Fix an issue where a (cached) shader program could be used from multiple

[movit] / util.h

diff --git a/util.h b/util.h
index 1fa4e7823efd65666dc1d6ba0acaa659b28b994d..45fe6ba922b9728c5ac827ddb4b4d63a13a8ad11 100644 (file)
--- a/util.h
+++ b/util.h
@@ -8,6 +8,7 @@
 #include <stdlib.h>
 #include <Eigen/Core>
 #include <string>
+#include "defs.h"
 
 #define BUFFER_OFFSET(i) ((char *)NULL + (i))
 
@@ -38,17 +39,44 @@ void print_3x3_matrix(const Eigen::Matrix3d &m);
 // Output a GLSL 3x3 matrix declaration.
 std::string output_glsl_mat3(const std::string &name, const Eigen::Matrix3d &m);
 
+// Output GLSL scalar, 2-length and 3-length vector declarations.
+std::string output_glsl_float(const std::string &name, float x);
+std::string output_glsl_vec2(const std::string &name, float x, float y);
+std::string output_glsl_vec3(const std::string &name, float x, float y, float z);
+
 // Calculate a / b, rounding up. Does not handle overflow correctly.
 unsigned div_round_up(unsigned a, unsigned b);
 
+enum CombineRoundingBehavior {
+       COMBINE_DO_NOT_ROUND = 0,
+       COMBINE_ROUND_TO_FP16 = 1,
+};
+
 // Calculate where to sample, and with what weight, if one wants to use
-// the GPU's bilinear hardware to sample w1 * x[0] + w2 * x[1].
+// the GPU's bilinear hardware to sample w1 * x[pos1] + w2 * x[pos2],
+// where pos1 and pos2 must be normalized coordinates describing neighboring
+// texels in the mipmap level at which you sample. <num_subtexels> is the
+// number of distinct accessible subtexels in the given mipmap level,
+// calculated by num_texels / movit_texel_subpixel_precision. It is a float
+// for performance reasons, even though it is expected to be a whole number.
+// <inv_num_subtexels> is simply its inverse (1/x).
 //
 // Note that since the GPU might have limited precision in its linear
 // interpolation, the effective weights might be different from the ones you
 // asked for. sum_sq_error, if not NULL, will contain the sum of the
 // (estimated) squared errors of the two weights.
-void combine_two_samples(float w1, float w2, float *offset, float *total_weight, float *sum_sq_error);
+//
+// The answer, in "offset", comes as a normalized coordinate,
+// so if e.g. w2 = 0, you have simply offset = pos1. If <rounding_behavior>
+// is COMBINE_ROUND_TO_FP16, the coordinate is assumed to be stored as a
+// rounded fp16 value. This enables more precise calculation of total_weight
+// and sum_sq_error.
+template<class DestFloat>
+void combine_two_samples(float w1, float w2, float pos1, float pos2, float num_subtexels, float inv_num_subtexels,
+                         DestFloat *offset, DestFloat *total_weight, float *sum_sq_error);
+
+// Create a VBO with the given data. Returns the VBO number.
+GLuint generate_vbo(GLint size, GLenum type, GLsizeiptr data_size, const GLvoid *data);
 
 // Create a VBO with the given data, and bind it to the vertex attribute
 // with name <attribute_name>. Returns the VBO number.
@@ -66,12 +94,15 @@ unsigned next_power_of_two(unsigned v);
 // back into anything you intend to pass into OpenGL.
 void *get_gl_context_identifier();
 
+// Used in the check_error() macro, below.
+void abort_gl_error(GLenum err, const char *filename, int line) DOES_NOT_RETURN;
+
 }  // namespace movit
 
 #ifdef NDEBUG
 #define check_error()
 #else
-#define check_error() { int err = glGetError(); if (err != GL_NO_ERROR) { printf("GL error 0x%x at %s:%d\n", err, __FILE__, __LINE__); abort(); } }
+#define check_error() { GLenum err = glGetError(); if (err != GL_NO_ERROR) { movit::abort_gl_error(err, __FILE__, __LINE__); } }
 #endif
 
 // CHECK() is like assert(), but retains any side effects no matter the compilation mode.