When using reverse prime we do 2 copies, 1 from the primary GPU's
framebuffer to a shared pixmap and 1 from the shared pixmap to the
secondary GPU's framebuffer.
This means that on the primary GPU side the copy MUST be finished,
before we start the second copy (before the secondary GPU's driver
starts processing the damage on the shared pixmap).
This fixes secondary outputs sometimes showning (some) old fb contents,
because of the 2 copies racing with each other, for an example of
what this looks like see:
https://fedorapeople.org/~jwrdegoede/IMG_20160915_130555.jpg
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
In Xwayland's xwl_unrealize_cursor(), the x_cursor is cleared up only
when a device value is provided to the UnrealizeCursor() routine, but
if the device is NULL as called from FreeCursor(), the corresponding
x_cursor for the xwl_seat is left untouched.
This might cause a segfault when trying to access the unrealized
cursor's devPrivates in xwl_seat_set_cursor().
A possible occurrence of this is the client changing the cursor, the
Xserver calling FreeCursor() which does UnrealizeCursor() and then
the Wayland server sending a pointer enter event, which invokes
xwl_seat_set_cursor() while the seat's x_cursor has just been
unrealized.
To avoid this, walk through all the xwl_seats and clear up all x_cursor
matching the cursor being unrealized.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
wl_display_flush() can fail with EAGAIN and Xwayland would make this a
fatal error.
When this happens, it means that Xwayland has flooded the Wayland file
descriptor, either because the Wayland compositor cannot cope or more
likely because of a deadlock situation where the Wayland compositor is
blocking, waiting for an X reply while Xwayland tries to write data to
the Wayland file descriptor.
The general consensus to avoid the deadlock is for the Wayland
compositor to never issue blocking X11 roundtrips, but in practice
blocking rountrips can occur in various places, including Xlib calls
themselves so this is not always achievable without major surgery in the
Wayland compositor/Window manager.
What this patch does is to avoid dispatching to the Wayland file
descriptor until it becomes available for writing again, while at the
same time continue processing X11 requests to release the deadlock.
This is not perfect, as there is still the possibility of another X
client hammering the connection and we'll still fail writing to the
Wayland connection eventually, but this improves things enough to avoid
a 100% repeatable crash with vlc and gtkperf.
Also, it is worth considering that window managers and Wayland
compositors such as mutter already have a higher priority than other
regular X clients thanks to XSyncSetPriority(), mitigating the risk.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1278159
Bugzilla: https://bugzilla.gnome.org/show_bug.cgi?id=763400
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Keeping the shm fd open beyond pixmap creation means we can easily
reach the open file descriptor limit if an X client asks us to create
that many pixmaps. Instead, let's get the wl_buffer immediatly so that
we can destroy the shm pool and close the fd before being asked to
create more.
Tested-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
glUniform4ui is available starting in GL{,ES} 3.0. Technically it's
also in EXT_gpu_shader4, but that's not worth supporting. There was also
a MESA_shading_language_130 spec proposed at one point; if that ever
gets finished, we can update epoxy to know about it and fix up the
feature check.
Signed-off-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Once I introduced a second X server being tested, I found that startx
hit races in choosing a display.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
I was confused by the behavior I'd written before. keithp and
mattst88 responded with shock that I would have made it so surprising,
as well.
v2: Point to index.html instead of problems.html, which won't exist if
we had no problems.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
The "dmesg-fail" line was matching our "fail" regex, so if you didn't
have those we would ignore fails.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
The normal xinit is racy because it doesn't use -displayfd. This
implements the bare minimum for testing purposes, using -displayfd to
sequence starting the client, and avoids adding yet another dependency
to the server.
v2: Fix asprintf error checks.
v3: Add error checking for fork(), clarify calloc() arg.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
This speeds up headless testing of Xephyr -glamor with softpipe from
"a test per minute or so" to "a test every few seconds".
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Even if the pixmap's storage has alpha, it may have been uploaded with
garbage in the alpha channel, so we need to force the shader to set
alpha to 1. This was broken way back in
355334fcd9.
Fixes rendercheck -t composite -f x8r8g8b8.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
The copy optimization in d37329cba4
replicated a bug from last time we did a copy optimization: CopyArea
is only defined for matching depths. This is only a problem at 15 vs
16 depth right now (24 vs 32 would also have matching Render formats,
but they should work) but be strict in case we store other depths
differently in the future.
Fixes rendercheck -t blend -o src -f x4r4g4b4,x3r4g4b4
v2: Drop excessive src->depth == dst->depth check that snuck in.
v3: Switch back to src->depth == dst->depth
v4: Touch up commit message (s/bpp/depth).
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
This function is used by the modesetting driver to implement DRI2 and
shouldn't fail on systems that don't support DRI3.
v2: Drop stale commit message wording, fix compiler warning (by anholt)
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
v2: Fix "orignal" too (review feedback by ajax, change by anholt)_
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Eric Anholt <eric@anholt.net>
This code is safe. If the data race fails, the result is that we take the
lock and recheck.
==================
WARNING: ThreadSanitizer: data race (pid=31401)
Read of size 1 at 0x00010f5d2500 by thread T11:
#0 wait_for_mieq_init darwinEvents.c:102 (X11.bin+0x00010003155a)
#1 -[X11Application(Private) sendX11NSEvent:] X11Application.m:1330 (X11.bin+0x00010001d652)
#2 __28-[X11Application sendEvent:]_block_invoke X11Application.m:476 (X11.bin+0x00010001887f)
#3 __tsan::invoke_and_release_block(void*) <null>:144 (libclang_rt.tsan_osx_dynamic.dylib+0x00000005d97b)
#4 _dispatch_client_callout <null>:33 (libdispatch.dylib+0x0000000020ef)
Previous write of size 1 at 0x00010f5d2500 by thread T8:
[failed to restore the stack]
Location is global 'mieqInitialized' at 0x00010f5d2500 (X11.bin+0x000100599500)
Thread T11 (tid=4367138, running) created by thread T-1
[failed to restore the stack]
Thread T8 (tid=4367130, running) created by main thread at:
#0 pthread_create <null>:144 (libclang_rt.tsan_osx_dynamic.dylib+0x000000024490)
#1 create_thread quartzStartup.c:78 (X11.bin+0x000100039d2d)
#2 QuartzInitServer quartzStartup.c:95 (X11.bin+0x000100039b96)
#3 X11ApplicationMain X11Application.m:1238 (X11.bin+0x00010001cd54)
#4 X11ControllerMain X11Controller.m:984 (X11.bin+0x00010002a5b2)
#5 server_main quartzStartup.c:136 (X11.bin+0x000100039fbb)
#6 do_start_x11_server bundle-main.c:436 (X11.bin+0x000100002e25)
#7 _Xstart_x11_server mach_startupServer.c:189 (X11.bin+0x000100004e09)
#8 mach_startup_server mach_startupServer.c:399 (X11.bin+0x0001000056a4)
#9 mach_msg_server mach_msg.c:563 (libsystem_kernel.dylib+0x000000012186)
#10 start <null>:29 (libdyld.dylib+0x000000005254)
SUMMARY: ThreadSanitizer: data race darwinEvents.c:102 in wait_for_mieq_init
==================
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
This allows us to remove darwinEvents_lock() and darwinEvents_unlock()
and remove the serverRunning hack from dix
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
AppKit handles crashes on app launch with their own dialog now, so we shouldn't need to do this ourselves.
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
When a client is marked as write blocked, clear any old 'write ready'
bit in the osfds structure so that a new indication of write ready
(which is marked as edge trigggered) will trigger the callback.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Tested-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Tested-by: Matthieu Herrb <matthieu@herrb.eu>
When a client with pending output is ready (has request data pending),
FlushAllOutput will skip it to get all of the requests processed
before sending any queued output. That means FlushAllOutput is going
to return with some output pending to a client which isn't known to be
write blocked. And that means NewOutputPending needs to be set so that
FlushAllOutput will get called again to actually go flush this client.
It might be interesting to try just flushing the client to send any
queued data along the way. This patch just restores the server
behavior to what it was before the ospoll changes.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Tested-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Tested-by: Matthieu Herrb <matthieu@herrb.eu>
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
RedirectKey() action had been broken by commit 2e6190.
A dropped check caused over-intense autorepeat of keysyms enriched
with the action.
Previous to this commit, the check wrapped the entire switch() block,
which was dropped with the move to a separate function.
Restore the checking.
Signed-off-by: Mihail Konev <k.mvc@ya.ru>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This function will be called when a pointer is grabbed non-root window
set as the 'confineTo'. This will enable the ddx to handle the
confinement their own way.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
This new vfunc will be called, if set, after a client has issued a
WarpPointer request. This is necessary for implementing pointer warp
emulation in Xwayland.
Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Fixes XRRGetOutputPrimary and xrandr not reporting a primary output after
startup. This was especially confusing when an output was explicitly
marked as primary using Option "Primary" in Section "Monitor".
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Move ms_flush_drm_events out of GLAMOR ifdef.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=97586
Signed-off-by: Qiang Yu <Qiang.Yu@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
pClient does not contain a live value after the transition to lists
https://bugs.freedesktop.org/show_bug.cgi?id=97765
Application Specific Information:
X.Org X Server 1.18.99.1 Build Date: 20160910
=================================================================
==16921==ERROR: AddressSanitizer: global-buffer-overflow on address 0x000108ce3834 at pc 0x000108880766 bp 0x7000045f76c0 sp 0x7000045f76b8
READ of size 4 at 0x000108ce3834 thread T6
#0 0x108880765 in SmartScheduleClient dispatch.c:365
#1 0x10887ecc5 in Dispatch dispatch.c:422
#2 0x1088c05f1 in dix_main main.c:301
#3 0x1082aabba in server_thread quartzStartup.c:66
#4 0x7fffc5f16aaa in _pthread_body (libsystem_pthread.dylib+0x3aaa)
#5 0x7fffc5f169f6 in _pthread_start (libsystem_pthread.dylib+0x39f6)
#6 0x7fffc5f161fc in thread_start (libsystem_pthread.dylib+0x31fc)
Regressed-in: 8f1edf4bd3
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
The dbus teardown code is called when the server fatal errors even if
that is before dbus has ever been initialized. By statically
initializing the value of bus_info.fd, we avoid calling RemoveNotifyFd
on stdin.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Julien Cristau <jcristau@debian.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
If windowsdriproto headers are available, build a Windows-DRI extension,
which supports requests to enable local clients to directly render GL to a
Windows drawable:
- a query to check if WGL is being used on a screen
- a query to map a fbconfigID to a native pixelformatindex
- a query to map a drawable to a native handle
Windows-DRI can only be useful if we are using WGL, so make an note if WGL
is active on a screen.
Make validGlxDrawable() public
Adjust glxWinSetPixelFormat() so it doesn't require a context, just a
screen and config.
That enables factoring out the deferred drawable creation code as
glxWinDeferredCreateDrawable()
Enhance glxWinDeferredCreateDrawable(), so that pixmaps are placed into a
file mapping, so they exist in memory which can be shared with the direct
rendering process.
Currently, this file mapping is accessed by a name generated from the XID.
This will not be unique across multiple server instances. It would perhaps
be better, although more complicated, to use an anonymous file mapping, and
then duplicate the handle for the direct rendering process.
Use glxWinDeferredCreateDrawable() to ensure the native handle exists for
the Windows-DRI query to map a drawable to native handle.
v2:
Various printf format warning fixes
v3:
Fix format warnings on x86
Move some uninteresting windows-dri output to debug log level
v4:
check for windowsdriproto when --enable-windowsdri
use windowsdriproto_CFLAGS
Signed-off-by: Jon Turney <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>
This ensures that the deviceProc is never called while the input
thread is processing data from the device.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This keeps the input driver SetProperty function from being called
while input events are being processed.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Autoconf logic borrowed from glib
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Tested-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
If RTLD_DI_SETSIGNAL is set to let us turn runtime linker/loader errors
into catchable signals, then we should only show the errors when catching
that signal, instead of tossing out red herrings to distract people with
unrelated crashes long after their last failed symbol lookup (especially
when using drivers built to support multiple API's by checking which
symbols are available before calling them).
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Aaron Plattner <aplattner@nvidia.com>
O_CLOEXEC is not a file bit. It is not setable with F_SETFL. One must
use it when calling open(2). To set it cloexec on an existing fd,
F_SETFD and FD_CLOEXEC must be used.
This also fixes a build failure regression on configurations that don't
have O_CLOEXEC defined.
cf: http://pubs.opengroup.org/onlinepubs/9699919799/functions/fcntl.html
Regressed-in: 30ac756798
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Reviewed-by: Julien Cristau <jcristau@debian.org>
