José Fonseca [Mon, 16 Apr 2012 09:59:25 +0000 (10:59 +0100)]
tracediff2: Handle variations in number of arguments.
José Fonseca [Mon, 16 Apr 2012 09:57:45 +0000 (10:57 +0100)]
cdecl: Always generate double quotes.
For consitency.
José Fonseca [Sun, 15 Apr 2012 15:13:51 +0000 (16:13 +0100)]
Cleanup how pointers are handled.
In particular prevent high order bits to be lost when processing on
64bits traces on 32bits platforms.
José Fonseca [Sun, 15 Apr 2012 15:07:25 +0000 (16:07 +0100)]
Move the last bits of code out of the top source dir.
José Fonseca [Sun, 15 Apr 2012 14:04:10 +0000 (15:04 +0100)]
Remove codegen.py.
Unused.
José Fonseca [Sun, 15 Apr 2012 14:03:31 +0000 (15:03 +0100)]
Drop GLsync_
José Fonseca [Sun, 15 Apr 2012 13:33:00 +0000 (14:33 +0100)]
Factor the proc address wrapping into gltrace.py
José Fonseca [Sun, 15 Apr 2012 08:37:46 +0000 (09:37 +0100)]
Name trace targets consistently.
José Fonseca [Sun, 15 Apr 2012 08:31:18 +0000 (09:31 +0100)]
Cleanup a bit more IID handling.
José Fonseca [Sun, 15 Apr 2012 07:42:25 +0000 (08:42 +0100)]
Handle REFIIDs on functions too.
José Fonseca [Sat, 14 Apr 2012 21:26:09 +0000 (22:26 +0100)]
Fix MacOSX build.
José Fonseca [Sat, 14 Apr 2012 21:02:42 +0000 (22:02 +0100)]
Use ObjPointers consistenty.
José Fonseca [Sat, 14 Apr 2012 19:36:09 +0000 (20:36 +0100)]
Fix IDirect3DVertexDeclaration9::GetDeclaration
José Fonseca [Sat, 14 Apr 2012 19:28:58 +0000 (20:28 +0100)]
Fix max units computation.
My thinko while tweaking Imre's
e68b8d68741f2559a4be34d53b8aa3c7ed761471 patch.
José Fonseca [Sat, 14 Apr 2012 19:25:52 +0000 (20:25 +0100)]
Properly (un)wrap array arguments.
José Fonseca [Sat, 14 Apr 2012 17:13:25 +0000 (18:13 +0100)]
Move dispatch to its own subdirectory.
José Fonseca [Sat, 14 Apr 2012 16:28:40 +0000 (17:28 +0100)]
Don't change behavior of glproc code via RETRACE define.
So that we don't have to build the same code twice.
José Fonseca [Sat, 14 Apr 2012 16:24:14 +0000 (17:24 +0100)]
Don't link against d3d9 in runtime.
José Fonseca [Sat, 14 Apr 2012 16:22:57 +0000 (17:22 +0100)]
Move retracers to their own directory.
José Fonseca [Sat, 14 Apr 2012 14:55:40 +0000 (15:55 +0100)]
Move tracers to wrappers subdirectory.
José Fonseca [Sat, 14 Apr 2012 14:17:12 +0000 (15:17 +0100)]
Don't abuse NotImplementedError.
So that we can distinguish between types we can't support, and those we
can support, but haven't.
José Fonseca [Sat, 14 Apr 2012 13:58:49 +0000 (14:58 +0100)]
Add d3d11.def file.
José Fonseca [Sat, 14 Apr 2012 13:56:45 +0000 (14:56 +0100)]
Minimal D3D11 support.
Michal Krol [Fri, 13 Apr 2012 17:21:23 +0000 (18:21 +0100)]
Fix dumping of depth textures.
José Fonseca [Fri, 13 Apr 2012 16:36:19 +0000 (17:36 +0100)]
Merge branch 'd3dretrace'
José Fonseca [Fri, 13 Apr 2012 13:57:08 +0000 (14:57 +0100)]
Prevent derreference after free when retracing glFeedbackBuffer/glSelectBuffer.
José Fonseca [Fri, 13 Apr 2012 12:57:23 +0000 (13:57 +0100)]
Skip functions without sideeffects alltogether.
Function body of functions with no sideeffects is no longer necessary,
now that glReadPixels is considered to have sideeffects.
José Fonseca [Fri, 13 Apr 2012 12:49:41 +0000 (13:49 +0100)]
A few minor corrections to GL specs.
José Fonseca [Fri, 13 Apr 2012 08:03:28 +0000 (09:03 +0100)]
Find the x64 DXSDK libraries.
Imre Deak [Fri, 30 Mar 2012 12:45:38 +0000 (15:45 +0300)]
Android: hard code the trace file path to be under /data
On Android for Dalvik applications cwd is preset to that of the zygote
(app_process) process and it's the root directory. Since root is r/o
mounted by default it's more convenient to have the trace file in /data
which is writeable at least by root by default.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Imre Deak [Mon, 2 Apr 2012 13:47:48 +0000 (16:47 +0300)]
gles: fix lookup for GLESv1 functions
Get the function from the GLESv1 library in case it's not found in
GLESv2.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Imre Deak [Mon, 9 Apr 2012 16:35:41 +0000 (19:35 +0300)]
gles: sanity check GL_COMBINED_TEXTURE_IMAGE_UNITS
At least the Android software GL implementation doesn't return the
proper value for this, but rather returns 0. The GL(ES) specification
mandates a minimum value of 2, so use this as a fall-back value.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Imre Deak [Fri, 30 Mar 2012 12:48:44 +0000 (15:48 +0300)]
gles: don't call unsupported glGetHandleARB
GLES doesn't support this call, so do not call it.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Imre Deak [Mon, 9 Apr 2012 16:35:06 +0000 (19:35 +0300)]
gles: fix GLES version string matching
Some implementations return a version string starting with "OpenGL ES-",
fix the detection logic accordingly.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
José Fonseca [Thu, 12 Apr 2012 06:13:01 +0000 (07:13 +0100)]
Trace d3d10.1 too.
Dan McCabe [Wed, 21 Mar 2012 16:53:45 +0000 (09:53 -0700)]
Add trim support to qapitrace GUI app.
This patch adds support for trimming of traces via the qapitrace GUI.
We enhance the GUI by adding a Trim entry to the Trace menu. When the
user selects either a frame or a call, the "apitrace trim" command
will be invoked to trim all calls after the selected call or frame.
New trace files are created automatically and named according to the
last call number in the trimmed trace.
For example, if the original trace file is:
/path/foo.trace
and the trace is trimmed after call 1234, the name of the trimmed trace
file will be named:
/path/foo.1234.trim.trace
Upon trimming, the trimmed trace file will be loaded into qapitrace.
First, we enhance the Trace menu of the GUI app.
Next, we add a TrimProcess class. This is modelled on the TraceProcess
class, but takes into account differences and simplifications.
Next, we tie the TrimProcess class into the main window by accessing
that class and its members appropriately as well as tieing in message
communication with that class.
Finally, we add a reference to the source of TrimProcess to the make
system.
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Carl Worth [Mon, 9 Apr 2012 20:45:33 +0000 (13:45 -0700)]
apitrace diff-images: Print one line for each file being compared.
This is important so that the test suite can verify that "apitrace
diff-images" is actually doing something. The tricky case is when no
images are found to compare. In this case, diff-images exits with
value 0, (no differences), which is valid, but the test suite wants to
know that no comparison happened since that's unexpected, (and should
flag a failure in the test suite).
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Carl Worth [Mon, 9 Apr 2012 20:45:33 +0000 (13:45 -0700)]
apitrace diff-images: Print one line for each file being compared.
This is important so that the test suite can verify that "apitrace
diff-images" is actually doing something. The tricky case is when no
images are found to compare. In this case, diff-images exits with
value 0, (no differences), which is valid, but the test suite wants to
know that no comparison happened since that's unexpected, (and should
flag a failure in the test suite).
Carl Worth [Mon, 9 Apr 2012 18:24:52 +0000 (11:24 -0700)]
Don't show images where there is no difference.
This makes for a much more compact view for the common case where
image mismatches are rare, and makes the mismatches stand out more
making them easier to notice.
To avoid hiding information irretrievably, the filename is made into a
link to the image so that it can be easily seen if desired.
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Carl Worth [Tue, 3 Apr 2012 23:02:25 +0000 (16:02 -0700)]
snapdiff: Provide indication whether images match or not
This indication is provided in two ways:
1. The exit value of the script will be 1 if there are any image
differences, (0 for no image differences).
2. The filename cell in the resulting HTML table will be colored red
if there are image differences (and green for no image
differences).
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Carl Worth [Tue, 3 Apr 2012 22:58:50 +0000 (15:58 -0700)]
os_posix: Fix return value for os::execute()
Previously, this was returning the status value from waitpid directly.
That is incorrect as the actual exit status from the executed program
is likely to be in high-order bits of that status value. Instead, use
the WEXITSTATUS macro to return the actual exit status value.
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Carl Worth [Tue, 3 Apr 2012 22:18:00 +0000 (15:18 -0700)]
snapdiff: Add filename column to HTML report
Basically, I want one cell that will always be present for turning the
cell green or red based on whether the images match or not.
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Carl Worth [Mon, 9 Apr 2012 18:27:22 +0000 (11:27 -0700)]
snapdiff: Remove debugging print statement.
I assume this print statement is just some leftover from
debugging. The output isn't meaningful to the user.
Carl Worth [Wed, 28 Mar 2012 01:13:14 +0000 (18:13 -0700)]
cli: Add a simple implementation of "apitrace dump-images"
This merely calls out to the existing glretrace command with
appropriate options for generating snapshots. This at least provides
the image-dumping capability within the unified command-line
interface.
In the future, it will likely make sens to make dump-images do things
in-process.
José Fonseca [Wed, 11 Apr 2012 11:04:00 +0000 (12:04 +0100)]
Fix D3D10 flags.
José Fonseca [Sat, 7 Apr 2012 09:49:35 +0000 (10:49 +0100)]
Avoid output blobs.
Not really supported.
José Fonseca [Sat, 7 Apr 2012 09:23:33 +0000 (10:23 +0100)]
Debug output of region mappings.
José Fonseca [Mon, 2 Apr 2012 06:10:48 +0000 (08:10 +0200)]
Fix type mismatch in enum deserialization.
José Fonseca [Thu, 5 Apr 2012 18:55:53 +0000 (19:55 +0100)]
Handle glMapObjectBufferATI/glUnmapObjectBufferATI.
José Fonseca [Sun, 1 Apr 2012 20:47:30 +0000 (22:47 +0200)]
Add missing os_time include.
José Fonseca [Sat, 7 Apr 2012 09:50:53 +0000 (10:50 +0100)]
Separate allocation and extraction in retracing.
José Fonseca [Sat, 7 Apr 2012 09:50:17 +0000 (10:50 +0100)]
Recognize zero SizeToLock as full buffer.
José Fonseca [Sat, 7 Apr 2012 09:49:35 +0000 (10:49 +0100)]
Avoid output blobs.
Not really supported.
José Fonseca [Sat, 7 Apr 2012 09:42:50 +0000 (10:42 +0100)]
Allow InOut arguments.
José Fonseca [Sat, 7 Apr 2012 09:23:33 +0000 (10:23 +0100)]
Debug output of region mappings.
José Fonseca [Sat, 7 Apr 2012 09:13:37 +0000 (10:13 +0100)]
Use ANSI escape codes when inside WINE.
José Fonseca [Fri, 6 Apr 2012 06:49:45 +0000 (07:49 +0100)]
Create window on d3dretrace.
José Fonseca [Thu, 5 Apr 2012 21:58:43 +0000 (22:58 +0100)]
Revert "Improve IUnknown::QueryInterface"
This reverts commit
7f3688bb956928bfef266ebbe0b3ccbf6e984559.
José Fonseca [Thu, 5 Apr 2012 19:06:42 +0000 (20:06 +0100)]
Disable BlockingIODevice with QJSON.
It looks like it doesn't work reliably in all cases, at least with
certain versions.
Kenneth Graunke [Thu, 5 Apr 2012 17:51:26 +0000 (10:51 -0700)]
Fix crash when using apitrace trim's -o option.
The -o option takes an argument (the filename), so the getopt string
needs to be 'o:', not 'o'. Otherwise, no argument is parsed, optarg
remains NULL, and we crash on "output = optarg".
José Fonseca [Thu, 5 Apr 2012 18:56:19 +0000 (19:56 +0100)]
Remove unneeded include.
José Fonseca [Thu, 5 Apr 2012 18:56:08 +0000 (19:56 +0100)]
Add missing dependencies to glproc.
José Fonseca [Thu, 5 Apr 2012 18:59:56 +0000 (19:59 +0100)]
Avoid rebuilding types unnecessarily.
José Fonseca [Thu, 5 Apr 2012 18:58:38 +0000 (19:58 +0100)]
Avoid conflicting tmps.
José Fonseca [Thu, 5 Apr 2012 18:58:20 +0000 (19:58 +0100)]
Warn on null this pointer.
José Fonseca [Thu, 5 Apr 2012 18:58:00 +0000 (19:58 +0100)]
Improve IUnknown::QueryInterface
José Fonseca [Thu, 5 Apr 2012 18:57:45 +0000 (19:57 +0100)]
Fix handles.
José Fonseca [Thu, 5 Apr 2012 18:57:12 +0000 (19:57 +0100)]
glMapObjectBufferATI
José Fonseca [Thu, 5 Apr 2012 18:57:02 +0000 (19:57 +0100)]
More D3D9 spec tweaks.
José Fonseca [Thu, 5 Apr 2012 18:56:19 +0000 (19:56 +0100)]
Remove unneeded include.
José Fonseca [Thu, 5 Apr 2012 18:56:08 +0000 (19:56 +0100)]
Add missing dependencies to glproc.
José Fonseca [Thu, 5 Apr 2012 18:55:53 +0000 (19:55 +0100)]
Handle glMapObjectBufferATI/glUnmapObjectBufferATI.
José Fonseca [Thu, 5 Apr 2012 10:23:11 +0000 (11:23 +0100)]
Several improvements to C decl parser.
José Fonseca [Thu, 5 Apr 2012 10:22:53 +0000 (11:22 +0100)]
More verbose unknown IID warnings.
José Fonseca [Thu, 5 Apr 2012 10:20:22 +0000 (11:20 +0100)]
Trace ID3D10Debug interface.
José Fonseca [Thu, 5 Apr 2012 10:18:27 +0000 (11:18 +0100)]
Recognize more D3D10 flags.
José Fonseca [Thu, 5 Apr 2012 06:10:30 +0000 (07:10 +0100)]
Make obj pointers first class citizens.
José Fonseca [Mon, 2 Apr 2012 20:00:53 +0000 (21:00 +0100)]
Fix glretrace build.
José Fonseca [Mon, 2 Apr 2012 06:10:48 +0000 (08:10 +0200)]
Fix type mismatch in enum deserialization.
José Fonseca [Mon, 2 Apr 2012 06:10:06 +0000 (08:10 +0200)]
Deserialize structures.
José Fonseca [Mon, 2 Apr 2012 06:07:32 +0000 (08:07 +0200)]
Rename ConstRemover.
José Fonseca [Sun, 1 Apr 2012 20:48:45 +0000 (22:48 +0200)]
Fix D33D9 link failures.
José Fonseca [Sun, 1 Apr 2012 20:48:24 +0000 (22:48 +0200)]
Fix zeroing of unsupported arguments for complex types.
José Fonseca [Sun, 1 Apr 2012 20:47:30 +0000 (22:47 +0200)]
Add missing os_time include.
José Fonseca [Sun, 1 Apr 2012 20:47:11 +0000 (22:47 +0200)]
Describe C++ references accurately.
José Fonseca [Sun, 1 Apr 2012 20:45:09 +0000 (22:45 +0200)]
Adjust D3D9's sideeffect settings.
José Fonseca [Fri, 30 Mar 2012 20:22:15 +0000 (21:22 +0100)]
Allow to include EGL on MacOSX.
José Fonseca [Fri, 30 Mar 2012 19:41:25 +0000 (20:41 +0100)]
Merge branch 'master' into d3dretrace
Conflicts:
retrace.py
José Fonseca [Fri, 30 Mar 2012 18:58:20 +0000 (19:58 +0100)]
Fix build without X11.
José Fonseca [Fri, 30 Mar 2012 06:15:17 +0000 (07:15 +0100)]
Build retrace sources only once.
José Fonseca [Thu, 29 Mar 2012 23:18:16 +0000 (00:18 +0100)]
Drop TRACE_EGL.
Also unnecessary.
José Fonseca [Thu, 29 Mar 2012 22:38:55 +0000 (23:38 +0100)]
Drop HAVE_EGL.
Unnecessary now that we budle EGL/GLES headers
José Fonseca [Thu, 29 Mar 2012 19:13:10 +0000 (20:13 +0100)]
Fix typo in wgl function call detection.
Alexandros Frantzis [Mon, 28 Nov 2011 10:04:05 +0000 (12:04 +0200)]
Use either glGetTexImage() or bind the texture to a FBO to get the contents.
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
José Fonseca [Thu, 29 Mar 2012 11:38:05 +0000 (12:38 +0100)]
Infer texture size on OpenGL ES via TexSubImage.
José Fonseca [Thu, 29 Mar 2012 09:42:28 +0000 (10:42 +0100)]
More code simplification.
José Fonseca [Thu, 29 Mar 2012 09:28:45 +0000 (10:28 +0100)]
Simplify image size determination.
José Fonseca [Thu, 29 Mar 2012 08:53:15 +0000 (09:53 +0100)]
Split image dumping to a separate source file.
José Fonseca [Thu, 29 Mar 2012 08:42:41 +0000 (09:42 +0100)]
Split gl shader dumping into a separate file.
José Fonseca [Thu, 29 Mar 2012 08:42:10 +0000 (09:42 +0100)]
Rename glstate.py to glstate_params.py.
To match generated source file name.