multi-thread fix: lock each entry points with a static var

we compare the current VP8GetCPUInfo pointer to the last used.
This is less code overall and each implementation is still
testable separately (by just changing VP8GetCPUInfo, but not
a separate threads!)

(cherry picked from commit a437694a17e2ee4c68bf0abba8e45e0dfd2a8e27)

Conflicts:
	src/dsp/alpha_processing.c
	src/dsp/argb.c
	src/dsp/dec.c
	src/dsp/enc.c
	src/dsp/lossless.c
	src/dsp/upsampling.c
	src/dsp/yuv.c

Change-Id: Ia13fa8ffc4561a884508f6ab71ed0d1b9f1ce59b

6 files changed, 30 insertions, 1 deletions

diff --git a/src/dsp/alpha_processing.c b/src/dsp/alpha_processing.c
index d0f7a6cc..ac88c60d 100644
--- a/src/dsp/alpha_processing.c
+++ b/src/dsp/alpha_processing.c
@@ -311,7 +311,11 @@ int (*WebPExtractAlpha)(const uint8_t*, int, int, int, uint8_t*, int);
 
 extern void WebPInitAlphaProcessingSSE2(void);
 
+static volatile VP8CPUInfo last_cpuinfo_used = (VP8CPUInfo)&last_cpuinfo_used;
+
 void WebPInitAlphaProcessing(void) {
+  if (last_cpuinfo_used == VP8GetCPUInfo) return;
+
   WebPMultARGBRow = MultARGBRow;
   WebPMultRow = MultRow;
   WebPApplyAlphaMultiply = ApplyAlphaMultiply;
@@ -326,4 +330,5 @@ void WebPInitAlphaProcessing(void) {
     }
 #endif
   }
+  last_cpuinfo_used = VP8GetCPUInfo;
 }
diff --git a/src/dsp/dec.c b/src/dsp/dec.c
index 65a2a885..35f3cf2a 100644
--- a/src/dsp/dec.c
+++ b/src/dsp/dec.c
@@ -688,7 +688,11 @@ extern void VP8DspInitSSE2(void);
 extern void VP8DspInitNEON(void);
 extern void VP8DspInitMIPS32(void);
 
+static volatile VP8CPUInfo last_cpuinfo_used = (VP8CPUInfo)&last_cpuinfo_used;
+
 void VP8DspInit(void) {
+  if (last_cpuinfo_used == VP8GetCPUInfo) return;
+
   VP8InitClipTables();
 
   VP8TransformWHT = TransformWHT;
@@ -727,5 +731,5 @@ void VP8DspInit(void) {
     }
 #endif
   }
+  last_cpuinfo_used = VP8GetCPUInfo;
 }
-
diff --git a/src/dsp/enc.c b/src/dsp/enc.c
index e4ea8cb8..a5fcf5dd 100644
--- a/src/dsp/enc.c
+++ b/src/dsp/enc.c
@@ -692,7 +692,11 @@ extern void VP8EncDspInitAVX2(void);
 extern void VP8EncDspInitNEON(void);
 extern void VP8EncDspInitMIPS32(void);
 
+static volatile VP8CPUInfo last_cpuinfo_used = (VP8CPUInfo)&last_cpuinfo_used;
+
 void VP8EncDspInit(void) {
+  if (last_cpuinfo_used == VP8GetCPUInfo) return;
+
   VP8DspInit();  // common inverse transforms
   InitTables();
 
@@ -737,5 +741,6 @@ void VP8EncDspInit(void) {
     }
 #endif
   }
+  last_cpuinfo_used = VP8GetCPUInfo;
 }
 
diff --git a/src/dsp/lossless.c b/src/dsp/lossless.c
index a1bf3584..7c5784db 100644
--- a/src/dsp/lossless.c
+++ b/src/dsp/lossless.c
@@ -1590,7 +1590,11 @@ extern void VP8LDspInitSSE2(void);
 extern void VP8LDspInitNEON(void);
 extern void VP8LDspInitMIPS32(void);
 
+static volatile VP8CPUInfo last_cpuinfo_used = (VP8CPUInfo)&last_cpuinfo_used;
+
 void VP8LDspInit(void) {
+  if (last_cpuinfo_used == VP8GetCPUInfo) return;
+
   memcpy(VP8LPredictors, kPredictorsC, sizeof(VP8LPredictors));
 
   VP8LSubtractGreenFromBlueAndRed = VP8LSubtractGreenFromBlueAndRed_C;
@@ -1634,6 +1638,7 @@ void VP8LDspInit(void) {
     }
 #endif
   }
+  last_cpuinfo_used = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------
diff --git a/src/dsp/upsampling.c b/src/dsp/upsampling.c
index 2b1656bf..c787782d 100644
--- a/src/dsp/upsampling.c
+++ b/src/dsp/upsampling.c
@@ -189,7 +189,11 @@ const WebPYUV444Converter WebPYUV444Converters[MODE_LAST] = {
 extern void WebPInitUpsamplersSSE2(void);
 extern void WebPInitUpsamplersNEON(void);
 
+static volatile VP8CPUInfo last_cpuinfo_used2 = (VP8CPUInfo)&last_cpuinfo_used2;
+
 void WebPInitUpsamplers(void) {
+  if (last_cpuinfo_used2 == VP8GetCPUInfo) return;
+
 #ifdef FANCY_UPSAMPLING
   WebPUpsamplers[MODE_RGB]       = UpsampleRgbLinePair;
   WebPUpsamplers[MODE_RGBA]      = UpsampleRgbaLinePair;
@@ -217,6 +221,7 @@ void WebPInitUpsamplers(void) {
 #endif
   }
 #endif  // FANCY_UPSAMPLING
+  last_cpuinfo_used2 = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------
diff --git a/src/dsp/yuv.c b/src/dsp/yuv.c
index d7cb4ebc..8213074d 100644
--- a/src/dsp/yuv.c
+++ b/src/dsp/yuv.c
@@ -123,7 +123,11 @@ WebPSamplerRowFunc WebPSamplers[MODE_LAST];
 extern void WebPInitSamplersSSE2(void);
 extern void WebPInitSamplersMIPS32(void);
 
+static volatile VP8CPUInfo last_cpuinfo_used = (VP8CPUInfo)&last_cpuinfo_used;
+
 void WebPInitSamplers(void) {
+  if (last_cpuinfo_used == VP8GetCPUInfo) return;
+
   WebPSamplers[MODE_RGB]       = YuvToRgbRow;
   WebPSamplers[MODE_RGBA]      = YuvToRgbaRow;
   WebPSamplers[MODE_BGR]       = YuvToBgrRow;
@@ -149,6 +153,7 @@ void WebPInitSamplers(void) {
     }
 #endif  // WEBP_USE_MIPS32
   }
+  last_cpuinfo_used = VP8GetCPUInfo;
 }
 
 //-----------------------------------------------------------------------------