linux-kernel - [PATCH v4 15/23] amdkfd: Add module parameter of scheduling policy

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1411591537-31636-16-git-send-email-oded.gabbay@amd.com>
Date:	Wed, 24 Sep 2014 23:45:29 +0300
From:	Oded Gabbay <oded.gabbay@....com>
To:	David Airlie <airlied@...ux.ie>,
	Alex Deucher <alexander.deucher@....com>,
	Jerome Glisse <j.glisse@...il.com>
CC:	<linux-kernel@...r.kernel.org>, <dri-devel@...ts.freedesktop.org>,
	"John Bridgman" <John.Bridgman@....com>,
	Andrew Lewycky <Andrew.Lewycky@....com>,
	Joerg Roedel <joro@...tes.org>, Ben Goz <ben.goz@....com>,
	Oded Gabbay <oded.gabbay@....com>,
	Christian König <christian.koenig@....com>
Subject: [PATCH v4 15/23] amdkfd: Add module parameter of scheduling policy

From: Ben Goz <ben.goz@....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

Signed-off-by: Ben Goz <ben.goz@....com>
Signed-off-by: Oded Gabbay <oded.gabbay@....com>
---
 drivers/gpu/drm/radeon/amdkfd/kfd_module.c | 12 ++++++++++++
 drivers/gpu/drm/radeon/amdkfd/kfd_priv.h   | 29 +++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+)

diff --git a/drivers/gpu/drm/radeon/amdkfd/kfd_module.c b/drivers/gpu/drm/radeon/amdkfd/kfd_module.c
index a31bf03..5c58031 100644
--- a/drivers/gpu/drm/radeon/amdkfd/kfd_module.c
+++ b/drivers/gpu/drm/radeon/amdkfd/kfd_module.c
@@ -45,6 +45,11 @@ static const struct kgd2kfd_calls kgd2kfd = {
 	.resume		= kgd2kfd_resume,
 };
 
+int sched_policy = KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION;
+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,
@@ -79,6 +84,13 @@ static int __init kfd_module_init(void)
 	int err;
 
 	/* 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/radeon/amdkfd/kfd_priv.h b/drivers/gpu/drm/radeon/amdkfd/kfd_priv.h
index 66980df..5835d07 100644
--- a/drivers/gpu/drm/radeon/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/radeon/amdkfd/kfd_priv.h
@@ -65,6 +65,35 @@ extern int max_num_of_queues_per_process;
 #define KFD_MAX_NUM_OF_QUEUES_PER_PROCESS_DEFAULT 128
 #define KFD_MAX_NUM_OF_QUEUES_PER_PROCESS 1024
 
+/* 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
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/