All of lore.kernel.org
 help / color / mirror / Atom feed
From: Oded Gabbay <oded.gabbay@amd.com>
To: David Airlie <airlied@linux.ie>,
	Alex Deucher <alexander.deucher@amd.com>,
	Jerome Glisse <j.glisse@gmail.com>
Cc: <linux-kernel@vger.kernel.org>, <dri-devel@lists.freedesktop.org>,
	"John Bridgman" <John.Bridgman@amd.com>,
	Ben Goz <ben.goz@amd.com>, Oded Gabbay <oded.gabbay@amd.com>,
	Evgeny Pinchuk <evgeny.pinchuk@amd.com>,
	Jay Cornwall <jay.cornwall@amd.com>,
	Alexey Skidanov <Alexey.Skidanov@amd.com>,
	"Andrew Lewycky" <Andrew.Lewycky@amd.com>
Subject: [PATCH v5 15/24] amdkfd: Add module parameter of scheduling policy
Date: Sat, 8 Nov 2014 20:37:36 +0200	[thread overview]
Message-ID: <1415471865-1893-16-git-send-email-oded.gabbay@amd.com> (raw)
In-Reply-To: <1415471865-1893-1-git-send-email-oded.gabbay@amd.com>

From: Ben Goz <ben.goz@amd.com>

This patch adds a new parameter to the amdkfd driver. This parameter enables
the user to select the scheduling policy of the CP. The choices are:

* CP Scheduling with support for over-subscription
* CP Scheduling without support for over-subscription
* Without CP Scheduling

Note that the third option (Without CP scheduling) is only for debug purposes
and bringup of new H/W. As such, it is _not_ guaranteed to work at all times on
all H/W versions.

v3: Fixed description of parameter, changed the permissions to read_only, added
a verification of the value and added documentation

v5: Set default sched_policy to HWS as it is now supported by firmware

Signed-off-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_module.c | 18 +++++++++++++++---
 drivers/gpu/drm/amd/amdkfd/kfd_priv.h   | 30 ++++++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_module.c b/drivers/gpu/drm/amd/amdkfd/kfd_module.c
index 56a71d8..0858dce 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_module.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_module.c
@@ -29,10 +29,10 @@
 #define KFD_DRIVER_AUTHOR	"AMD Inc. and others"
 
 #define KFD_DRIVER_DESC		"Standalone HSA driver for AMD's GPUs"
-#define KFD_DRIVER_DATE		"20140804"
+#define KFD_DRIVER_DATE		"20140807"
 #define KFD_DRIVER_MAJOR	0
-#define KFD_DRIVER_MINOR	6
-#define KFD_DRIVER_PATCHLEVEL	4
+#define KFD_DRIVER_MINOR	8
+#define KFD_DRIVER_PATCHLEVEL	0
 
 const struct kfd2kgd_calls *kfd2kgd;
 static const struct kgd2kfd_calls kgd2kfd = {
@@ -45,6 +45,11 @@ static const struct kgd2kfd_calls kgd2kfd = {
 	.resume		= kgd2kfd_resume,
 };
 
+int sched_policy = KFD_SCHED_POLICY_HWS;
+module_param(sched_policy, int, 0444);
+MODULE_PARM_DESC(sched_policy,
+	"Kernel cmdline parameter that defines the amdkfd scheduling policy");
+
 int max_num_of_processes = KFD_MAX_NUM_OF_PROCESSES_DEFAULT;
 module_param(max_num_of_processes, int, 0444);
 MODULE_PARM_DESC(max_num_of_processes,
@@ -88,6 +93,13 @@ static int __init kfd_module_init(void)
 	kfd2kgd = NULL;
 
 	/* Verify module parameters */
+	if ((sched_policy < KFD_SCHED_POLICY_HWS) ||
+		(sched_policy > KFD_SCHED_POLICY_NO_HWS)) {
+		pr_err("kfd: sched_policy has invalid value\n");
+		return -1;
+	}
+
+	/* Verify module parameters */
 	if ((max_num_of_processes < 0) ||
 		(max_num_of_processes > KFD_MAX_NUM_OF_PROCESSES)) {
 		pr_err("kfd: max_num_of_processes must be between 0 to KFD_MAX_NUM_OF_PROCESSES\n");
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index bba3c2a..32bca8a 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -69,6 +69,36 @@ extern int max_num_of_queues_per_process;
 
 #define KFD_KERNEL_QUEUE_SIZE 2048
 
+/* Kernel module parameter to specify the scheduling policy */
+extern int sched_policy;
+
+/**
+ * enum kfd_sched_policy
+ *
+ * @KFD_SCHED_POLICY_HWS: H/W scheduling policy known as command processor (cp)
+ * scheduling. In this scheduling mode we're using the firmware code to
+ * schedule the user mode queues and kernel queues such as HIQ and DIQ.
+ * the HIQ queue is used as a special queue that dispatches the configuration
+ * to the cp and the user mode queues list that are currently running.
+ * the DIQ queue is a debugging queue that dispatches debugging commands to the
+ * firmware.
+ * in this scheduling mode user mode queues over subscription feature is
+ * enabled.
+ *
+ * @KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION: The same as above but the over
+ * subscription feature disabled.
+ *
+ * @KFD_SCHED_POLICY_NO_HWS: no H/W scheduling policy is a mode which directly
+ * set the command processor registers and sets the queues "manually". This
+ * mode is used *ONLY* for debugging proposes.
+ *
+ */
+enum kfd_sched_policy {
+	KFD_SCHED_POLICY_HWS = 0,
+	KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION,
+	KFD_SCHED_POLICY_NO_HWS
+};
+
 enum cache_policy {
 	cache_policy_coherent,
 	cache_policy_noncoherent
-- 
2.1.0


WARNING: multiple messages have this Message-ID (diff)
From: Oded Gabbay <oded.gabbay@amd.com>
To: David Airlie <airlied@linux.ie>,
	Alex Deucher <alexander.deucher@amd.com>,
	Jerome Glisse <j.glisse@gmail.com>
Cc: Jay Cornwall <jay.cornwall@amd.com>,
	Andrew Lewycky <Andrew.Lewycky@amd.com>,
	linux-kernel@vger.kernel.org,
	Evgeny Pinchuk <evgeny.pinchuk@amd.com>,
	dri-devel@lists.freedesktop.org
Subject: [PATCH v5 15/24] amdkfd: Add module parameter of scheduling policy
Date: Sat, 8 Nov 2014 20:37:36 +0200	[thread overview]
Message-ID: <1415471865-1893-16-git-send-email-oded.gabbay@amd.com> (raw)
In-Reply-To: <1415471865-1893-1-git-send-email-oded.gabbay@amd.com>

From: Ben Goz <ben.goz@amd.com>

This patch adds a new parameter to the amdkfd driver. This parameter enables
the user to select the scheduling policy of the CP. The choices are:

* CP Scheduling with support for over-subscription
* CP Scheduling without support for over-subscription
* Without CP Scheduling

Note that the third option (Without CP scheduling) is only for debug purposes
and bringup of new H/W. As such, it is _not_ guaranteed to work at all times on
all H/W versions.

v3: Fixed description of parameter, changed the permissions to read_only, added
a verification of the value and added documentation

v5: Set default sched_policy to HWS as it is now supported by firmware

Signed-off-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_module.c | 18 +++++++++++++++---
 drivers/gpu/drm/amd/amdkfd/kfd_priv.h   | 30 ++++++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_module.c b/drivers/gpu/drm/amd/amdkfd/kfd_module.c
index 56a71d8..0858dce 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_module.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_module.c
@@ -29,10 +29,10 @@
 #define KFD_DRIVER_AUTHOR	"AMD Inc. and others"
 
 #define KFD_DRIVER_DESC		"Standalone HSA driver for AMD's GPUs"
-#define KFD_DRIVER_DATE		"20140804"
+#define KFD_DRIVER_DATE		"20140807"
 #define KFD_DRIVER_MAJOR	0
-#define KFD_DRIVER_MINOR	6
-#define KFD_DRIVER_PATCHLEVEL	4
+#define KFD_DRIVER_MINOR	8
+#define KFD_DRIVER_PATCHLEVEL	0
 
 const struct kfd2kgd_calls *kfd2kgd;
 static const struct kgd2kfd_calls kgd2kfd = {
@@ -45,6 +45,11 @@ static const struct kgd2kfd_calls kgd2kfd = {
 	.resume		= kgd2kfd_resume,
 };
 
+int sched_policy = KFD_SCHED_POLICY_HWS;
+module_param(sched_policy, int, 0444);
+MODULE_PARM_DESC(sched_policy,
+	"Kernel cmdline parameter that defines the amdkfd scheduling policy");
+
 int max_num_of_processes = KFD_MAX_NUM_OF_PROCESSES_DEFAULT;
 module_param(max_num_of_processes, int, 0444);
 MODULE_PARM_DESC(max_num_of_processes,
@@ -88,6 +93,13 @@ static int __init kfd_module_init(void)
 	kfd2kgd = NULL;
 
 	/* Verify module parameters */
+	if ((sched_policy < KFD_SCHED_POLICY_HWS) ||
+		(sched_policy > KFD_SCHED_POLICY_NO_HWS)) {
+		pr_err("kfd: sched_policy has invalid value\n");
+		return -1;
+	}
+
+	/* Verify module parameters */
 	if ((max_num_of_processes < 0) ||
 		(max_num_of_processes > KFD_MAX_NUM_OF_PROCESSES)) {
 		pr_err("kfd: max_num_of_processes must be between 0 to KFD_MAX_NUM_OF_PROCESSES\n");
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index bba3c2a..32bca8a 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -69,6 +69,36 @@ extern int max_num_of_queues_per_process;
 
 #define KFD_KERNEL_QUEUE_SIZE 2048
 
+/* Kernel module parameter to specify the scheduling policy */
+extern int sched_policy;
+
+/**
+ * enum kfd_sched_policy
+ *
+ * @KFD_SCHED_POLICY_HWS: H/W scheduling policy known as command processor (cp)
+ * scheduling. In this scheduling mode we're using the firmware code to
+ * schedule the user mode queues and kernel queues such as HIQ and DIQ.
+ * the HIQ queue is used as a special queue that dispatches the configuration
+ * to the cp and the user mode queues list that are currently running.
+ * the DIQ queue is a debugging queue that dispatches debugging commands to the
+ * firmware.
+ * in this scheduling mode user mode queues over subscription feature is
+ * enabled.
+ *
+ * @KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION: The same as above but the over
+ * subscription feature disabled.
+ *
+ * @KFD_SCHED_POLICY_NO_HWS: no H/W scheduling policy is a mode which directly
+ * set the command processor registers and sets the queues "manually". This
+ * mode is used *ONLY* for debugging proposes.
+ *
+ */
+enum kfd_sched_policy {
+	KFD_SCHED_POLICY_HWS = 0,
+	KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION,
+	KFD_SCHED_POLICY_NO_HWS
+};
+
 enum cache_policy {
 	cache_policy_coherent,
 	cache_policy_noncoherent
-- 
2.1.0

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

  parent reply	other threads:[~2014-11-08 18:43 UTC|newest]

Thread overview: 70+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-11-08 18:37 [PATCH v5 00/24] AMDKFD Kernel Driver Oded Gabbay
2014-11-08 18:37 ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 01/24] drm/radeon: reduce number of free VMIDs and pipes in KV Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 02/24] drm/radeon/cik: Don't touch int of pipes 1-7 Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 03/24] drm/radeon: Report doorbell configuration to amdkfd Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 04/24] drm/radeon: adding synchronization for GRBM GFX Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 05/24] drm/radeon: Add radeon <--> amdkfd interface Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 06/24] Update MAINTAINERS and CREDITS files with amdkfd info Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:46   ` Joe Perches
2014-11-08 18:46     ` Joe Perches
2014-11-08 18:59     ` Oded Gabbay
2014-11-08 18:59       ` Oded Gabbay
2014-11-08 20:54       ` Joe Perches
2014-11-08 20:54         ` Joe Perches
2014-11-08 19:01   ` Josh Triplett
2014-11-08 20:49     ` Joe Perches
2014-11-08 20:49       ` Joe Perches
2014-11-09  3:50       ` Josh Triplett
2014-11-08 18:37 ` [PATCH v5 07/24] amdkfd: Add IOCTL set definitions of amdkfd Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 08/24] amdkfd: Add amdkfd skeleton driver Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-21 10:24   ` Paul Bolle
2014-11-21 19:34     ` Oded Gabbay
2014-11-21 19:34       ` Oded Gabbay
2014-11-21 20:18       ` Borislav Petkov
2014-11-21 20:18         ` Borislav Petkov
2014-11-08 18:37 ` [PATCH v5 09/24] amdkfd: Add topology module to amdkfd Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 10/24] amdkfd: Add basic modules " Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 11/24] amdkfd: Add binding/unbinding calls to amd_iommu driver Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 12/24] amdkfd: Add queue module Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 13/24] amdkfd: Add mqd_manager module Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 14/24] amdkfd: Add kernel queue module Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` Oded Gabbay [this message]
2014-11-08 18:37   ` [PATCH v5 15/24] amdkfd: Add module parameter of scheduling policy Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 16/24] amdkfd: Add packet manager module Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 17/24] amdkfd: Add process queue " Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 18/24] amdkfd: Add device " Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 19/24] amdkfd: Add interrupt handling module Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:37 ` [PATCH v5 20/24] amdkfd: Implement the create/destroy/update queue IOCTLs Oded Gabbay
2014-11-08 18:37   ` Oded Gabbay
2014-11-08 18:50 ` [PATCH v5 21/24] amdkfd: Implement the Set Memory Policy IOCTL Oded Gabbay
2014-11-08 18:50   ` Oded Gabbay
2014-11-08 18:54 ` [PATCH v5 22/24] amdkfd: Implement the Get Clock Counters IOCTL Oded Gabbay
2014-11-08 18:54   ` Oded Gabbay
2014-11-08 18:54 ` [PATCH v5 23/24] amdkfd: Implement the Get Process Aperture IOCTL Oded Gabbay
2014-11-08 18:54   ` Oded Gabbay
2014-11-08 18:55 ` [PATCH v5 24/24] amdkfd: Implement the Get Version IOCTL Oded Gabbay
2014-11-08 18:55   ` Oded Gabbay
2014-11-10  2:34 ` [PATCH v5 00/24] AMDKFD Kernel Driver Dave Airlie
2014-11-10  2:34   ` Dave Airlie
2014-11-10  7:36   ` Oded Gabbay
2014-11-10  7:36     ` Oded Gabbay
2014-11-10 21:16     ` Dave Airlie
2014-11-10 21:16       ` Dave Airlie
2014-11-10 21:20       ` Alex Deucher
2014-11-10 21:20         ` Alex Deucher

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=1415471865-1893-16-git-send-email-oded.gabbay@amd.com \
    --to=oded.gabbay@amd.com \
    --cc=Alexey.Skidanov@amd.com \
    --cc=Andrew.Lewycky@amd.com \
    --cc=John.Bridgman@amd.com \
    --cc=airlied@linux.ie \
    --cc=alexander.deucher@amd.com \
    --cc=ben.goz@amd.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=evgeny.pinchuk@amd.com \
    --cc=j.glisse@gmail.com \
    --cc=jay.cornwall@amd.com \
    --cc=linux-kernel@vger.kernel.org \
    /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.