If there is one platform device, which is not paused nor resumed,
systemd_logind_vtenter() will never get called.
This break suspend/resume, and switching to VT on system with Nvidia
proprietary driver.
This is a regression introduced by f5bd039633
So now call systemd_logind_vtenter() if there are no paused
platform devices.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1271
Fixes: f5bd0396 - xf86/logind: fix call systemd_logind_vtenter after receiving drm device resume
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Tested-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
This is minor, but that error message says a wrong function name.
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Initially reported downstream in Gentoo. Manifests with errors like:
```
gnu/bin/ld: hw/xfree86/common/libxorg_common.a(xf86fbBus.c.o): in function `xf86ClaimFbSlot':
xf86fbBus.c:(.text+0x20): undefined reference to `sbusSlotClaimed'
/usr/lib/gcc/sparc-unknown-linux-gnu/11.2.0/../../../../sparc-unknown-linux-gnu/bin/ld: xf86fbBus.c:(.text+0x2c): undefined reference to `sbusSlotClaimed'
```
While we use the headers in meson.build, we don't reference xf86sbusBus.c
which defines the missing symbols like sbusSlotClaimed.
Bug: https://bugs.gentoo.org/828513
Signed-off-by: Sam James <sam@gentoo.org>
This commit allows X11 clients running through Xwayland to lease
non-desktop connectors from the Wayland compositor by implementing
support for drm-lease-v1.
In order to not deadlock with the Wayland compositor if its response
to a lease request is delayed, the new interface in _rrScrPriv
introduced in the last commit is used, which makes it possible to
block the X11 client while a response is pending.
Leasing normal outputs is not yet supported, all connectors offered
for lease will be advertised as non-desktop.
Co-authored-by: Xaver Hugl <xaver.hugl@gmail.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
Add a new interface to _rrScrPriv to make it possible for the server to
delay answering a lease request, at the cost of blocking the client. This
is needed for implementing drm-lease-v1, as the Wayland protocol has no
defined time table for responding to lease requests.
Signed-off-by: Xaver Hugl <xaver.hugl@gmail.com>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
This was overlooked when converting the function to use libxcvt.
Bring back name initialization from old code.
This was causing a segfault in xf86LookupMode() if modes where
name is NULL are present the modePool list.
Signed-off-by: Matthieu Herrb <matthieu@herrb.eu>
The compositor may send us wl_seat and its capabilities before sending
e.g. relative_pointer_manager or pointer_gesture interfaces. This would
result in devices being created in capabilities handler, but listeners
not, because the interfaces weren't available at the time. So we
manually attempt to setup listeners again.
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
The implementation is relatively straightforward because both wayland
and Xorg use libinput semantics for touchpad gestures.
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
Currently, when given the choice, Xwayland will pick the GBM backend
over the EGLstream backend if both are available, unless the command
line option “-eglstream” is specified.
The NVIDIA proprietary driver had no support for GBM until driver series
495, but starting with the driver series 495, both can be used.
But there are other requirements with the rest of the stack, typically
Mesa, egl-wayland, libglvnd as documented in the NVIDIA driver.
So if the NVIDIA driver series 495 gets installed, Xwayland will pick
the GBM backend even if EGLstream is available and may fail to render
properly.
To avoid that issue, prefer EGLstream if EGLstream and all the Wayland
interfaces are available, and fallback to GBM automatically unless
“-eglstream” was specified.
With this, the compositor, given the choice, can decide which actual
backend Xwayland would use by advertising (or not) the Wayland
"wl_eglstream_controller" interface.
This change has no impact on compositors which do not have support for
EGLstream in the first place.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
Add (verbose) statements to trace the actual backend used with glamor.
That can be useful for debugging.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
On a normal startup sequence, the Xwayland glamor backend would log
an error whenever a required Wayland protocol is missing.
Those are not really errors though, more informational messages along
the glamor backend selection process.
Demote those errors to verbose messages to reduce the verbosity of
Xwayland at startup by default.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
If no EGLstream capable device is found at startup, Xwayland's EGLstream
backend will log an error message "glamor: No eglstream capable devices
found".
However, considering that the vast majority of drivers do not implement
EGLstream, the lack of EGLstream capable device is more of the norm than
the exception.
Change the error message to a log verbose message.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
When switching to VT, the ioctl DRM_DROP_MASTER must be done before
the ioctl VT_RELDISP. Otherwise the kernel can't change the modesetting
reliably, and this leads to the console not showing up in some cases, like
after unplugging a docking station with a DP or HDMI monitor.
Before doing the VT_RELDISP, send a dbus message to logind, to
pause the drm device, so logind will do the ioctl DRM_DROP_MASTER.
With this patch, it changes the order logind will send the resume
event, and drm will be sent last instead of first.
so there is a also fix to call systemd_logind_vtenter() at the right time.
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
logind send the resume event for input devices and drm device,
in any order. if we call vt_enter before logind resume the drm device,
it leads to a driver error, because logind has not done the
DRM_IOCTL_SET_MASTER on it.
Keep the old workaround to make sure we call systemd_logind_vtenter at
least once if there are no platform device
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Quite a lot of applications currently expect the screen DPI exposed by
the X server to be 96 even when the real display DPI is different.
Additionally, currently Xwayland completely ignores any hardware
information and sets the DPI to 96. Accordingly the new behavior, even
if it fixes a bug, should not be enabled automatically to all users.
A better solution would be to make the default DPI stay as is and enable
the correct behavior with a command line option (maybe -dpi auto, or
similar). For now let's just revert the bug fix.
This reverts commit 05b3c681ea.
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
We were storing the pointer to struct glamor_context. However, glamor
itself is storing the EGLContext pointer since the commit below. Since
the two values could never be equal, this resulted in constant
superfluous eglMakeCurrent calls. The implicit glFlush triggered by
those couldn't be good for performance.
Fixes: 7c88977d33 "glamor: Store the actual EGL/GLX context pointer in lastGLContext"
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
The code to clear a cursor pending frame callback was duplicated in
multiple places in the code.
Introduce a new xwl_cursor_clear_frame_cb() function and remove the
duplicated code.
No functional change.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Carlos Garnacho <carlosg@gnome.org>
It just make more sense to keep xwl_cursor_release() with the rest of
the cursor code.
No functional change.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Carlos Garnacho <carlosg@gnome.org>
Two different functions in xwayland-cursor.c and xwayland-input.c use
the same name xwl_seat_update_cursor() which is confusing when reading
the code.
Rename xwl_seat_update_cursor() to xwl_seat_update_all_cursors() in
xwayland-cursor.c to help with readability of the code.
No functional change.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Carlos Garnacho <carlosg@gnome.org>
Passing -noTouchPointerEmulation results in an error about the
flag not being recognized.
Signed-off-by: Simon Ser <contact@emersion.fr>
Fixes: 7d34b1f2b7 ("xwayland: add -noTouchPointerEmulation")
If the tablet tool is moved out of proximity before the cursor's pending
frame callback is received, any further attempts to update the cursor
will fail because the frame callback is still pending.
Make sure to clear any cursor pending frame when the tool gets in
proximity again, similar to what we do when the pointer re-enters a
surface, so that the cursor updates aren't discarded.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
See-also: https://gitlab.gnome.org/GNOME/mutter/-/issues/1969
Reviewed-by: Carlos Garnacho <carlosg@gnome.org>
Some clients (typically Java, but maybe others) rely on ConfigureNotify
or RRScreenChangeNotify events to tell that the XRandR request is
successful.
When emulated XRandR is used in Xwayland, compute the emulated root size
and send the expected ConfigureNotify and RRScreenChangeNotify events
with the emulated size of the root window to the asking X11 client.
Note that the root window size does not actually change, as XRandR
emulation is achieved by scaling the client window using viewports in
Wayland, so this event is sort of misleading.
Also, because Xwayland is using viewports, emulating XRandR does not
reconfigure the outputs location, meaning that the actual size of the
root window which encompasses all the outputs together may not change
in a multi-monitor setup. To work around this limitation, when using an
emulated mode, we report the size of that emulated mode alone as the
root size for the configure notify event.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
SimpleDRM 'devices' are a fallback device, and do not have a busid
so they are getting skipped. This will allow simpledrm to work
with the modesetting driver
The "sync crtc" is the crtc used to drive the display timing of a
drawable under DRI2 and DRI3/Present. If a drawable intersects
multiple video outputs, then normally the crtc is chosen which has
the largest intersection area with the drawable.
If multiple outputs / crtc's have exacty the same intersection
area then the crtc chosen was simply the first one with maximum
intersection. Iow. the choice was random, depending on plugging
order of displays.
This adds the ability to choose a preferred output in such a tie
situation. The RandR output marked as "primary output" is chosen
on such a tie.
This new behaviour and its implementation is consistent with other
video ddx drivers. See amdgpu-ddx, ati-ddx and nouveau-ddx for
reference. This commit is a straightforward port from amdgpu-ddx.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
In a setup with both VRR capable and non-VRR capable displays,
it was so far inconsistent if the driver would allow use of
VRR support or not, as "is_connector_vrr_capable" was set to
whatever the capabilities of the last added drm output were.
Iow. the plugging order of monitors determined the outcome.
Fix this: Now if at least one display is VRR capable, the driver
will treat an X-Screen as capable for VRR, plugging order no
longer matters.
Tested with a dual-display setup with one VRR monitor and one
non-VRR monitor. This is also beneficial with the new Option
"AsyncFlipSecondaries".
When we are at it, also add some so far missing description of
the "VariableRefresh" driver option, copied from amdgpu-ddx.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
A lut size of 4096 slots has been verified to work correctly,
as tested with amdgpu-kms. Intel Tigerlake Gen12 hw has a very
large GAMMA_LUT size of 262145 slots, but also issues with its
current GAMMA_LUT implementation, as of Linux 5.14.
Therefore we keep GAMMA_LUT off for large lut's. This currently
excludes Intel Icelake, Tigerlake and later.
This can be overriden via the "UseGammaLUT" boolean xorg.conf option
to force use of GAMMA_LUT on or off.
See following link for the Tigerlake situation:
https://gitlab.freedesktop.org/drm/intel/-/issues/3916#note_1085315
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
With the GBM backend becoming usable with different drivers such as
NVIDIA, set the GLVND vendor to the same value as the GBM backend name.
Mesa implementation however returns "drm" so we need to special case
this value - Basically, for anything other than "drm" we simply assume
that the GBM backend name is the same as the vendor.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: James Jones <jajones@nvidia.com>
Tested-by: James Jones <jajones@nvidia.com>
Xwayland was passing GBM bos directly to
eglCreateImageKHR using the EGL_NATIVE_PIXMAP_KHR
target. Given the EGL GBM platform spec claims it
is invalid to create a EGLSurface from a native
pixmap on the GBM platform, implying there is no
mapping between GBM objects and EGL's concept of
native pixmaps, this seems a bit questionable.
This change modifies the bo import function to
extract all the required data from the bo and then
imports it as a dma-buf instead when the dma-buf +
modifiers path is available.
Signed-off-by: James Jones <jajones@nvidia.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Xwayland's xwl_shm_create_pixmap() computes the size of the shared
memory pool to create using a size_t, yet the Wayland protocol uses an
integer for that size.
If the pool size becomes larger than INT32_MAX, we end up asking Wayland
to create a shared memory pool of negative size which in turn will raise
a protocol error which terminates the Wayland connection, and therefore
Xwayland.
Avoid that issue early by return a NULL pixmap in that case, which will
trigger a BadAlloc error, but leave Xwayland alive.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
This reverts commit 617f591fc4.
The problem described in that commit exists, but the two
preceeding commits with improvements to the servers RandR
code should avoid the mentioned problems while allowing the
use of GAMMA_LUT's instead of legacy gamma lut.
Use of legacy gamma lut's is not a good fix, because it will reduce
color output precision of gpu's with more than 1024 GAMMA_LUT
slots, e.g., AMD, ARM MALI and KOMEDA with 4096 slot luts,
and some Mediathek parts with 512 slot luts. On KOMEDA, legacy
lut's are completely unsupported by the kms driver, so gamma
correction gets disabled.
The situation is especially bad on Intel Icelake and later:
Use of legacy gamma tables will cause the kms driver to switch
to hardware legacy lut's with 256 slots, 8 bit wide, without
interpolation. This way color output precision is restricted to
8 bpc and any deep color / HDR output (10 bpc, fp16, fixed point 16)
becomes impossible. The latest Intel gen gpu's would have worse
color precision than parts which are more than 10 years old.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
The assumption in the upsampling code was that the crtc->gamma_size
size of the crtc's gamma table is a power of two. This is true for
almost all current driver + gpu combos at least on Linux, with typical
sizes of 256, 512, 1024 or 4096 slots.
However, Intel Gen-11 Icelake and later are outliers, as their gamma
table has 2^18 + 1 slots, very big and not a power of two!
Try to make upsampling behave at least reasonable: Replicate the
last gamma value to fill up remaining crtc->gamma_red/green/blue
slots, which would normally stay uninitialized. This is important,
because while the intel display driver does not actually use all
2^18+1 values passed as part of a GAMMA_LUT, it does need the
very last slot, which would not get initialized by the old code.
This should hopefully create reasonable behaviour with Icelake+
but is untested on the actual Intel hw due to lack of suitable
hw.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
If randrp->palette_size is zero, the memcpy() path can read past the
end of the randr_crtc's gammaRed/Green/Blue tables if the hw crtc's
gamma_size is greater than the randr_crtc's gammaSize.
Avoid this by clamping the to-be-copied size to the smaller of both
sizes.
Note that during regular server startup, the memcpy() path is only
taken initially twice, but then a suitable palette is created for
use during a session. Therefore during an actual running X-Session,
the xf86RandR12CrtcComputeGamma() will be used, which makes sure that
data is properly up- or down-sampled for mismatching source and
target crtc gamma sizes.
This should avoid reading past randr_crtc gamma memory for gpu's
with big crtc->gamma_size, e.g., AMD/MALI/KOMEDA 4096 slots, or
Intel Icelake and later with 262145 slots.
Tested against modesetting-ddx and amdgpu-ddx under screen color
depth 24 (8 bpc) and 30 (10 bpc) to make sure that clamping happens
properly.
This is an alternative fix for the one attempted in commit
617f591fc4.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
We turn this on if the GL underneath us can enable GL_FRAMEBUFFER_SRGB.
We do try to generate both capable and incapable configs, which is to
keep llvmpipe working until the client side gets smarter about its srgb
capabilities.
xwl_present_reset_timer checks if the pending flip is synchronous, so
we need to call it after adding the pending flip to the flip queue.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1219
Fixes: b2a06e0700 "xwayland/present: Drop sync_flip member of struct xwl_present_window"
Tested-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Rotation is broken for all drm drivers not providing hardware rotation
support. Drivers that give direct access to vram and not needing dirty
updates still work but only by accident. The problem is caused by
modesetting not sending the correct fb_id to drmModeDirtyFB() and
passing the damage rects in the rotated state and not as the crtc
expects them. This patch takes care of both problems.
Signed-off-by: Patrik Jakobsson <pjakobsson@suse.de>
This is not actually a change for xwayland with gbm, or for xfree86 with
big-GL, but we do change them as well to use EGL_NO_CONFIG_KHR
explicitly.
Reviewed-by: Emma Anholt <emma@anholt.net>
There's no real benefit to using GLX, and the other DDXes are using EGL
already, so let's converge on EGL so we can concentrate the fixes in one
place.
We go to some effort to avoid being the thing that requires libX11 here.
We prefer EGL_EXT_platform_xcb over _x11, and if forced to use the
latter we'll ask the dynamic linker for XGetXCBConnection and
XOpenDisplay rather than link against xlib stuff ourselves. Xephyr is
now a pure XCB application if it can be.
Reviewed-by: Emma Anholt <emma@anholt.net>
Due to a typo in tablet_pad_group(), we would allocate a variable
("group") and test another one ("pad") for allocation success.
Spotted by covscan.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Fixes: commit 8475e63 - "xwayland: add tablet pad support"
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
On screen init, if any of the private type registration fails we would
return FALSE without actually freeing the xwl_screen we just allocated.
This is not a serious leak as failure at that point would lead to the
premature termination of Xwayland at startup, but covscan complains and
it's easy enough to fix.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
The Linux version of xf86EnableIO calls a helper function called hwEnableIO().
Except on Alpha, this function reads /proc/ioports looking for the 'keyboard'
and 'timer' ports, extracts the port ranges, and enables access to them. It does
this by reading 4 bytes from the string for the start port number and 4 bytes
for the last port number, passing those to atoi(). However, it doesn't add a
fifth byte for a NUL terminator, so some implementations of atoi() read past the
end of this string, triggering an AddressSanitizer error:
==1383==ERROR: AddressSanitizer: stack-buffer-overflow on address 0x7fff71fd5b74 at pc 0x7fe1be0de3e0 bp 0x7fff71fd5ae0 sp 0x7fff71fd5288
READ of size 5 at 0x7fff71fd5b74 thread T0
#0 0x7fe1be0de3df in __interceptor_atoi /build/gcc/src/gcc/libsanitizer/asan/asan_interceptors.cpp:520
#1 0x564971adcc45 in hwEnableIO ../hw/xfree86/os-support/linux/lnx_video.c:138
#2 0x564971adce87 in xf86EnableIO ../hw/xfree86/os-support/linux/lnx_video.c:174
#3 0x5649719f6a30 in InitOutput ../hw/xfree86/common/xf86Init.c:439
#4 0x564971585924 in dix_main ../dix/main.c:190
#5 0x564971b6246e in main ../dix/stubmain.c:34
#6 0x7fe1bdab6b24 in __libc_start_main (/usr/lib/libc.so.6+0x27b24)
#7 0x564971490e9d in _start (/home/aaron/git/x/xserver/build.asan/hw/xfree86/Xorg+0xb2e9d)
Address 0x7fff71fd5b74 is located in stack of thread T0 at offset 100 in frame
#0 0x564971adc96a in hwEnableIO ../hw/xfree86/os-support/linux/lnx_video.c:118
This frame has 3 object(s):
[32, 40) 'n' (line 120)
[64, 72) 'buf' (line 122)
[96, 100) 'target' (line 122) <== Memory access at offset 100 overflows this variable
HINT: this may be a false positive if your program uses some custom stack unwind mechanism, swapcontext or vfork
(longjmp and C++ exceptions *are* supported)
SUMMARY: AddressSanitizer: stack-buffer-overflow /build/gcc/src/gcc/libsanitizer/asan/asan_interceptors.cpp:520 in __interceptor_atoi
Shadow bytes around the buggy address:
0x10006e3f2b10: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10006e3f2b20: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10006e3f2b30: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10006e3f2b40: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10006e3f2b50: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
=>0x10006e3f2b60: 00 00 f1 f1 f1 f1 00 f2 f2 f2 00 f2 f2 f2[04]f3
0x10006e3f2b70: f3 f3 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x10006e3f2b80: 00 00 00 00 00 00 00 00 00 00 00 00 f1 f1 f1 f1
0x10006e3f2b90: f1 f1 f8 f2 00 f2 f2 f2 f8 f3 f3 f3 00 00 00 00
0x10006e3f2ba0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 f1 f1
0x10006e3f2bb0: f1 f1 00 f3 f3 f3 00 00 00 00 00 00 00 00 00 00
Shadow byte legend (one shadow byte represents 8 application bytes):
Addressable: 00
Partially addressable: 01 02 03 04 05 06 07
Heap left redzone: fa
Freed heap region: fd
Stack left redzone: f1
Stack mid redzone: f2
Stack right redzone: f3
Stack after return: f5
Stack use after scope: f8
Global redzone: f9
Global init order: f6
Poisoned by user: f7
Container overflow: fc
Array cookie: ac
Intra object redzone: bb
ASan internal: fe
Left alloca redzone: ca
Right alloca redzone: cb
Shadow gap: cc
==1383==ABORTING
Fix this by NUL-terminating the string.
Fixes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1193#note_1053306
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
GAMMA_LUT sizes other than 1024 cause a crash during startup if the memcpy()
calls in xf86RandR12CrtcSetGamma() read past the end of the legacy X11 /
XVidMode gamma ramp.
This is a problem on Intel ICL / GEN11 platforms because they report a GAMMA_LUT
size of 262145. Since it's not clear that the modesetting driver will generate a
proper gamma ramp at that size even if xf86RandR12CrtcSetGamma() is fixed, just
disable use of GAMMA_LUT for sizes other than 1024 for now. This will cause the
modesetting driver to disable the CTM property and fall back to the legacy gamma
LUT.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Fixes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1193
Tested-by: Mark Herbert
Whenever an unredirected fullscreen window uses pageflipping for a
DRI3/Present PresentPixmap() operation and the X-Screen has more than
one active output, multiple crtc's need to execute pageflips. Only
after the last flip has completed can the PresentPixmap operation
as a whole complete.
If a sync_flip is requested for the present, then the current
implementation will synchronize each pageflip to the vblank of
its associated crtc. This provides tear-free image presentation
across all outputs, but introduces a different artifact, if not
all outputs run at the same refresh rate with perfect synchrony:
The slowest output throttles the presentation rate, and present
completion is delayed to flip completion of the "latest" output
to complete. This means degraded performance, e.g., a dual-display
setup with a 144 Hz monitor and a 60 Hz monitor will always be
throttled to at most 60 fps. It also means non-constant present
rate if refresh cycles drift against each other, creating complex
"beat patterns", tremors, stutters and periodic slowdowns - quite
irritating!
Such a scenario will be especially annoying if one uses multiple
outputs in "mirror mode" aka "clone mode". One output will usually
be the "production output" with the highest quality and fastest
display attached, whereas a secondary mirror output just has a
cheaper display for monitoring attached. Users care about perfect
and perfectly timed tear-free presentation on the "production output",
but cares less about quality on the secondary "mirror output". They
are willing to trade quality on secondary outputs away in exchange
for better presentation timing on the "production output".
One example use case for such production + monitoring displays are
neuroscience / medical science applications where one high quality
display device is used to present visual animations to test subjects
or patients in a fMRI scanner room (production display), whereas
an operator monitors the same visual animations from a control room
on a lower quality display. Presentation timing needs to be perfect,
and animations high-speed and tear-free for the production display,
whereas quality and timing don't matter for the monitoring display.
This commit gives users the option to choose such a trade-off as
opt-in:
It adds a new boolean option "AsyncFlipSecondaries" to the device section
of xorg.conf. If this option is specified as true, then DRI3 pageflip
behaviour changes as follows:
1. The "reference crtc" for a windows PresentPixmap operation does a
vblank synced flip, or a DRM_MODE_PAGE_FLIP_ASYNC non-synchronized
flip, as requested by the caller, just as in the past. Typically
flips will be requested to be vblank synchronized for tear-free
presentation. The "reference crtc" is the one chosen by the caller
to drive presentation timing (as specified by PresentPixmap()'s
"target_msc", "divisor", "remainder" parameters and implemented by
vblank events) and to deliver Present completion timestamps (msc
and ust) extracted from its pageflip completion event.
2. All other crtc's, which also page-flip in a multi-display configuration,
will try to flip with DRM_MODE_PAGE_FLIP_ASYNC, ie. immediately and
not synchronized to vblank. This allows the PresentPixmap operation
to complete with little delay compared to a single-display present,
especially if the different crtc's run at different video refresh
rates or their refresh cycles are not perfectly synchronized, but
drift against each other. The downside is potential tearing artifacts
on all outputs apart from the one of the "reference crtc".
Successfully tested on a AMD gpu with single-display, dual-display and
triple-display setups, and with single-X-Screen as well as dual-X-Screen
"ZaphodHeads" configurations.
Please consider merging this commit for the upcoming server 1.21 branch.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
It turns out xdmx currently crashes when any client attempts to use GL
and it has been in such state for about 14 years. There was a patch to
fix the problem [1] 4 years ago, but it never got merged. The last
activity on any bugs referring to xdmx has been more than 4 years ago.
Given such situation, I find it unlikely that anyone is still using xdmx
and just having the code is a drain of resources.
[1]: https://lists.x.org/archives/xorg-devel/2017-June/053919.html
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
When using DRI3+Present with PRIME render offload, sometimes there is
a mismatch between the stride of the to-be-presented Pixmap and the
frontbuffer. The current code would reject a pageflip present in this
case if atomic modesetting is not enabled, ie. always, as atomic
modesetting is disabled by default due to brokeness in the current
modesetting-ddx.
Fullscreen presents without page flipping however trigger the copy
path as fallback, which causes not only unreliable presentation timing
and degraded performance, but also massive tearing artifacts due to
rendering to the framebuffer without any hardware sync to vblank.
Tearing is extra awful on modesetting-ddx because glamor afaics seems
to use drawing of a textured triangle strip for the copy implementation,
not a dedicated blitter engine. The rasterization pattern creates extra
awful tearing artifacts.
We can do better: According to a tip from Michel Daenzer (thanks!),
at least atomic modesetting capable kms drivers should be able to
reliably change scanout stride during a pageflip, even if atomic
modesetting is not actually enabled for the modesetting client.
This commit adds detection logic to find out if the underlying kms
driver is atomic_modeset_capable, and if so, it no longer rejects
page flip presents on mismatched stride between new Pixmap and
frontbuffer.
We (ab)use a call to drmSetClientCap(ms->fd, DRM_CLIENT_CAP_ATOMIC, 0);
for this purpose. The call itself has no practical effect, as it
requests disabling atomic mode, although atomic mode is disabled by
default. However, the return value of drmSetClientCap() tells us if the
underlying kms driver is atomic modesetting capable: An atomic driver
will return 0 for success. A legacy non-atomic driver will return a
non-zero error code, either -EINVAL for early atomic Linux versions
4.0 - 4.19 (or for non-atomic Linux 3.x and earlier), or -EOPNOTSUPP
for Linux 4.20 and later.
Testing on a MacBookPro 2017 with Intel Kabylake display server gpu +
AMD Polaris11 as prime renderoffload gpu, X-Server master + Mesa 21.0.3
show improvement from unbearable tearing to perfect, despite a stride
mismatch between display gpu and Pixmap of 11776 Bytes vs. 11520
Bytes. That this is correct behaviour was also confirmed by comparing the
behaviour and .check_flip implementation of the patched modesetting-ddx
against the current intel-ddx SNA Present implementation.
Please consider merging this patch before the server-1.21 branch point.
This patch could also be cherry-picked into the server 1.20 branch to
fix the same limitation.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
In some scenarios, the Wayland compositor might have more knowledge
than the X11 server and may be able to perform pointer emulation for
touch events better. Add a command-line switch to allow compositors
to turn Xwayland pointer emulation off.
Signed-off-by: Simon Ser <contact@emersion.fr>
A misplaced error check can cause this failure scenario, and does
so reliably as tested on Ubuntu 21.04 with KDE Plasma 5 desktop
within the first few seconds of login session startup, rendering
VRR under modesetting-ddx unusable:
1. Some X11 client application changes some window property.
2. ms_change_property() is called as part of the property change
handling call chain (client->requestVector[X_ChangeProperty]).
It removes itself temporarily from the call chain - or so it
thinks, hooking up saved_change_property instead.
3. ret = saved_change_property(client) is called and fails
temporarily for some non-critical reason.
4. The misplaced error check returns early (error abort), without
first restoring ms_change_property() as initial X_ChangeProperty
handler in the call chain again.
-> Now ms_change_property() has removed itself permanently from the
property handler call chain for the remainder of the X session
and VRR property changes on windows are no longer handled, ie.
VRR no longer gets enabled/disabled in response to window VRR
property changes.
Place the error check at the proper place, just as it is correctly
done by amdgpu-ddx, and in modesetting-ddx ms_delete_property()
function.
Verified to fix VRR handling with an AMD gpu under KDE desktop
session.
Please consider merging before branching the server 1.21 branch.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
The xf86CVTMode() was implemented in a standalone source file because it
was being used for both the xfree86 API and the standalone cvt utility.
Now that the cvt utility is removed (as part of libxcvt) we can move the
small xf86CVTMode() function with the rest of the xf86Modes sources.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1142
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
The cvt utility is now replaced by the standalone version found in
libxcvt, no need to build the one in xfree86 anymore.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1142
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Replace the local implementation of the VESA CVT standard timing
modelines generator with the one from libxct to avoid code duplication.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1142
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Xwayland is using a copy of the CVT generator found in Xorg.
Rather than duplicating the code within the xserver tree, use the
libxcvt implementation instead.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1142
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
If there is an explicit configuration, assign the RandR provider
of the GPUDevice to the screen it was specified for.
If there is no configuration (default case) the screen number is
still 0 so it doesn't change behaviour.
The result is e.g:
# DISPLAY=:0.2 xrandr --listproviders
Providers: number : 2
Provider 0: id: 0xd2 cap: 0x2, Sink Output crtcs: 1 outputs: 1 associated providers: 0 name:modesetting
Provider 1: id: 0xfd cap: 0xb, Source Output, Sink Output, Sink Offload crtcs: 2 outputs: 2 associated providers: 0 name:Intel
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
We are handling two cases here: the active flip or the pending flip.
For the pending flip (event->pending == TRUE), we called
xwl_present_release_pixmap.
For the active flip (event->pending == FALSE), we called
xwl_present_release_event. However, xwl_present_flip_notify_vblank
already unhooked event->vblank.event_queue. So this was effectively the
same as calling xwl_present_release_pixmap.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Use present_vblank_rec::event_queue instead.
The changes in xwl_present_execute shouldn't really be needed, since
we should never hit queue_vblank in present_execute_wait. But let's be
safe rather than sorry, plus this simplifies the code.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Can just call xwl_present_execute directly.
This allows dropping the window member from struct xwl_present_window as
well.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
We clear the vblank->pixmap field, so next time xwl_present_execute
falls through to present_execute_post.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
This allows for various simplifications.
Use the pointer to the struct memory as the event ID. In contrast to
the SCMD code for Xorg (where pending DRM events cannot be cancelled),
this is safe here, because we can destroy pending Wayland callbacks. So
we can't get a callback with a stale pointer to freed memory.
Remove xwl_present_window::release_list in favour of
present_vblank_rec::window_list.
Remove xwl_present_event::xwl_present_window in favour of
present_vblank_rec::window.
xwl_present_free_event is never called for a NULL pointer anymore, no
need to check.
v2:
* Restore DestroyWindow wrapping order to make sure
present_destroy_window doesn't call xwl_present_abort_vblank.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
We can call xwl_present_free_event unconditionally from
xwl_present_abort_vblank, since the sync_callback is already destroyed
in xwl_present_cleanup.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Mainly into xwl_present_check_flip, and call that from
present_wnmd_check_flip_window.
No need for them to be separate anymore.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
This will allow eliminating indirections and making the Xwayland Present
code more efficient and easier to follow.
While this technically changes the Xorg video driver ABI, I don't know
of any drivers using the dropped present_wnmd_* symbols, and I doubt a
Xorg driver could make use of them as is anyway.
(As a bonus, Xorg no longer links any Xwayland specific Present code)
v2:
* Wrap DestroyWindow before initializing Present, so that
present_destroy_window runs before xwl_present_cleanup. Avoids crash
due to present_destroy_window calling xwl_present_* functions when
xwl_present_window was already freed. (Olivier Fourdan)
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Not sure why we'd need to abandon a pending stream for a pixmap just
because it's no longer a window pixmap. Let's try not to.
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
This makes sure RandR events are sent to interested clients as needed.
This was happening implicitly in some cases, but not in others, e.g. if
the root window size didn't change.
If this were to call RRTellChanged more often than necessary in some
cases, that should be harmless, as it only sends events if something
has actually changed since last time.
Should fix https://bugzilla.redhat.com/show_bug.cgi?id=1979892 .
v2:
* Call RRTellChanged at the very end of update_screen_size, just in
case.
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
To avoid an EGL stream in the wrong state, if the window pixmap changed
before the stream was connected, we would still keep the pending stream
but mark it as invalid. Once the callback is received, the pending would
be simply discarded.
But all of this is actually to avoid a bug in egl-wayland, there should
not be any problem with Xwayland destroying an EGL stream while the
compositor is still using it.
With that bug now fixed in egl-wayland 1.1.7, we can safely drop all
that logic from Xwayland EGLstream backend.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1189
If the pixmap does not actually change in set_window_pixmap(), there is
no need to invalidate the pending stream, if there's one.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Suggested-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
When building xserver with slibtool (https://dev.midipix.org/cross/slibtool)
the build will fail.
/usr/bin/ld: cannot find -ldmxconfig
This is because xserver creates libdmxconfig.a internally and then links with
the linker flag -ldmxconfig. However according to automake documentation the
-lfoo linker flags should only be used for external dependencies and all
internal libraries should be linked with the libtool archive file (.la) or
the static archive (.a) when the former is not available.
GNU libtool is far more permissive and happens to silently obscure this issue
while slibtool fails because it instead sees '-L./.libs -ldmxconfig'.
xf86_platform_devices[i].pdev may be NULL in cases we fail to parse the
busid in config_udev_odev_setup_attribs() (see also [1], [2]) such as
when udev does not give use ID_PATH. This in turn leads to
platform_find_pci_info() being not called and pdev being NULL.
[1]: https://gitlab.freedesktop.org/xorg/xserver/-/issues/993
[2]: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1076
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
This is the only place where we don't check whether
primaryBus.id.plat->pdev is not NULL before accessing its members.
It may be NULL in cases we fail to parse the busid in
config_udev_odev_setup_attribs() (see also [1], [2]) such as when udev
does not give use ID_PATH. This in turn leads to
platform_find_pci_info() being not called and pdev being NULL in one of
the items within the xf86_platform_devices array. For this to cause a
crash we only need it to become the primaryBus device.
[1]: https://gitlab.freedesktop.org/xorg/xserver/-/issues/993
[2]: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1076
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
screenp->displays[count] (passed to configDisplay() in
configScreen()) is NULL if there is no Virtual setting
in the configuration.
Fixes: f8a6be04d0 ("xfree86: Change
displays array to pointers array to fix invalid pointer issues
after table reallocation")
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Physical dimmension of display can be obtained not just by configuration or
DDC, but also directly from kernel via drmModeGetConnector(). Until now
xserver silently discarded these values even when no configuration nor EDID
were present and fallbacked to default DPI.
There are rare cases when xf86SetDepthBpp is resizing displays array in confScreen.
As that array is shared between set of ScrnInfoRec's then realloc might invalidate chached DispPtr display values in
otheres ScrnInfoRec objects.
If we will change displays array as an array of pointers to DispRec then cached DispRec pointers in ScrnInfoRec
won't be invalid after reallocation of displays array.
Signed-off-by: Łukasz Spintzyk <lukasz.spintzyk@synaptics.com>
In certain circumstances we will have a lot of flip errors without a
reasonable way to prevent them. In such case we reduce the number of
logged messages to at least not fill the error logs.
The details are as follows:
At least on i915 hardware support for async page flip support depends on
the used modifiers which themselves can change dynamically for a screen.
This results in the following problems:
- We can't know about whether a particular CRTC will be able to do an
async flip without hardcoding the same logic as the kernel as there's no
interface to query this information.
- There is no way to give this information to an application, because
the protocol of the present extension does not specify anything about
changing of the capabilities on runtime or the need to re-query them.
Even if the above was solved, the only benefit would be avoiding a
roundtrip to the kernel and reduced amount of error logs. The former
does not seem to be a good enough benefit compared to the amount of work
that would need to be done. The latter is solved in this commit.
Reviewed-by: Eero Tamminen <eero.t.tamminen@intel.com>
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
In xwl_glamor_eglstream_get_wl_buffer_for_pixmap. This can likely be hit
now with an SHM pixmap via the Present flip path. There might be other
corner cases.
Fixes: f3eb1684fa "xwayland: enable MIT-SHM shared pixmaps"
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
This mode for displays running on evdi/udl as side effect of failed glamor_egl_init
reverse_prime_offload_mode was initialized to FALSE
After Mesa upgrade to 21.0.0 GL_RENDERER is not llvmpipe that results in successful glamor_egl_init
and reverse_prime_offload_mode enabled.
This commit is explicitly disabling reverse_prime_offload_mode for evdi and udl drivers
Signed-off-by: Łukasz Spintzyk <lukasz.spintzyk@synaptics.com>
When the command line option "-terminate" is used, it could be
interesting to give it an optional grace period to let the Xserver
running for a little longer in case a new connection occurs.
This adds an optional parameter to the "-terminate" command line option
for this purpose.
v2: Use a delay in seconds instead of milliseconds
(Martin Peres <martin.peres@mupuf.org>)
v3: Clarify man page entry, ensure terminateDelay is always >= 0,
simplify TimerFree(). (Peter Hutterer <peter.hutterer@who-t.net>)
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
That will dramatically affect performance, might as well log when we
cannot use GL_OES_EGL_image with the NVIDIA closed-source driver.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
If the EGLStream backend is able to use hardware acceleration with the
NVIDIA closed source driver, we should use the "nvidia" GLX
implementation instead of the one from Mesa to take advantage of the
NVIDIA hardware accelerated rendering.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
If Xwayland's EGLstream backend supports hardware acceleration with the
NVIDIA closed-source driver, the GLX library also needs to be one
shipped by NVIDIA, that's what GLVND is for.
Add a new member to the xwl_screen that the backend can optionally set
to the preferred GLVND vendor to use.
If not set, "mesa" is assumed.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
When eglSwapBuffers inserts a new frame into a window's stream, there may be a
delay before the state of the consumer end of the stream is updated to reflect
this. If the subsequent wl_surface_attach, wl_surface_damage, wl_surface_commit
calls are received by the compositor before then, it will (typically) re-use
the previous frame acquired from the stream instead of the latest one.
This can leave the window displaying out-of-date contents, which might never be
updated thereafter.
To fix this, after calling eglSwapBuffers, xwl_glamor_eglstream_post_damage
should call eglStreamFlushNV. This call will block until it can be guaranteed
that the state of the consumer end of the stream has been updated to reflect
that a new frame is available.
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1171
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
Allow X11 clients to create shared pixmaps via the MIT-SHM
extension under Xwayland. Tested with a wlroots patch [1].
Also add a few assertions to make sure we have wl_buffers where we
need them.
[1]: https://github.com/swaywm/wlroots/pull/2875
Signed-off-by: Simon Ser <contact@emersion.fr>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
As of commit 098e0f52 xwl_glamor_eglstream_allow_commits will not allow commits
if the xwl_pixmap does not have an EGLSurface. This is valid for pixmaps backed
by an EGLStream, however pixmaps backed by a dma-buf for OpenGL or Vulkan
rendering will never have an EGLSurface. Unlike EGLStream backed pixmaps,
though, glamor will render directly to the buffer that Xwayland passes to the
compositor. Hence, they don't require the intermediate copy in
xwl_glamor_eglstream_post_damage that EGLStream backed pixmaps do, so there is
no need for an EGLSurface.
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
The EGLstream backend's post damage function uses a shader and
glDrawArrays() to copy the data from the glamor's pixmap texture prior
to do the eglSwapBuffers().
However, glDrawArrays() can be affected by the GL state, and therefore
not reliably produce the expected copy, causing the content of the
buffer to be corrupted.
Make sure to set the ALU to GXCopy prior to call glDrawArrays() to get
the expected result.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Suggested-by: Michel Dänzer <mdaenzer@redhat.com>
Currently, the EGLstream backend would increment the pixmap refcount for
each commit, and decrease that refcount on the wl_buffer release
callback.
But that's relying on the compositor sending us a release callback for
each commit, otherwise the pixmap refcount will keep increasing and the
pixmap will be leaked.
So instead, increment the refcount on the pixmap only when we have not
received a release notification for the wl_buffer, to avoid increasing
the pixmap refcount more than once without a corresponding release
event.
This way, if the pixmap is still in use when released on the X11 side,
the EGL stream will be kept until the compositor releases it.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Suggested-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
EGLstream's post_damage() would unconditionally return success
regardless of the actual status of the eglSwapBuffers().
Yet, if eglSwapBuffers() fails, we should not post the corresponding
damage as they wouldn't match the actual content of the buffer.
Use the eglSwapBuffers() return value as the return value for
post_damage() and do not take a refrence on the pixmap if it fails.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Use calloc() instead of malloc() like the rest of the code.
Also fix the arguments of calloc() calls to match the definition which
is calloc(size_t nmemb, size_t size).
This is a cleanup patch, no functional change.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
The EGL surface for the xwl_pixmap is created once the stream is ready
and valid.
If the pixmap's EGL surface fails, for whatever reason, the xwl_pixmap
will be unusable and will end up as an invalid wl_buffer.
Make sure we do not allow commits in that case and recreate the
xwl_pixmap/stream.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
Now that the pending stream is associated with the xwl_pixmap for
EGLStream and the xwl_pixmap itself is associated to the pixmap, we have
a reliable way to get to those data from any pending stream.
As a result, the list of pending streams that we keep in the EGLStream
global structure becomes useless.
So we can drop the pending stream's xwl_pixmap and also the list of
pending streams altogether, and save us a walk though that list for each
callback.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Suggested-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Commit affc47452 - "xwayland: Drop the separate refcount for the
xwl_pixmap" removed the separate reference counter for the xwl_pixmap
which holds the EGLStream.
While that works fine for the common case, if the window's pixmap is
changed before the stream is ready, the older pixmap will be destroyed
and the xwl_pixmap along with it, even if the compositor is still using
the stream.
The code that was removed with commit affc47452 was taking care of that
by increasing the separate reference counter for the xwl_pixmap, but it
no longer the case.
As a result, we may end up with the EGL stream in the wrong state when
trying to use it, which will cascade down into all sort of issues.
To avoid the problem, increase the reference count on the pixmap when it
is marked as invalid in EGLStream's SetWindowPixmap().
This way, the xwl_pixmap and the EGLStream are kept until released by
the compositor, even when the pixmap changes before stream is ready.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Fixes: affc47452 xwayland: Drop the separate refcount for the xwl_pixmap
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
Previously, we would have pending streams associated with top level X11
windows, keeping temporary accounting for the pending streams before
they get fully initialized for the xwl_pixmap which would be associated
with X11 pixmaps.
If the window content changes before the stream is ready, the
corresponding pending stream would be marked as invalid and the pending
stream would be eventually removed once the stream becomes ready.
Since commit affc47452 - "xwayland: Drop the separate refcount for the
xwl_pixmap", we no longer keep a separate reference counter for the
xwl_pixmap, but rather tie it to the X11 pixmap lifespan. Yet, the
pending stream would still be associated with the X11 toplevel window.
Dissociate the pending streams from the X11 toplevel window, to keep it
tied only to the xwl_pixmap so that we can have:
- pixmap <-> xwl_pixmap
- xwl_pixmap <-> pending stream
Of course, the pending streams remain temporary and get removed as soon
as the ready callback is triggered, but the pending streams are not
linked to the X11 window anymore which can change their content, and
therefore their X11 pixmap at any time.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
eglCreateStreamKHR() can fail and return EGL_NO_STREAM_KHR, in which
case there is no point in trying to create a buffer from it.
Similarly, eglCreateStreamProducerSurfaceKHR() also fail and return
EGL_NO_SURFACE, which in turn will be used in eglMakeCurrent() as
draw/read surface, and therefore would mean no draw/read buffer.
In those cases, log the error, and bail out early. That won't solve the
issue but will help with investigating the root cause of issues with
EGLStream backend.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
Some functions are called "callback" whereas they are not longer
callback functions or "unref" while they no longer deal with a reference
counter anymore, which is quite confusing. Rename those functions to be
more explicit.
Also, the pending streams can be destroyed in different places, move the
common code to separate function to avoid duplicating code and help with
readability of the code.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
The EGLStream backend would sometime generate GL errors trying to draw
to the framebuffer, which gives an invalid buffer, which in turn would
generate a Wayland error from the compositor which is fatal to the
client.
Check the framebuffer status and bail out early if it's not complete,
to avoid getting into trouble later.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
If the glamor backend failed to post damage, the caller should do the
same to avoid a failure to attach the buffer to the Wayland surface.
Change the API of Xwayland's glamor backend post_damage() to return a
status so that xwl_window_post_damage() can tell whether the callee
failed.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
If the buffer is NULL, do not even try to attach it, and risk a Wayland
protocol error which would be fatal to us.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Martin Peres <martin.peres@mupuf.org>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
EGLStream wl_eglstream_display_create_stream() may fail, yet Xwayland
would try to attach the buffer which may cause a fatal Wayland protocol
error raised by the compositor.
Check if the buffer creation worked, and fail gracefully otherwise (like
wayland-eglsurface does).
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Martin Peres <martin.peres@mupuf.org>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1156
If a window is being used for direct rendering with OpenGL or Vulkan, and is
using the flipping path for presentation, it's pixmap will be set to a dma-buf
backed pixmap created by the client-side GL driver. However, this means that
xwl_glamor_eglstream_post_damage won't work since it requires that the pixmap
has an EGLSurface that it can render to, which dma-buf backed pixmaps do not.
In this case, though, xwl_glamor_eglstream_post_damage is not necessary since
glamor will have rendered directly to the pixmap, so we can simply pass it
directly to the compositor. There's no need for the intermediate copy we
normally do in that function.
Therefore, this change adds an early-return case to post_damage for dma-buf
backed pixmaps, and removes the corresponding asserts from that function and
allow_commits.
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Copied TGL PCI ID from MESA iris_pci_ids.h
This update brings in a significant number of new platform ID's
Sync up until commit f02ae698
Signed-off-by: Mazlan, Hazwan Arif <hazwan.arif.mazlan@intel.com>
On FreeBSD 13.0-CURRENT for PowerPC64 big-endian (BE), X was
crashing in some cases. For instance, when twm was started
and the background was clicked to open its menu, X crashed
with a segmentation fault, trying to dereference a null pointer
at CreatePicture().
There were 2 issues with xorg-server handling of RGB masks that
caused the pointer above to be null and thus the crash:
- wrong use of ffs() to get the RGB offsets from the masks
- overflow when shifting a 16-bit integer
This change fixes both issues. They happen when the system is BE
but has a video adapter using a little-endian (LE) ARGB32
framebuffer. In order to display the correct colors, this setup
requires a BE RGBA32 color format to be used by X, by setting
the RGB masks appropriately, that didn't work properly because of
the issues above.
A lot of that code is the same as in xf86-amdgpu and xf86-nouveau drivers. By removing that functions from
ms namespace we can move that code to common implementation.
Signed-off-by: Łukasz Spintzyk <lukasz.spintzyk@synaptics.com>
Instead of getting the current msc value from the window crtc,
which not exist take the last saved msc value saved in
the window_priv struct
Signed-off-by: Łukasz Spintzyk <lukasz.spintzyk@synaptics.com>
Provides an implementation for the pixmap_from_buffers DRI3 function for
xwayland's eglstream backend. This will be used by the NVIDIA GLX driver
to pass buffers from client applications to the server. These can then
be presented using the PRESENT extension.
To hopefully make this less error-prone, we also introduce a "type"
field for this struct to distinguish between xwl_pixmaps for the new
DRI3-created pixmaps and those for the existing glamor-created pixmaps.
Additionally, the patch enables wnmd present mode with the eglstream backend.
This involves creating a wl_buffer for the provided dma-buf before importing it
into EGL and passing this to the compositor so it can be scanned out directly
if possible.
Since both backends now support this present mode, the HAS_PRESENT_FLIP flag is
no longer needed, so it can be removed.
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
This is preliminary work for hardware accelerated rendering with the
NVIDIA driver.
This exposes a new glamor backend function, check_flip, which can be
used to control whether flipping is supported for the given pixmap.
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
This is preliminary work for hardware accelerated rendering with the
NVIDIA driver.
This moves the modifiers and formats functions previously only available
to the GBM backend to the common glamor code so that it can be used by
both the GBM and EGLStream backends.
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
This is preliminary work for hardware accelerated rendering with the
NVIDIA driver.
The EGLStream backend can possibly also use the dmabuf interface, so
move the relevant code from the GBM specific source to the common bits.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
According to the "VESA Enhanced EDID Standard", all GTF compliant
displays are continuous frequency.
The GTF support flags in 'Display Range Limits Descriptor' cannot be
used if the display is a non-continuous frequency multi-mode display.
Closes#1105
Signed-off-by: Pascal VITOUX <vitoux.pascal@gmail.com>
Fixes: 6a79a737 ("xfree86: add drm modes on non-GTF panels")
Reviewed-by: Aaron Ma aaron.ma@canonical.com
Rather than trying to create a gamma ramp array of the appropriate size in
drmmode_crtc_init when the GAMMA_LUT property should be used, just flag the crtc
as wanting to use the GAMMA_LUT property and then replace the gamma ramp later,
right before calling xf86HandleColormaps. This avoids a problem during initial
startup where xf86RandR12CreateObjects12 hard-codes a gamma ramp size of 256,
causing xf86RandR12CrtcSetGamma to read past the end of the DIX layer's RandR
gamma ramp array:
PreInit
drmmode_pre_init
drmmode_crtc_init
crtc->gamma_size = 1024
ScreenInit
xf86CrtcScreenInit
xf86RandR12Init
xf86RandR12Init12
xf86RandR12CreateObjects12
RRCrtcCreate
randr_crtc->gammaSize = 0
xf86RandR12InitGamma(pScrn, 256)
RRCrtcGammaSetSize
randr_crtc->gammaSize = 256
xf86RandR12InitGamma
xf86RandR12CrtcInitGamma
RRCrtcGammaSet
xf86RandR12CrtcSetGamma
// crtc->gamma_size is 1024 here, while randr_crtc->gammaRed
// is a 256-element array.
memcpy(crtc->gamma_red, randr_crtc->gammaRed, crtc->gamma_size * sizeof(crtc->gamma_red[0]));
drmmode_setup_colormap
xf86HandleColormaps
xf86RandR12InitGamma
RRCrtcGammaSetSize
randr_crtc->gammaSize = 1024
Fixes: 245b9db0 - modesetting: Use GAMMA_LUT when available
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1126
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Reviewed-by: Robert Morell <rmorell@nvidia.com>
This brings the behavior closer than what we currently have with
autotools-based build system.
Meson does not currently have native support for symlinks.
See https://github.com/mesonbuild/meson/issues/1602.
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
Meson gets confused when there are two targets of the same name within
the same directory, so we use a different intermediate name.
This is a problem with the Xorg SUID wrapper which has the same filename
as the real Xorg executable and is configured in the same meson.build
file. This commit works around this by using a different filename in
build stage and renaming only when installing.
Signed-off-by: Povilas Kanapickas <povilas@radix.lt>
This bumps the minimum Wayland version to 1.5 (released in 2014).
Signed-off-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Martin Peres <martin.peres@mupuf.org>
When using autoconf/automake to build Xwayland, the actual path to
Xwayland is not fully qualified and refers to the "exec_prefix".
As a result, the path provided by the generated pkg-config file is wrong
when using autoconf to build the Xserver.
Fix the xwayland.pc file to also set the variable "prefix" and
"exec_prefix" so that the path to Xwayland is fully resolved.
Add those variables to the meson build as well for good measure.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
Since commit 20c78f38, we use the relative pointer for enter/leave
events.
However, sprite_check_lost_focus() which verifies whether the pointer has
left an Xwayland surface still explicitly check for the absolute
pointer.
As a result, no LeaveNotify event is emitted anymore now when the
pointer crosses from an Xwayland surface to a Wayland native one.
Make sure to check the last slave device against get_pointer_event() as
well, not just the absolute pointer.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Martin Peres <martin.peres@mupuf.org>
Acked-by: Roman Gilg <subdiff@gmail.com>
Tested-by: Roman Gilg <subdiff@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Fixes: 20c78f38 - xwayland: use get_pointer_device() for enter/leave
handling too
The code path added by commit 69e4b8e6 (xfree86: attempt to autoconfig
gpu slave devices (v3)) assumes that it will only be run if the primary
device on the screen is the first device in xf86configptr->conf_device_lst.
While this is true most of the time, there are two specific cases where
this assumption fails.
First, if the first device in conf_device_lst is assigned to a different
seat than the running X server, it will be skipped by the previous
FIND_SUITABLE macro usage. Second, if the primary device was explicitly
assigned to the screen but auto_gpu_device is still set and no secondary
devices were explicitly listed, that device may not be the first device
in conf_device_lst.
When the first device in conf_device_lst is not the primary device
assigned to the screen, two problems emerge. First, the first device in
conf_device_lst will never be assigned to the screen as a secondary
device. Second, the primary device is additionally assigned to the
screen as a secondary device. The combination of these problems causes
certain otherwise valid configurations to be invalid. For example, if a
primary device is assigned to a screen and a secondary device is listed
in the config but not explicitly assigned to the screen, then one order
of the device sections results in a usable PRIME or Reverse PRIME setup
and the other order does not.
This commit removes the assumption that the primary device is the first
device in conf_device_lst by starting the loop from the start of
conf_device_lst and skipping the primary device when it is encountered.
Signed-off-by: Jacob Cherry <jcherry@nvidia.com>
Xwayland won't emulate XWarpPointer requests if the cursor is visible,
this is to avoid having the cursor jumping on screen and preventing
random X11 clients from controlling the pointer in Wayland, while
allowing games which use that mechanism with a hidden cursor to work in
Xwayland.
There are, however, games which tend to do it in the wrong order, i.e.
show the cursor before moving the pointer, and because Xwayland will not
allow an X11 client to move the pointer while the cursor is visible, the
requests will fail.
Add a workaround for such X11 clients, when the cursor is being shown,
keep it invisible until the cursor is actually moved. This way, X11
clients which show their cursor just before moving it would still have a
chance to succeed.
v2: Add a timeout to show the cursor for well behaved clients.
v3: Some cleanup (Michel)
v4: Do not cancel cursor delay when updating the cursor to avoid
delaying cursor visibility indefinitely if the client keeps
settings different cursors (Michel)
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Tested-by: Jaap Buurman jaapbuurman@gmail.com
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/734
A change during the 1.20 development cycle resulted in fbconfigs being walked
and deallocated individually during __glXScreenDestroy. This change
now avoids a use-after-free caused by that change.
==50859==ERROR: AddressSanitizer: heap-use-after-free on address 0x00010d3819c8 at pc 0x0001009d4230 bp 0x00016feca7a0 sp 0x00016feca798
READ of size 8 at 0x00010d3819c8 thread T5
#0 0x1009d422c in __glXScreenDestroy glxscreens.c:448
#1 0x10091cc98 in __glXAquaScreenDestroy indirect.c:510
#2 0x1009d2734 in glxCloseScreen glxscreens.c:169
#3 0x100740a24 in dix_main main.c:325
#4 0x10023ed50 in server_thread quartzStartup.c:65
#5 0x199ae7fd0 in _pthread_start+0x13c (libsystem_pthread.dylib:arm64e+0x6fd0)
#6 0x199ae2d38 in thread_start+0x4 (libsystem_pthread.dylib:arm64e+0x1d38)
0x00010d3819c8 is located 200 bytes inside of 12800-byte region [0x00010d381900,0x00010d384b00)
freed by thread T5 here:
#0 0x101477ba8 in wrap_free+0x98 (libclang_rt.asan_osx_dynamic.dylib:arm64e+0x3fba8)
#1 0x1009d4240 in __glXScreenDestroy glxscreens.c:449
#2 0x10091cc98 in __glXAquaScreenDestroy indirect.c:510
#3 0x1009d2734 in glxCloseScreen glxscreens.c:169
#4 0x100740a24 in dix_main main.c:325
#5 0x10023ed50 in server_thread quartzStartup.c:65
#6 0x199ae7fd0 in _pthread_start+0x13c (libsystem_pthread.dylib:arm64e+0x6fd0)
#7 0x199ae2d38 in thread_start+0x4 (libsystem_pthread.dylib:arm64e+0x1d38)
previously allocated by thread T5 here:
#0 0x101477e38 in wrap_calloc+0x9c (libclang_rt.asan_osx_dynamic.dylib:arm64e+0x3fe38)
#1 0x100925a40 in __glXAquaCreateVisualConfigs visualConfigs.c:116
#2 0x10091cb24 in __glXAquaScreenProbe+0x224 (X11.bin:arm64+0x100730b24)
#3 0x1009cd840 in xorgGlxServerInit glxext.c:528
#4 0x10074539c in _CallCallbacks dixutils.c:743
#5 0x100932a70 in CallCallbacks callback.h:83
#6 0x100932478 in GlxExtensionInit vndext.c:244
#7 0x10020a364 in InitExtensions miinitext.c:267
#8 0x10073fe7c in dix_main main.c:197
#9 0x10023ed50 in server_thread quartzStartup.c:65
#10 0x199ae7fd0 in _pthread_start+0x13c (libsystem_pthread.dylib:arm64e+0x6fd0)
#11 0x199ae2d38 in thread_start+0x4 (libsystem_pthread.dylib:arm64e+0x1d38)
Regressed-in: 4b0a3cbab1
CC: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
It violates @autoreleasepool best practices, and this helps collapse quartzCocoa.m into quartz.c
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
There is a place where this code was called on the main thread.
We're using a rather nasty anti-pattern to just call a block inline rather
than synchonously calling it on the main thread if we're already on the main
thread. This code could use a good overhaul, but I don't have time to rip
it apart right now. This will address the immediate issue.
Fixes: https://github.com/XQuartz/XQuartz/issues/40
Fixes: https://github.com/XQuartz/XQuartz/issues/48
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
In Weston, clicking the window decoration of an Xwayland client gives us a
wl_pointer.button event immediately followed by a wl_pointer.leave event.
The leave event does not contain any button state information, so the button
remains logically down in the DIX.
Once the pointer button is released, a wl_pointer.enter event is sent with
the current button state (zero). This needs to trigger a ButtonRelease event
but for that we need to ensure that the device is the same as the one we send
ButtonPress events through.
Fixes a regression introduced in a4095162ca.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
Trying to change the acceleration/threshold on Xwayland cannot work, and
the corresponding handler xwl_pointer_control() is a no-op.
Yet, an X11 client trying to change those on the touch device may
possibly cause a crash because the touch device in Xwayland doesn't set
that.
Initialize the touch device's PtrFeedback to make sure that just cannot
happen.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1137
We are using the relative pointer for motion events, but buttons and
axis events still go through the absolute pointer device.
That means additional DeviceChanged events that could be avoided if the
buttons and axis events were coming from the same device as motion
events.
Route those events to the relative pointer if available so that motion,
buttons and axis events come from the same device (most of the time).
Suggested-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Related: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1130
The relative pointer only has 2 axis, if we want to route the mouse
wheel events to that device, we need to add the axis definition, similar
to what is done for the absolute pointer.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Related: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1130
This is a cleanup patch, no functional change.
Split the function dispatch_pointer_motion_event() into three separate
simpler functions, relative motion with a warp emulator, relative motion
and absolute motion.
This makes the code a lot easier to read for me, rather than having
everything in a single function with nested if/else conditions.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Xwayland supports relative motion events from the Wayland compositor via
the relative-pointer protocol, and converts those to the absolute range
in device units for raw events.
Some X11 clients however wrongly assume relative values in the axis
values even for devices explicitly labeled as absolute. While this is a
bug in the client, such applications would work fine in plain Xorg but
not with Xwayland.
To avoid that issue, use the relative values for raw events without
conversion, so that such application continue to work in Xwayland.
Thanks Peter for figuring out the root cause.
v2: Don't duplicate relative and absolute events (Peter)
v3: Use POINTER_RAWONLY (Peter)
Suggested-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1130
This add a new flag POINTER_RAWONLY for GetPointerEvents() which does
pretty much the opposite of POINTER_NORAW.
Basically, this tells GetPointerEvents() that we only want the
DeviceChanged events and any raw events for this motion but no actual
motion events.
This is preliminary work for Xwayland to be able to use relative motion
events for raw events. Xwayland would use absolute events for raw
events, but some X11 clients (wrongly) assume raw events to be always
relative.
To allow such clients to work with Xwayland, it needs to switch to
relative raw events (if those are available from the Wayland
compositor).
However, Xwayland cannot use relative motion events for actual pointer
location because that would cause a drift over time, the pointer being
actually controlled by the Wayland compositor.
So Xwayland needs to be able to send only relative raw events, hence
this API.
Bump the ABI_XINPUT_VERSION minor version to reflect that API addition.
v2: Actually avoid sending motion events (Peter)
v3: Keep sending raw emulated events with RAWONLY (Peter)
Suggested-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Related: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1130
When an X11 client issues an active grab on the keyboard, Xwayland
forward this to the Wayland compositor using the Xwayland specific
protocol "xwayland-keyboard-grab" if it can find the corresponding
Xwayland window.
Some X11 clients (typically older games) however try to issue the
keyboard grab on the X11 root window, which has obviously no matching
Xwayland window. In such a case, the grab is simply ignored and the game
will not work as expected.
To workaround that issue, if an X11 client issues a keyboard grab on the
root window, Xwayland will search for a toplevel window belonging to the
same X11 client that it can use as the grab window instead.
This way, the grab can be forwarded to the Wayland compositor that can
either grant or deny the request based on the window and its internal
policies.
The heuristic picks the first realized toplevel window belonging to the
client so that the Wayland compositor will send it the keyboard events,
and the Xserver grab mechanism will then take care of routing the events
to the expected X11 window by itself.
v2: Make the test more clear (Dor Askayo <dor.askayo@gmail.com>)
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Jonas Ådahl <jadahl@gmail.com>
See-also: https://gitlab.gnome.org/GNOME/mutter/-/issues/1249
Not all extensions can be enabled or disabled at runtime, list the
extensions which can from the help message rather than on error only.
v2:
* Print the header message in the ListStaticExtensions() (Peter
Hutterer)
* Do not export ListStaticExtensions() as Xserver API
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
The definition relies on IOPortBase, which is only ever set in
hw/xfree86/os-support/bsd/arm_video.c
This caused build failures on linux/mips with GCC 10, due to this
change (from https://gcc.gnu.org/gcc-10/changes.html#c):
"GCC now defaults to -fno-common. As a result, global variable accesses
are more efficient on various targets. In C, global variables with
multiple tentative definitions now result in linker errors. With
-fcommon such definitions are silently merged during linking."
As a result anything including compiler.h would get its own definition
of IOPortBase and the linker would error out.
Since commit b3f3d65e, xwayland now supports the command line option
"-listenfd" for passing file descriptors and marked "-listen" as
deprecated for this specific purpose.
Add a new pkg-config variable "have_listenfd" to the xwayland.pc so that
compositors can know this is available and use listenfd in place of the
deprecated option.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Jonas Ådahl <jadahl@gmail.com>
There are currently no callers that make use of the "created" output parameter
of xwl_glamor_pixmap_get_wl_buffer. Remove it, along with the corresponding
argument of the associated EGL backend entrypoint.
Just a small code cleanup, there is no need to allocate a variable only
to check the return value of eglInitialize().
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Tony Lindgren <tony@atomide.com>
Glamor requires at least big GL 2.1 or GLES2, therefore Xwayland tries
to initialize first GL and then GLES2 if that fails.
It does that all in one single function which makes the code slightly
complicated, move the initialization of big-GL and GLES2 to separate
functions to help with readability of the code.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Tony Lindgren <tony@atomide.com>
We currently bail out early for GLES only devices, and call
epoxy_gl_version() too early for GLES only that will make GLES only
devices return NULL for glGetString(GL_RENDERER).
Let's also add a check to see if we need to recreate the context to
avoid pointless warnings for GLES only devices as suggested by
Olivier Fourdan <ofourdan@redhat.com>.
Fixes: a506b4ec - xwayland: make context current to check GL version
Signed-off-by: Tony Lindgren <tony@atomide.com>
The window buffer mechanism would free the pixmap and its corresponding
Wayland buffer as soon as window buffers are disposed.
Typically when the X11 window is unrealized, the current window buffer
is still used by the Wayland compositor and yet Xwayland will destroy
the buffer.
As a matter of fact, Xwayland should not destroy the Wayland buffer
before the wl_buffer.release event is received.
Add a reference counter to the window buffer similar to the to pixmap
reference counter to keep the buffer around until the release callback
is received.
Increase that reference counter on the buffer which will be attached to
the surface, and drop that reference when receiving the release callback
notification.
v2: Use a specific reference counter on the buffer rather than relying
on the pixmap refcnt (Michel Dänzer <mdaenzer@redhat.com>)
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Acked-by: Martin Peres <martin.peres@mupuf.org>
The cursor code would destroy the buffer as soon as the cursor is
unrealized on X11 side.
Yet, the Wayland compositor may still be using the buffer as long as a
released callback has not been received.
Increase the reference counter on the pixmap to hold a reference on the
pixmap when attaching it to the surface and use the new pixmap release
callback mechanism to release that reference when the buffer is
released.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Acked-by: Martin Peres <martin.peres@mupuf.org>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
The seat and tablet cursor functions are very similar, factorize the
commonalities to simplify the code and reduce the copy/paste.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Commit 6a5a4e6037 removed the option to
configure useSIGIO option. Indeed, the xfree86 SIGIO support was
reworked to use internal versions of OsBlockSIGIO and OsReleaseSIGIO.
As a result, useSIGIO is no longer needed and can dropped
Fixes: 6a5a4e60 - Remove SIGIO support for input [v5]
Closes: xorg/xserver#1107
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Prabhu Sundararaj <prabhu.sundararaj@nxp.com>
Signed-off-by: Mylène Josserand <mylene.josserand@free-electrons.com>
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
LogMessage logs only when the XLOG_VERBOSITY is >= 1, but by default
XLOG_VERBOSITY is 0, so for example warning about deprected -listen
parameter is never shown when running "Xwayland -listen 32 -help".
Signed-off-by: Mariusz Ceier <mceier+freedesktop@gmail.com>
By default, the macro DebugPresent() is a no-op but it can be enabled
at build time for debugging purpose.
However, doing so prevents the code to build because one debug statement
tries to make use of a non-existent variable:
present.c: In function ‘ms_present_queue_vblank’:
present.c:147:18: error: ‘vbl’ undeclared (first use in this function)
147 | vbl.request.sequence));
| ^~~
present.c:49:32: note: in definition of macro ‘DebugPresent’
49 | #define DebugPresent(x) ErrorF x
| ^
Fix the build with DebugPresent() by removing the vbl variable from the
debug message.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
With !155, the device bus ID received via udev is constructed
properly with the "usb:" prefix. But, it is not enough to
make the following line to work in Section "Device":
BusID "usb:0:1.2:1.0"
Introduce BUS_USB, so the prefix can be distinguished from BUS_PCI
and check the supplied BusID value against device->attribs->busid
in xf86PlatformDeviceCheckBusID().
Signed-off-by: Böszörményi Zoltán <zboszor@pr.hu>
Provide an actual definition of noDriExtension where used, rather than a
tentative definition in a header, to fix compilation with -fno-common
(the default with gcc 10).
EGLStream implementation in Xwayland keeps a list of pending streams for
a window.
If the windows's pixmap is destroyed while there is a pending stream,
the pending stream will point to freed memory once the callback is
triggered.
Make sure to cancel the pending stream if there's one when the pixmap is
destroyed.
v2:
* Use xorg_list_for_each_entry() instead of the safe variant (Michel
Dänzer <mdaenzer@redhat.com>)
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Tested-by: Karol Szuster <karolsz9898@gmail.com>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
Closes https://gitlab.freedesktop.org/xorg/xserver/-/issues/1096
Commit 77658741 - "xwayland: Add buffer release callback" added an API
to deal with Wayland buffer release callbacks.
The EGLstream implementation has its own wl_buffer callback, move that
to the buffer release API instead so we don't have to deal with Wayland
buffers directly and match the other Xwayland pixmap backend
implementations.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Tested-by: Erik Kurzinger <ekurzinger@nvidia.com>
Resolves warnings from Oracle Parfait static analyser:
Error: Misleading macro
Misleading macro [misleading-macro]:
misleading evaluation of ternary '?:' operator in expansion of macro V_ADDR_RB due to missing parentheses
at line 392 of hw/xfree86/int10/generic.c.
'|' operator has higher precedence than ternary '?:' operator inside macro body at line 431
low precedence ternary '?:' operator is hidden by expansion of macro V_ADDR_RB at line 431
Misleading macro [misleading-macro]:
misleading evaluation of ternary '?:' operator in expansion of macro V_ADDR_RB due to missing parentheses
at line 392 of hw/xfree86/int10/generic.c.
'<<' operator has higher precedence than ternary '?:' operator inside macro body at line 431
low precedence ternary '?:' operator is hidden by expansion of macro V_ADDR_RB at line 431
Misleading macro [misleading-macro]:
misleading evaluation of ternary '?:' operator in expansion of macro V_ADDR_RB due to missing parentheses
at line 392 of hw/xfree86/int10/generic.c.
'<<' operator has higher precedence than ternary '?:' operator inside macro body at line 442
low precedence ternary '?:' operator is hidden by expansion of macro V_ADDR_RB at line 442
Misleading macro [misleading-macro]:
misleading evaluation of ternary '?:' operator in expansion of macro V_ADDR_RB due to missing parentheses
at line 392 of hw/xfree86/int10/generic.c.
'<<' operator has higher precedence than ternary '?:' operator inside macro body at line 443
low precedence ternary '?:' operator is hidden by expansion of macro V_ADDR_RB at line 443
Misleading macro [misleading-macro]:
misleading evaluation of ternary '?:' operator in expansion of macro V_ADDR_RB due to missing parentheses
at line 392 of hw/xfree86/int10/generic.c.
'|' operator has higher precedence than ternary '?:' operator inside macro body at line 443
low precedence ternary '?:' operator is hidden by expansion of macro V_ADDR_RB at line 441
Misleading macro [misleading-macro]:
misleading evaluation of ternary '?:' operator in expansion of macro V_ADDR_RB due to missing parentheses
at line 392 of hw/xfree86/int10/generic.c.
'<<' operator has higher precedence than ternary '?:' operator inside macro body at line 443
low precedence ternary '?:' operator is hidden by expansion of macro V_ADDR_RB at line 443
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Resolves warning from Oracle Parfait static analyser:
Error: Unchecked result
Unchecked result [unchecked-result-call-X]:
Unchecked return value from call to XOpenDisplay. Value display must be ch
ecked to ensure this function was successful.
at line 73 of hw/dmx/examples/xbell.c in function 'main'.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
It runs XTS via piglit on (non-rootless) Xwayland on weston using the
headless backend.
Xwayland might use glamor if enabled in the build, but we're making sure
it uses software rendering.
v2:
* Use weston-info to wait for weston to be ready, instead of just a
fixed sleep. (Martin Peres)
v3:
* Build wayland 1.18 & weston 9.0 locally, since the packages in Debian
buster are too old for current Xwayland.
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
As Xwayland is usually spawned by the Wayland server/compositor, its
command line options are not always adjustable.
Yet, if EGLStream is not supported in a given Xwayland build, the option
will do nothing (yet we must still accept it otherwise Xwayland would
refuse to run if the Wayland compositor uses it).
If Xwayland was built without support for EGLStream, there is not point
in showing the option in the help message though.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Martin Peres <martin.peres@mupuf.org>
The command line options "-shm" is used to instruct Xwayland to prefer
shared-memory for passing buffers to the Wayland server, rather than
using glamor and DRI3.
The option was there from the beginning, yet not documented in the
"-help" message.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Martin Peres <martin.peres@mupuf.org>
One general assumption in Xwayland is that the xwl_window remains the
same for all the child windows of the toplevel window.
When mapping a new X11 window, ensure_surface_for_window() checks for an
existing xwl_window by using xwl_window_get() which will just check for
the registered xwl_window for the window.
That means that a client mapping a child window of an existing window
with a xwl_window will get another different xwl_window.
If an X11 client issues a Present request on the parent window, hence
placed underneath its child window of the same size, the Wayland
compositor may not send the frame callback event for the parent's
Wayland surface which is reckoned to be not visible, obscured behind
the other Wayland surface for the child X11 window.
That bug affects some games running in wine which may get 1 fps because
the repaint occurs only on timeout with a long interval (as with, e.g.
https://bugs.winehq.org/show_bug.cgi?id=47066)
Fix ensure_surface_for_window() by using xwl_window_from_window() which
will walk the window tree, so that a child window won't get another
xwl_window than its parent.
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1099
See-also: https://bugs.winehq.org/show_bug.cgi?id=47066
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
When running non-rootless, Xwayland requires that the Wayland compositor
supports the XDG-WM-Base protocol.
Check for XDG-WM-Base protocol support at startup and exit cleanly if
missing rather than segfaulting later in ensure_surface_for_window()
while trying to use xdg_wm_base_get_xdg_surface().
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Instead, bump the pixmap's refcount at the bottom of post_damage to
reflect the compositor's hold on the buffer, and "destroy" the pixmap in
the buffer release callback (which will dec the pixmap's refcount and
free if necessary).
Signed-off-by: Adam Jackson <ajax@redhat.com>
When the "CTM" (color transform matrix) modesetting property is available,
create a corresponding RandR property.
To match the format of the property available in the amdgpu driver, expose it as
an array of 18 32-bit XA_INTEGERs representing a 3x3 matrix in row-major order,
where each entry is a S31.32 sign-magnitude fixed-point number with the
fractional part listed first.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: James Jones <jajones@nvidia.com>
If the kernel exposes GAMMA_LUT and GAMMA_LUT_SIZE properties and the size is
not what the server has pre-configured for the crtc, free the old gamma ramp
memory allocated by the server and replace it with new allocations of the
appropriate size.
In addition, when GAMMA_LUT is available, use drmModeCreatePropertyBlob() and
drmModeObjectSetProperty() to set the gamma ramp rather than using the legacy
drmModeCrtcSetGamma() function.
Add a new option "UseGammaLUT" to allow disabling this new behavior and falling
back to drmModeCrtcSetGamma() unconditionally.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Modeset properties can be set even when ms->atomic_modeset is disabled by using
the drmModeObjectSetProperty() function.
This will be necessary in a later change in order to set the GAMMA_LUT and CTM
properties.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Jocelyn Falempe
Michel Dänzer
Andrea Monaco
Sam James
Matthieu Herrb
Drew DeVault
Xaver Hugl
Povilas Kanapickas
Olivier Fourdan
Dave Airlie
nerdopolis
Jon Turney
Matthieu Herrb
Simon Ser
nerdopolis
Mario Kleiner
James Jones
Adam Jackson
Patrik Jakobsson
Aaron Plattner
Zoltán Böszörményi
Łukasz Spintzyk
orbea
Zoltán Böszörményi
Daniel Strnad
Lukasz Spintzyk
Erik Kurzinger
Mazlan, Hazwan Arif
luporl
Edouard Gaulué
Jeremy Huddleston Sequoia
Pascal VITOUX
Jacob Cherry
Peter Hutterer
Jim DeLaHunt
Misha Gusarov
Julien Cristau
Tony Lindgren
Prabhu Sundararaj
Mariusz Ceier
Alan Coopersmith