RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-25 05:27:43 +00:00
Code Issues Activity Actions

LibGL: Improve glFrustum precision and error handling

Browse source 
Do not convert to float too early. Additionally, handle some error
cases for the input parameters.
This commit is contained in:
Jelle Raaijmakers 2022-02-21 01:31:24 +01:00 • committed by Linus Groh
parent d92047c74d
commit 5cf967e4f2
1 changed files with 8 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

15
Userland/Libraries/LibGL/SoftwareGLContext.cpp
View file

@ -382,17 +382,18 @@ void SoftwareGLContext::gl_frustum(GLdouble left, GLdouble right, GLdouble botto
APPEND_TO_CALL_LIST_AND_RETURN_IF_NEEDED(gl_frustum, left, right, bottom, top, near_val, far_val);
RETURN_WITH_ERROR_IF(m_in_draw_state, GL_INVALID_OPERATION);
RETURN_WITH_ERROR_IF(near_val < 0 || far_val < 0, GL_INVALID_VALUE);
RETURN_WITH_ERROR_IF(left == right || bottom == top || near_val == far_val, GL_INVALID_VALUE);
// Let's do some math!
// FIXME: Are we losing too much precision by doing this?
float a = static_cast<float>((right + left) / (right - left));
float b = static_cast<float>((top + bottom) / (top - bottom));
float c = static_cast<float>(-((far_val + near_val) / (far_val - near_val)));
float d = static_cast<float>(-((2 * (far_val * near_val)) / (far_val - near_val)));
auto a = static_cast<float>((right + left) / (right - left));
auto b = static_cast<float>((top + bottom) / (top - bottom));
auto c = static_cast<float>(-((far_val + near_val) / (far_val - near_val)));
auto d = static_cast<float>(-((2 * far_val * near_val) / (far_val - near_val)));
FloatMatrix4x4 frustum {
((2 * (float)near_val) / ((float)right - (float)left)), 0, a, 0,
0, ((2 * (float)near_val) / ((float)top - (float)bottom)), b, 0,
static_cast<float>(2 * near_val / (right - left)), 0, a, 0,
0, static_cast<float>(2 * near_val / (top - bottom)), b, 0,
0, 0, c, d,
0, 0, -1, 0
};