sha/asm/sha1-x86_64.pl: fix crash in SHAEXT code on Windows.

RT#4530

(Imported from upstream's 7123aa81e9fb19afb11fdf3850662c5f7ff1f19c.)

We've yet to enable this code, but this confirms that we do indeed need
to get our future all-variants stuff working on Windows as well as
Linux and find an AVX2-capable CI setup on each.

The crash here is caused by some win64-only code using %rax as a frame
pointer (perlasm injects a mov rax,rsp in the prologue of every win64
function).

Change-Id: Ifbe59ceb6ae29266d9cf8a461920344a32b6e555
Reviewed-on: https://boringssl-review.googlesource.com/10366
CQ-Verified: CQ bot account: commit-bot@chromium.org <commit-bot@chromium.org>
Commit-Queue: David Benjamin <davidben@google.com>
Commit-Queue: Adam Langley <agl@google.com>
Reviewed-by: Adam Langley <agl@google.com>

crypto/sha/asm/sha1-x86_64.pl

diff --git a/crypto/sha/asm/sha1-x86_64.pl b/crypto/sha/asm/sha1-x86_64.pl
index cdc7285..ff960bb 100644
--- a/crypto/sha/asm/sha1-x86_64.pl
+++ b/crypto/sha/asm/sha1-x86_64.pl
@@ -364,9 +364,9 @@
 .align	16
 .Loop_shaext:
 	dec		$num
-	lea		0x40($inp),%rax		# next input block
+	lea		0x40($inp),%r8		# next input block
 	paddd		@MSG[0],$E
-	cmovne		%rax,$inp
+	cmovne		%r8,$inp
 	movdqa		$ABCD,$ABCD_SAVE	# offload $ABCD
 ___
 for($i=0;$i<20-4;$i+=2) {