X-Git-Url: https://git.distorted.org.uk/~mdw/catacomb/blobdiff_plain/0f23f75ff53acadf80e9d3dfd2dfd14cb526074f..729a52ce3a79c1edd4d72334660a5dd80998b3ef:/symm/salsa20-x86ish-sse2.S

diff --git a/symm/salsa20-x86ish-sse2.S b/symm/salsa20-x86ish-sse2.S
index a168d79a..47401b7a 100644
--- a/symm/salsa20-x86ish-sse2.S
+++ b/symm/salsa20-x86ish-sse2.S
@@ -42,7 +42,7 @@
 /// Main code.
 
 	.arch pentium4
-	.section .text
+	.text
 
 FUNC(salsa20_core_x86ish_sse2)
 
@@ -146,13 +146,13 @@ FUNC(salsa20_core_x86ish_sse2)
 	movdqu	xmm2, [IN + 32]
 	movdqu	xmm3, [IN + 48]
 
-	## Take a copy for later.
+	// Take a copy for later.
 	movdqa	SAVE0, xmm0
 	movdqa	SAVE1, xmm1
 	movdqa	SAVE2, xmm2
 	movdqa	SAVE3, xmm3
 
-loop:
+0:
 	// Apply a column quarterround to each of the columns simultaneously.
 	// Alas, there doesn't seem to be a packed doubleword rotate, so we
 	// have to synthesize it.
@@ -256,7 +256,7 @@ loop:
 	// Decrement the loop counter and see if we should go round again.
 	// Later processors fuse this pair into a single uop.
 	sub	NR, 2
-	ja	loop
+	ja	0b
 
 	// Almost there.  Firstly, the feedforward addition, and then we have
 	// to write out the result.  Here we have to undo the permutation