[PATCH i-g-t 0/7] kms_flip: cleanups

[PATCH i-g-t 0/7] kms_flip: cleanups

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 1097 bytes --]

The last few weeks, I have been dive into the page flip and vblank with
the goal to learn how the user space interact with DRM; because of this,
I am studying the kms_flip code. As a result, I noticed a bunch of
cleanups that could be useful for improving the overall quality of
kms_flip. This patchset can be seen in two parts:

 1. Eliminates unreachable tests
 2. Basic cleanups with the goal to improve the code readability

In particular, the patch number three enables kms_flip to work without
any issue in any non-intel drivers.

Rodrigo Siqueira (7):
  kms_flip: Removes unreachable code related to TEST_RPM
  kms_flip: Removes unreachable code related to TEST_TS_CONT
  kms_flip: Fix warning related to i915 gem dependence
  kms_flip: Remove unreachable condition in wait_for_events
  kms_flip: remove magic constant in run_test_on_crtc_set()
  kms_flip: Rework set_mode()
  kms_flip: Standardize return value for fb_is_bound

 tests/kms_flip.c | 88 +++++++++++++++++-------------------------------
 1 file changed, 31 insertions(+), 57 deletions(-)

-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 1/7] kms_flip: Removes unreachable code related to TEST_RPM

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 2201 bytes --]

This commit removes the code related to TEST_RPM test because the
kms_flip never sets this flags, i.e., TEST_RPM is not used. Take a look
at commit 07a3fccf to see why this flag is never set.

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 15 ---------------
 1 file changed, 15 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index 798fc4e8..f769d30e 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -68,7 +68,6 @@
 #define TEST_ENOENT		(1 << 22)
 #define TEST_FENCE_STRESS	(1 << 23)
 #define TEST_VBLANK_RACE	(1 << 24)
-#define TEST_RPM		(1 << 25)
 #define TEST_SUSPEND		(1 << 26)
 #define TEST_TS_CONT		(1 << 27)
 #define TEST_BO_TOOBIG		(1 << 28)
@@ -809,9 +808,6 @@ static unsigned int run_test_step(struct test_output *o)
 			     "failed to disable output: %s\n",
 			     strerror(errno));
 
-	if (o->flags & TEST_RPM)
-		igt_assert(igt_wait_for_pm_status(IGT_RUNTIME_PM_STATUS_SUSPENDED));
-
 	if (o->flags & TEST_SUSPEND)
 		igt_system_suspend_autoresume(SUSPEND_STATE_MEM,
 					      SUSPEND_TEST_NONE);
@@ -1321,9 +1317,6 @@ static int run_test(int duration, int flags)
 
 	igt_require((flags & TEST_HANG) == 0 || !is_wedged(drm_fd));
 
-	if (flags & TEST_RPM)
-		igt_require(igt_setup_runtime_pm());
-
 	resources = drmModeGetResources(drm_fd);
 	igt_require(resources);
 
@@ -1570,10 +1563,6 @@ int main(int argc, char **argv)
 		if (tests[i].flags & TEST_NO_2X_OUTPUT)
 			continue;
 
-		/* code doesn't disable all crtcs, so skip rpm tests */
-		if (tests[i].flags & TEST_RPM)
-			continue;
-
 		igt_subtest_f( "2x-%s", tests[i].name)
 			run_pair(tests[i].duration, tests[i].flags);
 	}
@@ -1592,10 +1581,6 @@ int main(int argc, char **argv)
 		if (tests[i].flags & TEST_NO_2X_OUTPUT)
 			continue;
 
-		/* code doesn't disable all crtcs, so skip rpm tests */
-		if (tests[i].flags & TEST_RPM)
-			continue;
-
 		igt_subtest_f( "2x-%s-interruptible", tests[i].name)
 			run_pair(tests[i].duration, tests[i].flags);
 	}
-- 
2.21.0


-- 
Rodrigo Siqueira
https://siqueira.tech
Graduate Student
Department of Computer Science
University of São Paulo

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 2/7] kms_flip: Removes unreachable code related to TEST_TS_CONT

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 1668 bytes --]

This commit removes the code related to TEST_TS_CONT test because the
kms_flip never sets this flags, i.e., TEST_TS_CONT is not used. Take a
look at commit 07a3fccf to see why this flag is never set.

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 16 ----------------
 1 file changed, 16 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index f769d30e..0d261b77 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -69,7 +69,6 @@
 #define TEST_FENCE_STRESS	(1 << 23)
 #define TEST_VBLANK_RACE	(1 << 24)
 #define TEST_SUSPEND		(1 << 26)
-#define TEST_TS_CONT		(1 << 27)
 #define TEST_BO_TOOBIG		(1 << 28)
 
 #define TEST_BASIC		(1 << 30)
@@ -497,21 +496,6 @@ static void check_state(const struct test_output *o, const struct event_state *e
 			     "unexpected %s seq %u, should be >= %u\n",
 			     es->name, es->current_seq, es->last_seq + o->seq_step);
 
-	/* Check that the vblank frame didn't wrap unexpectedly. */
-	if (o->flags & TEST_TS_CONT) {
-		/* Ignore seq_step here since vblank waits time out immediately
-		 * when we kill the crtc. */
-		igt_assert_f(es->current_seq - es->last_seq >= 0,
-			     "unexpected %s seq %u, should be >= %u\n",
-			     es->name, es->current_seq, es->last_seq);
-		igt_assert_f(es->current_seq - es->last_seq <= 150,
-			     "unexpected %s seq %u, should be < %u\n",
-			     es->name, es->current_seq, es->last_seq + 150);
-
-		igt_debug("testing ts continuity: Current frame %u, old frame %u\n",
-			  es->current_seq, es->last_seq);
-	}
-
 	if (o->flags & TEST_CHECK_TS) {
 		double elapsed, expected;
 
-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 3/7] kms_flip: Fix warning related to i915 gem dependence

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 1400 bytes --]

In the kms_flip tests has an igt_fixture that allocates memory for i915
driver with “drm_intel_bufmgr_gem_init()”, which produces the following
warning:

 IGT-Version: 1.23-g8d81c2c2 (x86_64) (Linux: 5.0.0-rc7-VKMS-RULES+ x86_64)
 Using monotonic timestamps
 DRM_IOCTL_I915_GEM_APERTURE failed: Invalid argument
 Assuming 131072kB available aperture size.
 May lead to reduced performance or incorrect rendering.
 get chip id failed: -1 [22]

This commit handles this specific dependence with the i915 driver which
make the kms_flip logs better.

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index 0d261b77..57138ec1 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -1528,10 +1528,12 @@ int main(int argc, char **argv)
 		igt_install_exit_handler(kms_flip_exit_handler);
 		get_timestamp_format();
 
-		bufmgr = drm_intel_bufmgr_gem_init(drm_fd, 4096);
-		if (bufmgr) {
-			devid = intel_get_drm_devid(drm_fd);
-			batch = intel_batchbuffer_alloc(bufmgr, devid);
+		if (is_i915_device(drm_fd)) {
+			bufmgr = drm_intel_bufmgr_gem_init(drm_fd, 4096);
+			if (bufmgr) {
+				devid = intel_get_drm_devid(drm_fd);
+				batch = intel_batchbuffer_alloc(bufmgr, devid);
+			}
 		}
 	}
 
-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 4/7] kms_flip: Remove unreachable condition in wait_for_events

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 1145 bytes --]

In the function wait_for_events() has a double check in the select()
function return as described below:

  igt_assert_f(ret >= 0,
        "select error (errno %i)\n", errno);
  igt_assert_f(ret > 0,
        "select timed out or error (ret %d)\n", ret);

Note that the second assert condition will not be reached because of the
first assertion. This commit removes the code duplication and update the
error message.

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index 57138ec1..9ef77de9 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -1014,9 +1014,7 @@ static unsigned int wait_for_events(struct test_output *o)
 		} while (ret < 0 && errno == EINTR);
 
 		igt_assert_f(ret >= 0,
-			     "select error (errno %i)\n", errno);
-		igt_assert_f(ret > 0,
-			     "select timed out or error (ret %d)\n", ret);
+			     "select error: %s (%d))\n", strerror(errno), ret);
 		igt_assert_f(!FD_ISSET(0, &fds),
 			     "no fds active, breaking\n");
 
-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 5/7] kms_flip: Remove magic constant in run_test_on_crtc_set()

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 2041 bytes --]

The function run_test_on_crtc_set() expects a parameter named crtc_count
which slight changes the behavior of the test. If this crtc_count is
‘1’, the test will run in a single CRTC; otherwise, it will run in two
different CRTC. However, this function uses hardcoded literal 1 and 2
for each case. This patch creates two constant with the goal to improve
the readability.

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index 9ef77de9..42ae3ebc 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -76,6 +76,9 @@
 #define EVENT_FLIP		(1 << 0)
 #define EVENT_VBLANK		(1 << 1)
 
+#define RUN_TEST		1
+#define RUN_PAIR		2
+
 #ifndef DRM_CAP_TIMESTAMP_MONOTONIC
 #define DRM_CAP_TIMESTAMP_MONOTONIC 6
 #endif
@@ -1164,7 +1167,7 @@ static void run_test_on_crtc_set(struct test_output *o, int *crtc_idxs,
 	int i;
 
 	switch (crtc_count) {
-	case 1:
+	case RUN_TEST:
 		connector_find_preferred_mode(o->_connector[0], crtc_idxs[0], o);
 		if (!o->mode_valid)
 			return;
@@ -1175,7 +1178,7 @@ static void run_test_on_crtc_set(struct test_output *o, int *crtc_idxs,
 			 kmstest_connector_type_str(o->kconnector[0]->connector_type),
 			 o->kconnector[0]->connector_type_id);
 		break;
-	case 2:
+	case RUN_PAIR:
 		connector_find_compatible_mode(crtc_idxs[0], crtc_idxs[1], o);
 		if (!o->mode_valid)
 			return;
@@ -1341,7 +1344,7 @@ static int run_test(int duration, int flags)
 			o.depth = 24;
 
 			crtc_idx = n;
-			run_test_on_crtc_set(&o, &crtc_idx, 1, duration);
+			run_test_on_crtc_set(&o, &crtc_idx, RUN_TEST, duration);
 		}
 	}
 
@@ -1410,7 +1413,8 @@ static int run_pair(int duration, int flags)
 					crtc_idxs[0] = n;
 					crtc_idxs[1] = m;
 
-					run_test_on_crtc_set(&o, crtc_idxs, 2,
+					run_test_on_crtc_set(&o, crtc_idxs,
+							     RUN_PAIR,
 							     duration);
 				}
 			}
-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 6/7] kms_flip: Rework set_mode()

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 1304 bytes --]

This patch removes the duplicate code inside the function set_mode().

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index 42ae3ebc..de3ab600 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -609,22 +609,24 @@ static bool is_wedged(int fd)
 
 static int set_mode(struct test_output *o, uint32_t fb, int x, int y)
 {
-	int n;
+	int n, ret;
 
 	for (n = o->count - 1; n >= 0; n--) {
+		uint32_t buffer_id = fb, x_crtc = x, y_crtc = y;
+		uint32_t *conn = &o->_connector[n];
+		int count = 1;
+		drmModeModeInfoPtr mode = &o->kmode[n];
+
 		if (fb == 0) {
-			int ret = drmModeSetCrtc(drm_fd, o->_crtc[n],
-						 0, 0, 0,
-						 0, 0, 0);
-			if (ret)
-				return ret;
-		} else {
-			int ret = drmModeSetCrtc(drm_fd, o->_crtc[n],
-						 fb, x, y,
-						 &o->_connector[n], 1, &o->kmode[n]);
-			if (ret)
-				return ret;
+			buffer_id = x_crtc = y_crtc = count = 0;
+			conn = NULL; mode = NULL;
 		}
+
+		ret = drmModeSetCrtc(drm_fd, o->_crtc[n],
+				     buffer_id, x_crtc, y_crtc,
+				     conn, count, mode);
+		if (ret)
+			return ret;
 	}
 
 	return 0;
-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH i-g-t 7/7] kms_flip: Standardize return value for fb_is_bound

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 968 bytes --]

The function fb_is_bound() mix integer value with booleans for handling
the return value. This commit standardizes the return value of
fb_is_bound() for using only booleans.

Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
---
 tests/kms_flip.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/tests/kms_flip.c b/tests/kms_flip.c
index de3ab600..abfdd363 100755
--- a/tests/kms_flip.c
+++ b/tests/kms_flip.c
@@ -947,8 +947,7 @@ static void paint_flip_mode(struct igt_fb *fb, bool odd_frame)
 	igt_put_cairo_ctx(drm_fd, fb, cr);
 }
 
-static int
-fb_is_bound(struct test_output *o, int fb)
+static bool fb_is_bound(struct test_output *o, int fb)
 {
 	int n;
 
@@ -958,7 +957,7 @@ fb_is_bound(struct test_output *o, int fb)
 		};
 
 		if (drmIoctl(drm_fd, DRM_IOCTL_MODE_GETCRTC, &mode))
-			return 0;
+			return false;
 
 		if (!mode.mode_valid || mode.fb_id != fb)
 			return false;
-- 
2.21.0

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH i-g-t 4/7] kms_flip: Remove unreachable condition in wait_for_events

[Petri Latvala]

On Mon, Mar 04, 2019 at 12:30:55PM -0300, Rodrigo Siqueira wrote:
> In the function wait_for_events() has a double check in the select()
> function return as described below:
> 
>   igt_assert_f(ret >= 0,
>         "select error (errno %i)\n", errno);
>   igt_assert_f(ret > 0,
>         "select timed out or error (ret %d)\n", ret);
> 
> Note that the second assert condition will not be reached because of the
> first assertion. This commit removes the code duplication and update the
> error message.

It will be reached if ret equals 0.

If select() returns 0, errno is unspecified, the main reason for the
separation.


-- 
Petri Latvala
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [igt-dev] [PATCH i-g-t 7/7] kms_flip: Standardize return value for fb_is_bound

[Maarten Lankhorst]

Op 04-03-2019 om 16:32 schreef Rodrigo Siqueira:
> The function fb_is_bound() mix integer value with booleans for handling
> the return value. This commit standardizes the return value of
> fb_is_bound() for using only booleans.
>
> Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
> ---
>  tests/kms_flip.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/tests/kms_flip.c b/tests/kms_flip.c
> index de3ab600..abfdd363 100755
> --- a/tests/kms_flip.c
> +++ b/tests/kms_flip.c
> @@ -947,8 +947,7 @@ static void paint_flip_mode(struct igt_fb *fb, bool odd_frame)
>  	igt_put_cairo_ctx(drm_fd, fb, cr);
>  }
>  
> -static int
> -fb_is_bound(struct test_output *o, int fb)
> +static bool fb_is_bound(struct test_output *o, int fb)
>  {
>  	int n;
>  
> @@ -958,7 +957,7 @@ fb_is_bound(struct test_output *o, int fb)
>  		};
>  
>  		if (drmIoctl(drm_fd, DRM_IOCTL_MODE_GETCRTC, &mode))
> -			return 0;
> +			return false;
>  
>  		if (!mode.mode_valid || mode.fb_id != fb)
>  			return false;
>
> _______________________________________________
> igt-dev mailing list
> igt-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/igt-dev

Nice cleanup. Hope we can eventually convert kms_flip to igt_display at some point for readability.

For all except patch 4:

Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>

:)

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [igt-dev] [PATCH i-g-t 7/7] kms_flip: Standardize return value for fb_is_bound

[Rodrigo Siqueira]

[-- Attachment #1.1: Type: text/plain, Size: 2108 bytes --]

On 03/06, Maarten Lankhorst wrote:
> Op 04-03-2019 om 16:32 schreef Rodrigo Siqueira:
> > The function fb_is_bound() mix integer value with booleans for handling
> > the return value. This commit standardizes the return value of
> > fb_is_bound() for using only booleans.
> >
> > Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
> > ---
> >  tests/kms_flip.c | 5 ++---
> >  1 file changed, 2 insertions(+), 3 deletions(-)
> >
> > diff --git a/tests/kms_flip.c b/tests/kms_flip.c
> > index de3ab600..abfdd363 100755
> > --- a/tests/kms_flip.c
> > +++ b/tests/kms_flip.c
> > @@ -947,8 +947,7 @@ static void paint_flip_mode(struct igt_fb *fb, bool odd_frame)
> >  	igt_put_cairo_ctx(drm_fd, fb, cr);
> >  }
> >  
> > -static int
> > -fb_is_bound(struct test_output *o, int fb)
> > +static bool fb_is_bound(struct test_output *o, int fb)
> >  {
> >  	int n;
> >  
> > @@ -958,7 +957,7 @@ fb_is_bound(struct test_output *o, int fb)
> >  		};
> >  
> >  		if (drmIoctl(drm_fd, DRM_IOCTL_MODE_GETCRTC, &mode))
> > -			return 0;
> > +			return false;
> >  
> >  		if (!mode.mode_valid || mode.fb_id != fb)
> >  			return false;
> >
> > _______________________________________________
> > igt-dev mailing list
> > igt-dev@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/igt-dev
> 
> Nice cleanup. Hope we can eventually convert kms_flip to igt_display at some point for readability.

Hi Petri and Maarten,

First of all, thanks for your review and the clarification in the patch
4 (I really missed the ret = 0, thanks for highlight this case :)

Maarten, could you give some extra details about the idea of converting
kms_flip to igt_display? Could you point out some links or code
reference? I'm just curious to know a little more about this subject.

Best Regards
Rodrigo Siqueira

> For all except patch 4:
> 
> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> 
> :)
>


-- 
Rodrigo Siqueira
https://siqueira.tech
Graduate Student
Department of Computer Science
University of São Paulo

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx