Driver workarounds
List of OpenGL driver workarounds used by Magnum.
Driver workarounds used by a particular app are listed in the engine startup log such as here:
Renderer: GeForce GT 740M/PCIe/SSE2 by NVIDIA Corporation OpenGL version: 4.6.0 NVIDIA 390.25 Using optional features: GL_ARB_ES2_compatibility GL_ARB_direct_state_access ... Using driver workarounds: no-forward-compatible-core-context no-layout-qualifiers-on-old-glsl nv-zero-context-profile-mask ...
These identifiers correspond to the strings in the listing below. For debugging and diagnostic purposes it's possible to disable particular workarounds by passing their identifier string to the --magnum-disable-workarounds
command-line option. See Command-line options for more information.
#if defined(CORRADE_TARGET_ANDROID) && defined(MAGNUM_TARGET_GLES) /* Android Emulator can run with a SwiftShader GPU and thus needs some of the SwiftShader context creation workarounds. However, it's impossible to detect, as EGL_VERSION is always "1.4 Android META-EGL" and EGL_VENDOR always "Android". As there's nothing that would hint at SwiftShader being used, we conservatively assume every emulator can be a SwiftShader. But that's not easy either, the only vague hint that we're dealing with an emulator is the HOSTNAME env var, which is set to e.g. generic_x86, but to e.g. HWVTR on a device, so try that. */ "android-generic-hostname-might-be-swiftshader"_s, #endif #if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL) /* ANGLE's shader linker insists on returning a message consisting of a single newline on success, causing annoying noise in the console. Similar to "intel-windows-chatty-shader-compiler". Not present on WebGL, apparently browsers filter the noise out on their own. */ "angle-chatty-shader-compiler"_s, #endif #ifdef MAGNUM_TARGET_GLES /* ANGLE has a buggy bounds validation when drawing a mesh with instanced attributes added (with divisor set) using non-instanced glDraw*() APIs (in particular, when instance count is 1). This should be allowed according to the GL spec, which describes e.g. glDrawElements() as a special case of the "virtual" glDrawElementsOneInstance(). To work around the validation bug, gl*Draw*Instanced() is used unconditionally for all meshes that have instanced attributes. A test that triggers this issue is in MeshGLTest::drawInstancedAttributeSingleInstance(). */ "angle-instanced-attributes-always-draw-instanced"_s, #endif #if defined(CORRADE_TARGET_APPLE) && !defined(MAGNUM_TARGET_GLES) /* Calling glBufferData(), glMapBuffer(), glMapBufferRange() or glUnmapBuffer() on ANY buffer when ANY buffer is attached to a currently bound GL_TEXTURE_BUFFER crashes in gleUpdateCtxDirtyStateForBufStampChange deep inside Apple's GLengine. This can be worked around by unbinding all buffer textures before attempting to do such operation. A previous iteration of this workaround was to remember if a buffer is attached to a buffer texture, temporarily detaching it, calling given data-modifying API and then attaching it back with the same parameters. Unfortunately we also had to cache the internal texture format, as GL_TEXTURE_INTERNAL_FORMAT query is broken for buffer textures as well, returning always GL_R8 (the spec-mandated default). "Fortunately" macOS doesn't support ARB_texture_buffer_range so we didn't need to store also offset/size, only texture ID and its internal format, wasting 8 bytes per Buffer instance. HOWEVER, then we discovered this is not enough and also completely unrelated buffers suffer from the same crash. Fixing that properly in a similar manner would mean going through all live buffer texture instances and temporarily detaching their buffer when doing *any* data modification on *any* buffer, which would have extreme perf implications. So FORTUNATELY unbinding the textures worked around this too, and is a much nicer workaround after all. */ "apple-buffer-texture-unbind-on-buffer-modify"_s, #endif #if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_GLES2) /* Qualcomm Adreno drivers V@0615.65 (and possibly others) report __VERSION__ as 300 even for GLSL ES 3.10 and 3.20, breaking version-dependent shader code. */ "adreno-glsl-version-stuck-at-300"_s, #endif #if defined(CORRADE_TARGET_ANDROID) && defined(MAGNUM_TARGET_GLES) /* glBeginQuery() with GL_TIME_ELAPSED causes a GL_OUT_OF_MEMORY error when running from the Android shell (through ADB). No such error happens in an APK. Detecting using the $SHELL environment variable and disabling GL_EXT_disjoint_timer_query in that case. */ "arm-mali-timer-queries-oom-in-shell"_s, #endif #if !defined(MAGNUM_TARGET_GLES) && defined(CORRADE_TARGET_WINDOWS) /* ARB_direct_state_access on AMD Windows drivers has broken glTextureSubImage3D() / glGetTextureImage() on cube map textures (but not cube map arrays), always failing with erros like `glTextureSubImage3D has generated an error (GL_INVALID_VALUE)` if Z size or offset is larger than 1. Working around that by up/downloading slice-by-slice using non-DSA APIs, similarly to the svga3d-texture-upload-slice-by-slice workaround. The compressed image up/ download is affected as well, but we lack APIs for easy format-dependent slicing and offset calculation, so those currently still fail. */ "amd-windows-cubemap-image3d-slice-by-slice"_s, /* AMD Windows drivers have broken the DSA glCopyTextureSubImage3D(), returning GL_INVALID_VALUE. The non-DSA code path works. */ "amd-windows-broken-dsa-cubemap-copy"_s, /* AMD Windows glCreateQueries() works for everything except GL_TRANSFORM_FEEDBACK_[STREAM_]OVERFLOW, probably they just forgot to adapt it to this new GL 4.6 addition. Calling the non-DSA code path in that case instead. Similar to "mesa-dsa-createquery-except-pipeline-stats". */ "amd-windows-dsa-createquery-except-xfb-overflow"_s, #endif #if !defined(MAGNUM_TARGET_GLES) && !defined(CORRADE_TARGET_APPLE) /* Creating core context with specific version on AMD and NV proprietary drivers on Linux/Windows and Intel drivers on Windows causes the context to be forced to given version instead of selecting latest available version */ "no-forward-compatible-core-context"_s, #endif #if !defined(MAGNUM_TARGET_GLES) && defined(CORRADE_TARGET_WINDOWS) /* On Windows Intel drivers ARB_shading_language_420pack is exposed in GLSL even though the extension (e.g. binding keyword) is not supported */ "intel-windows-glsl-exposes-unsupported-shading-language-420pack"_s, #endif #ifndef MAGNUM_TARGET_GLES /* Mesa glCreateQueries() works for everything except stuff from GL 4.6 ARB_pipeline_statistics_query, probably just forgotten. Calling the non-DSA code path in that case instead. Similar to "amd-windows-dsa-createquery-except-xfb-overflow". */ "mesa-dsa-createquery-except-pipeline-stats"_s, /* Forward-compatible GL contexts on Mesa still report line width range as [1, 7], but setting wide line width fails. According to the specs the max value on forward compatible contexts should be 1.0, so patching it. */ "mesa-forward-compatible-line-width-range"_s, #endif #if !defined(MAGNUM_TARGET_GLES2) && defined(CORRADE_TARGET_WINDOWS) /* On Windows NVidia drivers the glTransformFeedbackVaryings() does not make a copy of its char* arguments so it fails at link time when the original char arrays are not in scope anymore. Enabling *synchronous* debug output circumvents this bug. Can be triggered by running TransformFeedbackGLTest with GL_KHR_debug extension disabled. */ "nv-windows-dangling-transform-feedback-varying-names"_s, #endif #ifndef MAGNUM_TARGET_GLES /* Layout qualifier causes compiler error with GLSL 1.20 on Mesa, GLSL 1.30 on NVidia and 1.40 on macOS. Everything is fine when using a newer GLSL version. */ "no-layout-qualifiers-on-old-glsl"_s, /* NVidia drivers (358.16) report compressed block size from internal format query in bits instead of bytes */ "nv-compressed-block-size-in-bits"_s, /* NVidia drivers (358.16) report different compressed image size for cubemaps based on whether the texture is immutable or not and not based on whether I'm querying all faces (ARB_DSA) or a single face (non-DSA, EXT_DSA) */ "nv-cubemap-inconsistent-compressed-image-size"_s, /* NVidia drivers (358.16) return only the first slice of compressed cube map image when querying all six slices using the ARB_DSA API */ "nv-cubemap-broken-full-compressed-image-query"_s, /* NVidia drivers return 0 when asked for GL_CONTEXT_PROFILE_MASK, so it needs to be worked around by asking for GL_ARB_compatibility */ "nv-zero-context-profile-mask"_s, /* (Headless) EGL contexts for desktop GL on NVidia 384 and 390 drivers don't have correct statically linked GL 1.0 and 1.1 functions (such as glGetString()) and one has to retrieve them explicitly using eglGetProcAddress(). Doesn't seem to happen on pre-384 and 396, but it's not possible to get driver version through EGL, so enabling this unconditionally on all EGL NV contexts. */ "nv-egl-incorrect-gl11-function-pointers"_s, /* On NV driver 450.80.02, eglQueryDeviceAttribEXT() segfaults when querying GPUs that the user does not have access to (i.e. via cgroup). Instead, always call eglQueryDeviceStringEXT() as that doesn't segfault and sets an error that can be retrieved via eglGetError() to see if the user has access to that device. On well-behaved driver versions, eglQueryDeviceAttribEXT() returns false instead of segfaulting. */ "nv-egl-crashy-query-device-attrib"_s, #endif #ifndef MAGNUM_TARGET_GLES /* SVGA3D (VMware host GL driver) glDrawArrays() draws nothing when the vertex buffer memory is initialized using glNamedBufferData() from ARB_DSA. Using the non-DSA glBufferData() works. */ "svga3d-broken-dsa-bufferdata"_s, /* SVGA3D does out-of-bound writes in some cases of glGetTexSubImage(), leading to memory corruption on client machines. That's nasty, so the whole ARB_get_texture_sub_image is disabled. */ "svga3d-gettexsubimage-oob-write"_s, #endif /* SVGA3D has broken handling of glTex[ture][Sub]Image*D() for 1D arrays, 2D arrays, 3D textures and cube map textures where it uploads just the first slice in the last dimension. This is only with copies from host memory, not with buffer images. Seems to be fixed in Mesa 13, but I have no such system to verify that on. https://github.com/mesa3d/mesa/commit/2aa9ff0cda1f6ad97c83d5583fab7a84efabe19e */ "svga3d-texture-upload-slice-by-slice"_s, #if defined(CORRADE_TARGET_EMSCRIPTEN) && defined(__EMSCRIPTEN_PTHREADS__) /* Shader sources containing UTF-8 characters are converted to empty strings when running on Emscripten with -s USE_PTHREADS=1. Working around that by replacing all chars > 127 with spaces. Relevant code: https://github.com/kripken/emscripten/blob/7f89560101843198787530731f40a65288f6f15f/src/fetch-worker.js#L54-L58 */ "emscripten-pthreads-broken-unicode-shader-sources"_s, #endif #if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_WEBGL) /* Empty EGL_CONTEXT_FLAGS_KHR cause SwiftShader 3.3 to fail context creation with EGL_BAD_ATTRIBUTE. Not sending the flags then. Relevant code: https://github.com/google/swiftshader/blob/5fb5e817a20d3e60f29f7338493f922b5ac9d7c4/src/OpenGL/libEGL/libEGL.cpp#L794-L810 Version 4.1 suffers from the same thing, but 4.0 on Android not, for some reason. */ "swiftshader-no-empty-egl-context-flags"_s, /* SwiftShader 3.3.0.1 crashes deep inside eglMakeCurrent() when using EGL_NO_SURFACE. Supplying a 32x32 PBuffer to work around that. (Android's) SwiftShader 4.0 needs it too, but doesn't crash, only fails to make the context current with EGL_BAD_MATCH. Version 4.1 doesn't seem to need this workaround anymore. */ "swiftshader-egl-context-needs-pbuffer"_s, #endif #if defined(MAGNUM_TARGET_GLES2) && !defined(MAGNUM_TARGET_WEBGL) /* SwiftShader 4.1.0 on ES2 contexts reports GL_ANGLE_instanced_arrays and GL_EXT_instanced_arrays but has no glDrawArraysInstancedANGLE / glDrawArraysInstancedEXT nor glDrawElementsInstancedANGLE / glDrawElementsInstancedEXT entrypoints, only the unsuffixed versions for ES3. OTOH, glVertexAttribDivisor is there for both ANGLE and EXT. Relevant code: https://github.com/google/swiftshader/blob/ad5c2952ca88730c07e04f6f1566194b66860c26/src/OpenGL/libGLESv2/libGLESv2.cpp#L6352-L6357 Disabling the two extensions on ES2 contexts to avoid nullptr crashes. */ "swiftshader-no-es2-draw-instanced-entrypoints"_s, /* SwiftShader 4.1.0 on ES2 contexts reports GL_OES_texture_3D but from all its entrypoints only glTexImage3DOES is present, all others are present only in the ES3 unsuffixed versions. Relevant code: https://github.com/google/swiftshader/blob/ad5c2952ca88730c07e04f6f1566194b66860c26/src/OpenGL/libGLESv2/libGLESv2.cpp#L6504 Disabling the extension on ES2 contexts to avoid nullptr crashes. */ "swiftshader-no-es2-oes-texture-3d-entrypoints"_s, #endif #if defined(MAGNUM_TARGET_GLES) && !defined(MAGNUM_TARGET_GLES2) && !defined(MAGNUM_TARGET_WEBGL) /* SwiftShader 4.1.0 has special handling for binding buffers to the transform feedback target, requiring an XFB object to be active when a buffer is bound to GL_TRANSFORM_FEEDBACK_BUFFER and ignoring the glBindBuffer() call otherwise. No other driver does that. As a workaround, setting Buffer::TargetHint::TransformFeedback will make it use Buffer::TargetHint::Array instead, as that works okay. */ "swiftshader-broken-xfb-buffer-binding-target"_s, /* SwiftShader 4.1.0 does implement gl_VertexID for ES3 contexts, but in practice it doesn't work, returning a constant value. In order to make this easier to check, there's a dummy MAGNUM_shader_vertex_id extension that's defined on all GL 3.0+ and GLES 3.0+ / WebGL 2+ contexts *except* for SwiftShader. */ "swiftshader-broken-shader-vertex-id"_s, #endif #ifndef MAGNUM_TARGET_GLES /* Even with the DSA variant, where GL_IMPLEMENTATION_COLOR_READ_* is passed to glGetNamedFramebufferParameter(), Mesa complains that the framebuffer is not bound for reading. Relevant code: https://github.com/mesa3d/mesa/blob/212c0c630a849e4737e2808a993d708cbb2f18f7/src/mesa/main/framebuffer.c#L841-L843 Workaround is to explicitly bind the framebuffer for reading. */ "mesa-implementation-color-read-format-dsa-explicit-binding"_s, #endif #if !defined(MAGNUM_TARGET_GLES2) && defined(CORRADE_TARGET_WINDOWS) /* Intel drivers on Windows return GL_UNSIGNED_BYTE for *both* GL_IMPLEMENTATION_COLOR_READ_FORMAT and _TYPE when using either glGetNamedFramebufferParameter() or glGetFramebufferParameter(), independently on what's the actual framebuffer format. Using glGetInteger() makes it return GL_RGBA and GL_UNSIGNED_BYTE for RGBA8 framebuffers, and cause an "Error has been generated. GL error GL_INVALID_OPERATION in GetIntegerv: (ID: 2576729458) Generic error" when it is not. Since glGetInteger() is actually able to return a correct value in *one circumstance*, it's preferrable to the other random shit the driver is doing. */ "intel-windows-implementation-color-read-format-completely-broken"_s, #endif #if !defined(MAGNUM_TARGET_GLES) && defined(CORRADE_TARGET_WINDOWS) /* Intel drivers on Windows have some synchronization / memory alignment bug in the DSA glNamedBufferData() when the same buffer is set as an index buffer to a mesh right after or repeatedly. Calling glBindBuffer() right before or after the data upload fixes the issue. The above is reproducible with the 2019.01 ImGui example, and used to be worked around in a more hopeful way. However, the reports about things going *bad* in heavier ImGui-based apps didn't stop with that and none of my tests were able to reproduce anything. Since I lost patience already, I'm disabling the DSA code paths for everything related to buffers. (Two weeks pass.) But wait! while that fixed all issues for *some* users, it made things completely broken elsewhere, causing an app to render just a clear color and nothing else. The cancer apparently spread further, so I'm disabling all VAO-related DSA code paths as well now. Workarounds listed separately, in case someone might want to dig further or experience the misery of only one of them being active. To save you time experimenting: - (Epilepsy warning!) With the former disabled and no matter whether the second is disabled or not, the ImGui example (or any other ImGui-based app, really), the screen will start flickering heavily under *some* circumstances. This is known since drivers 24 at least. - With the former enabled and the second disabled, you might either experience a total doom, where just the framebuffer clear color is visible, or your app is totally fine. This is reproducible with drivers 25 or 26 at least. Note that modifying the code to enable this workaround on other drivers (AMD on Windows, e.g.) doesn't break anything, so it's not like the workaround would be incomplete with some code paths still relying on DSA that's not there. It's clearly Intel drivers fault. - With both enabled, things seem to be fine, and I hope it stays that way also for future driver updates. */ "intel-windows-crazy-broken-buffer-dsa"_s, "intel-windows-crazy-broken-vao-dsa"_s, /* ARB_direct_state_access implementation on Intel Windows drivers has broken *everything* related to cube map textures (but not cube map arrays) -- data upload, data queries, framebuffer attachment, framebuffer copies, all complaining about "Wrong <func> 6 provided for <target> 34067" and similar (GL_TEXTURE_CUBE_MAP is 34067). Using the non-DSA code paths as a workaround (for the 3D image up/download as well). */ "intel-windows-broken-dsa-for-cubemaps"_s, /* DSA glBindTextureUnit() on Intel Windows drivers simply doesn't work when passing 0 to it. Non-zero IDs work correctly except for cube maps. Using the non-DSA code path for unbinding and cube maps as a workaround. */ "intel-windows-half-baked-dsa-texture-bind"_s, /* DSA glNamedFramebufferTexture() on Intel Windows drivers doesn't work for layered cube map array attachments. Non-layered or non-array cube map attachment works. Using the non-DSA code path as a workaround. */ "intel-windows-broken-dsa-layered-cubemap-array-framebuffer-attachment"_s, /* DSA glClearNamedFramebuffer*() on Intel Windows drivers doesn't do anything. Using the non-DSA code path as a workaournd. See also "mesa-broken-dsa-framebuffer-clear" below. */ "intel-windows-broken-dsa-framebuffer-clear"_s, /* Using DSA glCreateQueries() on Intel Windows drivers breaks glBeginQueryIndexed(). Using the non-DSA glGenQueries() instead makes it work properly. See TransformFeedbackGLTest for a test. */ "intel-windows-broken-dsa-indexed-queries"_s, /* DSA-ified "vertex layout" glVertexArrayAttribIFormat() is broken when passing shorts instead of full 32bit ints. Using the old-style glVertexAttribIPointer() works correctly. No idea if the non-DSA glVertexAttribIFormat() works or not. A test that triggers this issue is in MeshGLTest::addVertexBufferIntWithShort(). */ "intel-windows-broken-dsa-integer-vertex-attributes"_s, /* Shader compiler on Intel Windows drivers insists on telling me "No errors." when it should just stay silent. See also "angle-chatty-shader-compiler". */ "intel-windows-chatty-shader-compiler"_s, /* When using more than just a vertex and fragment shader (geometry shader, e.g.), ARB_explicit_uniform_location on Intel silently uses wrong locations, blowing up with either a non-descript Error has been generated. GL error GL_INVALID_OPERATION in ProgramUniformMatrix4fv: (ID: 2052228270) Generic error or, if you are lucky, a highly-cryptic-but-still-better-than-nothing Error has been generated. GL error GL_INVALID_OPERATION in ProgramUniform4fv: (ID: 1725519030) GL error GL_INVALID_OPERATION: mismatched type setting uniform of location "3" in program 1, "" using shaders, 2, "", 3, "", 8, "" *unless* you have vertex uniform locations first, fragment locations second and geometry locations last. Another case is happening with color for a Flat3D shader -- because a (compiled out / unused) texture matrix was at location 1, setting color to location 2 didn't work, ending up with a Generic error again (driver version 27). Because this is impossible to prevent, the extension is completely disabled on all Intel Windows drivers. */ "intel-windows-explicit-uniform-location-is-less-explicit-than-you-hoped"_s, #endif #ifndef MAGNUM_TARGET_GLES /* Mesa 24 (or, 24.2 at least) crashes on exit deep inside X11 if the DSA glClearNamedFramebuffer() APIs are used. Not sure what's up, couldn't find any relevant changelog entry and unfortunately the previous version I had was only 23.3.5, so it could be anything in between. My hunch is that it's due to some new code that deals with framebuffer compression and which was only correctly cleaned up in the non-DSA code path. Or something. See also "intel-windows-broken-dsa-framebuffer-clear" above. */ "mesa-broken-dsa-framebuffer-clear"_s, #endif #ifndef MAGNUM_TARGET_GLES /* NVidia seems to be returning values for the default framebuffer when GL_IMPLEMENTATION_COLOR_READ_FORMAT and _TYPE is queried using glGetNamedFramebufferParameter(). Using either glGetInteger() or glGetFramebufferParameter() works correctly. */ "nv-implementation-color-read-format-dsa-broken"_s, #endif #ifndef MAGNUM_TARGET_GLES /* ApiTrace needs an explicit initial glViewport() call to initialize its framebuffer size, otherwise it assumes it's zero-sized. */ "apitrace-zero-initial-viewport"_s, #endif #if defined(MAGNUM_TARGET_WEBGL) && !defined(MAGNUM_TARGET_GLES2) /* While the EXT_disjoint_timer_query extension should be only on WebGL 1 and EXT_disjoint_timer_query_webgl2 only on WebGL 2, Firefox reports EXT_disjoint_timer_query on both. The entry points work correctly however, so this workaround makes Magnum pretend EXT_disjoint_timer_query_webgl2 is available when it detects EXT_disjoint_timer_query on WebGL 2 builds on Firefox. See also https://bugzilla.mozilla.org/show_bug.cgi?id=1328882, https://www.khronos.org/webgl/public-mailing-list/public_webgl/1705/msg00015.php and https://github.com/emscripten-core/emscripten/pull/9652 for the Emscripten-side part of this workaround. */ "firefox-fake-disjoint-timer-query-webgl2"_s, #endif #ifdef MAGNUM_TARGET_WEBGL /* Firefox 92+ says "WEBGL_debug_renderer_info is deprecated in Firefox and will be removed. Please use RENDERER." if attempting to use the unmasked renderer / vendor string. The information is provided through the regular APIs instead. Disabling the extension if present on the new versions to avoid console spam. */ "firefox-deprecated-debug-renderer-info"_s #endif
Chromium has a similar list: https://cs.chromium.org/chromium/src/gpu/config/gpu_