shithub: openh264

Info  •  Files  •  Log  •  Branches

Don't clobber q4-q7 in WelsIntra16x16Combined3Satd_neon

This is similar to what is done in other neon functions. This
function was missed since it isn't covered by the current
set of unittests.

--- a/codec/encoder/core/arm/intra_pred_sad_3_opt_neon.S

+++ b/codec/encoder/core/arm/intra_pred_sad_3_opt_neon.S

@@ -154,6 +154,7 @@

 WELS_ASM_FUNC_BEGIN WelsIntra16x16Combined3Satd_neon

     stmdb sp!, {r4-r7, lr}

+    vpush {q4-q7}

 	//Get the top line data to 'q15'(16 bytes)

 	sub  r7, r0, r1

@@ -258,8 +259,8 @@

     HDM_TRANSFORM_4X4_L0 d9, d13, d23, d19, d27, d15, d16, d17, d14

 	//Get the data from stack

-	ldr r5, [sp, #20] //the addr of Best_mode

-	ldr r6, [sp, #24] //the value of i_lambda

+	ldr r5, [sp, #84] //the addr of Best_mode

+	ldr r6, [sp, #88] //the value of i_lambda

 	//vadd.u16   d24, d25

 	vrshr.u16  d15, #1

@@ -291,6 +292,7 @@

     str r4, [r5]

+    vpop {q4-q7}

     ldmia sp!, {r4-r7, lr}

 WELS_ASM_FUNC_END