As seen in libc, prevent gcc's memory usage from exploding during

optimizing by moving each round into a separate function on sparc64.

2 files changed, 87 insertions, 1 deletions

diff --git a/net/bind9/distinfo b/net/bind9/distinfo
index 0802dd08395..2351a8bcd62 100644
--- a/net/bind9/distinfo
+++ b/net/bind9/distinfo
@@ -1,6 +1,7 @@
-$NetBSD: distinfo,v 1.6 2002/05/04 14:56:23 taca Exp $
+$NetBSD: distinfo,v 1.7 2002/06/27 07:33:26 kleink Exp $
 
 SHA1 (bind/9.2.1/bind-9.2.1.tar.gz) = 6ea711ce3b786b3c5fab89273b4132061745c846
 Size (bind/9.2.1/bind-9.2.1.tar.gz) = 5021044 bytes
 SHA1 (patch-ab) = 1d3d82e3e01ac38e8a1b1d98e018849e8fe071bf
 SHA1 (patch-ac) = 1696adfab7392e0f150d34931b808e51ad017938
+SHA1 (patch-ai) = 916d2cdadd420b34e908f3c128e864bf271ebb6a
diff --git a/net/bind9/patches/patch-ai b/net/bind9/patches/patch-ai
new file mode 100644
index 00000000000..9fc071cd977
--- /dev/null
+++ b/net/bind9/patches/patch-ai
@@ -0,0 +1,85 @@
+$NetBSD: patch-ai,v 1.1 2002/06/27 07:33:27 kleink Exp $
+
+--- lib/isc/sha1.c.orig	Tue Jan  9 22:56:28 2001
++++ lib/isc/sha1.c	Thu Jun 27 08:52:59 2002
+@@ -85,6 +85,59 @@
+ 	unsigned int l[16];
+ } CHAR64LONG16;
+ 
++#ifdef __sparc_v9__
++void do_R01(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *);
++void do_R2(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *);
++void do_R3(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *);
++void do_R4(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *);
++
++#define nR0(v,w,x,y,z,i) R0(*v,*w,*x,*y,*z,i)
++#define nR1(v,w,x,y,z,i) R1(*v,*w,*x,*y,*z,i)
++#define nR2(v,w,x,y,z,i) R2(*v,*w,*x,*y,*z,i)
++#define nR3(v,w,x,y,z,i) R3(*v,*w,*x,*y,*z,i)
++#define nR4(v,w,x,y,z,i) R4(*v,*w,*x,*y,*z,i)
++
++void
++do_R01(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *block)
++{
++    nR0(a,b,c,d,e, 0); nR0(e,a,b,c,d, 1); nR0(d,e,a,b,c, 2); nR0(c,d,e,a,b, 3);
++    nR0(b,c,d,e,a, 4); nR0(a,b,c,d,e, 5); nR0(e,a,b,c,d, 6); nR0(d,e,a,b,c, 7);
++    nR0(c,d,e,a,b, 8); nR0(b,c,d,e,a, 9); nR0(a,b,c,d,e,10); nR0(e,a,b,c,d,11);
++    nR0(d,e,a,b,c,12); nR0(c,d,e,a,b,13); nR0(b,c,d,e,a,14); nR0(a,b,c,d,e,15);
++    nR1(e,a,b,c,d,16); nR1(d,e,a,b,c,17); nR1(c,d,e,a,b,18); nR1(b,c,d,e,a,19);
++}
++
++void
++do_R2(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *block)
++{
++    nR2(a,b,c,d,e,20); nR2(e,a,b,c,d,21); nR2(d,e,a,b,c,22); nR2(c,d,e,a,b,23);
++    nR2(b,c,d,e,a,24); nR2(a,b,c,d,e,25); nR2(e,a,b,c,d,26); nR2(d,e,a,b,c,27);
++    nR2(c,d,e,a,b,28); nR2(b,c,d,e,a,29); nR2(a,b,c,d,e,30); nR2(e,a,b,c,d,31);
++    nR2(d,e,a,b,c,32); nR2(c,d,e,a,b,33); nR2(b,c,d,e,a,34); nR2(a,b,c,d,e,35);
++    nR2(e,a,b,c,d,36); nR2(d,e,a,b,c,37); nR2(c,d,e,a,b,38); nR2(b,c,d,e,a,39);
++}
++
++void
++do_R3(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *block)
++{
++    nR3(a,b,c,d,e,40); nR3(e,a,b,c,d,41); nR3(d,e,a,b,c,42); nR3(c,d,e,a,b,43);
++    nR3(b,c,d,e,a,44); nR3(a,b,c,d,e,45); nR3(e,a,b,c,d,46); nR3(d,e,a,b,c,47);
++    nR3(c,d,e,a,b,48); nR3(b,c,d,e,a,49); nR3(a,b,c,d,e,50); nR3(e,a,b,c,d,51);
++    nR3(d,e,a,b,c,52); nR3(c,d,e,a,b,53); nR3(b,c,d,e,a,54); nR3(a,b,c,d,e,55);
++    nR3(e,a,b,c,d,56); nR3(d,e,a,b,c,57); nR3(c,d,e,a,b,58); nR3(b,c,d,e,a,59);
++}
++
++void
++do_R4(u_int32_t *a, u_int32_t *b, u_int32_t *c, u_int32_t *d, u_int32_t *e, CHAR64LONG16 *block)
++{
++    nR4(a,b,c,d,e,60); nR4(e,a,b,c,d,61); nR4(d,e,a,b,c,62); nR4(c,d,e,a,b,63);
++    nR4(b,c,d,e,a,64); nR4(a,b,c,d,e,65); nR4(e,a,b,c,d,66); nR4(d,e,a,b,c,67);
++    nR4(c,d,e,a,b,68); nR4(b,c,d,e,a,69); nR4(a,b,c,d,e,70); nR4(e,a,b,c,d,71);
++    nR4(d,e,a,b,c,72); nR4(c,d,e,a,b,73); nR4(b,c,d,e,a,74); nR4(a,b,c,d,e,75);
++    nR4(e,a,b,c,d,76); nR4(d,e,a,b,c,77); nR4(c,d,e,a,b,78); nR4(b,c,d,e,a,79);
++}
++#endif
++
+ /*
+  * Hash a single 512-bit block. This is the core of the algorithm.
+  */
+@@ -108,6 +161,12 @@
+ 	e = state[4];
+ 
+ 	/* 4 rounds of 20 operations each. Loop unrolled. */
++#ifdef __sparc_v9__
++	do_R01(&a, &b, &c, &d, &e, block);
++	do_R2(&a, &b, &c, &d, &e, block);
++	do_R3(&a, &b, &c, &d, &e, block);
++	do_R4(&a, &b, &c, &d, &e, block);
++#else
+ 	R0(a,b,c,d,e, 0); R0(e,a,b,c,d, 1); R0(d,e,a,b,c, 2); R0(c,d,e,a,b, 3);
+ 	R0(b,c,d,e,a, 4); R0(a,b,c,d,e, 5); R0(e,a,b,c,d, 6); R0(d,e,a,b,c, 7);
+ 	R0(c,d,e,a,b, 8); R0(b,c,d,e,a, 9); R0(a,b,c,d,e,10); R0(e,a,b,c,d,11);
+@@ -128,6 +187,7 @@
+ 	R4(c,d,e,a,b,68); R4(b,c,d,e,a,69); R4(a,b,c,d,e,70); R4(e,a,b,c,d,71);
+ 	R4(d,e,a,b,c,72); R4(c,d,e,a,b,73); R4(b,c,d,e,a,74); R4(a,b,c,d,e,75);
+ 	R4(e,a,b,c,d,76); R4(d,e,a,b,c,77); R4(c,d,e,a,b,78); R4(b,c,d,e,a,79);
++#endif
+ 
+ 	/* Add the working vars back into context.state[] */
+ 	state[0] += a;