Remove threading dependencies with --disable-multithread

Avoid a pthreads dependency via pthread_once() when compiled with
--disable-multithread.

In addition, this synchronization is disabled for Win32 as well, even
though we can be sure that the required primatives exist, so that the
requirements on the application when built with --disable-multithread
are consistent across platforms.

Users using libvpx built with --disable-multithread in a multithreaded
context should provide their own synchronization. Updated the
documentation to vpx_codec_enc_init_ver() and vpx_codec_dec_init_ver()
to note this requirement. Moved the RTCD initialization call to match
this description, as previously it didn't happen until the first
frame.

Change-Id: Id576f6bce2758362188278d3085051c218a56d4a

build/make/rtcd.sh

@@ -211,6 +211,8 @@ common_top() {
 $(process_forward_decls)
 
 $(declare_function_pointers c $ALL_ARCHS)
+
+void ${symbol:-rtcd}(void);
 EOF
 }
 
@@ -231,11 +233,10 @@ x86() {
 
   cat <<EOF
 $(common_top)
-void ${symbol:-rtcd}(void);
 
 #ifdef RTCD_C
 #include "vpx_ports/x86.h"
-void ${symbol:-rtcd}(void)
+static void setup_rtcd_internal(void)
 {
     int flags = x86_simd_caps();
 
@@ -261,11 +262,9 @@ arm() {
 $(common_top)
 #include "vpx_config.h"
 
-void ${symbol:-rtcd}(void);
-
 #ifdef RTCD_C
 #include "vpx_ports/arm.h"
-void ${symbol:-rtcd}(void)
+static void setup_rtcd_internal(void)
 {
     int flags = arm_cpu_caps();
 
@@ -285,10 +284,8 @@ unoptimized() {
 $(common_top)
 #include "vpx_config.h"
 
-void ${symbol:-rtcd}(void);
-
 #ifdef RTCD_C
-void ${symbol:-rtcd}(void)
+static void setup_rtcd_internal(void)
 {
 $(set_function_pointers c)
 }

vp8/common/generic/systemdependent.c

@@ -83,57 +83,6 @@ static int get_cpu_count()
 #endif
 
 
-#if HAVE_PTHREAD_H
-#include <pthread.h>
-static void once(void (*func)(void))
-{
-    static pthread_once_t lock = PTHREAD_ONCE_INIT;
-    pthread_once(&lock, func);
-}
-
-
-#elif defined(_WIN32)
-static void once(void (*func)(void))
-{
-    /* Using a static initializer here rather than InitializeCriticalSection()
-     * since there's no race-free context in which to execute it. Protecting
-     * it with an atomic op like InterlockedCompareExchangePointer introduces
-     * an x86 dependency, and InitOnceExecuteOnce requires Vista.
-     */
-    static CRITICAL_SECTION lock = {(void *)-1, -1, 0, 0, 0, 0};
-    static int done;
-
-    EnterCriticalSection(&lock);
-
-    if (!done)
-    {
-        func();
-        done = 1;
-    }
-
-    LeaveCriticalSection(&lock);
-}
-
-
-#else
-/* No-op version that performs no synchronization. vpx_rtcd() is idempotent,
- * so as long as your platform provides atomic loads/stores of pointers
- * no synchronization is strictly necessary.
- */
-
-static void once(void (*func)(void))
-{
-    static int done;
-
-    if(!done)
-    {
-        func();
-        done = 1;
-    }
-}
-#endif
-
-
 void vp8_machine_specific_config(VP8_COMMON *ctx)
 {
 #if CONFIG_MULTITHREAD
@@ -145,6 +94,4 @@ void vp8_machine_specific_config(VP8_COMMON *ctx)
 #elif ARCH_X86 || ARCH_X86_64
     ctx->cpu_caps = x86_simd_caps();
 #endif
-
-    once(vpx_rtcd);
 }

vp8/common/rtcd.c

@@ -10,3 +10,60 @@
 #include "vpx_config.h"
 #define RTCD_C
 #include "vpx_rtcd.h"
+
+#if CONFIG_MULTITHREAD && HAVE_PTHREAD_H
+#include <pthread.h>
+static void once(void (*func)(void))
+{
+    static pthread_once_t lock = PTHREAD_ONCE_INIT;
+    pthread_once(&lock, func);
+}
+
+
+#elif CONFIG_MULTITHREAD && defined(_WIN32)
+#include <windows.h>
+static void once(void (*func)(void))
+{
+    /* Using a static initializer here rather than InitializeCriticalSection()
+     * since there's no race-free context in which to execute it. Protecting
+     * it with an atomic op like InterlockedCompareExchangePointer introduces
+     * an x86 dependency, and InitOnceExecuteOnce requires Vista.
+     */
+    static CRITICAL_SECTION lock = {(void *)-1, -1, 0, 0, 0, 0};
+    static int done;
+
+    EnterCriticalSection(&lock);
+
+    if (!done)
+    {
+        func();
+        done = 1;
+    }
+
+    LeaveCriticalSection(&lock);
+}
+
+
+#else
+/* No-op version that performs no synchronization. vpx_rtcd() is idempotent,
+ * so as long as your platform provides atomic loads/stores of pointers
+ * no synchronization is strictly necessary.
+ */
+
+static void once(void (*func)(void))
+{
+    static int done;
+
+    if(!done)
+    {
+        func();
+        done = 1;
+    }
+}
+#endif
+
+
+void vpx_rtcd()
+{
+    once(setup_rtcd_internal);
+}

vp8/vp8_cx_iface.c

@@ -9,6 +9,7 @@
  */
 
 
+#include "vpx_rtcd.h"
 #include "vpx/vpx_codec.h"
 #include "vpx/internal/vpx_codec_internal.h"
 #include "vpx_version.h"
@@ -568,6 +569,8 @@ static vpx_codec_err_t vp8e_init(vpx_codec_ctx_t *ctx,
 
     struct VP8_COMP *optr;
 
+    vpx_rtcd();
+
     if (!ctx->priv)
     {
         priv = calloc(1, sizeof(struct vpx_codec_alg_priv));

vp8/vp8_dx_iface.c

@@ -11,6 +11,7 @@
 
 #include <stdlib.h>
 #include <string.h>
+#include "vpx_rtcd.h"
 #include "vpx/vpx_decoder.h"
 #include "vpx/vp8dx.h"
 #include "vpx/internal/vpx_codec_internal.h"
@@ -187,6 +188,8 @@ static vpx_codec_err_t vp8_init(vpx_codec_ctx_t *ctx,
     vpx_codec_err_t        res = VPX_CODEC_OK;
     (void) data;
 
+    vpx_rtcd();
+
     /* This function only allocates space for the vpx_codec_alg_priv_t
      * structure. More memory may be required at the time the stream
      * information becomes known.

vpx/vpx_decoder.h

@@ -113,6 +113,10 @@ extern "C" {
      * function directly, to ensure that the ABI version number parameter
      * is properly initialized.
      *
+     * If the library was configured with --disable-multithread, this call
+     * is not thread safe and should be guarded with a lock if being used
+     * in a multithreaded context.
+     *
      * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
      * parameter), the storage pointed to by the cfg parameter must be
      * kept readable and stable until all memory maps have been set.

vpx/vpx_encoder.h

@@ -655,6 +655,10 @@ extern "C" {
      * function directly, to ensure that the ABI version number parameter
      * is properly initialized.
      *
+     * If the library was configured with --disable-multithread, this call
+     * is not thread safe and should be guarded with a lock if being used
+     * in a multithreaded context.
+     *
      * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags
      * parameter), the storage pointed to by the cfg parameter must be
      * kept readable and stable until all memory maps have been set.