From fd3bb67b964d360d01642da4af20bcaece52efbc Mon Sep 17 00:00:00 2001 From: Mark Wooding Date: Wed, 18 May 2016 10:29:03 +0100 Subject: [PATCH] symm/{salsa20,chacha}-x86ish-sse2.S: Use numeric labels for internal loops. --- symm/chacha-x86ish-sse2.S | 4 ++-- symm/salsa20-x86ish-sse2.S | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/symm/chacha-x86ish-sse2.S b/symm/chacha-x86ish-sse2.S index f36bf90f..86889968 100644 --- a/symm/chacha-x86ish-sse2.S +++ b/symm/chacha-x86ish-sse2.S @@ -131,7 +131,7 @@ FUNC(chacha_core_x86ish_sse2) 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. @@ -228,7 +228,7 @@ loop: // Decrement the loop counter and see if we should go round again. sub NR, 2 - ja loop + ja 0b // Almost there. Firstly, the feedforward addition. paddd xmm0, SAVE0 diff --git a/symm/salsa20-x86ish-sse2.S b/symm/salsa20-x86ish-sse2.S index b3ebf90c..d74836b3 100644 --- a/symm/salsa20-x86ish-sse2.S +++ b/symm/salsa20-x86ish-sse2.S @@ -152,7 +152,7 @@ FUNC(salsa20_core_x86ish_sse2) 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 -- 2.11.0