Do not use sha512-parisc for now, as it is subtly bugged - passes the sha

regress tests but causes tls ciphersuite using sha386 to fail; found the
hard way by henning@.
I can't see anything wrong in the generated assembly code yet, but building
a libcrypto with no assembler code but sha512_block_data_order() is enough
to trigger Henning's issue, so the bug lies there.
No ABI change; ok deraadt@

src/lib/libcrypto/crypto/arch/hppa/Makefile.inc

@@ -1,4 +1,4 @@
-# $OpenBSD: Makefile.inc,v 1.7 2014/11/17 20:31:21 miod Exp $
+# $OpenBSD: Makefile.inc,v 1.8 2015/03/05 20:35:28 miod Exp $
 
 # hppa-specific libcrypto build rules
 
@@ -30,10 +30,12 @@ SRCS+= rc4_enc.c rc4_skey.c
 # sha
 CFLAGS+= -DSHA1_ASM
 SSLASM+= sha sha1-parisc sha1-parisc
+.if 0	# sha512 subtly broken, sha256 probably as well
 CFLAGS+= -DSHA256_ASM
 SSLASM+= sha sha512-parisc sha256-parisc
 CFLAGS+= -DSHA512_ASM
 SSLASM+= sha sha512-parisc sha512-parisc
+.endif
 # whrlpool
 SRCS+= wp_block.c