Doing projective transforms required repositioning the cursor using the
hotspot, but that requires relocating the upper left corner in terms of said
hotspot.
Instead of using a separate function to notify DIX about transform changes,
add the transform to RRCrtcNotify so that the whole Crtc state changes
atomically.
RandR matrix computations lose too much precision in fixed point;
computations using the inverted matrix can be as much as 10 pixels off.
Convert them to double precision values and pass those around. These API
changes are fairly heavyweight; the official Render interface remains fixed
point, so the fixed point matrix comes along for the ride everywhere.
Add APIs to xf86RandR12 support and randr extension to record whether the
driver supports transforms, report that value in the RRGetCrtcTransform
reply.
New RRCrtcGetTransform function in DIX that DDX can use to get the pending
transform. The DDX code should be complete; the DIX code is just a stub at
this point.
Drivers that care about crtc positions on the screen to ensure that vblank
works correctly need to be notified when crtcs are changed.
Provide a hook in the mode setting code that is invoked whenever any
configuration is done to the screen.
Use this new hook in the DRI code so that DRI clients are notified and
receive updated information.
Signed-off-by: Keith Packard <keithp@keithp.com>
The xfree86 server previously hat NewInputDeviceRequest and InitInput, and
both basically did the same thing. Reduce NIDR to parameter checking and use
xf86NewInputDevice from both InitInput and NIDR to actually create the device.
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Add some commentary about future directions needed for the GLX drawable
creation and destruction code.
Match xalloc with xfree.
I made some minor formatting improvements.
(cherry picked from commit b772d64fce31d16b498c621096e39d5203994d6e)
In attach() check for pDraw being NULL, and also print an ErrorF message,
because we eventually want to track down why this is occuring.
It's unclear how this occurs, but as I noted in the 1.4 branch, I believe that
the DrawablePtr/struct _Drawable -> id is the member being accessed that causes
KERN_PROTECTION_FAILURE at 0x0000000000000004
This passes my tests using: env LIBGL_ALWAYS_INDIRECT=1 ./sometest.
I fixed a warning: caused by initializing the screen->base.visuals with the
configs. It is a ** not a *. It seems that some other part of GLX will
initialize this for us.
(cherry picked from commit 17f6a261fca6d5856069dce28bb4838261afc6bc)
When the linux kernel sets the NX bit vm86 segfaults when it tries to execute
code in memory that is not marked EXEC. Such code gets called whenever
we return from a VBIOS call to signal the calling program that the call
is actually finished and that we are not trapping for other reasons (like
IO accesses).
Use mprotect(2) to set these memory ranges PROT_EXEC.
There's little chance that we'll get the input devices at runtime without HAL,
we might as well force the server to add mouse/kbd devices automatically -
just like in the olden days.
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
These values need not be constrained to integer values.
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Added a configure option called --enable-standalone-xpbproxy which is useful for deveoping xpbproxy.
The 'active' switch in preferences just disables the in-server xpbproxy (not this standalone).
(cherry picked from commit 42944936326ef8732f622db9f75b79a92980550d)
This prevents visuals with odd sizes. The machine I use didn't have
this problem, but it shows up on some others.
(cherry picked from commit ed181382ddeb77019577d39b9c06b1cd839e18e4)
Use the settings queried from the system in xprScreen.c, rather than those 2 calls.
The 2 calls increased the total number of visuals a great deal (when using GLXEXT),
and not all of the visuals were usable with GLX. Some of the visuals aren't usable
with GLX still, such as DirectColor, but that seems to be acceptable based on my
understanding of the manual that states "a subset of visuals are made available
for OpenGL rendering."
(cherry picked from commit 373b8a5f32509722e06d8776109e6c3f06645ee0)
It was returning inverted values in comparison to the 1.4 branch. This resulted in
the windows not drawing due to a deep path of: RootlessReorderWindow ->
SCREENREC(pScreen)->imp->DoReorderWindow(winRec) - > xprDoReorderWindow ->
AppleWMDoReorderWindow.
(cherry picked from commit d1d398db762fe7d5720f9b5fafa2bbe8372fceac)
As reported in http://bugs.freedesktop.org/show_bug.cgi?id=18438
the server suggests reconfiguring HAL if AllowEmptyInput is enabled
and no input devices are known.
Instead of that notice, if HAL is disabled at configure time,
AllowEmptyInput is enabled in the config and no input devices are
found report those facts and recommend disabling AllowEmptyInput.
When using any XAAPixmapOps, we call into unknown but freshly
unwrapped callbacks (like fb ones). Unlike the XAA*Fallback calls,
we did so without syncing first, exposing us to all kinds of
synchronisation issues.
I believe that the rendering errors appeared now because *PaintWindow
vanished (e4d11e58), and we just use miPaintWindow instead. This
takes a less direct route to the hw and ends up at
PolyFillRectPixmap, which very often left drawing artifacts.
We now sync accordingly, and no longer get the rendering artifacts i
was methodically reproducing on radeonhd, radeon, unichrome...
Also, in order to allow driver authors to remove extensive syncing
or flushing to hide this issue, create XAA_VERSION_ defines, put
them in xaa.h and bump the patchlevel.
(novell bug #435791)
When setting the depth to 24, leave bpp unset so the logic to pick
a supported value is used instead of ignoring the driver's preference
and forcing 32 bpp.
GL/capabilities.c: Add handleDepthModes(), and extend
handleRendererDescription() for the various depth and multisampling flags.
Add initialization of the new config options to initConfig().
GL/capabilities.h: Add depth and multisample config members.
GL/visualConfigs.c: Add depth and multisampling support to the visual config
setup.
(cherry picked from commit f527381eea6a8ae6cd791475b2060d21fcf8efb2)
GL/capabilities.c: #if 0 (for now) any capabilities above 8 bits per channel,
because they introduce drawing problems.
GL/indirect.c: Comment out some visual setup code that shouldn't be running, and
actually seemed to cause some problems. The current visualConfigs.c code seems
to do a reasonable job of setting up visuals for XQuartz.
GL/visualConfigs.c: Make use of the proper visual .class. Eliminate depth 0.
It seems we really just want 24 for now, and 0 I think was a flaw in the original
code.
(cherry picked from commit 1e5f63f15e13a40a6e69a1505934d10d6990b6a2)
Works around a silly bug in the kernel that causes wakeup storms after
too many keypresses. Should fix the kernel bug too, but this at least
keeps the idle wakeup count below 1000/sec.
When a user specifies the position of an output for which no modes exist
(for whatever reason) assume that the width and height of this output
is 0. The result will be the same as if this output isn't taken into
consideration at all and thus should be sane. It will prevent a segfault
when trying to determine the width and height of a non-existent mode.
This is used in libGL to determine a preferred visual. glxinfo -v will also
report "visualCaveat=Slow" for such visuals.
(cherry picked from commit d5c17f585a240be9f9135975738e62951cde3353)
Also merge sun_bios.c into sun_vid.c and upstream Solaris patch to
keep aperture device open, to allow mappings to occur after X server
has given up uid 0.
Maybe one day I stop doing stupid patches like
a3a7c12fcf.
So, if X < low, reset to low, and _not_ to high.
If X > high, reset to high, and _not_ to low.
It seems that we were calling copy_completed: too soon. Now we return instead
of falling through to the free_propdata, and copy_completed:.
(cherry picked from commit 5797d5b662817a3369a74a673f4a7c995cde75cf)
Eterm ends up setting the type of the property to TARGETS, instead of ATOM.
That's why it wasn't working.
(cherry picked from commit 9be5998ede7427f14dd9597e117b3d6b427ba4e5)
This may work, unfortunately I don't have test apps that fail.
The way it works is by using an NSImage class initWithPasteboard:
method, which we then get the TIFFRepresentation of, and convert
to PNG or JPEG.
The TIFFRepresentation uses NSTIFFCompressionNone; which should be
lossless.
(cherry picked from commit 8d048cfa956f4a0860250cc836a6748912b37ad8)
This gives us a *lot* more visuals, and they should be more accurate.
Some of the visuals may have been wrong before. This may enable more
OpenGL programs to behave correctly.
I also suspect libGL needs a little work to handle some of these correctly.
(cherry picked from commit 98a084ce9c455260ecdb5f8eecb67f33dbb53b69)
This consists of two parts:
In the implicit server layout, ignore those drivers when looking for a core
device.
And after finishing the server layout, run through the list of devices and
remove any that use mouse or kbd.
AEI is mutually exclusive with the kbd and mouse drivers, so pick either - or.
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
The precedence of == is higher than that of &, so that code was
probably buggy.
xf86Init.c: In function 'DoModalias':
xf86Init.c:300: warning: suggest parentheses around comparison in operand of &
xf86Init.c:304: warning: suggest parentheses around comparison in operand of &
xf86Init.c:308: warning: suggest parentheses around comparison in operand of &
xf86Init.c:136: warning: function declaration isn't a prototype
xf86Init.c:243: warning: function declaration isn't a prototype
xf86Init.c:249: warning: function declaration isn't a prototype
xaaInit.c: In function 'XAAInit':
xaaInit.c:201: warning: implicit declaration of function 'miInitializeCompositeWrapper'
xaaInit.c:201: warning: nested extern declaration of 'miInitializeCompositeWrapper'
Add missing includes to fix the following warnings:
xf86DGA.c: In function 'DGAProcessKeyboardEvent':
xf86DGA.c:1050: warning: implicit declaration of function 'UpdateDeviceState'
xf86DGA.c:1050: warning: nested extern declaration of 'UpdateDeviceState'
xf86Xinput.c: In function 'xf86ActivateDevice':
xf86Xinput.c:303: warning: implicit declaration of function 'AssignTypeAndName'
xf86Xinput.c:303: warning: nested extern declaration of 'AssignTypeAndName'
xf86Xinput.c:311: warning: implicit declaration of function 'DeviceIsPointerType'
xf86Xinput.c:311: warning: nested extern declaration of 'DeviceIsPointerType'
xf86Xinput.c:324: warning: implicit declaration of function 'XkbSetExtension'
xf86Xinput.c:324: warning: nested extern declaration of 'XkbSetExtension'
Add automatic detection of the graphic driver to load for sbus devices.
This allows xorg to work on those devices without a "Device" section.
Debian bug#483942.
Signed-off-by: Julien Cristau <jcristau@debian.org>
We need to revisit the GL/indirect.c changes. I think indirect.c should
be using the same feature detection and GLX visual config code. The
indirect changes will require testing...
(cherry picked from commit 49b7a7c6f20ca85d4d624444eb83db2f91344591)
Also set AutoAddDevices and AutoEnableDevices to their defaults.
And in doing so, switch the rest of the defaults over to named intializers.
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
Add dump_prefs() for instrospection.
Make prefs_get_bool a little more clear.
In get_property don't assume that the Mac realloc(NULL, 0); returns
non-NULL. We shouldn't depend on implementation-defined behavior too
much.
Add more commentary in various spots.
Add TARGETS in the TARGETS response. I don't know why some X11 apps
do this, but it could be something that some weird toolkit/app tests
for.
In reload_preferences add preferences synchronization, so that xpbproxy
reloads the preferences, and picks up any changes.
Fix a typo/misspelling of the sync_clipboard_to_pasteboard preference.
(cherry picked from commit 165cbbb90c793d9335ea76f0274b1f60ad295903)
Usually, the console is set to RAW in the kbd driver. If we hotplug all input
devices (i.e. the evdev driver for keyboards) and the console is left as-is.
As a result, the evdev driver must put an EVIOCGRAB on the device to avoid
characters leaking onto the console. This again breaks many things, amongst
them lirc, in-kernel mouse button emulation and HAL.
This patch sets the console to RAW if AllowEmptyInput is on.
Use-cases:
1. AEI is off
1.1. Only kbd driver is used - behaviour as-is.
1.2. kbd and evdev driver is used: if evdev does not grab the device,
duplicate events are generated.
2. AEI is on
2.1. Only evdev driver is used - behaviour as-is, but evdev does not need
to grab the device anymore.
2.2. evdev and kbd are used: duplicate key events are generated if evdev
does not grab the device.
1.2 is a marginal use-case that can be fixed by adding a "grab" option to the
evdev driver (update of xorg.conf is needed).
2.2 is an issue. If we have no ServerLayout section, AEI is on, but devices
specified in the xorg.conf are still added [1], resulting in duplicate events.
This is a common configuration and needs sorting out.
[1] 2eaed4a10f
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
I've seen about one case in three years where this has actually been
correlated with the real cause of failure, and we've trained people to
freak out about X_WARNING, so let's be less alarmist.
Very cute, Samsung, not only do you claim to be 16cm by 9cm in the
global size record, you also claim to be 160mm by 90mm in the detailed
timings. Grrr.
If absolute events were posted, dixflags got set conditionally on whether the
valuators are different from the last posted set of values.
If dixflags are undefined however, the DIX interprets them as relative
valuators. Fix this by making sure defining dixflags is always defined.
X.Org Bug 17724 <http://bugs.freedesktop.org/show_bug.cgi?id=17734>
NIDR should be used to create a new SD from e.g. within a driver.
DIDR should be used to remove a device from the server.
Signed-off-by: Peter Hutterer <peter.hutterer@redhat.com>
Use ErrorF for an error message after an xcalloc failure, and return instead of falling through to GlxSetVisualConfigs, or abort()ing at the test branch.
(cherry picked from commit 1056700971fd5c034396ed6dbea15e092f0c6332)
Commit the darwin.c changes I missed in the last commit, for calling
setVisualConfigs().
(cherry picked from commit eb3c014e1710bf0b93bda10ddb9b795cd150d02d)
Add a setVisualConfigs that is called before the fbScreenInit, to setup the __GLXvisualConfigs.
(cherry picked from commit fc86f9e4482043eca76d9d7a96e166be1aabf674)
If you need to bail out the server, use Ctrl-Alt-Fx, or enable zapping
if it bothers you that much. If Ctrl-Alt-Fx is broken, nag me until
it's permanently fixed.
The nvidia driver currently uses these hooks to work around problems where RAC
will disable access to the hardware at unexpected times. This change restores
these hooks until we can come up with a better API for working around RAC.
This reverts commit c1df4fbede.
The nvidia driver currently uses these callbacks to work around problems where
RAC will disable access to the hardware at unexpected times. This change
restores these hooks until we can come up with a better API for working around
RAC.
This reverts commit d7c0ba2e9e.
Conflicts:
hw/xfree86/loader/xf86sym.c
Some BIOSes (hi XGI!) will attempt to enumerate the PCI bus by asking
for the config space of every possible device number. This despite
perfectly functional BIOS methods to enumerate the bus exactly.
If anyone can come up with an example of a bus where:
- both i/o and memory resources are addressable
- access to them can be controlled
- but they can't be controlled independently
then by all means, reinstate this logic.
instead of calling CFRunLoopRun() directly. The leak wasn't reproducible on
this machine, but someone was able to produce a leak trace with Instruments
that indicates it was leaking in the CFRunLoopRun() path.
x-input.m: dequeue and ignore events when pbproxy_active is false.
x-selection.h: add an is_active method that is used by x-input.m to ignore
events.
x-selection.m: Handle nearly every preference, except for primary_on_grab,
which I don't really understand yet.
(cherry picked from commit 4d51ad851e64da83cbdfb0a4a22428418a7bcf75)
Remove some unnecesssary headers.
Remove some dead code that was never called or used in pbproxy.
Make use of an NSAutoreleasePool in x_init. It could potentially cause a leak
on a startup without this.
Start adding reload_preferences to the x_selection class, as well as event
handling for that.
(cherry picked from commit 602e8ba8f7ee196696bc9e3cea6ecdf3200dcf5c)
Fix the usage of the NSString cStringUsingEncoding: - it doesn't NUL
terminate the string, which lead to a bus error. So, we use
lengthOfBytesUsingEncoding: to get the length in bytes instead of
strlen().
(cherry picked from commit 6333d619e747c3b6bd3ba7557e35c0e5f6daa40f)
2 of the paths leaked, when INCR transfers were done. Now we
are leak free according to the leaks program for all transfers
I have tried so far.
(cherry picked from commit aa98db576bc02f0765cb35d0282a894ba3778213)
NSAutoreleasePool. Now the usage is consistent. In x_input_run()
we create a pool, and release it after processing the XEvents.
Add some getpid() output to main for debugging. It needs a bit more
testing before the next release.
Don't retain the NSPasteboard as the old code did. That may have
contributed to the leak, and it made it so that we needed the
NSAutoreleasePool created in main().
Remove the _known_types, and _pasteboard instance variables from
the x_selection class. They aren't needed anymore.
The leaks program now indicates 0 leaks after some usage. I want
to test further, but this seems much better, and my memory usage
graph indicates it's not growing.
(cherry picked from commit b245d84a72ee3929546cd11a6eba3c60fb4a4d95)
macro causes a leak (according to the leaks program).
Attempt to fix several other leaks with release method calls.
For some reason the process still grows more than it should...
I will need to use some better methods than leaks, and malloc_history
I suspect. Whatever is leaking, it's hard to find. I need to isolate
the cases more.
Add a missing image/jpeg branch.
Remove read_prop_32 - it's not used.
(cherry picked from commit 63a680354dcb545fef935ac97596dd35ceaed960)
According to the press release:
Previous SGI contributions to the free and open source community
are now available under the new license. These contributions
include the SGI® OpenGL® Sample Implementation, the GLX™ API and
other GLX extensions.
[...]
"SGI has been one of the most ardent commercial supporters of free
and open source software, so it was important to us that we continue
to support the free software development community by releasing our
earlier OpenGL-related contributions under this new license," said
Steve Neuner, director of Linux, SGI. "This license ensures that all
existing user communities will benefit, and their work can proceed
unimpeded. Both Mesa and the X.org Project can continue to utilize
this code in free software distributions of GNU/Linux. Now more than
ever, software previously released by SGI under earlier GLX and SGI
Free Software License B is free."
"The GLX API" is here read to include the original GLX source release
from:
http://www.sgi.com/products/software/opensource/glx/download.html
ftp://ftp.sgi.com/sgi/opengl/glx/glx1_2.1.tgz
which includes glxext.c as included in XFree86, from which our copies
in glx/ and hw/dmx/glxProxy/ are derived.
Adam Jackson
Alan Coopersmith
Keith Packard
Peter Hutterer
Matthieu Herrb
Stefan Dirsch
Jeremy Huddleston
George Staplin
Egbert Eich
Rémi Cardona
Michel Dänzer
Ben Byer
Alex Deucher
Adam Jackson
James Cloos
Luc Verhaegen
Julien Cristau
George Peter Staplin
Egbert Eich
Daniel Stone
Bernhard R. Link
Alan Hourihane
Kristian Høgsberg
Nathaniel McCallum
Aaron Plattner
Maarten Maathuis
Zhenyu Wang
George Staplin
Tiago Vignatti
Dave Airlie
Roland Scheidegger