docs: Document shader capturing environment variables.
[mesa.git] / docs / envvars.html
1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
2 <html lang="en">
3 <head>
4 <meta http-equiv="content-type" content="text/html; charset=utf-8">
5 <title>Environment Variables</title>
6 <link rel="stylesheet" type="text/css" href="mesa.css">
7 </head>
8 <body>
9
10 <div class="header">
11 <h1>The Mesa 3D Graphics Library</h1>
12 </div>
13
14 <iframe src="contents.html"></iframe>
15 <div class="content">
16
17 <h1>Environment Variables</h1>
18
19 <p>
20 Normally, no environment variables need to be set. Most of the environment
21 variables used by Mesa/Gallium are for debugging purposes, but they can
22 sometimes be useful for debugging end-user issues.
23 </p>
24
25
26 <h2>LibGL environment variables</h2>
27
28 <ul>
29 <li>LIBGL_DEBUG - If defined debug information will be printed to stderr.
30 If set to 'verbose' additional information will be printed.
31 <li>LIBGL_DRIVERS_PATH - colon-separated list of paths to search for DRI drivers
32 <li>LIBGL_ALWAYS_INDIRECT - if set to `true`, forces an indirect rendering context/connection.
33 <li>LIBGL_ALWAYS_SOFTWARE - if set to `true`, always use software rendering
34 <li>LIBGL_NO_DRAWARRAYS - if set to `true`, do not use DrawArrays GLX protocol (for debugging)
35 <li>LIBGL_SHOW_FPS - print framerate to stdout based on the number of glXSwapBuffers
36 calls per second.
37 <li>LIBGL_DRI3_DISABLE - disable DRI3 if set to `true`.
38 </ul>
39
40
41
42 <h2>Core Mesa environment variables</h2>
43
44 <ul>
45 <li>MESA_NO_ASM - if set, disables all assembly language optimizations
46 <li>MESA_NO_MMX - if set, disables Intel MMX optimizations
47 <li>MESA_NO_3DNOW - if set, disables AMD 3DNow! optimizations
48 <li>MESA_NO_SSE - if set, disables Intel SSE optimizations
49 <li>MESA_NO_ERROR - if set to 1, error checking is disabled as per KHR_no_error.
50 This will result in undefined behaviour for invalid use of the api, but
51 can reduce CPU use for apps that are known to be error free.</li>
52 <li>MESA_DEBUG - if set, error messages are printed to stderr. For example,
53 if the application generates a GL_INVALID_ENUM error, a corresponding error
54 message indicating where the error occurred, and possibly why, will be
55 printed to stderr.<br>
56
57 For release builds, MESA_DEBUG defaults to off (no debug output).
58
59 MESA_DEBUG accepts the following comma-separated list of named
60 flags, which adds extra behaviour to just set MESA_DEBUG=1:
61 <ul>
62 <li>silent - turn off debug messages. Only useful for debug builds.</li>
63 <li>flush - flush after each drawing command</li>
64 <li>incomplete_tex - extra debug messages when a texture is incomplete</li>
65 <li>incomplete_fbo - extra debug messages when a fbo is incomplete</li>
66 <li>context - create a debug context (see GLX_CONTEXT_DEBUG_BIT_ARB) and
67 print error and performance messages to stderr (or MESA_LOG_FILE).</li>
68 </ul>
69 <li>MESA_LOG_FILE - specifies a file name for logging all errors, warnings,
70 etc., rather than stderr
71 <li>MESA_TEX_PROG - if set, implement conventional texture env modes with
72 fragment programs (intended for developers only)
73 <li>MESA_TNL_PROG - if set, implement conventional vertex transformation
74 operations with vertex programs (intended for developers only).
75 Setting this variable automatically sets the MESA_TEX_PROG variable as well.
76 <li>MESA_EXTENSION_OVERRIDE - can be used to enable/disable extensions.
77 A value such as "GL_EXT_foo -GL_EXT_bar" will enable the GL_EXT_foo extension
78 and disable the GL_EXT_bar extension.
79 <li>MESA_EXTENSION_MAX_YEAR - The GL_EXTENSIONS string returned by Mesa is sorted
80 by extension year.
81 If this variable is set to year X, only extensions defined on or before year
82 X will be reported.
83 This is to work-around a bug in some games where the extension string is
84 copied into a fixed-size buffer without truncating.
85 If the extension string is too long, the buffer overrun can cause the game
86 to crash.
87 This is a work-around for that.
88 <li>MESA_GL_VERSION_OVERRIDE - changes the value returned by
89 glGetString(GL_VERSION) and possibly the GL API type.
90 <ul>
91 <li> The format should be MAJOR.MINOR[FC]
92 <li> FC is an optional suffix that indicates a forward compatible context.
93 This is only valid for versions &gt;= 3.0.
94 <li> GL versions &lt; 3.0 are set to a compatibility (non-Core) profile
95 <li> GL versions = 3.0, see below
96 <li> GL versions &gt; 3.0 are set to a Core profile
97 <li> Examples: 2.1, 3.0, 3.0FC, 3.1, 3.1FC
98 <ul>
99 <li> 2.1 - select a compatibility (non-Core) profile with GL version 2.1
100 <li> 3.0 - select a compatibility (non-Core) profile with GL version 3.0
101 <li> 3.0FC - select a Core+Forward Compatible profile with GL version 3.0
102 <li> 3.1 - select a Core profile with GL version 3.1
103 <li> 3.1FC - select a Core+Forward Compatible profile with GL version 3.1
104 </ul>
105 <li> Mesa may not really implement all the features of the given version.
106 (for developers only)
107 </ul>
108 <li>MESA_GLES_VERSION_OVERRIDE - changes the value returned by
109 glGetString(GL_VERSION) for OpenGL ES.
110 <ul>
111 <li> The format should be MAJOR.MINOR
112 <li> Examples: 2.0, 3.0, 3.1
113 <li> Mesa may not really implement all the features of the given version.
114 (for developers only)
115 </ul>
116 <li>MESA_GLSL_VERSION_OVERRIDE - changes the value returned by
117 glGetString(GL_SHADING_LANGUAGE_VERSION). Valid values are integers, such as
118 "130". Mesa will not really implement all the features of the given language version
119 if it's higher than what's normally reported. (for developers only)
120 <li>MESA_GLSL_CACHE_DISABLE - if set to `true`, disables the GLSL shader cache
121 <li>MESA_GLSL_CACHE_MAX_SIZE - if set, determines the maximum size of
122 the on-disk cache of compiled GLSL programs. Should be set to a number
123 optionally followed by 'K', 'M', or 'G' to specify a size in
124 kilobytes, megabytes, or gigabytes. By default, gigabytes will be
125 assumed. And if unset, a maximum size of 1GB will be used. Note: A separate
126 cache might be created for each architecture that Mesa is installed for on
127 your system. For example under the default settings you may end up with a 1GB
128 cache for x86_64 and another 1GB cache for i386.
129 <li>MESA_GLSL_CACHE_DIR - if set, determines the directory to be used
130 for the on-disk cache of compiled GLSL programs. If this variable is
131 not set, then the cache will be stored in $XDG_CACHE_HOME/mesa (if
132 that variable is set), or else within .cache/mesa within the user's
133 home directory.
134 <li>MESA_GLSL - <a href="shading.html#envvars">shading language compiler options</a>
135 <li>MESA_NO_MINMAX_CACHE - when set, the minmax index cache is globally disabled.
136 <li>MESA_SHADER_CAPTURE_PATH - see <a href="shading.html#capture">Capturing Shaders</a></li>
137 <li>MESA_SHADER_DUMP_PATH and MESA_SHADER_READ_PATH - see <a href="shading.html#replacement">Experimenting with Shader Replacements</a></li>
138 </ul>
139
140
141 <h2>Mesa Xlib driver environment variables</h2>
142
143 <p>
144 The following are only applicable to the Mesa Xlib software driver.
145 See the <a href="xlibdriver.html">Xlib software driver page</a> for details.
146 </p>
147 <ul>
148 <li>MESA_RGB_VISUAL - specifies the X visual and depth for RGB mode
149 <li>MESA_CI_VISUAL - specifies the X visual and depth for CI mode
150 <li>MESA_BACK_BUFFER - specifies how to implement the back color buffer,
151 either "pixmap" or "ximage"
152 <li>MESA_GAMMA - gamma correction coefficients for red, green, blue channels
153 <li>MESA_XSYNC - enable synchronous X behavior (for debugging only)
154 <li>MESA_GLX_FORCE_CI - if set, force GLX to treat 8bpp visuals as CI visuals
155 <li>MESA_GLX_FORCE_ALPHA - if set, forces RGB windows to have an alpha channel.
156 <li>MESA_GLX_DEPTH_BITS - specifies default number of bits for depth buffer.
157 <li>MESA_GLX_ALPHA_BITS - specifies default number of bits for alpha channel.
158 </ul>
159
160
161 <h2>i945/i965 driver environment variables (non-Gallium)</h2>
162
163 <ul>
164 <li>INTEL_NO_HW - if set to 1, prevents batches from being submitted to the hardware.
165 This is useful for debugging hangs, etc.</li>
166 <li>INTEL_DEBUG - a comma-separated list of named flags, which do various things:
167 <ul>
168 <li>ann - annotate IR in assembly dumps</li>
169 <li>aub - dump batches into an AUB trace for use with simulation tools</li>
170 <li>bat - emit batch information</li>
171 <li>blit - emit messages about blit operations</li>
172 <li>blorp - emit messages about the blorp operations (blits &amp; clears)</li>
173 <li>buf - emit messages about buffer objects</li>
174 <li>clip - emit messages about the clip unit (for old gens, includes the CLIP program)</li>
175 <li>color - use color in output</li>
176 <li>cs - dump shader assembly for compute shaders</li>
177 <li>do32 - generate compute shader SIMD32 programs even if workgroup size doesn't exceed the SIMD16 limit</li>
178 <li>dri - emit messages about the DRI interface</li>
179 <li>fbo - emit messages about framebuffers</li>
180 <li>fs - dump shader assembly for fragment shaders</li>
181 <li>gs - dump shader assembly for geometry shaders</li>
182 <li>hex - print instruction hex dump with the disassembly</li>
183 <li>l3 - emit messages about the new L3 state during transitions</li>
184 <li>miptree - emit messages about miptrees</li>
185 <li>no8 - don't generate SIMD8 fragment shader</li>
186 <li>no16 - suppress generation of 16-wide fragment shaders. useful for debugging broken shaders</li>
187 <li>nocompact - disable instruction compaction</li>
188 <li>nodualobj - suppress generation of dual-object geometry shader code</li>
189 <li>norbc - disable single sampled render buffer compression</li>
190 <li>optimizer - dump shader assembly to files at each optimization pass and iteration that make progress</li>
191 <li>perf - emit messages about performance issues</li>
192 <li>perfmon - emit messages about AMD_performance_monitor</li>
193 <li>pix - emit messages about pixel operations</li>
194 <li>prim - emit messages about drawing primitives</li>
195 <li>sf - emit messages about the strips &amp; fans unit (for old gens, includes the SF program)</li>
196 <li>shader_time - record how much GPU time is spent in each shader</li>
197 <li>spill_fs - force spilling of all registers in the scalar backend (useful to debug spilling code)</li>
198 <li>spill_vec4 - force spilling of all registers in the vec4 backend (useful to debug spilling code)</li>
199 <li>state - emit messages about state flag tracking</li>
200 <li>sync - after sending each batch, emit a message and wait for that batch to finish rendering</li>
201 <li>tcs - dump shader assembly for tessellation control shaders</li>
202 <li>tes - dump shader assembly for tessellation evaluation shaders</li>
203 <li>tex - emit messages about textures.</li>
204 <li>urb - emit messages about URB setup</li>
205 <li>vert - emit messages about vertex assembly</li>
206 <li>vs - dump shader assembly for vertex shaders</li>
207 </ul>
208 <li>INTEL_SCALAR_VS (or TCS, TES, GS) - force scalar/vec4 mode for a shader stage (Gen8-9 only)</li>
209 <li>INTEL_PRECISE_TRIG - if set to 1, true or yes, then the driver prefers
210 accuracy over performance in trig functions.</li>
211 </ul>
212
213
214 <h2>Radeon driver environment variables (radeon, r200, and r300g)</h2>
215
216 <ul>
217 <li>RADEON_NO_TCL - if set, disable hardware-accelerated Transform/Clip/Lighting.
218 </ul>
219
220
221 <h2>EGL environment variables</h2>
222
223 <p>
224 Mesa EGL supports different sets of environment variables. See the
225 <a href="egl.html">Mesa EGL</a> page for the details.
226 </p>
227
228
229 <h2>Gallium environment variables</h2>
230
231 <ul>
232 <li>GALLIUM_HUD - draws various information on the screen, like framerate,
233 cpu load, driver statistics, performance counters, etc.
234 Set GALLIUM_HUD=help and run e.g. glxgears for more info.
235 <li>GALLIUM_HUD_PERIOD - sets the hud update rate in seconds (float). Use zero
236 to update every frame. The default period is 1/2 second.
237 <li>GALLIUM_HUD_VISIBLE - control default visibility, defaults to true.
238 <li>GALLIUM_HUD_TOGGLE_SIGNAL - toggle visibility via user specified signal.
239 Especially useful to toggle hud at specific points of application and
240 disable for unencumbered viewing the rest of the time. For example, set
241 GALLIUM_HUD_VISIBLE to false and GALLIUM_HUD_TOGGLE_SIGNAL to 10 (SIGUSR1).
242 Use kill -10 <pid> to toggle the hud as desired.
243 <li>GALLIUM_HUD_DUMP_DIR - specifies a directory for writing the displayed
244 hud values into files.
245 <li>GALLIUM_DRIVER - useful in combination with LIBGL_ALWAYS_SOFTWARE=true for
246 choosing one of the software renderers "softpipe", "llvmpipe" or "swr".
247 <li>GALLIUM_LOG_FILE - specifies a file for logging all errors, warnings, etc.
248 rather than stderr.
249 <li>GALLIUM_PRINT_OPTIONS - if non-zero, print all the Gallium environment
250 variables which are used, and their current values.
251 <li>GALLIUM_DUMP_CPU - if non-zero, print information about the CPU on start-up
252 <li>TGSI_PRINT_SANITY - if set, do extra sanity checking on TGSI shaders and
253 print any errors to stderr.
254 <LI>DRAW_FSE - ???
255 <LI>DRAW_NO_FSE - ???
256 <li>DRAW_USE_LLVM - if set to zero, the draw module will not use LLVM to execute
257 shaders, vertex fetch, etc.
258 <li>ST_DEBUG - controls debug output from the Mesa/Gallium state tracker.
259 Setting to "tgsi", for example, will print all the TGSI shaders.
260 See src/mesa/state_tracker/st_debug.c for other options.
261 </ul>
262
263 <h3>Clover state tracker environment variables</h3>
264
265 <ul>
266 <li>CLOVER_EXTRA_BUILD_OPTIONS - allows specifying additional compiler and linker
267 options. Specified options are appended after the options set by the OpenCL
268 program in clBuildProgram.
269 <li>CLOVER_EXTRA_COMPILE_OPTIONS - allows specifying additional compiler
270 options. Specified options are appended after the options set by the OpenCL
271 program in clCompileProgram.
272 <li>CLOVER_EXTRA_LINK_OPTIONS - allows specifying additional linker
273 options. Specified options are appended after the options set by the OpenCL
274 program in clLinkProgram.
275 </ul>
276
277
278 <h3>Softpipe driver environment variables</h3>
279 <ul>
280 <li>SOFTPIPE_DUMP_FS - if set, the softpipe driver will print fragment shaders
281 to stderr
282 <li>SOFTPIPE_DUMP_GS - if set, the softpipe driver will print geometry shaders
283 to stderr
284 <li>SOFTPIPE_NO_RAST - if set, rasterization is no-op'd. For profiling purposes.
285 <li>SOFTPIPE_USE_LLVM - if set, the softpipe driver will try to use LLVM JIT for
286 vertex shading processing.
287 </ul>
288
289
290 <h3>LLVMpipe driver environment variables</h3>
291 <ul>
292 <li>LP_NO_RAST - if set LLVMpipe will no-op rasterization
293 <li>LP_DEBUG - a comma-separated list of debug options is accepted. See the
294 source code for details.
295 <li>LP_PERF - a comma-separated list of options to selectively no-op various
296 parts of the driver. See the source code for details.
297 <li>LP_NUM_THREADS - an integer indicating how many threads to use for rendering.
298 Zero turns off threading completely. The default value is the number of CPU
299 cores present.
300 </ul>
301
302 <h3>VMware SVGA driver environment variables</h3>
303 <ul>
304 <li>SVGA_FORCE_SWTNL - force use of software vertex transformation
305 <li>SVGA_NO_SWTNL - don't allow software vertex transformation fallbacks
306 (will often result in incorrect rendering).
307 <li>SVGA_DEBUG - for dumping shaders, constant buffers, etc. See the code
308 for details.
309 <li>SVGA_EXTRA_LOGGING - if set, enables extra logging to the vmware.log file,
310 such as the OpenGL program's name and command line arguments.
311 <li>See the driver code for other, lesser-used variables.
312 </ul>
313
314
315 <h3>VA-API state tracker environment variables</h3>
316 <ul>
317 <li>VAAPI_MPEG4_ENABLED - enable MPEG4 for VA-API, disabled by default.
318 </ul>
319
320
321 <h3>VC4 driver environment variables</h3>
322 <ul>
323 <li>VC4_DEBUG - a comma-separated list of named flags, which do various things:
324 <ul>
325 <li>cl - dump command list during creation</li>
326 <li>qpu - dump generated QPU instructions</li>
327 <li>qir - dump QPU IR during program compile</li>
328 <li>nir - dump NIR during program compile</li>
329 <li>tgsi - dump TGSI during program compile</li>
330 <li>shaderdb - dump program compile information for shader-db analysis</li>
331 <li>perf - print during performance-related events</li>
332 <li>norast - skip actual hardware execution of commands</li>
333 <li>always_flush - flush after each draw call</li>
334 <li>always_sync - wait for finish after each flush</li>
335 <li>dump - write a GPU command stream trace file (VC4 simulator only)</li>
336 </ul>
337 </ul>
338
339
340 <p>
341 Other Gallium drivers have their own environment variables. These may change
342 frequently so the source code should be consulted for details.
343 </p>
344
345 </div>
346 </body>
347 </html>