[Git][xorg-team/driver/xserver-xorg-video-amdgpu][upstream-unstable] 6 commits: glamor: Set AMDGPU_CREATE_PIXMAP_SCANOUT on DRI2 and shared pixmaps

["Timo Aaltonen (@tjaalton)" <tjaalton@debian.org>]

Title: GitLab

Timo Aaltonen pushed to branch upstream-unstable at X Strike Force / driver / xserver-xorg-video-amdgpu

Commits:

• 69365526
  by Dor Askayo at 2021-08-27T14:36:03+03:00

  glamor: Set AMDGPU_CREATE_PIXMAP_SCANOUT on DRI2 and shared pixmaps
  
  The scanout usage flag wasn't applied for DRI2 and shared pixmaps
  in 0732f81a2c67, resulting in GPU artifacts in some scenarios.
  
  Fixes: 0732f81a2c67 ("glamor: Make pixmap scanout compatible if its
                        dimensions are")
  Closes: https://gitlab.freedesktop.org/xorg/driver/xf86-video-amdgpu/-/issues/41
  Signed-off-by: Dor Askayo <dor.askayo@gmail.com>
  Signed-off-by: Michel Dänzer <mdaenzer@redhat.com>
  Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
  

• 65c12736
  by Mario Kleiner at 2022-01-25T03:59:15+01:00

  Add option for non-vsynced flips for "secondary" outputs.
  
  This is a straightforward port of a patch with the same name
  "modesetting: Add option for non-vsynced flips for "secondary"
  outputs." from X-Server master / X-Server 21.1. See server MR 742.
  The description below is therefore identical to that X-Server commit:
  
  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 and dual-display
  setups, and with single-X-Screen as well as dual-X-Screen "ZaphodHeads"
  configurations.
  
  Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
  

• 8bc148f0
  by Alan Coopersmith at 2022-02-14T19:23:49+00:00

  Fix spelling/wording issues
  
  Found by using:
      codespell --builtin clear,rare,usage,informal,code,names
  
  Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
  

• c21bcdd1
  by Jesse Zhang at 2022-02-16T16:21:21+01:00

  glamor: unset AMDGPU_CREATE_PIXMAP_SCANOUT on shared pixmap
  
  While running multi-display test(for both APUs and DGPUs), if
  the screen setting mode is changed from "single mode" to "share mode",
  the screen shows tiled distortion, due to wrongly created pixmap.
  This is a regression.
  
  Fixes: 0732f81a2c67 ("glamor: Make pixmap scanout compatible if its dimensions are")
  Closes: https://gitlab.freedesktop.org/xorg/driver/xf86-video-amdgpu/-/issues/48
  
  Acked-by: Shashank Sharma <shashank.sharma@amd.com>
  Signed-off-by: Jesse Zhang <jesse.zhang@amd.com>
  Signed-off-by: Shashank Sharma <shashank.sharma@amd.com>
  

• 402bfdea
  by Alan Coopersmith at 2022-02-16T15:36:58+00:00

  Build xz tarballs instead of bzip2
  
  Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
  

• 533bd30c
  by Shashank Sharma at 2022-02-22T16:26:37+01:00

  Bump version for the 22.0.0 release
  
  This release includes some bug fixes and one minor feature.
  
  Signed-off-by: Shashank Sharma <shashank.sharma@amd.com>
  

9 changed files:

• configure.ac
• man/amdgpu.man
• src/amdgpu_drv.h
• src/amdgpu_glamor.c
• src/amdgpu_kms.c
• src/amdgpu_pixmap.c
• src/amdgpu_present.c
• src/drmmode_display.c
• src/simple_list.h

Changes:

configure.ac

---

...	...	@@ -23,7 +23,7 @@
23	23	# Initialize Autoconf
24	24	AC_PREREQ([2.60])
25	25	AC_INIT([xf86-video-amdgpu],
26		- [21.0.0],
	26	+ [22.0.0],
27	27	[https://bugs.freedesktop.org/enter_bug.cgi?product=xorg&component=Driver/AMDgpu],
28	28	[xf86-video-amdgpu])
29	29
...	...	@@ -34,7 +34,7 @@ AC_CONFIG_MACRO_DIR([m4])
34	34	AC_CONFIG_AUX_DIR(.)
35	35
36	36	# Initialize Automake
37		-AM_INIT_AUTOMAKE([foreign dist-bzip2])
	37	+AM_INIT_AUTOMAKE([foreign dist-xz])
38	38	AC_SYS_LARGEFILE
39	39
40	40	# Require X.Org macros 1.8 or later for MAN_SUBSTS set by XORG_MANPAGE_SECTIONS

man/amdgpu.man

---

...	...	@@ -90,6 +90,21 @@ when an suitable application is flipping via the Present extension.
90	90	The default is
91	91	.B off.
92	92	.TP
	93	+.BI "Option \*qAsyncFlipSecondaries\*q \*q" boolean \*q
	94	+Use async flips for secondary video outputs on multi-display setups. If a screen
	95	+has multiple displays attached and DRI3 page flipping is used, then only one of
	96	+the displays will have its page flip synchronized to vblank for tear-free
	97	+presentation. This is the display that is used for presentation timing and
	98	+timestamping, usually the one covering the biggest pixel area of the screen.
	99	+All other displays ("Secondaries") will not synchronize their flips. This may
	100	+cause some tearing on these displays, but it prevents a permanent or periodic
	101	+slowdown or irritating judder of animations if not all video outputs are running
	102	+synchronized with each other and with the same refresh rate. There is no perfect
	103	+solution apart from perfectly synchronized outputs, but this option may give
	104	+preferrable results if the displays in a multi-display setup mirror or clone
	105	+each other. The default is
	106	+.B off.
	107	+.TP
93	108	.BI "Option \*qAccelMethod\*q \*q" string \*q
94	109	Setting this option to
95	110	.B none

src/amdgpu_drv.h

---

...	...	@@ -165,6 +165,7 @@ typedef enum {
165	165	OPTION_TEAR_FREE,
166	166	OPTION_DELETE_DP12,
167	167	OPTION_VARIABLE_REFRESH,
	168	+ OPTION_ASYNC_FLIP_SECONDARIES,
168	169	} AMDGPUOpts;
169	170
170	171	static inline ScreenPtr
...	...	@@ -306,6 +307,8 @@ typedef struct {
306	307	/* kms pageflipping */
307	308	WindowPtr flip_window;
308	309	Bool allowPageFlip;
	310	+ Bool can_async_flip;
	311	+ Bool async_flip_secondaries;
309	312
310	313	/* cursor size */
311	314	int cursor_w;

src/amdgpu_glamor.c

---

...	...	@@ -209,6 +209,15 @@ amdgpu_glamor_create_pixmap(ScreenPtr screen, int w, int h, int depth,
209	209	if (!format)
210	210	return NULL;
211	211
	212	+ if (usage != CREATE_PIXMAP_USAGE_BACKING_PIXMAP &&
	213	+ usage != CREATE_PIXMAP_USAGE_SHARED &&
	214	+ !info->shadow_primary &&
	215	+ w >= scrn->virtualX &&
	216	+ w <= scrn->displayWidth &&
	217	+ h == scrn->virtualY &&
	218	+ format->bitsPerPixel == scrn->bitsPerPixel)
	219	+ usage |= AMDGPU_CREATE_PIXMAP_SCANOUT;
	220	+
212	221	if (!(usage & AMDGPU_CREATE_PIXMAP_SCANOUT) &&
213	222	!AMDGPU_CREATE_PIXMAP_SHARED(usage)) {
214	223	if (info->shadow_primary) {
...	...	@@ -218,15 +227,9 @@ amdgpu_glamor_create_pixmap(ScreenPtr screen, int w, int h, int depth,
218	227	usage |= AMDGPU_CREATE_PIXMAP_LINEAR |
219	228	AMDGPU_CREATE_PIXMAP_GTT;
220	229	} else if (usage != CREATE_PIXMAP_USAGE_BACKING_PIXMAP) {
221		- if (w < scrn->virtualX || w > scrn->displayWidth ||
222		- h != scrn->virtualY ||
223		- format->bitsPerPixel != scrn->bitsPerPixel) {
224		- pixmap = glamor_create_pixmap(screen, w, h, depth, usage);
225		- if (pixmap)
226		- return pixmap;
227		- } else {
228		- usage |= AMDGPU_CREATE_PIXMAP_SCANOUT;
229		- }
	230	+ pixmap = glamor_create_pixmap(screen, w, h, depth, usage);
	231	+ if (pixmap)
	232	+ return pixmap;
230	233	}
231	234	}
232	235

src/amdgpu_kms.c

---

...	...	@@ -87,6 +87,7 @@ const OptionInfoRec AMDGPUOptions_KMS[] = {
87	87	{OPTION_TEAR_FREE, "TearFree", OPTV_BOOLEAN, {0}, FALSE},
88	88	{OPTION_DELETE_DP12, "DeleteUnusedDP12Displays", OPTV_BOOLEAN, {0}, FALSE},
89	89	{OPTION_VARIABLE_REFRESH, "VariableRefresh", OPTV_BOOLEAN, {0}, FALSE },
	90	+ {OPTION_ASYNC_FLIP_SECONDARIES, "AsyncFlipSecondaries", OPTV_BOOLEAN, {0}, FALSE},
90	91	{-1, NULL, OPTV_NONE, {0}, FALSE}
91	92	};
92	93
...	...	@@ -1637,6 +1638,13 @@ Bool AMDGPUPreInit_KMS(ScrnInfoPtr pScrn, int flags)
1637	1638
1638	1639	xf86DrvMsg(pScrn->scrnIndex, from, "VariableRefresh: %sabled\n",
1639	1640	info->vrr_support ? "en" : "dis");
	1641	+
	1642	+ info->async_flip_secondaries = FALSE;
	1643	+ from = xf86GetOptValBool(info->Options, OPTION_ASYNC_FLIP_SECONDARIES,
	1644	+ &info->async_flip_secondaries) ? X_CONFIG : X_DEFAULT;
	1645	+
	1646	+ xf86DrvMsg(pScrn->scrnIndex, from, "AsyncFlipSecondaries: %sabled\n",
	1647	+ info->async_flip_secondaries ? "en" : "dis");
1640	1648	}
1641	1649	}
1642	1650

src/amdgpu_pixmap.c

---

...	...	@@ -38,7 +38,7 @@ amdgpu_pixmap_create(ScreenPtr screen, int w, int h, int depth, unsigned usage)
38	38	PixmapPtr pixmap;
39	39	AMDGPUInfoPtr info;
40	40
41		- /* only DRI2 pixmap is suppported */
	41	+ /* only DRI2 pixmap is supported */
42	42	if (!(usage & AMDGPU_CREATE_PIXMAP_DRI2))
43	43	return fbCreatePixmap(screen, w, h, depth, usage);
44	44

src/amdgpu_present.c

---

...	...	@@ -496,8 +496,13 @@ amdgpu_present_has_async_flip(ScreenPtr screen)
496	496	Bool
497	497	amdgpu_present_screen_init(ScreenPtr screen)
498	498	{
499		- if (amdgpu_present_has_async_flip(screen))
	499	+ ScrnInfoPtr scrn = xf86ScreenToScrn(screen);
	500	+ AMDGPUInfoPtr info = AMDGPUPTR(scrn);
	501	+
	502	+ if (amdgpu_present_has_async_flip(screen)) {
500	503	amdgpu_present_screen_info.capabilities |= PresentCapabilityAsync;
	504	+ info->can_async_flip = TRUE;
	505	+ }
501	506
502	507	if (!present_screen_init(screen, &amdgpu_present_screen_info)) {
503	508	xf86DrvMsg(xf86ScreenToScrn(screen)->scrnIndex, X_WARNING,

src/drmmode_display.c

---

...	...	@@ -3970,17 +3970,27 @@ Bool amdgpu_do_pageflip(ScrnInfoPtr scrn, ClientPtr client,
3970	3970	uint32_t target_msc)
3971	3971	{
3972	3972	AMDGPUEntPtr pAMDGPUEnt = AMDGPUEntPriv(scrn);
	3973	+ AMDGPUInfoPtr info = AMDGPUPTR(scrn);
3973	3974	xf86CrtcConfigPtr config = XF86_CRTC_CONFIG_PTR(scrn);
3974	3975	xf86CrtcPtr crtc = NULL;
3975	3976	drmmode_crtc_private_ptr drmmode_crtc = config->crtc[0]->driver_private;
3976	3977	int crtc_id;
3977	3978	uint32_t flip_flags = flip_sync == FLIP_ASYNC ? DRM_MODE_PAGE_FLIP_ASYNC : 0;
	3979	+ uint32_t sec_flip_flags = flip_flags;
3978	3980	drmmode_flipdata_ptr flipdata;
3979	3981	Bool handle_deferred = FALSE;
3980	3982	uintptr_t drm_queue_seq = 0;
3981	3983	struct drmmode_fb *fb;
3982	3984	int i = 0;
3983	3985
	3986	+ /*
	3987	+ * Flip secondary non-ref_crtc crtc's async if possible and requested
	3988	+ * by xorg.conf option "AsyncFlipSecondaries". Otherwise follow the lead
	3989	+ * of flip_sync.
	3990	+ */
	3991	+ if (info->can_async_flip && info->async_flip_secondaries)
	3992	+ sec_flip_flags |= DRM_MODE_PAGE_FLIP_ASYNC;
	3993	+
3984	3994	flipdata = calloc(1, sizeof(*flipdata) + drmmode_crtc->drmmode->count_crtcs *
3985	3995	sizeof(flipdata->fb[0]));
3986	3996	if (!flipdata) {
...	...	@@ -4081,7 +4091,7 @@ Bool amdgpu_do_pageflip(ScrnInfoPtr scrn, ClientPtr client,
4081	4091	if (drmmode_page_flip_target_relative(pAMDGPUEnt,
4082	4092	drmmode_crtc,
4083	4093	flipdata->fb[crtc_id]->handle,
4084		- flip_flags,
	4094	+ sec_flip_flags,
4085	4095	drm_queue_seq, 0) != 0)
4086	4096	goto flip_error;
4087	4097	}

src/simple_list.h

---

...	...	@@ -2,7 +2,7 @@
2	2	* \file simple_list.h
3	3	* Simple macros for type-safe, intrusive lists.
4	4	*
5		- * Intended to work with a list sentinal which is created as an empty
	5	+ * Intended to work with a list sentinel which is created as an empty
6	6	* list. Insert & delete are O(1).
7	7	*
8	8	* \author
...	...	@@ -107,12 +107,12 @@ do { \
107	107	/**
108	108	* Make a empty list empty.
109	109	*
110		- * \param sentinal list (sentinal element).
	110	+ * \param sentinel list (sentinel element).
111	111	*/
112		-#define make_empty_list(sentinal) \
	112	+#define make_empty_list(sentinel) \
113	113	do { \
114		- (sentinal)->next = sentinal; \
115		- (sentinal)->prev = sentinal; \
	114	+ (sentinel)->next = sentinel; \
	115	+ (sentinel)->prev = sentinel; \
116	116	} while (0)
117	117
118	118	/**

—
View it on GitLab.
You're receiving this email because of your account on salsa.debian.org. If you'd like to receive fewer emails, you can adjust your notification settings.