amd/common: Fix various non-critical integer overflows

[mesa.git] / src / util / u_math.h

diff --git a/src/util/u_math.h b/src/util/u_math.h
index a672486f02f41fd1f35fc845d683b056f76a54a8..b725489b520b1564677bef1cdec80cbdd97cb475 100644 (file)
--- a/src/util/u_math.h
+++ b/src/util/u_math.h
@@ -185,6 +185,23 @@ util_fast_pow(float x, float y)
 static inline int
 util_ifloor(float f)
 {
+#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__)
+   /*
+    * IEEE floor for computers that round to nearest or even.
+    * 'f' must be between -4194304 and 4194303.
+    * This floor operation is done by "(iround(f + .5) + iround(f - .5)) >> 1",
+    * but uses some IEEE specific tricks for better speed.
+    * Contributed by Josh Vanderhoof
+    */
+   int ai, bi;
+   double af, bf;
+   af = (3 << 22) + 0.5 + (double)f;
+   bf = (3 << 22) + 0.5 - (double)f;
+   /* GCC generates an extra fstp/fld without this. */
+   __asm__ ("fstps %0" : "=m" (ai) : "t" (af) : "st");
+   __asm__ ("fstps %0" : "=m" (bi) : "t" (bf) : "st");
+   return (ai - bi) >> 1;
+#else
    int ai, bi;
    double af, bf;
    union fi u;
@@ -193,6 +210,7 @@ util_ifloor(float f)
    u.f = (float) af;  ai = u.i;
    u.f = (float) bf;  bi = u.i;
    return (ai - bi) >> 1;
+#endif
 }
 
 
@@ -649,6 +667,9 @@ util_memcpy_cpu_to_le32(void * restrict dest, const void * restrict src, size_t
  */
 #define CLAMP( X, MIN, MAX )  ( (X)>(MIN) ? ((X)>(MAX) ? (MAX) : (X)) : (MIN) )
 
+/* Syntax sugar occuring frequently in graphics code */
+#define SATURATE( X ) CLAMP(X, 0.0f, 1.0f)
+
 #define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
 #define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
 
@@ -783,7 +804,25 @@ util_fpstate_set_denorms_to_zero(unsigned current_fpstate);
 void
 util_fpstate_set(unsigned fpstate);
 
-
+/**
+ * For indexed draw calls, return true if the vertex count to be drawn is
+ * much lower than the vertex count that has to be uploaded, meaning
+ * that the driver should flatten indices instead of trying to upload
+ * a too big range.
+ *
+ * This is used by vertex upload code in u_vbuf and glthread.
+ */
+static inline bool
+util_is_vbo_upload_ratio_too_large(unsigned draw_vertex_count,
+                                   unsigned upload_vertex_count)
+{
+   if (draw_vertex_count > 1024)
+      return upload_vertex_count > draw_vertex_count * 4;
+   else if (draw_vertex_count > 32)
+      return upload_vertex_count > draw_vertex_count * 8;
+   else
+      return upload_vertex_count > draw_vertex_count * 16;
+}
 
 #ifdef __cplusplus
 }