inspired from similar code in amdgpu, fixes a crash when xrandr(1)
is invoqued with X server 21.1.1
Signed-off-by: Matthieu Herrb <matthieu.herrb@laas.fr>
Based on xf86-video-amdgpu, but applying experience gained in the
meantime in other projects and taking advantage of new features
available with current versions of GitLab.
Fixes compile errors with glamor disabled:
../../src/radeon_present.c: In function ‘radeon_present_check_flip’:
../../src/radeon_present.c:281:21: error: invalid use of undefined type ‘struct radeon_pixmap’
281 | if (priv && priv->fb_failed)
| ^~
../../src/radeon_present.c:288:19: error: invalid use of undefined type ‘struct radeon_pixmap’
288 | if (priv && !priv->fb_failed) {
| ^~
../../src/radeon_present.c:292:10: error: invalid use of undefined type ‘struct radeon_pixmap’
292 | priv->fb_failed = TRUE;
| ^~
When the drmModeSetCursor2() call was replaced with bare drmIoctl() call in
92df7097, a bug was introduced. With the use of drmModeSetCursor2(),
the return value from drmIoctl() (which calls ioctl()) were mangled, if
they were negative, they were replaced by -errno by a wrapper function
in xf86drMode.c in libdrm. After replacing drmModeSetCursor2() with the
call to drmIoctl(), this mangling no longer happens, and we need to
explicitly check if the call to drmIoctl() fails, which is indicated by
returning -1, and then why it failed, by checking errno.
If the error indicated by errno is EINVAL, then we can't use the
DRM_IOCTL_MODE_CURSOR2 ioctl(), and need to fall back to the
DRM_IOCTL_MODE_CURSOR ioctl().
This bug can manifest itself by an invisible hw cursor on systems where the
DRM_IOCTL_MODE_CURSOR2 is not implemented by the graphics driver.
Credit also to Alexey Dokuchaev for help with developing the fix and
testing.
This fixes#190
Signed-off-by: Niclas Zeising <zeising@daemonic.se>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Commit d1d8e3c8d0 causes X server
to fail on startup when GPU acceleration is not working (or is
disabled). The reason is that `radeon_get_pixmap_bo` function
gets called too early (before EXA has been initialized) and
fails with an assert:
#0 __GI_raise (sig=<optimized out>) at ../sysdeps/unix/sysv/linux/raise.c:50
#1 0x76ab1c6c in __GI_abort () at abort.c:79
#2 0x76ac0b64 in __assert_fail_base (fmt=0x76bfbce4 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n", assertion=0x7658c80c "key->initialized", file=<optimized out>, line=121,
function=0x7658d040 <__PRETTY_FUNCTION__.10607> "dixGetPrivateAddr") at assert.c:92
#3 0x76ac0c0c in __GI___assert_fail (assertion=0x7658c80c "key->initialized", file=0x7658c9d0 "../include/privates.h", line=121,
function=0x7658d040 <__PRETTY_FUNCTION__.10607> "dixGetPrivateAddr") at assert.c:101
#4 0x76579e6c in dixGetPrivateAddr (key=<optimized out>, key=<optimized out>, privates=<optimized out>) at ../include/privates.h:121
#5 0x7657a954 in dixGetPrivateAddr (key=<optimized out>, key=<optimized out>, privates=<optimized out>) at exa.c:70
#6 dixGetPrivate (key=<optimized out>, privates=<optimized out>) at ../include/privates.h:136
#7 exaGetPixmapDriverPrivate (pPix=<optimized out>) at exa.c:68
#8 0x7623d460 in radeon_get_pixmap_bo (pPix=0x71c1b8) at radeon.h:804
#9 radeon_get_pixmap_handle (pixmap=0x71c1b8, handle=0x7fa22328) at radeon_bo_helper.c:357
#10 0x76244458 in radeon_pixmap_get_fb (pix=0x71c1b8) at radeon.h:886
#11 drmmode_set_mode_major (crtc=0x691860, mode=0x69191c, rotation=<optimized out>, x=<optimized out>, y=<optimized out>) at drmmode_display.c:918
#12 0x762467e8 in drmmode_set_desired_modes (pScrn=0x67c678, drmmode=<optimized out>, set_hw=1) at drmmode_display.c:3128
#13 0x0047bfa4 in MapWindow (client=0x669ec8, pWin=0x7206c0) at window.c:2722
#14 MapWindow (pWin=0x7206c0, client=0x669ec8) at window.c:2665
#15 0x00449650 in dix_main (argc=3, argv=0x7fa22604, envp=<optimized out>) at main.c:247
#16 0x76ab2198 in __libc_start_main (main=0x42db10 <main>, argc=3, argv=0x7fa22604, init=<optimized out>, fini=0x606434 <__libc_csu_fini>, rtld_fini=0x77229930 <_dl_fini>,
stack_end=0x7fa225e0) at libc-start.c:308
#17 0x0042db80 in __start () at ../sysdeps/mips/start.S:110
Don't call `exaGetPixmapDriverPrivate` if the acceleration (EXA) is not
enabled [yet] to avoid the problem.
Closes: https://gitlab.freedesktop.org/xorg/driver/xf86-video-ati/issues/188
Closes: https://bugzilla.altlinux.org/show_bug.cgi?id=37539
Without the 'extern' this looks like a definition not just a
declaration, in every file that includes the header. gcc 10 is stricter
about this kind of multiple definition.
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
FindClientResourcesByType finds pixmaps from all screens, but trying to
process ones from other screens here makes no sense and likely results
in a crash or memory corruption.
Fixes: 06a4654841 ("Make all active CRTCs scan out an all-black
framebuffer in LeaveVT")
Even with SW cursor, page flipping can be used while no X cursor is
visible.
Occurred to me in the context of xorg/xserver#828.
(Ported from amdgpu commit 87f41ace4920fd2069794211683659eb25b025a6)
This can legitimately fail if the pixmap's storage is shared from
another device, e.g. when using PRIME render offloading.
(Ported from amdgpu commit 7d3fef72e0c871e1677e9e544f4cae5e238b5c52)
All callers were passing TRUE.
(Ported from amdgpu commit ea19a5207054bb159fc7fb6d88e0ceb10c3da010)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
This way, the MSC will continue ticking at the rate of (the last mode
which was enabled for) that CRTC, instead of the client running
unthrottled.
(Ported from amdgpu commit 3109f088fdbd89c2ee8078625d4f073852492656)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
otherwise client would wait for reply forever and desktop appears hang.
Signed-off-by: Flora Cui <flora.cui@amd.com>
(Ported from amdgpu commit fb06fb814700a47464abd756e1111dcc76d0d776)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Even if glamor_gbm_bo_from_pixmap / glamor_fd_from_pixmap themselves
don't trigger any drawing, there could already be unflushed drawing to
the pixmap whose storage we share with a client.
(Ported from amdgpu commit 4b17533fcb30842caf0035ba593b7d986520cc85)
Acked-by: Alex Deucher <alexander.deucher@amd.com>
If get_fb_ptr returns NULL, try again after pixmap_get_handle, it should
work then.
Fixes spurious Present page flipping failures using "normal" pixmaps
which aren't shared with direct rendering clients, e.g. with a
compositor using the RENDER extension.
Bugzilla: https://bugs.freedesktop.org/110417
(Ported from amdgpu commit bf61e6d7ac1a5754b1026d7f80acf25ef622c491)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
This adds tiling support to the driver, it retrieves the tile info from
the kernel and translates it into the server format and exposes the
property.
(Ported from xserver commits 8fb8bbb3062f1a06621ab7030a9e89d5e8367b35
and 6abdb54a11dac4e8854ff94ecdcb90a14321ab31)
(Ported from amdgpu commit 6ee857726166f495abcd68e4ff60e3a09593d079)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 274703087f.
Reports of visual corruption were bisected to this, e.g.
https://bugs.archlinux.org/task/61941 . I can reproduce this with Turks,
but not with Bonaire. I assume it's a Mesa/glamor bug, but let's revert
for now.
Acked-by: Alex Deucher <alexander.deucher@amd.com>
drm_queue_handler just puts the event on the signalled list; without
calling drm_queue_handle_deferred, actual processing of the event may be
delayed indefinitely, e.g. until another event arrives from the kernel.
This could result in DRI2 clients hanging during DPMS off.
Fixes: ba83a866af "Add radeon_drm_handle_event wrapper for
drmHandleEvent"
(Ported from amdgpu commit 09be74a3d1dd9604336d9a27f98d132b262dcbaf)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
If they don't, flipping will result in corrupted display.
Test case:
* Run Xorg at 1920x1080 with no window manager
* glxgears -geometry 2048x1080
The Present extension code in xserver 1.21 will check for this.
(Ported from amdgpu commit a636f42b496b0604ca00a144690ece61d1a88a27)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
drm_wait_pending_flip stopped waiting if drm_handle_event returned 0,
but that might have processed only some unrelated DRM events. As long as
the flip is pending, we have to keep waiting for its completion event.
Noticed while working on the previous fix.
(Ported from amdgpu commit 9045fb310f88780e250e60b80431ca153330e61b)
drmHandleEvent can be interrupted by a signal in read(), in which case
it doesn't process any events but returns -1, which
drm_handle_event propagated to its callers. This could cause the
following failure cascade:
1. drm_wait_pending_flip stopped waiting for a pending flip.
2. Its caller cleared drmmode_crtc->flip_pending before the flip
completed.
3. Another flip was attempted but got an unexpected EBUSY error because
the previous flip was still pending.
4. TearFree was disabled due to the error.
The solution is to call drmHandleEvent if it was interrupted by a
signal. We can do that in drm_handle_event, because when that is called,
either it is known that there are events ready to be processed, or the
caller has to wait for events to arrive anyway.
Bugzilla: https://bugs.freedesktop.org/109364
(Ported from amdgpu commit 3ff2cc225f6bc08364ee007fa54e9d0150adaf11)
And only clear it if it matches the framebuffer of the completed flip
being processed.
Fixes
(WW) RADEON(0): flip queue failed: Device or resource busy
(WW) RADEON(0): Page flip failed: Device or resource busy
(EE) RADEON(0): present flip failed
due to clobbering drmmode_crtc->flip_pending.
Reproducer: Enable TearFree, run warzone2100 fullscreen, toggle
Vertical sync on/off under Video Options. Discovered while investigating
https://bugs.freedesktop.org/109364 .
(Ported from amdgpu commit e72a02ba1d35743fefd939458b9d8cddce86e7f5)
We were using a relative target of 0, meaning "complete the flip ASAP".
This could result in the flip sometimes, but not always completing in
the same vertical blank period where the corresponding drawing occurred,
potentially causing judder artifacts with applications updating their
window contents synchronized to the display refresh. A good way to test
this is the vsynctester.com site in a windowed browser, where the judder
results in the large "VSYNC" text intermittently appearing red or cyan
instead of the expected gray.
To avoid this, use a relative target MSC of 1, meaning that if a
vertical blank period is in progress, the flip will only complete in the
next one.
Reported by Julian Tempel and Brandon Wright in
https://bugs.freedesktop.org/106175 .
(Ported from amdgpu commit a1b479c7d0066c481af920f297d6af9009dda11e)
If the compositing manager uses direct rendering (as is usually the case
these days), the storage of a pixmap allocated by glamor_create_pixmap
needs to be reallocated for sharing it with the compositing manager.
Instead, allocate pixmap storage which can be shared directly.
(Ported from amdgpu commit bf326f2ea19daa6c8da23d6788ff301ae70b8e69)
To make sure the client can't use the shared pixmap storage for direct
rendering first, which could produce garbage.
Bugzilla: https://bugs.freedesktop.org/109235
(Ported from amdgpu commit ebd32b1c07208f8dbe853e089f5e4b7c6a7a658a)
To make sure the client can't use the shared pixmap storage for direct
rendering first, which could produce garbage.
Bugzilla: https://bugs.freedesktop.org/109235
(Ported from amdgpu commit d168532ee739f7e33a2798051e64ba445dd3859f)
There's no point in listening for hotplug events if RandR is disabled,
as there's no other mechanism for them to be propagated. We were already
mostly ignoring them in that case.
Inspired by
1a489142c8
(via https://bugs.freedesktop.org/109230#c11).
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Switch to the other buffer when xf86_config->cursor changes. Avoids
these issues possible when re-using the same buffer:
* The HW may intermittently display a mix of the old and new cursor
images.
* If the hotspot changes, the HW may intermittently display the new
cursor image at the location corresponding to the old image's hotspot.
Bugzilla: https://bugs.freedesktop.org/108832
(Ported from amdgpu commit 0d60233d26ec70d4e1faa343b438e33829c6d5e4)
The cursor position is updated to be consistent with the new hotspot in
the same ioctl call.
(Ported from amdgpu commit b04697de5270e8e45744a7025c24df1f454a4cf0)
This should be functionally equivalent to what drmModeSetCursor(2) do
behind the scenes, but allows for new tricks in following changes.
(Ported from amdgpu commit b344e1559e936046ef02c777fc4f6bcefa3830bc)
Specifically, after both the page flip and vblank ioctls failed, but
then the vblank ioctl started working again. This can happen
intermittently e.g. when hotplugging a DP display. Previously, TearFree
would stay disabled in that case until a modeset was triggered somehow.
Bugzilla: https://bugs.freedesktop.org/103791
(Ported from amdgpu commit bcfa6c258fdf41a9928f8a3c78fc528d0fafee25)
