4
Several factors combine to make efficient dispatch of OpenGL functions
5
fairly complicated. This document attempts to explain some of the issues
6
and introduce the reader to Mesa's implementation. Readers already
7
familiar with the issues around GL dispatch can safely skip ahead to the
8
:ref:`overview of Mesa's implementation <overview>`.
10
1. Complexity of GL Dispatch
11
----------------------------
13
Every GL application has at least one object called a GL *context*. This
14
object, which is an implicit parameter to every GL function, stores all
15
of the GL related state for the application. Every texture, every buffer
16
object, every enable, and much, much more is stored in the context.
17
Since an application can have more than one context, the context to be
18
used is selected by a window-system dependent function such as
19
``glXMakeContextCurrent``.
21
In environments that implement OpenGL with X-Windows using GLX, every GL
22
function, including the pointers returned by ``glXGetProcAddress``, are
23
*context independent*. This means that no matter what context is
24
currently active, the same ``glVertex3fv`` function is used.
26
This creates the first bit of dispatch complexity. An application can
27
have two GL contexts. One context is a direct rendering context where
28
function calls are routed directly to a driver loaded within the
29
application's address space. The other context is an indirect rendering
30
context where function calls are converted to GLX protocol and sent to a
31
server. The same ``glVertex3fv`` has to do the right thing depending on
32
which context is current.
34
Highly optimized drivers or GLX protocol implementations may want to
35
change the behavior of GL functions depending on current state. For
36
example, ``glFogCoordf`` may operate differently depending on whether or
39
In multi-threaded environments, it is possible for each thread to have a
40
different GL context current. This means that poor old ``glVertex3fv``
41
has to know which GL context is current in the thread where it is being
46
2. Overview of Mesa's Implementation
47
------------------------------------
49
Mesa uses two per-thread pointers. The first pointer stores the address
50
of the context current in the thread, and the second pointer stores the
51
address of the *dispatch table* associated with that context. The
52
dispatch table stores pointers to functions that actually implement
53
specific GL functions. Each time a new context is made current in a
54
thread, these pointers are updated.
56
The implementation of functions such as ``glVertex3fv`` becomes
59
- Fetch the current dispatch table pointer.
60
- Fetch the pointer to the real ``glVertex3fv`` function from the
62
- Call the real function.
64
This can be implemented in just a few lines of C code. The file
65
``src/mesa/glapi/glapitemp.h`` contains code very similar to this.
68
:caption: Sample dispatch function
70
void glVertex3f(GLfloat x, GLfloat y, GLfloat z)
72
const struct _glapi_table * const dispatch = GET_DISPATCH();
74
(*dispatch->Vertex3f)(x, y, z);
77
The problem with this simple implementation is the large amount of
78
overhead that it adds to every GL function call.
80
In a multithreaded environment, a naive implementation of
81
``GET_DISPATCH`` involves a call to ``pthread_getspecific`` or a similar
82
function. Mesa provides a wrapper function called
83
``_glapi_get_dispatch`` that is used by default.
88
A number of optimizations have been made over the years to diminish the
89
performance hit imposed by GL dispatch. This section describes these
90
optimizations. The benefits of each optimization and the situations
91
where each can or cannot be used are listed.
93
3.1. Dual dispatch table pointers
94
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
96
The vast majority of OpenGL applications use the API in a single
97
threaded manner. That is, the application has only one thread that makes
98
calls into the GL. In these cases, not only do the calls to
99
``pthread_getspecific`` hurt performance, but they are completely
100
unnecessary! It is possible to detect this common case and avoid these
103
Each time a new dispatch table is set, Mesa examines and records the ID
104
of the executing thread. If the same thread ID is always seen, Mesa
105
knows that the application is, from OpenGL's point of view, single
108
As long as an application is single threaded, Mesa stores a pointer to
109
the dispatch table in a global variable called ``_glapi_Dispatch``. The
110
pointer is also stored in a per-thread location via
111
``pthread_setspecific``. When Mesa detects that an application has
112
become multithreaded, ``NULL`` is stored in ``_glapi_Dispatch``.
114
Using this simple mechanism the dispatch functions can detect the
115
multithreaded case by comparing ``_glapi_Dispatch`` to ``NULL``. The
116
resulting implementation of ``GET_DISPATCH`` is slightly more complex,
117
but it avoids the expensive ``pthread_getspecific`` call in the common
121
:caption: Improved ``GET_DISPATCH`` Implementation
123
#define GET_DISPATCH() \
124
(_glapi_Dispatch != NULL) \
125
? _glapi_Dispatch : pthread_getspecific(&_glapi_Dispatch_key)
130
Starting with the 2.4.20 Linux kernel, each thread is allocated an area
131
of per-thread, global storage. Variables can be put in this area using
132
some extensions to GCC. By storing the dispatch table pointer in this
133
area, the expensive call to ``pthread_getspecific`` and the test of
134
``_glapi_Dispatch`` can be avoided.
136
The dispatch table pointer is stored in a new variable called
137
``_glapi_tls_Dispatch``. A new variable name is used so that a single
138
libGL can implement both interfaces. This allows the libGL to operate
139
with direct rendering drivers that use either interface. Once the
140
pointer is properly declared, ``GET_DISPACH`` becomes a simple variable
144
:caption: TLS ``GET_DISPATCH`` Implementation
146
extern __thread struct _glapi_table *_glapi_tls_Dispatch
147
__attribute__((tls_model("initial-exec")));
149
#define GET_DISPATCH() _glapi_tls_Dispatch
151
Use of this path is controlled by the preprocessor define
152
``USE_ELF_TLS``. Any platform capable of using ELF TLS should use this
153
as the default dispatch method.
155
Windows has a similar concept, and beginning with Windows Vista, shared
156
libraries can take advantage of compiler-assisted TLS. This TLS data
157
has no fixed size and does not compete with API-based TLS (``TlsAlloc``)
158
for the limited number of slots available there, and so ``USE_ELF_TLS`` can
159
be used on Windows too, even though it's not truly ELF.
161
3.3. Assembly Language Dispatch Stubs
162
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
164
Many platforms have difficulty properly optimizing the tail-call in the
165
dispatch stubs. Platforms like x86 that pass parameters on the stack
166
seem to have even more difficulty optimizing these routines. All of the
167
dispatch routines are very short, and it is trivial to create optimal
168
assembly language versions. The amount of optimization provided by using
169
assembly stubs varies from platform to platform and application to
170
application. However, by using the assembly stubs, many platforms can
171
use an additional space optimization (see :ref:`below <fixedsize>`).
173
The biggest hurdle to creating assembly stubs is handling the various
174
ways that the dispatch table pointer can be accessed. There are four
175
different methods that can be used:
177
#. Using ``_glapi_Dispatch`` directly in builds for non-multithreaded
179
#. Using ``_glapi_Dispatch`` and ``_glapi_get_dispatch`` in
180
multithreaded environments.
181
#. Using ``_glapi_Dispatch`` and ``pthread_getspecific`` in
182
multithreaded environments.
183
#. Using ``_glapi_tls_Dispatch`` directly in TLS enabled multithreaded
186
People wishing to implement assembly stubs for new platforms should
187
focus on #4 if the new platform supports TLS. Otherwise, implement #2
188
followed by #3. Environments that do not support multithreading are
189
uncommon and not terribly relevant.
191
Selection of the dispatch table pointer access method is controlled by a
192
few preprocessor defines.
194
- If ``USE_ELF_TLS`` is defined, method #3 is used.
195
- If ``HAVE_PTHREAD`` is defined, method #2 is used.
196
- If none of the preceding are defined, method #1 is used.
198
Two different techniques are used to handle the various different cases.
199
On x86 and SPARC, a macro called ``GL_STUB`` is used. In the preamble of
200
the assembly source file different implementations of the macro are
201
selected based on the defined preprocessor variables. The assembly code
202
then consists of a series of invocations of the macros such as:
205
:caption: SPARC Assembly Implementation of ``glColor3fv``
207
GL_STUB(Color3fv, _gloffset_Color3fv)
209
The benefit of this technique is that changes to the calling pattern
210
(i.e., addition of a new dispatch table pointer access method) require
211
fewer changed lines in the assembly code.
213
However, this technique can only be used on platforms where the function
214
implementation does not change based on the parameters passed to the
215
function. For example, since x86 passes all parameters on the stack, no
216
additional code is needed to save and restore function parameters around
217
a call to ``pthread_getspecific``. Since x86-64 passes parameters in
218
registers, varying amounts of code needs to be inserted around the call
219
to ``pthread_getspecific`` to save and restore the GL function's
222
The other technique, used by platforms like x86-64 that cannot use the
223
first technique, is to insert ``#ifdef`` within the assembly
224
implementation of each function. This makes the assembly file
225
considerably larger (e.g., 29,332 lines for ``glapi_x86-64.S`` versus
226
1,155 lines for ``glapi_x86.S``) and causes simple changes to the
227
function implementation to generate many lines of diffs. Since the
228
assembly files are typically generated by scripts, this isn't a
231
Once a new assembly file is created, it must be inserted in the build
232
system. There are two steps to this. The file must first be added to
233
``src/mesa/sources``. That gets the file built and linked. The second
234
step is to add the correct ``#ifdef`` magic to
235
``src/mesa/glapi/glapi_dispatch.c`` to prevent the C version of the
236
dispatch functions from being built.
240
3.4. Fixed-Length Dispatch Stubs
241
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
243
To implement ``glXGetProcAddress``, Mesa stores a table that associates
244
function names with pointers to those functions. This table is stored in
245
``src/mesa/glapi/glprocs.h``. For different reasons on different
246
platforms, storing all of those pointers is inefficient. On most
247
platforms, including all known platforms that support TLS, we can avoid
250
If the assembly stubs are all the same size, the pointer need not be
251
stored for every function. The location of the function can instead be
252
calculated by multiplying the size of the dispatch stub by the offset of
253
the function in the table. This value is then added to the address of
254
the first dispatch stub.
256
This path is activated by adding the correct ``#ifdef`` magic to
257
``src/mesa/glapi/glapi.c`` just before ``glprocs.h`` is included.