--- orig/icedtea6/ports/hotspot/src/cpu/zero/vm/cppInterpreter_arm.S	2009-10-08 10:49:32.000000000 +0100
+++ new/icedtea6/ports/hotspot/src/cpu/zero/vm/cppInterpreter_arm.S	2009-10-08 11:13:38.000000000 +0100
@@ -1058,7 +1058,7 @@
 asm_generate_method_entry:
 	mov	r3, r0
 	mov	r0, #0
-
+#ifdef PRODUCT
 	adrl	ip, dispatch_init_adcon
 	ldm	ip, {r1, r2}
 	add	r1, r1, ip
@@ -1085,6 +1085,7 @@
 	adrcc	ip, asm_method_table
 	ldrcc	r0, [ip, r3, lsl #2]
 1:
+#endif // PRODUCT
 	bx	lr
 asm_method_table:
 	.word	normal_entry
@@ -2162,8 +2163,8 @@
         bne     .r_loop2
         movs    lr, lr, lsl #1
 	DISPATCH_NEXT
-	DISPATCH_NEXT
 	rsbmi	tmp1, tmp1, #0
+	DISPATCH_NEXT
 	PUSH	tmp1
 	DISPATCH_FINISH