diff options
| author | Denise Cheng <dcheng@codeaurora.org> | 2010-10-07 14:40:05 -0400 |
|---|---|---|
| committer | Denise Cheng <dcheng@codeaurora.org> | 2010-10-07 14:40:05 -0400 |
| commit | 339e4fa918e0dfd5a11c78d0253fc3ceae7d7e1a (patch) | |
| tree | ebca0c82dc8f8d054f6df21aa66bebff907771d3 | |
| parent | 14585365d5a0e16cbcc7ab3877b3fbfe94ec78ef (diff) | |
| download | android_external_jpeg-froyo.tar.gz android_external_jpeg-froyo.tar.bz2 android_external_jpeg-froyo.zip | |
jpeg: Replacing VeNum YCC to RGB color conversion routines.froyo-stablefroyo
* New routines use the same algorithm as in the original IJG libjpeg
* Enable VeNum YCC to RGB565 color conversion
Change-Id: I8408949ee9668ea59376fc4a4d02156d875f9c7b
| -rw-r--r-- | Android.mk | 4 | ||||
| -rw-r--r-- | asm/armv7/jdcolor-armv7.S | 307 | ||||
| -rw-r--r-- | jdcolor.c | 6 | ||||
| -rw-r--r-- | jdmerge.c | 11 |
4 files changed, 161 insertions, 167 deletions
@@ -39,10 +39,6 @@ LOCAL_SRC_FILES += jidctvenum.c LOCAL_SRC_FILES += asm/armv7/jdcolor-armv7.S LOCAL_SRC_FILES += asm/armv7/jdidct-armv7.S LOCAL_CFLAGS += -DANDROID_JPEG_USE_VENUM -# Disable VeNum YCC to RGB565 color conversion by default because it does not -# work well with other android libraries (i.e. libFFTEm) due to different color -# conversion matrix used. -LOCAL_CFLAGS += -DANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565 else # ANDROID_JPEG_USE_VENUM, false LOCAL_SRC_FILES += jidctint.c jidctfst.S jidctred.c endif # ANDROID_JPEG_USE_VENUM diff --git a/asm/armv7/jdcolor-armv7.S b/asm/armv7/jdcolor-armv7.S index c4202d7..e106330 100644 --- a/asm/armv7/jdcolor-armv7.S +++ b/asm/armv7/jdcolor-armv7.S @@ -37,8 +37,8 @@ * - yyvup2rgb565_venum * - yvup2bgr888_venum * - yyvup2bgr888_venum -* - yvup2bgra8888_venum -* - yyvup2bgra8888_venum +* - yvup2abgr8888_venum +* - yyvup2abgr8888_venum * *-------------------------------------------------------------------------- */ @@ -58,33 +58,30 @@ p_rgb .req r3 p_bgr .req r3 length .req r12 -#if !defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) .global yvup2rgb565_venum .global yyvup2rgb565_venum -#endif .global yvup2bgr888_venum .global yyvup2bgr888_venum .global yvup2abgr8888_venum .global yyvup2abgr8888_venum @ coefficients in color conversion matrix multiplication -.equ COEFF_Y, 298 @ contribution of Y -.equ COEFF_V_RED, 409 @ contribution of V for red -.equ COEFF_U_GREEN, -100 @ contribution of U for green -.equ COEFF_V_GREEN, -208 @ contribution of V for green -.equ COEFF_U_BLUE, 516 @ contribution of U for blue +.equ COEFF_Y, 256 @ contribution of Y +.equ COEFF_V_RED, 359 @ contribution of V for red +.equ COEFF_U_GREEN, -88 @ contribution of U for green +.equ COEFF_V_GREEN, -183 @ contribution of V for green +.equ COEFF_U_BLUE, 454 @ contribution of U for blue @ Clamping constants 0x0 and 0xFF .equ COEFF_0, 0 .equ COEFF_255, 255 @ Bias coefficients for red, green and blue -.equ COEFF_BIAS_R, -56992 @ Red bias = -298*16 - 409*128 + 128 -.equ COEFF_BIAS_G, 34784 @ Green bias = -298*16 + 308*128 + 128 -.equ COEFF_BIAS_B, -70688 @ Blue bias = -298*16 - 516*128 + 128 +.equ COEFF_BIAS_R, -45824 @ Red bias = -359*128 + 128 +.equ COEFF_BIAS_G, 34816 @ Green bias = (88+183)*128 + 128 +.equ COEFF_BIAS_B, -57984 @ Blue bias = -454*128 + 128 -#if !defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) /*-------------------------------------------------------------------------- * FUNCTION : yvup2rgb565_venum *-------------------------------------------------------------------------- @@ -140,8 +137,8 @@ yvup2rgb565_venum: ADR R12, constants - VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 409 | -100 | -208 | 516 | 298 | 0 | 255 | 0 - VLD1.S32 {D30, D31}, [R12] @ Q15 : -56992 | 34784 | -70688 | X + VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 359 | -88 | -183 | 454 | 256 | 0 | 255 | 0 + VLD1.S32 {D30, D31}, [R12] @ Q15 : -45824 | 34816 | -57984 | X /*------------------------------------------------------------------------- * Load the 5th parameter via stack @@ -164,8 +161,8 @@ yvup2rgb565_venum: /*------------------------------------------------------------------------- * Read bias * ------------------------------------------------------------------------ */ - VDUP.S32 Q0, D30[0] @ Q0: -56992 | -56992 | -56992 | -56992 - VDUP.S32 Q1, D30[1] @ Q1: 34784 | 34784 | 34784 | 34784 + VDUP.S32 Q0, D30[0] @ Q0: -45824 | -45824 | -45824 | -45824 + VDUP.S32 Q1, D30[1] @ Q1: 34816 | 34816 | 34816 | 34816 VDUP.S32 Q2, D31[0] @ Q2: -70688 | -70688 | -70688 | -70688 @@ -190,24 +187,24 @@ loop_yvup2rgb565: * D20, D21: V0 V1 V2 V3 V4 V5 V6 V7 * D22, D23: U0 U1 U2 U3 U4 U5 U6 U7 * ------------------------------------------------------------------------ */ - VMOVL.U16.U8 Q9, D12 - VMOVL.U16.U8 Q10, D14 - VMOVL.U16.U8 Q11, D15 + VMOVL.U8 Q9, D12 + VMOVL.U8 Q10, D14 + VMOVL.U8 Q11, D15 /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q12, D20, D6[0] @ Q12: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q13, D22, D6[1] @ Q13: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q13, D20, D6[2] @ Q13: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q14, D22, D6[3] @ Q14: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q12, D20, D6[0] @ Q12: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q13, D22, D6[1] @ Q13: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q13, D20, D6[2] @ Q13: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q14, D22, D6[3] @ Q14: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q12, Q0 @ Q12 add Red bias -56992 - VADD.S32 Q13, Q1 @ Q13 add Green bias 34784 - VADD.S32 Q14, Q2 @ Q14 add Blue bias -70688 + VADD.S32 Q12, Q0 @ Q12 add Red bias -45824 + VADD.S32 Q13, Q1 @ Q13 add Green bias 34816 + VADD.S32 Q14, Q2 @ Q14 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -230,17 +227,17 @@ loop_yvup2rgb565: /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q12, D21, D6[0] @ Q12: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q13, D23, D6[1] @ Q13: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q13, D21, D6[2] @ Q13: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q14, D23, D6[3] @ Q14: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q12, D21, D6[0] @ Q12: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q13, D23, D6[1] @ Q13: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q13, D21, D6[2] @ Q13: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q14, D23, D6[3] @ Q14: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q12, Q0 @ Q12 add Red bias -56992 - VADD.S32 Q13, Q1 @ Q13 add Green bias 34784 - VADD.S32 Q14, Q2 @ Q14 add Blue bias -70688 + VADD.S32 Q12, Q0 @ Q12 add Red bias -45824 + VADD.S32 Q13, Q1 @ Q13 add Green bias 34816 + VADD.S32 Q14, Q2 @ Q14 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -388,10 +385,10 @@ yyvup2rgb565_venum: ADR R12, constants - VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 409 | -100 | -208 | 516 | 298 | 0 | 255 | 0 - VLD1.S32 {D30, D31}, [R12] @ Q15 : -56992 | 34784 | -70688 | X + VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 359 | -88 | -183 | 454 | 256 | 0 | 255 | 0 + VLD1.S32 {D30, D31}, [R12] @ Q15 : -45824 | 34816 | -57984 | X - /*------------------------------------------------------------------------- + /*------------------------------------------------------------------------- * Load the 5th parameter via stack * R0 ~ R3 are used to pass the first 4 parameters, the 5th and above * parameters are passed via stack @@ -412,8 +409,8 @@ yyvup2rgb565_venum: /*------------------------------------------------------------------------- * Read bias * ------------------------------------------------------------------------ */ - VDUP.S32 Q0, D30[0] @ Q0: -56992 | -56992 | -56992 | -56992 - VDUP.S32 Q1, D30[1] @ Q1: 34784 | 34784 | 34784 | 34784 + VDUP.S32 Q0, D30[0] @ Q0: -45824 | -45824 | -45824 | -45824 + VDUP.S32 Q1, D30[1] @ Q1: 34816 | 34816 | 34816 | 34816 VDUP.S32 Q2, D31[0] @ Q2: -70688 | -70688 | -70688 | -70688 @@ -439,25 +436,25 @@ loop_yyvup2rgb565: * D28, D29: V0 V1 V2 V3 V4 V5 V6 V7 * D30, D31: U0 U1 U2 U3 U4 U5 U6 U7 * ------------------------------------------------------------------------ */ - VMOVL.U16.U8 Q12, D12 - VMOVL.U16.U8 Q13, D13 - VMOVL.U16.U8 Q14, D14 - VMOVL.U16.U8 Q15, D15 + VMOVL.U8 Q12, D12 + VMOVL.U8 Q13, D13 + VMOVL.U8 Q14, D14 + VMOVL.U8 Q15, D15 /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q6, D28, D6[0] @ Q6: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q7, D30, D6[1] @ Q7: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q7, D28, D6[2] @ q7: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q8, D30, D6[3] @ q8: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q6, D28, D6[0] @ Q6: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q7, D30, D6[1] @ Q7: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q7, D28, D6[2] @ q7: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q8, D30, D6[3] @ q8: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q6, Q0 @ Q6 add Red bias -56992 - VADD.S32 Q7, Q1 @ Q7 add Green bias 34784 - VADD.S32 Q8, Q2 @ Q8 add Blue bias -70688 + VADD.S32 Q6, Q0 @ Q6 add Red bias -45824 + VADD.S32 Q7, Q1 @ Q7 add Green bias 34816 + VADD.S32 Q8, Q2 @ Q8 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -508,9 +505,10 @@ loop_yyvup2rgb565: * D22: 3 bits of Green + 5 bits of Blue * D23: 5 bits of Red + 3 bits of Green * ------------------------------------------------------------------------ */ - VSRI.8 D23, D22, #5 @ right shift G by 5 and insert to R + VSRI.8 D23, D22, #5 @ right shift G by 5 and insert to R VSHL.U8 D22, D22, #3 @ left shift G by 3 VSRI.8 D22, D21, #3 @ right shift B by 3 and insert to G + SUBS length, length, #8 @ check if the length is less than 8 BMI trailing_yyvup2rgb565 @ jump to trailing processing if remaining length is less than 8 @@ -528,17 +526,17 @@ loop_yyvup2rgb565: /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q6, D29, D6[0] @ Q6: 409*(V4,V5,V6,V7) Red - VMULL.S16 Q7, D31, D6[1] @ Q7: -100*(U4,U5,U6,U7) Green - VMLAL.S16 Q7, D29, D6[2] @ Q7: -100*(U4,U5,U6,U7) - 208*(V4,V5,V6,V7) - VMULL.S16 Q8, D31, D6[3] @ Q8: 516*(U4,U5,U6,U7) Blue + VMULL.S16 Q6, D29, D6[0] @ Q6: 359*(V4,V5,V6,V7) Red + VMULL.S16 Q7, D31, D6[1] @ Q7: -88*(U4,U5,U6,U7) Green + VMLAL.S16 Q7, D29, D6[2] @ Q7: -88*(U4,U5,U6,U7) - 183*(V4,V5,V6,V7) + VMULL.S16 Q8, D31, D6[3] @ Q8: 454*(U4,U5,U6,U7) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q6, Q0 @ Q6 add Red bias -56992 - VADD.S32 Q7, Q1 @ Q7 add Green bias 34784 - VADD.S32 Q8, Q2 @ Q8 add Blue bias -70688 + VADD.S32 Q6, Q0 @ Q6 add Red bias -45824 + VADD.S32 Q7, Q1 @ Q7 add Green bias 34816 + VADD.S32 Q8, Q2 @ Q8 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -569,6 +567,7 @@ loop_yyvup2rgb565: VSHRN.S32 D20, Q8, #8 @ D20: B8 B10 B12 B14 in 16-bit Q0 format VSHRN.S32 D21, Q11, #8 @ D21: B9 B11 B13 B15 in 16-bit Q0 format VZIP.16 D20, D21 @ Q10: B8 B9 B10 B11 B12 B13 B14 B15 + /*------------------------------------------------------------------------- * Clamp the value to be within [0~255] * ------------------------------------------------------------------------ */ @@ -588,7 +587,7 @@ loop_yyvup2rgb565: * D22: 3 bits of Green + 5 bits of Blue * D23: 5 bits of Red + 3 bits of Green * ------------------------------------------------------------------------ */ - VSRI.8 D23, D22, #5 @ right shift G by 5 and insert to R + VSRI.8 D23, D22, #5 @ right shift G by 5 and insert to R VSHL.U8 D22, D22, #3 @ left shift G by 3 VSRI.8 D22, D21, #3 @ right shift B by 3 and insert to G @@ -644,12 +643,11 @@ end_yyvup2rgb565: LDMFD SP!, {PC} @ end of yyvup2rgb565 -#endif constants: - .hword (COEFF_V_RED), (COEFF_U_GREEN), (COEFF_V_GREEN), (COEFF_U_BLUE) @ 409 | -100 | -208 | 516 - .hword (COEFF_Y), (COEFF_0), (COEFF_255) , (COEFF_0) @ 298 | 0 | 255 | 0 - .word (COEFF_BIAS_R), (COEFF_BIAS_G), (COEFF_BIAS_B) @ -56992 | 34784 | -70688 | X + .hword (COEFF_V_RED), (COEFF_U_GREEN), (COEFF_V_GREEN), (COEFF_U_BLUE) @ 359 | -88 | -183 | 454 + .hword (COEFF_Y), (COEFF_0), (COEFF_255) , (COEFF_0) @ 256 | 0 | 255 | 0 + .word (COEFF_BIAS_R), (COEFF_BIAS_G), (COEFF_BIAS_B) @ -45824 | 34816 | -57984 | X /*-------------------------------------------------------------------------- * FUNCTION : yvup2bgr888_venum @@ -707,8 +705,8 @@ yvup2bgr888_venum: ADR R12, constants - VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 409 | -100 | -208 | 516 | 298 | 0 | 255 | 0 - VLD1.S32 {D30, D31}, [R12] @ Q15 : -56992 | 34784 | -70688 | X + VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 359 | -88 | -183 | 454 | 256 | 0 | 255 | 0 + VLD1.S32 {D30, D31}, [R12] @ Q15 : -45824 | 34816 | -57984 | X /*------------------------------------------------------------------------- * Load the 5th parameter via stack @@ -731,9 +729,9 @@ yvup2bgr888_venum: /*------------------------------------------------------------------------- * Read bias * ------------------------------------------------------------------------ */ - VDUP.S32 Q0, D30[0] @ Q0: -56992 | -56992 | -56992 | -56992 - VDUP.S32 Q1, D30[1] @ Q1: 34784 | 34784 | 34784 | 34784 - VDUP.S32 Q2, D31[0] @ Q2: -70688 | -70688 | -70688 | -70688 + VDUP.S32 Q0, D30[0] @ Q0: -45824 | -45824 | -45824 | -45824 + VDUP.S32 Q1, D30[1] @ Q1: 34816 | 34816 | 34816 | 34816 + VDUP.S32 Q2, D31[0] @ Q2: -57984 | -57984 | -57984 | -57984 /*------------------------------------------------------------------------- @@ -757,24 +755,24 @@ loop_yvup2bgr888: * D20, D21: V0 V1 V2 V3 V4 V5 V6 V7 * D22, D23: U0 U1 U2 U3 U4 U5 U6 U7 * ------------------------------------------------------------------------ */ - VMOVL.U16.U8 Q9, D12 - VMOVL.U16.U8 Q10, D14 - VMOVL.U16.U8 Q11, D15 + VMOVL.U8 Q9, D12 + VMOVL.U8 Q10, D14 + VMOVL.U8 Q11, D15 /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q12, D20, D6[0] @ Q12: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q13, D22, D6[1] @ Q13: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q13, D20, D6[2] @ Q13: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q14, D22, D6[3] @ Q14: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q12, D20, D6[0] @ Q12: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q13, D22, D6[1] @ Q13: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q13, D20, D6[2] @ Q13: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q14, D22, D6[3] @ Q14: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q12, Q0 @ Q12 add Red bias -56992 - VADD.S32 Q13, Q1 @ Q13 add Green bias 34784 - VADD.S32 Q14, Q2 @ Q14 add Blue bias -70688 + VADD.S32 Q12, Q0 @ Q12 add Red bias -45824 + VADD.S32 Q13, Q1 @ Q13 add Green bias 34816 + VADD.S32 Q14, Q2 @ Q14 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -797,17 +795,17 @@ loop_yvup2bgr888: /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q12, D21, D6[0] @ Q12: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q13, D23, D6[1] @ Q13: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q13, D21, D6[2] @ Q13: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q14, D23, D6[3] @ Q14: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q12, D21, D6[0] @ Q12: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q13, D23, D6[1] @ Q13: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q13, D21, D6[2] @ Q13: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q14, D23, D6[3] @ Q14: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q12, Q0 @ Q12 add Red bias -56992 - VADD.S32 Q13, Q1 @ Q13 add Green bias 34784 - VADD.S32 Q14, Q2 @ Q14 add Blue bias -70688 + VADD.S32 Q12, Q0 @ Q12 add Red bias -45824 + VADD.S32 Q13, Q1 @ Q13 add Green bias 34816 + VADD.S32 Q14, Q2 @ Q14 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -884,10 +882,11 @@ trailing_yvup2bgr888: BEQ end_yvup2bgr888 @ done if 0 pixel left SUBS length, length, #1 @ update length counter - VST3.U8 {D26[6], D27[6], D28[6]}, [p_bgr]! @ store one more pixel + VST3.U8 {D26[6], D27[6], D28[6]}, [p_bgr]! @ store one more pixel end_yvup2bgr888: LDMFD SP!, {PC} + @ end of yvup2bgr888 @@ -946,8 +945,8 @@ yyvup2bgr888_venum: ADR R12, constants - VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 409 | -100 | -208 | 516 | 298 | 0 | 255 | 0 - VLD1.S32 {D30, D31}, [R12] @ Q15 : -56992 | 34784 | -70688 | X + VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 359 | -88 | -183 | 454 | 256 | 0 | 255 | 0 + VLD1.S32 {D30, D31}, [R12] @ Q15 : -45824 | 34816 | -57984 | X /*------------------------------------------------------------------------- * Load the 5th parameter via stack @@ -970,8 +969,8 @@ yyvup2bgr888_venum: /*------------------------------------------------------------------------- * Read bias * ------------------------------------------------------------------------ */ - VDUP.S32 Q0, D30[0] @ Q0: -56992 | -56992 | -56992 | -56992 - VDUP.S32 Q1, D30[1] @ Q1: 34784 | 34784 | 34784 | 34784 + VDUP.S32 Q0, D30[0] @ Q0: -45824 | -45824 | -45824 | -45824 + VDUP.S32 Q1, D30[1] @ Q1: 34816 | 34816 | 34816 | 34816 VDUP.S32 Q2, D31[0] @ Q2: -70688 | -70688 | -70688 | -70688 @@ -997,25 +996,25 @@ loop_yyvup2bgr888: * D28, D29: V0 V1 V2 V3 V4 V5 V6 V7 * D30, D31: U0 U1 U2 U3 U4 U5 U6 U7 * ------------------------------------------------------------------------ */ - VMOVL.U16.U8 Q12, D12 - VMOVL.U16.U8 Q13, D13 - VMOVL.U16.U8 Q14, D14 - VMOVL.U16.U8 Q15, D15 + VMOVL.U8 Q12, D12 + VMOVL.U8 Q13, D13 + VMOVL.U8 Q14, D14 + VMOVL.U8 Q15, D15 /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q6, D28, D6[0] @ Q6: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q7, D30, D6[1] @ Q7: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q7, D28, D6[2] @ q7: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q8, D30, D6[3] @ q8: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q6, D28, D6[0] @ Q6: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q7, D30, D6[1] @ Q7: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q7, D28, D6[2] @ q7: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q8, D30, D6[3] @ q8: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q6, Q0 @ Q6 add Red bias -56992 - VADD.S32 Q7, Q1 @ Q7 add Green bias 34784 - VADD.S32 Q8, Q2 @ Q8 add Blue bias -70688 + VADD.S32 Q6, Q0 @ Q6 add Red bias -45824 + VADD.S32 Q7, Q1 @ Q7 add Green bias 34816 + VADD.S32 Q8, Q2 @ Q8 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -1078,16 +1077,16 @@ loop_yyvup2bgr888: /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q6, D29, D6[0] @ Q6: 409*(V4,V5,V6,V7) Red - VMULL.S16 Q7, D31, D6[1] @ Q7: -100*(U4,U5,U6,U7) Green - VMLAL.S16 Q7, D29, D6[2] @ Q7: -100*(U4,U5,U6,U7) - 208*(V4,V5,V6,V7) - VMULL.S16 Q8, D31, D6[3] @ Q8: 516*(U4,U5,U6,U7) Blue + VMULL.S16 Q6, D29, D6[0] @ Q6: 359*(V4,V5,V6,V7) Red + VMULL.S16 Q7, D31, D6[1] @ Q7: -88*(U4,U5,U6,U7) Green + VMLAL.S16 Q7, D29, D6[2] @ Q7: -88*(U4,U5,U6,U7) - 183*(V4,V5,V6,V7) + VMULL.S16 Q8, D31, D6[3] @ Q8: 454*(U4,U5,U6,U7) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q6, Q0 @ Q6 add Red bias -56992 - VADD.S32 Q7, Q1 @ Q7 add Green bias 34784 + VADD.S32 Q6, Q0 @ Q6 add Red bias -45824 + VADD.S32 Q7, Q1 @ Q7 add Green bias 34816 VADD.S32 Q8, Q2 @ Q8 add Blue bias -70688 /*------------------------------------------------------------------------- @@ -1119,6 +1118,7 @@ loop_yyvup2bgr888: VSHRN.S32 D20, Q8, #8 @ D20: B8 B10 B12 B14 in 16-bit Q0 format VSHRN.S32 D21, Q11, #8 @ D21: B9 B11 B13 B15 in 16-bit Q0 format VZIP.16 D20, D21 @ Q10: B8 B9 B10 B11 B12 B13 B14 B15 + /*------------------------------------------------------------------------- * Clamp the value to be within [0~255] * ------------------------------------------------------------------------ */ @@ -1232,7 +1232,7 @@ end_yyvup2bgr888: * NOTES : *-------------------------------------------------------------------------- */ -.type yvup2abgr8888_venum, %function +.type yvup2abgr8888_venum, %function yvup2abgr8888_venum: /*------------------------------------------------------------------------- * Store stack registers @@ -1243,8 +1243,8 @@ yvup2abgr8888_venum: ADR R12, constants - VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 409 | -100 | -208 | 516 | 298 | 0 | 255 | 0 - VLD1.S32 {D30, D31}, [R12] @ Q15 : -56992 | 34784 | -70688 | X + VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 359 | -88 | -183 | 454 | 256 | 0 | 255 | 0 + VLD1.S32 {D30, D31}, [R12] @ Q15 : -45824 | 34816 | -57984 | X /*------------------------------------------------------------------------- * Load the 5th parameter via stack @@ -1267,8 +1267,8 @@ yvup2abgr8888_venum: /*------------------------------------------------------------------------- * Read bias * ------------------------------------------------------------------------ */ - VDUP.S32 Q0, D30[0] @ Q0: -56992 | -56992 | -56992 | -56992 - VDUP.S32 Q1, D30[1] @ Q1: 34784 | 34784 | 34784 | 34784 + VDUP.S32 Q0, D30[0] @ Q0: -45824 | -45824 | -45824 | -45824 + VDUP.S32 Q1, D30[1] @ Q1: 34816 | 34816 | 34816 | 34816 VDUP.S32 Q2, D31[0] @ Q2: -70688 | -70688 | -70688 | -70688 @@ -1293,24 +1293,24 @@ loop_yvup2abgr: * D20, D21: V0 V1 V2 V3 V4 V5 V6 V7 * D22, D23: U0 U1 U2 U3 U4 U5 U6 U7 * ------------------------------------------------------------------------ */ - VMOVL.U16.U8 Q9, D12 - VMOVL.U16.U8 Q10, D14 - VMOVL.U16.U8 Q11, D15 + VMOVL.U8 Q9, D12 + VMOVL.U8 Q10, D14 + VMOVL.U8 Q11, D15 /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q12, D20, D6[0] @ Q12: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q13, D22, D6[1] @ Q13: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q13, D20, D6[2] @ Q13: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q14, D22, D6[3] @ Q14: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q12, D20, D6[0] @ Q12: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q13, D22, D6[1] @ Q13: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q13, D20, D6[2] @ Q13: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q14, D22, D6[3] @ Q14: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q12, Q0 @ Q12 add Red bias -56992 - VADD.S32 Q13, Q1 @ Q13 add Green bias 34784 - VADD.S32 Q14, Q2 @ Q14 add Blue bias -70688 + VADD.S32 Q12, Q0 @ Q12 add Red bias -45824 + VADD.S32 Q13, Q1 @ Q13 add Green bias 34816 + VADD.S32 Q14, Q2 @ Q14 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -1333,17 +1333,17 @@ loop_yvup2abgr: /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q12, D21, D6[0] @ Q12: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q13, D23, D6[1] @ Q13: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q13, D21, D6[2] @ Q13: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q14, D23, D6[3] @ Q14: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q12, D21, D6[0] @ Q12: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q13, D23, D6[1] @ Q13: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q13, D21, D6[2] @ Q13: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q14, D23, D6[3] @ Q14: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q12, Q0 @ Q12 add Red bias -56992 - VADD.S32 Q13, Q1 @ Q13 add Green bias 34784 - VADD.S32 Q14, Q2 @ Q14 add Blue bias -70688 + VADD.S32 Q12, Q0 @ Q12 add Red bias -45824 + VADD.S32 Q13, Q1 @ Q13 add Green bias 34816 + VADD.S32 Q14, Q2 @ Q14 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -1425,7 +1425,7 @@ trailing_yvup2abgr: BEQ end_yvup2abgr @ done if 0 pixel left SUBS length, length, #1 @ update length counter - VST4.U8 {D26[6], D27[6], D28[6], D29[6]}, [p_bgr]! @ store one more pixel + VST4.U8 {D26[6], D27[6], D28[6], D29[6]}, [p_bgr]! @ store one more pixel end_yvup2abgr: LDMFD SP!, {PC} @@ -1475,7 +1475,7 @@ end_yvup2abgr: * NOTES : *-------------------------------------------------------------------------- */ -.type yyvup2abgr8888_venum, %function +.type yyvup2abgr8888_venum, %function yyvup2abgr8888_venum: /*------------------------------------------------------------------------- * Store stack registers @@ -1486,8 +1486,8 @@ yyvup2abgr8888_venum: ADR R12, constants - VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 409 | -100 | -208 | 516 | 298 | 0 | 255 | 0 - VLD1.S32 {D30, D31}, [R12] @ Q15 : -56992 | 34784 | -70688 | X + VLD1.S16 {D6, D7}, [R12]! @ D6, D7: 359 | -88 | -183 | 454 | 256 | 0 | 255 | 0 + VLD1.S32 {D30, D31}, [R12] @ Q15 : -45824 | 34816 | -57984 | X /*------------------------------------------------------------------------- * Load the 5th parameter via stack @@ -1510,8 +1510,8 @@ yyvup2abgr8888_venum: /*------------------------------------------------------------------------- * Read bias * ------------------------------------------------------------------------ */ - VDUP.S32 Q0, D30[0] @ Q0: -56992 | -56992 | -56992 | -56992 - VDUP.S32 Q1, D30[1] @ Q1: 34784 | 34784 | 34784 | 34784 + VDUP.S32 Q0, D30[0] @ Q0: -45824 | -45824 | -45824 | -45824 + VDUP.S32 Q1, D30[1] @ Q1: 34816 | 34816 | 34816 | 34816 VDUP.S32 Q2, D31[0] @ Q2: -70688 | -70688 | -70688 | -70688 @@ -1537,25 +1537,25 @@ loop_yyvup2abgr: * D28, D29: V0 V1 V2 V3 V4 V5 V6 V7 * D30, D31: U0 U1 U2 U3 U4 U5 U6 U7 * ------------------------------------------------------------------------ */ - VMOVL.U16.U8 Q12, D12 - VMOVL.U16.U8 Q13, D13 - VMOVL.U16.U8 Q14, D14 - VMOVL.U16.U8 Q15, D15 + VMOVL.U8 Q12, D12 + VMOVL.U8 Q13, D13 + VMOVL.U8 Q14, D14 + VMOVL.U8 Q15, D15 /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q6, D28, D6[0] @ Q6: 409*(V0,V1,V2,V3) Red - VMULL.S16 Q7, D30, D6[1] @ Q7: -100*(U0,U1,U2,U3) Green - VMLAL.S16 Q7, D28, D6[2] @ q7: -100*(U0,U1,U2,U3) - 208*(V0,V1,V2,V3) - VMULL.S16 Q8, D30, D6[3] @ q8: 516*(U0,U1,U2,U3) Blue + VMULL.S16 Q6, D28, D6[0] @ Q6: 359*(V0,V1,V2,V3) Red + VMULL.S16 Q7, D30, D6[1] @ Q7: -88*(U0,U1,U2,U3) Green + VMLAL.S16 Q7, D28, D6[2] @ Q7: -88*(U0,U1,U2,U3) - 183*(V0,V1,V2,V3) + VMULL.S16 Q8, D30, D6[3] @ Q8: 454*(U0,U1,U2,U3) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q6, Q0 @ Q6 add Red bias -56992 - VADD.S32 Q7, Q1 @ Q7 add Green bias 34784 - VADD.S32 Q8, Q2 @ Q8 add Blue bias -70688 + VADD.S32 Q6, Q0 @ Q6 add Red bias -45824 + VADD.S32 Q7, Q1 @ Q7 add Green bias 34816 + VADD.S32 Q8, Q2 @ Q8 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -1624,17 +1624,17 @@ loop_yyvup2abgr: /*------------------------------------------------------------------------- * Multiply contribution from chrominance, results are in 32-bit * ------------------------------------------------------------------------ */ - VMULL.S16 Q6, D29, D6[0] @ Q6: 409*(V4,V5,V6,V7) Red - VMULL.S16 Q7, D31, D6[1] @ Q7: -100*(U4,U5,U6,U7) Green - VMLAL.S16 Q7, D29, D6[2] @ Q7: -100*(U4,U5,U6,U7) - 208*(V4,V5,V6,V7) - VMULL.S16 Q8, D31, D6[3] @ Q8: 516*(U4,U5,U6,U7) Blue + VMULL.S16 Q6, D29, D6[0] @ Q6: 359*(V4,V5,V6,V7) Red + VMULL.S16 Q7, D31, D6[1] @ Q7: -88*(U4,U5,U6,U7) Green + VMLAL.S16 Q7, D29, D6[2] @ Q7: -88*(U4,U5,U6,U7) - 183*(V4,V5,V6,V7) + VMULL.S16 Q8, D31, D6[3] @ Q8: 454*(U4,U5,U6,U7) Blue /*------------------------------------------------------------------------- * Add bias * ------------------------------------------------------------------------ */ - VADD.S32 Q6, Q0 @ Q6 add Red bias -56992 - VADD.S32 Q7, Q1 @ Q7 add Green bias 34784 - VADD.S32 Q8, Q2 @ Q8 add Blue bias -70688 + VADD.S32 Q6, Q0 @ Q6 add Red bias -45824 + VADD.S32 Q7, Q1 @ Q7 add Green bias 34816 + VADD.S32 Q8, Q2 @ Q8 add Blue bias -57984 /*------------------------------------------------------------------------- * Calculate Red, Green, Blue @@ -1665,6 +1665,7 @@ loop_yyvup2abgr: VSHRN.S32 D20, Q8, #8 @ D20: B8 B10 B12 B14 in 16-bit Q0 format VSHRN.S32 D21, Q11, #8 @ D21: B9 B11 B13 B15 in 16-bit Q0 format VZIP.16 D20, D21 @ Q10: B8 B9 B10 B11 B12 B13 B14 B15 + /*------------------------------------------------------------------------- * Clamp the value to be within [0~255] * ------------------------------------------------------------------------ */ @@ -277,7 +277,7 @@ METHODDEF(void) ycc_rgb_565_convert (j_decompress_ptr cinfo, JSAMPIMAGE input_buf, JDIMENSION input_row, JSAMPARRAY output_buf, int num_rows) -#if defined(ANDROID_JPEG_USE_VENUM) && !defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if defined(ANDROID_JPEG_USE_VENUM) /* * Converts YCC->RGB565 using VeNum instructions. */ @@ -916,7 +916,7 @@ jinit_color_deconverter (j_decompress_ptr cinfo) if (cinfo->dither_mode == JDITHER_NONE) { if (cinfo->jpeg_color_space == JCS_YCbCr) { cconvert->pub.color_convert = ycc_rgb_565_convert; -#if !defined(ANDROID_JPEG_USE_VENUM) || defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if !defined(ANDROID_JPEG_USE_VENUM) build_ycc_rgb_table(cinfo); #endif /* ANDROID_JPEG_USE_VENUM */ } else if (cinfo->jpeg_color_space == JCS_GRAYSCALE) { @@ -928,7 +928,7 @@ jinit_color_deconverter (j_decompress_ptr cinfo) } else { /* only ordered dither is supported */ if (cinfo->jpeg_color_space == JCS_YCbCr) { -#if defined(ANDROID_JPEG_USE_VENUM) && !defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if defined(ANDROID_JPEG_USE_VENUM) /* Use VeNum routine even if dithering option is selected. */ cconvert->pub.color_convert = ycc_rgb_565_convert; #else @@ -347,7 +347,7 @@ METHODDEF(void) h2v1_merged_upsample_565 (j_decompress_ptr cinfo, JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr, JSAMPARRAY output_buf) -#if defined(ANDROID_JPEG_USE_VENUM) && !defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if defined(ANDROID_JPEG_USE_VENUM) { my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample; JSAMPROW inptr0, inptr1, inptr2; @@ -628,7 +628,7 @@ METHODDEF(void) h2v2_merged_upsample_565 (j_decompress_ptr cinfo, JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr, JSAMPARRAY output_buf) -#if defined(ANDROID_JPEG_USE_VENUM) && !defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if defined(ANDROID_JPEG_USE_VENUM) { my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample; JSAMPROW outptr0, outptr1; @@ -854,7 +854,7 @@ jinit_merged_upsampler (j_decompress_ptr cinfo) upsample->upmethod = h2v2_merged_upsample; #ifdef ANDROID_RGB if (cinfo->out_color_space == JCS_RGB_565) { -#if !defined(ANDROID_JPEG_USE_VENUM) || defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if !defined(ANDROID_JPEG_USE_VENUM) if (cinfo->dither_mode != JDITHER_NONE) { upsample->upmethod = h2v2_merged_upsample_565D; } else @@ -891,10 +891,7 @@ jinit_merged_upsampler (j_decompress_ptr cinfo) upsample->spare_row = NULL; } -#ifdef ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565 - if (cinfo->out_color_space == JCS_RGB_565) -#endif -#if !defined(ANDROID_JPEG_USE_VENUM) || defined(ANDROID_JPEG_DISABLE_VENUM_YCC_RGB_565) +#if !defined(ANDROID_JPEG_USE_VENUM) build_ycc_rgb_table(cinfo); #endif } |