All of lore.kernel.org
 help / color / mirror / Atom feed
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Intel-gfx@lists.freedesktop.org
Cc: cgroups@vger.kernel.org, linux-kernel@vger.kernel.org,
	"Tejun Heo" <tj@kernel.org>,
	"Johannes Weiner" <hannes@cmpxchg.org>,
	"Zefan Li" <lizefan.x@bytedance.com>,
	"Dave Airlie" <airlied@redhat.com>,
	"Daniel Vetter" <daniel.vetter@ffwll.ch>,
	"Rob Clark" <robdclark@chromium.org>,
	"Stéphane Marchesin" <marcheu@chromium.org>,
	"T . J . Mercier" <tjmercier@google.com>,
	Kenny.Ho@amd.com, "Christian König" <christian.koenig@amd.com>,
	"Brian Welty" <brian.welty@intel.com>,
	"Tvrtko Ursulin" <tvrtko.ursulin@intel.com>
Subject: [RFC 09/17] cgroup/drm: Introduce weight based drm cgroup control
Date: Wed, 19 Oct 2022 18:32:46 +0100	[thread overview]
Message-ID: <20221019173254.3361334-10-tvrtko.ursulin@linux.intel.com> (raw)
In-Reply-To: <20221019173254.3361334-1-tvrtko.ursulin@linux.intel.com>

From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

Similar to CPU scheduling, implement a concept of weight in the drm cgroup
controller.

Uses the same range and default as the CPU controller - CGROUP_WEIGHT_MIN,
CGROUP_WEIGHT_DFL and CGROUP_WEIGHT_MAX.

Later each cgroup is assigned a time budget proportionaly based on the
relative weights of it's siblings. This time budget is in turn split by
the group's children and so on.

Children of the root cgroup will be exempt from split budgets and
therefore compete for the GPU time independently and without weight based
control.

This will be used to implement a soft, or best effort signal from drm
cgroup to drm core notifying about groups which are over their allotted
budget.

No guarantees that the limit can be enforced are provided or implied.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
---
 kernel/cgroup/drm.c | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/kernel/cgroup/drm.c b/kernel/cgroup/drm.c
index 01954c3a2087..4b6f88d8236e 100644
--- a/kernel/cgroup/drm.c
+++ b/kernel/cgroup/drm.c
@@ -17,6 +17,7 @@ struct drm_cgroup_state {
 
 	int priority;
 	int effective_priority;
+	unsigned int weight;
 };
 
 static DEFINE_MUTEX(drmcg_mutex);
@@ -163,6 +164,33 @@ drmcs_write_priority(struct cgroup_subsys_state *css, struct cftype *cftype,
 	return 0;
 }
 
+static u64
+drmcs_read_weight(struct cgroup_subsys_state *css, struct cftype *cft)
+{
+	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
+
+	return drmcs->weight;
+}
+
+static int
+drmcs_write_weight(struct cgroup_subsys_state *css, struct cftype *cftype,
+		   u64 weight)
+{
+	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
+	int ret;
+
+	if (weight < CGROUP_WEIGHT_MIN || weight > CGROUP_WEIGHT_MAX)
+		return -ERANGE;
+
+	ret = mutex_lock_interruptible(&drmcg_mutex);
+	if (ret)
+		return ret;
+	drmcs->weight = weight;
+	mutex_unlock(&drmcg_mutex);
+
+	return 0;
+}
+
 static int drmcs_online(struct cgroup_subsys_state *css)
 {
 	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
@@ -210,6 +238,8 @@ drmcs_alloc(struct cgroup_subsys_state *parent_css)
 	if (!drmcs)
 		return ERR_PTR(-ENOMEM);
 
+	drmcs->weight = CGROUP_WEIGHT_DFL;
+
 	return &drmcs->css;
 }
 
@@ -230,6 +260,12 @@ struct cftype files[] = {
 		.flags = CFTYPE_NOT_ON_ROOT,
 		.read_s64 = drmcs_read_effective_priority,
 	},
+	{
+		.name = "weight",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.read_u64 = drmcs_read_weight,
+		.write_u64 = drmcs_write_weight,
+	},
 	{ } /* Zero entry terminates. */
 };
 
-- 
2.34.1


WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Intel-gfx@lists.freedesktop.org
Cc: "Rob Clark" <robdclark@chromium.org>,
	Kenny.Ho@amd.com, "Daniel Vetter" <daniel.vetter@ffwll.ch>,
	"Johannes Weiner" <hannes@cmpxchg.org>,
	linux-kernel@vger.kernel.org,
	"Stéphane Marchesin" <marcheu@chromium.org>,
	"Christian König" <christian.koenig@amd.com>,
	"Zefan Li" <lizefan.x@bytedance.com>,
	"Dave Airlie" <airlied@redhat.com>, "Tejun Heo" <tj@kernel.org>,
	cgroups@vger.kernel.org, "T . J . Mercier" <tjmercier@google.com>
Subject: [Intel-gfx] [RFC 09/17] cgroup/drm: Introduce weight based drm cgroup control
Date: Wed, 19 Oct 2022 18:32:46 +0100	[thread overview]
Message-ID: <20221019173254.3361334-10-tvrtko.ursulin@linux.intel.com> (raw)
In-Reply-To: <20221019173254.3361334-1-tvrtko.ursulin@linux.intel.com>

From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

Similar to CPU scheduling, implement a concept of weight in the drm cgroup
controller.

Uses the same range and default as the CPU controller - CGROUP_WEIGHT_MIN,
CGROUP_WEIGHT_DFL and CGROUP_WEIGHT_MAX.

Later each cgroup is assigned a time budget proportionaly based on the
relative weights of it's siblings. This time budget is in turn split by
the group's children and so on.

Children of the root cgroup will be exempt from split budgets and
therefore compete for the GPU time independently and without weight based
control.

This will be used to implement a soft, or best effort signal from drm
cgroup to drm core notifying about groups which are over their allotted
budget.

No guarantees that the limit can be enforced are provided or implied.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
---
 kernel/cgroup/drm.c | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/kernel/cgroup/drm.c b/kernel/cgroup/drm.c
index 01954c3a2087..4b6f88d8236e 100644
--- a/kernel/cgroup/drm.c
+++ b/kernel/cgroup/drm.c
@@ -17,6 +17,7 @@ struct drm_cgroup_state {
 
 	int priority;
 	int effective_priority;
+	unsigned int weight;
 };
 
 static DEFINE_MUTEX(drmcg_mutex);
@@ -163,6 +164,33 @@ drmcs_write_priority(struct cgroup_subsys_state *css, struct cftype *cftype,
 	return 0;
 }
 
+static u64
+drmcs_read_weight(struct cgroup_subsys_state *css, struct cftype *cft)
+{
+	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
+
+	return drmcs->weight;
+}
+
+static int
+drmcs_write_weight(struct cgroup_subsys_state *css, struct cftype *cftype,
+		   u64 weight)
+{
+	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
+	int ret;
+
+	if (weight < CGROUP_WEIGHT_MIN || weight > CGROUP_WEIGHT_MAX)
+		return -ERANGE;
+
+	ret = mutex_lock_interruptible(&drmcg_mutex);
+	if (ret)
+		return ret;
+	drmcs->weight = weight;
+	mutex_unlock(&drmcg_mutex);
+
+	return 0;
+}
+
 static int drmcs_online(struct cgroup_subsys_state *css)
 {
 	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
@@ -210,6 +238,8 @@ drmcs_alloc(struct cgroup_subsys_state *parent_css)
 	if (!drmcs)
 		return ERR_PTR(-ENOMEM);
 
+	drmcs->weight = CGROUP_WEIGHT_DFL;
+
 	return &drmcs->css;
 }
 
@@ -230,6 +260,12 @@ struct cftype files[] = {
 		.flags = CFTYPE_NOT_ON_ROOT,
 		.read_s64 = drmcs_read_effective_priority,
 	},
+	{
+		.name = "weight",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.read_u64 = drmcs_read_weight,
+		.write_u64 = drmcs_write_weight,
+	},
 	{ } /* Zero entry terminates. */
 };
 
-- 
2.34.1


WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Intel-gfx@lists.freedesktop.org
Cc: "Rob Clark" <robdclark@chromium.org>,
	Kenny.Ho@amd.com, "Daniel Vetter" <daniel.vetter@ffwll.ch>,
	"Johannes Weiner" <hannes@cmpxchg.org>,
	linux-kernel@vger.kernel.org,
	"Stéphane Marchesin" <marcheu@chromium.org>,
	"Christian König" <christian.koenig@amd.com>,
	"Zefan Li" <lizefan.x@bytedance.com>,
	"Dave Airlie" <airlied@redhat.com>, "Tejun Heo" <tj@kernel.org>,
	cgroups@vger.kernel.org, "T . J . Mercier" <tjmercier@google.com>
Subject: [RFC 09/17] cgroup/drm: Introduce weight based drm cgroup control
Date: Wed, 19 Oct 2022 18:32:46 +0100	[thread overview]
Message-ID: <20221019173254.3361334-10-tvrtko.ursulin@linux.intel.com> (raw)
In-Reply-To: <20221019173254.3361334-1-tvrtko.ursulin@linux.intel.com>

From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

Similar to CPU scheduling, implement a concept of weight in the drm cgroup
controller.

Uses the same range and default as the CPU controller - CGROUP_WEIGHT_MIN,
CGROUP_WEIGHT_DFL and CGROUP_WEIGHT_MAX.

Later each cgroup is assigned a time budget proportionaly based on the
relative weights of it's siblings. This time budget is in turn split by
the group's children and so on.

Children of the root cgroup will be exempt from split budgets and
therefore compete for the GPU time independently and without weight based
control.

This will be used to implement a soft, or best effort signal from drm
cgroup to drm core notifying about groups which are over their allotted
budget.

No guarantees that the limit can be enforced are provided or implied.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
---
 kernel/cgroup/drm.c | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/kernel/cgroup/drm.c b/kernel/cgroup/drm.c
index 01954c3a2087..4b6f88d8236e 100644
--- a/kernel/cgroup/drm.c
+++ b/kernel/cgroup/drm.c
@@ -17,6 +17,7 @@ struct drm_cgroup_state {
 
 	int priority;
 	int effective_priority;
+	unsigned int weight;
 };
 
 static DEFINE_MUTEX(drmcg_mutex);
@@ -163,6 +164,33 @@ drmcs_write_priority(struct cgroup_subsys_state *css, struct cftype *cftype,
 	return 0;
 }
 
+static u64
+drmcs_read_weight(struct cgroup_subsys_state *css, struct cftype *cft)
+{
+	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
+
+	return drmcs->weight;
+}
+
+static int
+drmcs_write_weight(struct cgroup_subsys_state *css, struct cftype *cftype,
+		   u64 weight)
+{
+	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
+	int ret;
+
+	if (weight < CGROUP_WEIGHT_MIN || weight > CGROUP_WEIGHT_MAX)
+		return -ERANGE;
+
+	ret = mutex_lock_interruptible(&drmcg_mutex);
+	if (ret)
+		return ret;
+	drmcs->weight = weight;
+	mutex_unlock(&drmcg_mutex);
+
+	return 0;
+}
+
 static int drmcs_online(struct cgroup_subsys_state *css)
 {
 	struct drm_cgroup_state *drmcs = css_to_drmcs(css);
@@ -210,6 +238,8 @@ drmcs_alloc(struct cgroup_subsys_state *parent_css)
 	if (!drmcs)
 		return ERR_PTR(-ENOMEM);
 
+	drmcs->weight = CGROUP_WEIGHT_DFL;
+
 	return &drmcs->css;
 }
 
@@ -230,6 +260,12 @@ struct cftype files[] = {
 		.flags = CFTYPE_NOT_ON_ROOT,
 		.read_s64 = drmcs_read_effective_priority,
 	},
+	{
+		.name = "weight",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.read_u64 = drmcs_read_weight,
+		.write_u64 = drmcs_write_weight,
+	},
 	{ } /* Zero entry terminates. */
 };
 
-- 
2.34.1


  parent reply	other threads:[~2022-10-19 17:35 UTC|newest]

Thread overview: 86+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-10-19 17:32 [RFC 00/17] DRM scheduling cgroup controller Tvrtko Ursulin
2022-10-19 17:32 ` Tvrtko Ursulin
2022-10-19 17:32 ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 01/17] cgroup: Add the DRM " Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 02/17] drm: Track clients per owning process Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-20  6:40   ` Christian König
2022-10-20  6:40     ` Christian König
2022-10-20  6:40     ` [Intel-gfx] " Christian König
2022-10-20  7:34     ` Tvrtko Ursulin
2022-10-20  7:34       ` Tvrtko Ursulin
2022-10-20  7:34       ` [Intel-gfx] " Tvrtko Ursulin
2022-10-20 11:33       ` Christian König
2022-10-20 11:33         ` Christian König
2022-10-20 11:33         ` [Intel-gfx] " Christian König
2022-10-27 14:35         ` Tvrtko Ursulin
2022-10-27 14:35           ` Tvrtko Ursulin
2022-10-27 14:35           ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 03/17] cgroup/drm: Support cgroup priority control Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 04/17] drm/cgroup: Allow safe external access to file_priv Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 05/17] drm: Connect priority updates to drm core Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-20  9:50   ` kernel test robot
2022-10-19 17:32 ` [RFC 06/17] drm: Only track clients which are providing drm_cgroup_ops Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [Intel-gfx] [RFC 07/17] drm/i915: i915 priority Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 08/17] drm: Allow for migration of clients Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` Tvrtko Ursulin [this message]
2022-10-19 17:32   ` [RFC 09/17] cgroup/drm: Introduce weight based drm cgroup control Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 10/17] drm: Add ability to query drm cgroup GPU time Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 11/17] drm: Add over budget signalling callback Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 12/17] cgroup/drm: Client exit hook Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 13/17] cgroup/drm: Ability to periodically scan cgroups for over budget GPU usage Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-21 22:52   ` T.J. Mercier
2022-10-21 22:52     ` T.J. Mercier
2022-10-21 22:52     ` [Intel-gfx] " T.J. Mercier
2022-10-27 14:45     ` Tvrtko Ursulin
2022-10-27 14:45       ` Tvrtko Ursulin
2022-10-27 14:45       ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [Intel-gfx] [RFC 14/17] cgroup/drm: Show group budget signaling capability in sysfs Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32 ` [RFC 15/17] drm/i915: Migrate client to new owner on context create Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` [Intel-gfx] " Tvrtko Ursulin
2022-10-19 17:32 ` [Intel-gfx] [RFC 16/17] drm/i915: Wire up with drm controller GPU time query Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32 ` [Intel-gfx] [RFC 17/17] drm/i915: Implement cgroup controller over budget throttling Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 17:32   ` Tvrtko Ursulin
2022-10-19 18:45 ` [RFC 00/17] DRM scheduling cgroup controller Tejun Heo
2022-10-19 18:45   ` Tejun Heo
2022-10-19 18:45   ` [Intel-gfx] " Tejun Heo
2022-10-27 14:32   ` Tvrtko Ursulin
2022-10-27 14:32     ` Tvrtko Ursulin
2022-10-27 14:32     ` [Intel-gfx] " Tvrtko Ursulin
2022-10-31 20:20     ` Tejun Heo
2022-10-31 20:20       ` Tejun Heo
2022-10-31 20:20       ` [Intel-gfx] " Tejun Heo
2022-11-09 16:59       ` Tvrtko Ursulin
2022-11-09 16:59         ` Tvrtko Ursulin
2022-11-09 16:59         ` Tvrtko Ursulin
2022-10-19 19:25 ` [Intel-gfx] ✗ Fi.CI.BUILD: failure for " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20221019173254.3361334-10-tvrtko.ursulin@linux.intel.com \
    --to=tvrtko.ursulin@linux.intel.com \
    --cc=Intel-gfx@lists.freedesktop.org \
    --cc=Kenny.Ho@amd.com \
    --cc=airlied@redhat.com \
    --cc=brian.welty@intel.com \
    --cc=cgroups@vger.kernel.org \
    --cc=christian.koenig@amd.com \
    --cc=daniel.vetter@ffwll.ch \
    --cc=hannes@cmpxchg.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lizefan.x@bytedance.com \
    --cc=marcheu@chromium.org \
    --cc=robdclark@chromium.org \
    --cc=tj@kernel.org \
    --cc=tjmercier@google.com \
    --cc=tvrtko.ursulin@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.