Don't make the shared software winsys rely on internal knowledge about
the cell driver's texture twiddling.
This is just a sketch and hasn't even been compile tested.
Just use flush_frontbuffer directly. The flush_frontbuffer routine has
been somewhat devalued recently, but it is actually just the right
interface for our needs.
It is in pipe_screen, meaning that any wrapping (eg trace module)
will get properly unwrapped before we try and use the pipe_surface
argument for real.
If a particular co-state-tracker needs to implement this itself, it
should organize a way to allow the winsys to call back up to its
level, rather than hijacking the driver-supplied implementation.
Currently there are still at least two functions bundled up inside the
winsys concept:
a) that of a backend resource manager, sometimes capable of performing
present() operations,
b) the initialization code/routine for the whole driver stack.
The inclusion of (b) makes it difficult to share implementations of
(a) between different drivers. For instance, a clean xlib winsys
could be of use for software-rasterized VG, GLES, EGL, etc, stacks.
But that is only true as long as there is no dependency from the
winsys to higher level code, as would be the case when we include (b)
in this component.
This change creates a new gallium/targets subtree, specifically for
implementing the glue needed to build individual driver stacks, and
moves that code out of a single example winsys, namely xlib.
Other drivers continue to build unchanged, but hopefully can migrate
to this structure over time.
Promote the llvmpipe winsys more or less unchanged to
state_trackers/sw_winsys.h.
Some minor breakages:
- softpipe::texture_blanket is broken, but scheduled for removal anyway.
- haven't fixed up g3vdl yet.
Introduce a new shared usage and rename primary to scanout.
The display target usage is more of a windows concept and
doesn't mean the same thing as shared. Display target means
that the surface should be presentable, for softpipe this
means that it should be backed by a hardware buffer.
This avoids exposing the ms driver structure to the winsys,
and nicely encapsulates driver customizable stuff.
In the future more things might be customizable by the winsys, like
throttling, 3D readback etc.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Since the execbuffer change actually changed size off the ioctl
struct and not just a reuse of padded bits, we can't support
old kernels as easily as the scanout change was.
There are so many state trackers and winsys that rely on it for debug
building, and trace can really build anywhere and is thin so there's no
point is making it an option.
Based on Xavier Chantry <chantry.xavier@gmail.com>'s patch:
"I also removed the trace check in xlib SConscript which is now useless, and
that script seems to deal with the cell driver, so I added cell in the check
for supported drivers."
This reverts commit fff5be8e7b.
Probably went too soon with this, dileX reported OA not working for him
it works here fine, but the optimisations I wanted aren't working properly
yet so I'll fix that now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
This creates a cleaner winsys and drop the simple screen stuff.
It makes r300g use pb_bufmgr structs and adds usage of the cached
bufmgr for vertex/index buffers.
It also avoids mapping too often.
I'm not 100% sure this is perfect but it won't find its own bugs.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Windows doesn't really expect things to fail at this point -- it
will try many times until it finally gives up, worse, something bad
happens.
The WGL state tracker will gracefully decline to do work even when it
is loaded.
Squashed commit of the following:
commit 009598122adf5cad42bf9af067eccaf8e89e1233
Author: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Mon Feb 8 11:03:26 2010 +0000
nouveau: fix gallium-screen-context branch fallout
commit 7b2ffc2019d72e833afea7eebf3e80121187375d
Merge: 51e190e c036d13
Author: Keith Whitwell <keithw@vmware.com>
Date: Fri Feb 5 09:55:02 2010 +0000
Merge commit 'origin/master' into gallium-screen-context
Conflicts:
src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
This branch has got a pretty tortured history now, I expect
a squash merge will be appropriate when it is done.
commit 51e190e95acf120f72768fafb29e9721e358df1b
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 17:58:02 2010 +0000
gallium: fix some build issues
commit f524bdaa723fb181637ad30c6ad708aeedabe25b
Merge: f906212 3aba0a2
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 17:51:32 2010 +0000
Merge commit 'origin/master' into gallium-screen-context
commit f9062126883199eabf045714c717cd35f4b7b313
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 17:17:12 2010 +0000
gallium/docs: small description of screen::create_context
commit efcb37bd3d5ed37f06c6105bd2d750b374ec0927
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:42:42 2010 +0000
drm/radeon: more dead create_context wrapper removal
commit 6badc0dd9e06cf2ec936940bcf12b9ef5324b301
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:42:30 2010 +0000
drm/i965: more dead create_context wrapper removal
commit cf04ebd5a54b18b2d894cfdab2b0f2fd55626ffc
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:42:05 2010 +0000
st/python: more dead create_context wrapper removal
commit 444f114c3516abf71c430e6e9d0d2ae3b80679d3
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:37:58 2010 +0000
idenity: wrapped context creation
commit 5a6d09cb9e468d1ee6c8d54b887618819d8d94f2
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:28:47 2010 +0000
ws/gdi: remove dead context_create wrapper
commit 132b55f4bec39386ac625f09aaa11f609664024c
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:27:52 2010 +0000
ws/gdi: remove dead context_create wrapper
commit 56d2d21a0cdcb197a364049d354c2f15a4fc026a
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:25:38 2010 +0000
st/xorg: use screen::context_create
commit 838c5cfe56b2af6c644909bed3c5e7cdd64c336a
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:23:20 2010 +0000
glx/xlib: simplify creation of trace-wrapped contexts
Trace screen knows how to properly wrap context creation in the
wrapped screen, so nothing special to do here.
commit c99404c03ebaec4175f08a2f363e43c9085f2635
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:18:24 2010 +0000
st/python: no need to special case context creation for trace
commit 193a527a682b6877bb1faecd8092df4dfd055a18
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:15:30 2010 +0000
drm/radeon: remove dead create_context declaration
commit bb984eecc25cf23bc77e1c818b81165ba1a07c9a
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:14:58 2010 +0000
nv/drm: remove dead create_context ref
commit e809313a44287dc4e16c28e9e74ef3b2b5271aa1
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:12:51 2010 +0000
st/egl: remove a layer of wrappers around screen::create_context
commit 39caa6672a04122f185c22e17aab86d1c40938bf
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:05:28 2010 +0000
r300g: fill in screen::context_create
commit 407f12556d16ba0656774d8a1d9ebda22f82f473
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 16:04:04 2010 +0000
cell: adapt for screen::create_context, untested
commit d02b0c6ce321a04c76fdabb09e0e4380ce1c1376
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:50:24 2010 +0000
drm/nv: adapt for screen::create_context
All contexts now created directly through the screen, so remove
equivalent code here.
Remove apparently un-needed array of contexts in the winsys.
commit 53eec5b1349aa1b6892a75a7bff7e7530957aeae
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:50:08 2010 +0000
stw: adapt for screen::create_context, untested
commit c6a64de3eb381bc9a88e9fbdecbf87d77925aaf5
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:49:20 2010 +0000
trace: expose the wrapped context's priv data
If we are going to keep this priv idea, really want an accessor
function for it so that trace and other drivers can wrap that.
commit 75d6104e11d86ec2b0749627ed58e35f856ee6eb
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:47:55 2010 +0000
nv30: adapt to screen::context_create
commit 12f5deb6ed9723e9b5d34577052b8365813ca14e
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:44:47 2010 +0000
nv40: adapt to screen::context_create
commit 14baccaa3b6bbb3b91056126f6521828e786dc62
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:35:27 2010 +0000
nv50: adapt to screen::create_context
Not build tested. Need to figure out how to build nouveau.
commit a0e94505ccd2d7f3e604465a2ac302f1286b73b6
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:22:27 2010 +0000
llvmpipe: update for screen::create_context, untested
commit 0eae17107c950346030e4f7e0ec232f868d3893d
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 15:16:57 2010 +0000
xlib/llvmpipe: remove dead winsys context creation path
commit 2f69f9ffaa7e2a01d2483277246ed13051ae4ca3
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 14:58:27 2010 +0000
gallium: convert most code to use screen::create_context
I wish I could build all of gallium at once to find breakages.
commit d7b57f4061b82322cbcae176125913d9f0dea6c1
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 12:46:21 2010 +0000
glx: permit building with older protocol headers
I'd like to be able to build mesa on current distro releases without
having to upgrade from the standard dri2proto and glproto headers. With
this change I'm able to build on ancient releases such as Ubuntu 9-10...
In general, it would be nice to be able to build-test mesa to check for
unintended breakages without having to follow the external dependencies
of every group working on the codebase.
commit 57adedd6fb06c98572ed8d4aef19203df4c4eea2
Merge: da71847 e1906ae
Author: Keith Whitwell <keithw@vmware.com>
Date: Thu Feb 4 11:38:15 2010 +0000
Merge commit 'origin/master' into gallium-screen-context
Conflicts:
src/gallium/drivers/softpipe/sp_video_context.h
src/gallium/drivers/trace/tr_context.c
src/gallium/state_trackers/wgl/shared/stw_context.c
src/gallium/winsys/gdi/gdi_softpipe_winsys.c
commit da71847ea6414d7e352c6094f8963bb4eda344dc
Author: José Fonseca <jfonseca@vmware.com>
Date: Sat May 2 08:57:39 2009 +0100
wgl: Use pipe_screen::context_create.
commit 2595a188f93fd903600ef5d8517737ee0592035d
Author: José Fonseca <jfonseca@vmware.com>
Date: Sat May 2 08:56:47 2009 +0100
trace: Implement pipe_screen::context_create.
commit f3640e4ae37a5260cbfba999d079f827de0a313a
Author: José Fonseca <jfonseca@vmware.com>
Date: Sat May 2 08:56:17 2009 +0100
softpipe: Implement pipe_screen::context_create.
commit 347266bddc8bd39c711bacb2193793759d0f3696
Author: José Fonseca <jfonseca@vmware.com>
Date: Sat May 2 08:55:31 2009 +0100
gallium: New pipe_screen::context_create callback.
The previous memory leak fix didn't always work properly.
Now check the xm_buffer::smh field (now documented!) to
see if the buffer points to shared memory.
With the addition egl_x11_swrast, egl_softpipe is sort of deprecated.
The new driver serves the same purpose as egl_softpipe does. It is
based on egl_g3d and provides more features.
The egl_g3d state tracker has support for KMS, and the support is based
on the egl state tracker. As egl_g3d provides more features, it should
be better to keep only egl_g3d to unify the efforts.
This is a minimum change to keep things building and running minimally -- it
might cause applications to fail to allocate buffers due to out of
GMR memory.
A proper update will be commited later.
The new EGL drivers use egl_g3d and respect EGL_DISPLAYS. They are
named after the display supported and the DRM name.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
st_api.c is supposed to define st_api_OpenGL to advertise OpenGL
support. However, the linker discards the symbol because it has no
user. It is better to leave this to other state trackers that link to
libmesagallium.a.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
The call to _eglSwapBuffers is unnecessary, and the function is missing
when -fvisibility=hidden. Also remove the extraneous braces and indent
the block.
And not just softpipe.
It is particularly convenient to use llvmpipe instead, since it is much
faster. It also allows to use rbug with all xlib drivers.
As requested by just about everybody. I'm going back to actually reading
patches before ACKing and pushing them.
This reverts commit 417ce06306.
Conflicts:
src/gallium/winsys/drm/radeon/core/radeon_buffer.c
Well, technically non-r[345]xx. At any rate...
$ glxgears
libGL: OpenDriver: trying /home/simpson/mesa/lib/gallium/r600_dri.so
131 frames in 5.0 seconds = 26.107 FPS
I'm sure you can see where this is going. :3
Remove x86 specific hacks. Not that they will ever be used on
none x86 arches, but they are built by default. And the way the
flags where added was a hack.
Add the vmwgfx_drm.h header for now, this allows the svga driver
to be enabled by default without forcing people to install the
vmwgfx_drm.h header on their system.
To be removed once vmwgfx_drm.h is in libdrm.
This is done by stopping linking to libmesagallium.a and removing DRI
related stuff. The state tracker an application links to decides the
API supported.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
When GLX_INDIRECT_RENDERING is defined, some symbols are used in
libglapi.a but are not defined. Define them through the help of
glapitemp.h.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Makes it easier to dump as we get all of the information
about the upload in a single hit.
Opens the window to simplification in the driver if these
relocation arrays can be maintained statically rather than
being recreated whenever we check for a new upload.
Still needs some cleanup to avoid uglyness introduced with the
delta values.
The winsys now inserts the presumed offset into referring buffers from
inside of bo_emit_reloc(). Remove the many locally coded places where
this was happening in the driver and eliminate the worry of getting it
wrong.
No longer need to expose offset values to the driver at all, so no need
to worry about what to do in the driver when they change. Just use
zero values wherever we had offsets previously -- the relocations will
fix it all up for us.
Any allocation that may fail should be checked, and propogate the
error upwards. At the highest level we will flush batch and retry.
This is an alternate strategy to what the original DRI driver did of
attempting to flush batch from the lowest levels (eg inside
BEGIN_BATCH). The trouble with that strategy was that flushes could
occur at unexpected times, and additionally there was a need for a
wierd notification mechanism to propogate the 'lost context' state
back up to higher levels.
Propogating the errors directly gives us a lot of flexibility how to
deal with these states, at the expense of a lot more checking in the
code.
Will add some sanity checks later to make sure that out-of-memory
conditions are properly escalated and not lost halfway up the stack.
But there are some missing symbols, "nm -u i965_dri.so"
[SNIP]
U brw_surface_bo
U brw_surface_pitch
U brw_texture_blanket_winsys_buffer
U brw_texture_get_winsys_buffer
U brw_update_dirty_counts
[SNIP]
Create a dummy winsys that just debug-prints on calls into the winsys
functions. Will use this to get to the point where we are generating
sane-looking debug dumps and diassembly.
Also fix various warnings generated with the new compiler flags set in
this config.
The intel_xorg file looks like it's got quite a bit of code that could
be lifted up into the xorg state tracker -- should really just have a
list of pci ids and a pointer to a screen create func.
1: add rv530 support
- num z pipes cap
- add proper start/finish query options for rv530
2: convert to use linked list properly.
3: add flushing required check.
4: initial Z top disabling support.
TODO:
make it actually work on my rv530.
The driver never work with real hardware and has
bitrotted for quite some time now, might as well
drop it. If somebody wants to look at it just use git.
Same as in src/gallium/winsys/drm/intel/xorg/Makefile
Thanks MrCooper for explanations on IRC
[ Summary amended by Michel Dänzer to clarify that this is related to the xorg
state tracker ]
Most use cases just got the buffer from the texture
and then called into one of the get_handle functions.
Also with this patch it would be easier to move to a
generic function for getting handles from textures
and textures from handles, that is exposed via the screen.
When a single-buffered window was resized the new window size was never
detected. This fix that, but there's still a bug which causes window
contents corruption for certain window sizes...
This reverts commit 6a40d1e9d9.
Turns out that we *do* need these for OQ after all. Go figure.
Conflicts:
src/gallium/winsys/drm/radeon/core/radeon_r300.h
Move some fields of _EGLDriver to _EGLDisplay. It also becomes
unnecessary to pass _EGLDisplay to drivers when _eglMain is called.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
The motivation is so that drivers do not need to look up and check for
bad display, context, and etc. It also becomes unnecessary for drivers
to call the link functions.
This commit makes eglapi.[ch] do the lookup and check. As a result, the
driver API is overhauled, and almost all sources and drivers need
update. The updates are mainly find and replace with human brains.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
This adds support for eglBindTexImage and eglReleaseTexImage. They rely
on the state tracker to do the real work.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
IsBound tells if a context or surface is current. What it does not tell
is, to which thread a context is current, or to which context a surface
is current. This commit replaces IsBound by a pointer to the binding
thread or context.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
A context or surface that is neither linked to a display nor current to
a thread should be destroyed. Therefore, an unlinked context or surface
implies a pending delete automatically.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
This commit uses the newly introduced link functions to manage EGL
contexts and surfaces. As a result of this, the API for drivers are
changed. All drivers are updated for the change.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
The special feature of bin/minstall to copy symlinks is only ever needed
when installing libraries which may have .so symlinks. All the headers
and directories can use a normal install program.
These two modes are separated as $(INSTALL) and $(MINSTALL) to allow the
user (or autoconf) to override installing normal files as they please.
An autoconf check for the install program has been added and will be
used in preference to minstall when available.
Fixes bug 16053.
This is only used for debuging the gem backend on i965
chipset using the softpipe pipe driver.
Usage: "export INTEL_SOFTPIPE=y" and point LIBGL_DRIVERS_PATH
to "$MESA/lib/gallium" where $MESA is the mesa root.