aarch64: Fix integral_init4/8h_neon

The stride is the number of uint16_t elements and thus needs
to be shifted.

This issue had slipped unnoticed since checkasm didn't actually
verify the output of these functions.

diff --git a/common/aarch64/mc-a.S b/common/aarch64/mc-a.S
index a4488a417285869d28dbc17a37ffd9f9fdf036a1..b6b588eed95bb44e8549ab01a0d2a5cf5e20b472 100644 (file)
--- a/common/aarch64/mc-a.S
+++ b/common/aarch64/mc-a.S
@@ -1403,7 +1403,7 @@ endfunc
 .endm
 
 function integral_init4h_neon, export=1
-    sub         x3,  x0,  x2
+    sub         x3,  x0,  x2, lsl #1
     ld1        {v6.8b,v7.8b}, [x1], #16
 1:
     subs        x2,  x2,  #16
@@ -1438,7 +1438,7 @@ endfunc
 .endm
 
 function integral_init8h_neon, export=1
-    sub         x3,  x0,  x2
+    sub         x3,  x0,  x2, lsl #1
     ld1        {v16.8b,v17.8b}, [x1], #16
 1:
     subs        x2,  x2,  #16