All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 1/4] habanalabs: separate common code to dedicated folders
@ 2021-01-03 20:29 Oded Gabbay
  2021-01-03 20:29 ` [PATCH 2/4] habanalabs: increment ctx ref from within a cs allocation Oded Gabbay
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: Oded Gabbay @ 2021-01-03 20:29 UTC (permalink / raw)
  To: linux-kernel; +Cc: Ofir Bitton

From: Ofir Bitton <obitton@habana.ai>

We separate some of the common code source files to different
folders for a better maintainability and testability.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/common/Makefile           | 10 ++++++++--
 drivers/misc/habanalabs/common/mmu/Makefile       |  2 ++
 drivers/misc/habanalabs/common/{ => mmu}/mmu.c    |  2 +-
 drivers/misc/habanalabs/common/{ => mmu}/mmu_v1.c |  4 ++--
 drivers/misc/habanalabs/common/pci/Makefile       |  2 ++
 drivers/misc/habanalabs/common/{ => pci}/pci.c    |  4 ++--
 6 files changed, 17 insertions(+), 7 deletions(-)
 create mode 100644 drivers/misc/habanalabs/common/mmu/Makefile
 rename drivers/misc/habanalabs/common/{ => mmu}/mmu.c (99%)
 rename drivers/misc/habanalabs/common/{ => mmu}/mmu_v1.c (99%)
 create mode 100644 drivers/misc/habanalabs/common/pci/Makefile
 rename drivers/misc/habanalabs/common/{ => pci}/pci.c (99%)

diff --git a/drivers/misc/habanalabs/common/Makefile b/drivers/misc/habanalabs/common/Makefile
index eccd8c7dc62d..5d8b48288cf4 100644
--- a/drivers/misc/habanalabs/common/Makefile
+++ b/drivers/misc/habanalabs/common/Makefile
@@ -1,7 +1,13 @@
 # SPDX-License-Identifier: GPL-2.0-only
+
+include $(src)/common/mmu/Makefile
+habanalabs-y += $(HL_COMMON_MMU_FILES)
+
+include $(src)/common/pci/Makefile
+habanalabs-y += $(HL_COMMON_PCI_FILES)
+
 HL_COMMON_FILES := common/habanalabs_drv.o common/device.o common/context.o \
 		common/asid.o common/habanalabs_ioctl.o \
 		common/command_buffer.o common/hw_queue.o common/irq.o \
 		common/sysfs.o common/hwmon.o common/memory.o \
-		common/command_submission.o common/mmu.o common/mmu_v1.o \
-		common/firmware_if.o common/pci.o
+		common/command_submission.o common/firmware_if.o
diff --git a/drivers/misc/habanalabs/common/mmu/Makefile b/drivers/misc/habanalabs/common/mmu/Makefile
new file mode 100644
index 000000000000..d852c3874658
--- /dev/null
+++ b/drivers/misc/habanalabs/common/mmu/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+HL_COMMON_MMU_FILES := common/mmu/mmu.o common/mmu/mmu_v1.o
diff --git a/drivers/misc/habanalabs/common/mmu.c b/drivers/misc/habanalabs/common/mmu/mmu.c
similarity index 99%
rename from drivers/misc/habanalabs/common/mmu.c
rename to drivers/misc/habanalabs/common/mmu/mmu.c
index ec11111aee8a..105d556e3028 100644
--- a/drivers/misc/habanalabs/common/mmu.c
+++ b/drivers/misc/habanalabs/common/mmu/mmu.c
@@ -7,7 +7,7 @@
 
 #include <linux/slab.h>
 
-#include "habanalabs.h"
+#include "../habanalabs.h"
 
 static bool is_dram_va(struct hl_device *hdev, u64 virt_addr)
 {
diff --git a/drivers/misc/habanalabs/common/mmu_v1.c b/drivers/misc/habanalabs/common/mmu/mmu_v1.c
similarity index 99%
rename from drivers/misc/habanalabs/common/mmu_v1.c
rename to drivers/misc/habanalabs/common/mmu/mmu_v1.c
index 2ce6ea89d4fa..9d37ac9bf316 100644
--- a/drivers/misc/habanalabs/common/mmu_v1.c
+++ b/drivers/misc/habanalabs/common/mmu/mmu_v1.c
@@ -5,8 +5,8 @@
  * All Rights Reserved.
  */
 
-#include "habanalabs.h"
-#include "../include/hw_ip/mmu/mmu_general.h"
+#include "../habanalabs.h"
+#include "../../include/hw_ip/mmu/mmu_general.h"
 
 #include <linux/slab.h>
 
diff --git a/drivers/misc/habanalabs/common/pci/Makefile b/drivers/misc/habanalabs/common/pci/Makefile
new file mode 100644
index 000000000000..dc922a686683
--- /dev/null
+++ b/drivers/misc/habanalabs/common/pci/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+HL_COMMON_PCI_FILES := common/pci/pci.o
diff --git a/drivers/misc/habanalabs/common/pci.c b/drivers/misc/habanalabs/common/pci/pci.c
similarity index 99%
rename from drivers/misc/habanalabs/common/pci.c
rename to drivers/misc/habanalabs/common/pci/pci.c
index b4725e6101f6..c56ec1574127 100644
--- a/drivers/misc/habanalabs/common/pci.c
+++ b/drivers/misc/habanalabs/common/pci/pci.c
@@ -5,8 +5,8 @@
  * All Rights Reserved.
  */
 
-#include "habanalabs.h"
-#include "../include/hw_ip/pci/pci_general.h"
+#include "../habanalabs.h"
+#include "../../include/hw_ip/pci/pci_general.h"
 
 #include <linux/pci.h>
 
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread
* [PATCH 2/4] habanalabs: increment ctx ref from within a cs allocation
  2021-01-03 20:29 [PATCH 1/4] habanalabs: separate common code to dedicated folders Oded Gabbay
@ 2021-01-03 20:29 ` Oded Gabbay
  2021-01-03 20:29 ` [PATCH 3/4] habanalabs: add driver support for internal cb scheduling Oded Gabbay
  2021-01-03 20:29 ` [PATCH 4/4] habanalabs/gaudi: remove PCI access to SM block Oded Gabbay
  2 siblings, 0 replies; 4+ messages in thread
From: Oded Gabbay @ 2021-01-03 20:29 UTC (permalink / raw)
  To: linux-kernel; +Cc: Ofir Bitton

From: Ofir Bitton <obitton@habana.ai>

A CS must increment the relevant context reference count.
We want to increment the reference inside the CS allocation function
as opposed for today where we increment it outside.
This is logical since we want to avoid explicitly incrementing
the context every time we call the CS allocate function.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 .../misc/habanalabs/common/command_submission.c   | 15 +++++----------
 1 file changed, 5 insertions(+), 10 deletions(-)

diff --git a/drivers/misc/habanalabs/common/command_submission.c b/drivers/misc/habanalabs/common/command_submission.c
index f7fac82ac41d..3affb350070c 100644
--- a/drivers/misc/habanalabs/common/command_submission.c
+++ b/drivers/misc/habanalabs/common/command_submission.c
@@ -479,6 +479,9 @@ static int allocate_cs(struct hl_device *hdev, struct hl_ctx *ctx,
 		return -ENOMEM;
 	}
 
+	/* increment refcnt for context */
+	hl_ctx_get(hdev, ctx);
+
 	cs->ctx = ctx;
 	cs->submitted = false;
 	cs->completed = false;
@@ -550,6 +553,7 @@ static int allocate_cs(struct hl_device *hdev, struct hl_ctx *ctx,
 	kfree(cs_cmpl);
 free_cs:
 	kfree(cs);
+	hl_ctx_put(ctx);
 	return rc;
 }
 
@@ -827,14 +831,9 @@ static int cs_ioctl_default(struct hl_fpriv *hpriv, void __user *chunks,
 	if (rc)
 		goto out;
 
-	/* increment refcnt for context */
-	hl_ctx_get(hdev, hpriv->ctx);
-
 	rc = allocate_cs(hdev, hpriv->ctx, CS_TYPE_DEFAULT, &cs);
-	if (rc) {
-		hl_ctx_put(hpriv->ctx);
+	if (rc)
 		goto free_cs_chunk_array;
-	}
 
 	cs->timestamp = !!(flags & HL_CS_FLAGS_TIMESTAMP);
 	*cs_seq = cs->sequence;
@@ -1276,15 +1275,11 @@ static int cs_ioctl_signal_wait(struct hl_fpriv *hpriv, enum hl_cs_type cs_type,
 		}
 	}
 
-	/* increment refcnt for context */
-	hl_ctx_get(hdev, ctx);
-
 	rc = allocate_cs(hdev, ctx, cs_type, &cs);
 	if (rc) {
 		if (cs_type == CS_TYPE_WAIT ||
 			cs_type == CS_TYPE_COLLECTIVE_WAIT)
 			hl_fence_put(sig_fence);
-		hl_ctx_put(ctx);
 		goto free_cs_chunk_array;
 	}
 
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread
* [PATCH 3/4] habanalabs: add driver support for internal cb scheduling
  2021-01-03 20:29 [PATCH 1/4] habanalabs: separate common code to dedicated folders Oded Gabbay
  2021-01-03 20:29 ` [PATCH 2/4] habanalabs: increment ctx ref from within a cs allocation Oded Gabbay
@ 2021-01-03 20:29 ` Oded Gabbay
  2021-01-03 20:29 ` [PATCH 4/4] habanalabs/gaudi: remove PCI access to SM block Oded Gabbay
  2 siblings, 0 replies; 4+ messages in thread
From: Oded Gabbay @ 2021-01-03 20:29 UTC (permalink / raw)
  To: linux-kernel; +Cc: Ofir Bitton

From: Ofir Bitton <obitton@habana.ai>

In order to support scnenarios in which driver needs access to
HW components but it cannot access them directly, we add support for
scheduling command buffers internally.
These command buffers will be transmitted upon next user command
submission context.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 .../habanalabs/common/command_submission.c    | 141 ++++++++++++++++++
 drivers/misc/habanalabs/common/context.c      |   3 +
 drivers/misc/habanalabs/common/device.c       |   2 +
 drivers/misc/habanalabs/common/habanalabs.h   |  26 ++++
 4 files changed, 172 insertions(+)

diff --git a/drivers/misc/habanalabs/common/command_submission.c b/drivers/misc/habanalabs/common/command_submission.c
index 3affb350070c..2dd69d6d4782 100644
--- a/drivers/misc/habanalabs/common/command_submission.c
+++ b/drivers/misc/habanalabs/common/command_submission.c
@@ -585,6 +585,18 @@ void hl_cs_rollback_all(struct hl_device *hdev)
 	}
 }
 
+void hl_pending_cb_rollback_all(struct hl_device *hdev)
+{
+	struct hl_pending_cb *pending_cb, *tmp;
+
+	list_for_each_entry_safe(pending_cb, tmp,
+			&hdev->kernel_ctx->pending_cb_list, cb_node) {
+		list_del(&pending_cb->cb_node);
+		hl_cb_put(pending_cb->cb);
+		kfree(pending_cb);
+	}
+}
+
 static void job_wq_completion(struct work_struct *work)
 {
 	struct hl_cs_job *job = container_of(work, struct hl_cs_job,
@@ -954,6 +966,131 @@ static int cs_ioctl_default(struct hl_fpriv *hpriv, void __user *chunks,
 	return rc;
 }
 
+static int pending_cb_create_job(struct hl_device *hdev, struct hl_ctx *ctx,
+		struct hl_cs *cs, struct hl_cb *cb, u32 size, u32 hw_queue_id)
+{
+	struct hw_queue_properties *hw_queue_prop;
+	struct hl_cs_counters_atomic *cntr;
+	struct hl_cs_job *job;
+
+	hw_queue_prop = &hdev->asic_prop.hw_queues_props[hw_queue_id];
+	cntr = &hdev->aggregated_cs_counters;
+
+	job = hl_cs_allocate_job(hdev, hw_queue_prop->type, true);
+	if (!job) {
+		atomic64_inc(&ctx->cs_counters.out_of_mem_drop_cnt);
+		atomic64_inc(&cntr->out_of_mem_drop_cnt);
+		dev_err(hdev->dev, "Failed to allocate a new job\n");
+		return -ENOMEM;
+	}
+
+	job->id = 0;
+	job->cs = cs;
+	job->user_cb = cb;
+	atomic_inc(&job->user_cb->cs_cnt);
+	job->user_cb_size = size;
+	job->hw_queue_id = hw_queue_id;
+	job->patched_cb = job->user_cb;
+	job->job_cb_size = job->user_cb_size;
+
+	/* increment refcount as for external queues we get completion */
+	cs_get(cs);
+
+	cs->jobs_in_queue_cnt[job->hw_queue_id]++;
+
+	list_add_tail(&job->cs_node, &cs->job_list);
+
+	hl_debugfs_add_job(hdev, job);
+
+	return 0;
+}
+
+static int hl_submit_pending_cb(struct hl_fpriv *hpriv)
+{
+	struct hw_queue_properties *hw_queue_prop;
+	struct hl_device *hdev = hpriv->hdev;
+	struct hl_ctx *ctx = hpriv->ctx;
+	struct hl_pending_cb *pending_cb, *tmp;
+	struct list_head local_cb_list;
+	struct hl_cs *cs;
+	struct hl_cb *cb;
+	u32 hw_queue_id;
+	u32 cb_size;
+	int process_list, rc = 0;
+
+	if (list_empty(&ctx->pending_cb_list))
+		return 0;
+
+	process_list = atomic_cmpxchg(&ctx->thread_pending_cb_token, 1, 0);
+
+	/* Only a single thread is allowed to process the list */
+	if (!process_list)
+		return 0;
+
+	if (list_empty(&ctx->pending_cb_list))
+		goto free_pending_cb_token;
+
+	/* move all list elements to a local list */
+	INIT_LIST_HEAD(&local_cb_list);
+	spin_lock(&ctx->pending_cb_lock);
+	list_for_each_entry_safe(pending_cb, tmp, &ctx->pending_cb_list,
+								cb_node)
+		list_move_tail(&pending_cb->cb_node, &local_cb_list);
+	spin_unlock(&ctx->pending_cb_lock);
+
+	rc = allocate_cs(hdev, ctx, CS_TYPE_DEFAULT, &cs);
+	if (rc)
+		goto add_list_elements;
+
+	hl_debugfs_add_cs(cs);
+
+	/* Iterate through pending cb list, create jobs and add to CS */
+	list_for_each_entry(pending_cb, &local_cb_list, cb_node) {
+		cb = pending_cb->cb;
+		cb_size = pending_cb->cb_size;
+		hw_queue_id = pending_cb->hw_queue_id;
+		hw_queue_prop = &hdev->asic_prop.hw_queues_props[hw_queue_id];
+
+		rc = pending_cb_create_job(hdev, ctx, cs, cb, cb_size,
+								hw_queue_id);
+		if (rc)
+			goto free_cs_object;
+	}
+
+	rc = hl_hw_queue_schedule_cs(cs);
+	if (rc) {
+		if (rc != -EAGAIN)
+			dev_err(hdev->dev,
+				"Failed to submit CS %d.%llu (%d)\n",
+				ctx->asid, cs->sequence, rc);
+		goto free_cs_object;
+	}
+
+	/* pending cb was scheduled successfully */
+	list_for_each_entry_safe(pending_cb, tmp, &local_cb_list, cb_node) {
+		list_del(&pending_cb->cb_node);
+		kfree(pending_cb);
+	}
+
+	cs_put(cs);
+
+	goto free_pending_cb_token;
+
+free_cs_object:
+	cs_rollback(hdev, cs);
+	cs_put(cs);
+add_list_elements:
+	spin_lock(&ctx->pending_cb_lock);
+	list_for_each_entry_safe_reverse(pending_cb, tmp, &local_cb_list,
+								cb_node)
+		list_move(&pending_cb->cb_node, &ctx->pending_cb_list);
+	spin_unlock(&ctx->pending_cb_lock);
+free_pending_cb_token:
+	atomic_set(&ctx->thread_pending_cb_token, 1);
+
+	return rc;
+}
+
 static int hl_cs_ctx_switch(struct hl_fpriv *hpriv, union hl_cs_args *args,
 				u64 *cs_seq)
 {
@@ -1353,6 +1490,10 @@ int hl_cs_ioctl(struct hl_fpriv *hpriv, void *data)
 	if (rc)
 		goto out;
 
+	rc = hl_submit_pending_cb(hpriv);
+	if (rc)
+		goto out;
+
 	cs_type = hl_cs_get_cs_type(args->in.cs_flags &
 					~HL_CS_FLAGS_FORCE_RESTORE);
 	chunks = (void __user *) (uintptr_t) args->in.chunks_execute;
diff --git a/drivers/misc/habanalabs/common/context.c b/drivers/misc/habanalabs/common/context.c
index 3d86b83f4ca6..a5003461b577 100644
--- a/drivers/misc/habanalabs/common/context.c
+++ b/drivers/misc/habanalabs/common/context.c
@@ -142,8 +142,11 @@ int hl_ctx_init(struct hl_device *hdev, struct hl_ctx *ctx, bool is_kernel_ctx)
 	kref_init(&ctx->refcount);
 
 	ctx->cs_sequence = 1;
+	INIT_LIST_HEAD(&ctx->pending_cb_list);
+	spin_lock_init(&ctx->pending_cb_lock);
 	spin_lock_init(&ctx->cs_lock);
 	atomic_set(&ctx->thread_ctx_switch_token, 1);
+	atomic_set(&ctx->thread_pending_cb_token, 1);
 	ctx->thread_ctx_switch_wait_token = 0;
 	ctx->cs_pending = kcalloc(hdev->asic_prop.max_pending_cs,
 				sizeof(struct hl_fence *),
diff --git a/drivers/misc/habanalabs/common/device.c b/drivers/misc/habanalabs/common/device.c
index 9ef77a2a34ec..66ca8bd8fc70 100644
--- a/drivers/misc/habanalabs/common/device.c
+++ b/drivers/misc/habanalabs/common/device.c
@@ -998,6 +998,7 @@ int hl_device_reset(struct hl_device *hdev, bool hard_reset,
 
 	/* Go over all the queues, release all CS and their jobs */
 	hl_cs_rollback_all(hdev);
+	hl_pending_cb_rollback_all(hdev);
 
 kill_processes:
 	if (hard_reset) {
@@ -1519,6 +1520,7 @@ void hl_device_fini(struct hl_device *hdev)
 
 	/* Go over all the queues, release all CS and their jobs */
 	hl_cs_rollback_all(hdev);
+	hl_pending_cb_rollback_all(hdev);
 
 	/* Kill processes here after CS rollback. This is because the process
 	 * can't really exit until all its CSs are done, which is what we
diff --git a/drivers/misc/habanalabs/common/habanalabs.h b/drivers/misc/habanalabs/common/habanalabs.h
index 537545df531a..c4ca55467bdd 100644
--- a/drivers/misc/habanalabs/common/habanalabs.h
+++ b/drivers/misc/habanalabs/common/habanalabs.h
@@ -1016,6 +1016,20 @@ struct hl_cs_counters_atomic {
 	atomic64_t validation_drop_cnt;
 };
 
+/**
+ * struct hl_pending_cb - pending command buffer structure
+ * @cb_node: cb node in pending cb list
+ * @cb: command buffer to send in next submission
+ * @cb_size: command buffer size
+ * @hw_queue_id: destination queue id
+ */
+struct hl_pending_cb {
+	struct list_head	cb_node;
+	struct hl_cb		*cb;
+	u32			cb_size;
+	u32			hw_queue_id;
+};
+
 /**
  * struct hl_ctx - user/kernel context.
  * @mem_hash: holds mapping from virtual address to virtual memory area
@@ -1031,6 +1045,8 @@ struct hl_cs_counters_atomic {
  * @mmu_lock: protects the MMU page tables. Any change to the PGT, modifying the
  *            MMU hash or walking the PGT requires talking this lock.
  * @debugfs_list: node in debugfs list of contexts.
+ * pending_cb_list: list of pending command buffers waiting to be sent upon
+ *                  next user command submission context.
  * @cs_counters: context command submission counters.
  * @cb_va_pool: device VA pool for command buffers which are mapped to the
  *              device's MMU.
@@ -1039,11 +1055,17 @@ struct hl_cs_counters_atomic {
  *			index to cs_pending array.
  * @dram_default_hops: array that holds all hops addresses needed for default
  *                     DRAM mapping.
+ * @pending_cb_lock: spinlock to protect pending cb list
  * @cs_lock: spinlock to protect cs_sequence.
  * @dram_phys_mem: amount of used physical DRAM memory by this context.
  * @thread_ctx_switch_token: token to prevent multiple threads of the same
  *				context	from running the context switch phase.
  *				Only a single thread should run it.
+ * @thread_pending_cb_token: token to prevent multiple threads from processing
+ *				the pending CB list. Only a single thread should
+ *				process the list since it is protected by a
+ *				spinlock and we don't want to halt the entire
+ *				command submission sequence.
  * @thread_ctx_switch_wait_token: token to prevent the threads that didn't run
  *				the context switch phase from moving to their
  *				execution phase before the context switch phase
@@ -1062,13 +1084,16 @@ struct hl_ctx {
 	struct mutex			mem_hash_lock;
 	struct mutex			mmu_lock;
 	struct list_head		debugfs_list;
+	struct list_head		pending_cb_list;
 	struct hl_cs_counters_atomic	cs_counters;
 	struct gen_pool			*cb_va_pool;
 	u64				cs_sequence;
 	u64				*dram_default_hops;
+	spinlock_t			pending_cb_lock;
 	spinlock_t			cs_lock;
 	atomic64_t			dram_phys_mem;
 	atomic_t			thread_ctx_switch_token;
+	atomic_t			thread_pending_cb_token;
 	u32				thread_ctx_switch_wait_token;
 	u32				asid;
 	u32				handle;
@@ -2143,6 +2168,7 @@ int hl_cb_va_pool_init(struct hl_ctx *ctx);
 void hl_cb_va_pool_fini(struct hl_ctx *ctx);
 
 void hl_cs_rollback_all(struct hl_device *hdev);
+void hl_pending_cb_rollback_all(struct hl_device *hdev);
 struct hl_cs_job *hl_cs_allocate_job(struct hl_device *hdev,
 		enum hl_queue_type queue_type, bool is_kernel_allocated_cb);
 void hl_sob_reset_error(struct kref *ref);
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread
* [PATCH 4/4] habanalabs/gaudi: remove PCI access to SM block
  2021-01-03 20:29 [PATCH 1/4] habanalabs: separate common code to dedicated folders Oded Gabbay
  2021-01-03 20:29 ` [PATCH 2/4] habanalabs: increment ctx ref from within a cs allocation Oded Gabbay
  2021-01-03 20:29 ` [PATCH 3/4] habanalabs: add driver support for internal cb scheduling Oded Gabbay
@ 2021-01-03 20:29 ` Oded Gabbay
  2 siblings, 0 replies; 4+ messages in thread
From: Oded Gabbay @ 2021-01-03 20:29 UTC (permalink / raw)
  To: linux-kernel; +Cc: Ofir Bitton

From: Ofir Bitton <obitton@habana.ai>

due to HW limitation we must remove all direct access to SM
registers, in order to do that we will access SM registers using
the HW QMANS.
When possible and no user context is present, we can directly access
the HW QMANS, whenever there is an active user, driver will
prepare a pending command buffer list which will be sent upon
user submissions.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/gaudi/gaudi.c         | 246 ++++++++++++++++--
 drivers/misc/habanalabs/gaudi/gaudiP.h        |   2 +
 .../habanalabs/include/gaudi/gaudi_packets.h  |   3 +
 3 files changed, 225 insertions(+), 26 deletions(-)

diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c b/drivers/misc/habanalabs/gaudi/gaudi.c
index f9a99872a56e..611a691206f1 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -354,6 +354,10 @@ static int gaudi_send_job_on_qman0(struct hl_device *hdev,
 					struct hl_cs_job *job);
 static int gaudi_memset_device_memory(struct hl_device *hdev, u64 addr,
 					u32 size, u64 val);
+static int gaudi_memset_registers(struct hl_device *hdev, u64 reg_base,
+					u32 num_regs, u32 val);
+static int gaudi_schedule_register_memset(struct hl_device *hdev,
+		u32 hw_queue_id, u64 reg_base, u32 num_regs, u32 val);
 static int gaudi_run_tpc_kernel(struct hl_device *hdev, u64 tpc_kernel,
 				u32 tpc_id);
 static int gaudi_mmu_clear_pgt_range(struct hl_device *hdev);
@@ -913,11 +917,17 @@ static void gaudi_sob_group_hw_reset(struct kref *ref)
 	struct gaudi_hw_sob_group *hw_sob_group =
 		container_of(ref, struct gaudi_hw_sob_group, kref);
 	struct hl_device *hdev = hw_sob_group->hdev;
-	int i;
+	u64 base_addr;
+	int rc;
 
-	for (i = 0 ; i < NUMBER_OF_SOBS_IN_GRP ; i++)
-		WREG32(mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_SOB_OBJ_0 +
-				(hw_sob_group->base_sob_id + i) * 4, 0);
+	base_addr = CFG_BASE + mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_SOB_OBJ_0 +
+			hw_sob_group->base_sob_id * 4;
+	rc = gaudi_schedule_register_memset(hdev, hw_sob_group->queue_id,
+			base_addr, NUMBER_OF_SOBS_IN_GRP, 0);
+	if (rc)
+		dev_err(hdev->dev,
+			"failed resetting sob group - sob base %u, count %u",
+			hw_sob_group->base_sob_id, NUMBER_OF_SOBS_IN_GRP);
 
 	kref_init(&hw_sob_group->kref);
 }
@@ -1008,6 +1018,8 @@ static void gaudi_collective_master_init_job(struct hl_device *hdev,
 		cprop->hw_sob_group[sob_group_offset].base_sob_id;
 	master_monitor = prop->collective_mstr_mon_id[0];
 
+	cprop->hw_sob_group[sob_group_offset].queue_id = queue_id;
+
 	dev_dbg(hdev->dev,
 		"Generate master wait CBs, sob %d (mask %#x), val:0x%x, mon %u, q %d\n",
 		master_sob_base, cprop->mstr_sob_mask[0],
@@ -5582,31 +5594,204 @@ static int gaudi_memset_device_memory(struct hl_device *hdev, u64 addr,
 	return rc;
 }
 
-static void gaudi_restore_sm_registers(struct hl_device *hdev)
+static int gaudi_memset_registers(struct hl_device *hdev, u64 reg_base,
+					u32 num_regs, u32 val)
+{
+	struct packet_msg_long *pkt;
+	struct hl_cs_job *job;
+	u32 cb_size, ctl;
+	struct hl_cb *cb;
+	int i, rc;
+
+	cb_size = (sizeof(*pkt) * num_regs) + sizeof(struct packet_msg_prot);
+
+	if (cb_size > SZ_2M) {
+		dev_err(hdev->dev, "CB size must be smaller than %uMB", SZ_2M);
+		return -ENOMEM;
+	}
+
+	cb = hl_cb_kernel_create(hdev, cb_size, false);
+	if (!cb)
+		return -EFAULT;
+
+	pkt = cb->kernel_address;
+
+	ctl = FIELD_PREP(GAUDI_PKT_LONG_CTL_OP_MASK, 0); /* write the value */
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_OPCODE_MASK, PACKET_MSG_LONG);
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_EB_MASK, 1);
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_RB_MASK, 1);
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_MB_MASK, 1);
+
+	for (i = 0; i < num_regs ; i++, pkt++) {
+		pkt->ctl = cpu_to_le32(ctl);
+		pkt->value = cpu_to_le32(val);
+		pkt->addr = cpu_to_le64(reg_base + (i * 4));
+	}
+
+	job = hl_cs_allocate_job(hdev, QUEUE_TYPE_EXT, true);
+	if (!job) {
+		dev_err(hdev->dev, "Failed to allocate a new job\n");
+		rc = -ENOMEM;
+		goto release_cb;
+	}
+
+	job->id = 0;
+	job->user_cb = cb;
+	atomic_inc(&job->user_cb->cs_cnt);
+	job->user_cb_size = cb_size;
+	job->hw_queue_id = GAUDI_QUEUE_ID_DMA_0_0;
+	job->patched_cb = job->user_cb;
+	job->job_cb_size = cb_size;
+
+	hl_debugfs_add_job(hdev, job);
+
+	rc = gaudi_send_job_on_qman0(hdev, job);
+	hl_debugfs_remove_job(hdev, job);
+	kfree(job);
+	atomic_dec(&cb->cs_cnt);
+
+release_cb:
+	hl_cb_put(cb);
+	hl_cb_destroy(hdev, &hdev->kernel_cb_mgr, cb->id << PAGE_SHIFT);
+
+	return rc;
+}
+
+static int gaudi_schedule_register_memset(struct hl_device *hdev,
+		u32 hw_queue_id, u64 reg_base, u32 num_regs, u32 val)
 {
+	struct hl_ctx *ctx = hdev->compute_ctx;
+	struct hl_pending_cb *pending_cb;
+	struct packet_msg_long *pkt;
+	u32 cb_size, ctl;
+	struct hl_cb *cb;
 	int i;
 
-	for (i = 0 ; i < NUM_OF_SOB_IN_BLOCK << 2 ; i += 4) {
-		WREG32(mmSYNC_MNGR_E_N_SYNC_MNGR_OBJS_SOB_OBJ_0 + i, 0);
-		WREG32(mmSYNC_MNGR_E_S_SYNC_MNGR_OBJS_SOB_OBJ_0 + i, 0);
-		WREG32(mmSYNC_MNGR_W_N_SYNC_MNGR_OBJS_SOB_OBJ_0 + i, 0);
+	/* If no compute context available memset registers directly */
+	if (!ctx)
+		return gaudi_memset_registers(hdev, reg_base, num_regs, val);
+
+	cb_size = (sizeof(*pkt) * num_regs) +
+			sizeof(struct packet_msg_prot) * 2;
+
+	if (cb_size > SZ_2M) {
+		dev_err(hdev->dev, "CB size must be smaller than %uMB", SZ_2M);
+		return -ENOMEM;
 	}
 
-	for (i = 0 ; i < NUM_OF_MONITORS_IN_BLOCK << 2 ; i += 4) {
-		WREG32(mmSYNC_MNGR_E_N_SYNC_MNGR_OBJS_MON_STATUS_0 + i, 0);
-		WREG32(mmSYNC_MNGR_E_S_SYNC_MNGR_OBJS_MON_STATUS_0 + i, 0);
-		WREG32(mmSYNC_MNGR_W_N_SYNC_MNGR_OBJS_MON_STATUS_0 + i, 0);
+	pending_cb = kzalloc(sizeof(*pending_cb), GFP_KERNEL);
+	if (!pending_cb)
+		return -ENOMEM;
+
+	cb = hl_cb_kernel_create(hdev, cb_size, false);
+	if (!cb) {
+		kfree(pending_cb);
+		return -EFAULT;
 	}
 
-	i = GAUDI_FIRST_AVAILABLE_W_S_SYNC_OBJECT * 4;
+	pkt = cb->kernel_address;
+
+	ctl = FIELD_PREP(GAUDI_PKT_LONG_CTL_OP_MASK, 0); /* write the value */
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_OPCODE_MASK, PACKET_MSG_LONG);
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_EB_MASK, 1);
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_RB_MASK, 1);
+	ctl |= FIELD_PREP(GAUDI_PKT_CTL_MB_MASK, 1);
+
+	for (i = 0; i < num_regs ; i++, pkt++) {
+		pkt->ctl = cpu_to_le32(ctl);
+		pkt->value = cpu_to_le32(val);
+		pkt->addr = cpu_to_le64(reg_base + (i * 4));
+	}
 
-	for (; i < NUM_OF_SOB_IN_BLOCK << 2 ; i += 4)
-		WREG32(mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_SOB_OBJ_0 + i, 0);
+	hl_cb_destroy(hdev, &hdev->kernel_cb_mgr, cb->id << PAGE_SHIFT);
 
-	i = GAUDI_FIRST_AVAILABLE_W_S_MONITOR * 4;
+	pending_cb->cb = cb;
+	pending_cb->cb_size = cb_size;
+	/* The queue ID MUST be an external queue ID. Otherwise, we will
+	 * have undefined behavior
+	 */
+	pending_cb->hw_queue_id = hw_queue_id;
 
-	for (; i < NUM_OF_MONITORS_IN_BLOCK << 2 ; i += 4)
-		WREG32(mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_MON_STATUS_0 + i, 0);
+	spin_lock(&ctx->pending_cb_lock);
+	list_add_tail(&pending_cb->cb_node, &ctx->pending_cb_list);
+	spin_unlock(&ctx->pending_cb_lock);
+
+	return 0;
+}
+
+static int gaudi_restore_sm_registers(struct hl_device *hdev)
+{
+	u64 base_addr;
+	u32 num_regs;
+	int rc;
+
+	base_addr = CFG_BASE + mmSYNC_MNGR_E_N_SYNC_MNGR_OBJS_SOB_OBJ_0;
+	num_regs = NUM_OF_SOB_IN_BLOCK;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_E_S_SYNC_MNGR_OBJS_SOB_OBJ_0;
+	num_regs = NUM_OF_SOB_IN_BLOCK;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_W_N_SYNC_MNGR_OBJS_SOB_OBJ_0;
+	num_regs = NUM_OF_SOB_IN_BLOCK;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_E_N_SYNC_MNGR_OBJS_MON_STATUS_0;
+	num_regs = NUM_OF_MONITORS_IN_BLOCK;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_E_S_SYNC_MNGR_OBJS_MON_STATUS_0;
+	num_regs = NUM_OF_MONITORS_IN_BLOCK;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_W_N_SYNC_MNGR_OBJS_MON_STATUS_0;
+	num_regs = NUM_OF_MONITORS_IN_BLOCK;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_SOB_OBJ_0 +
+			(GAUDI_FIRST_AVAILABLE_W_S_SYNC_OBJECT * 4);
+	num_regs = NUM_OF_SOB_IN_BLOCK - GAUDI_FIRST_AVAILABLE_W_S_SYNC_OBJECT;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	base_addr = CFG_BASE +  mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_MON_STATUS_0 +
+			(GAUDI_FIRST_AVAILABLE_W_S_MONITOR * 4);
+	num_regs = NUM_OF_MONITORS_IN_BLOCK - GAUDI_FIRST_AVAILABLE_W_S_MONITOR;
+	rc = gaudi_memset_registers(hdev, base_addr, num_regs, 0);
+	if (rc) {
+		dev_err(hdev->dev, "failed resetting SM registers");
+		return -ENOMEM;
+	}
+
+	return 0;
 }
 
 static void gaudi_restore_dma_registers(struct hl_device *hdev)
@@ -5663,18 +5848,23 @@ static void gaudi_restore_qm_registers(struct hl_device *hdev)
 	}
 }
 
-static void gaudi_restore_user_registers(struct hl_device *hdev)
+static int gaudi_restore_user_registers(struct hl_device *hdev)
 {
-	gaudi_restore_sm_registers(hdev);
+	int rc;
+
+	rc = gaudi_restore_sm_registers(hdev);
+	if (rc)
+		return rc;
+
 	gaudi_restore_dma_registers(hdev);
 	gaudi_restore_qm_registers(hdev);
+
+	return 0;
 }
 
 static int gaudi_context_switch(struct hl_device *hdev, u32 asid)
 {
-	gaudi_restore_user_registers(hdev);
-
-	return 0;
+	return gaudi_restore_user_registers(hdev);
 }
 
 static int gaudi_mmu_clear_pgt_range(struct hl_device *hdev)
@@ -8202,12 +8392,16 @@ static u32 gaudi_gen_wait_cb(struct hl_device *hdev,
 static void gaudi_reset_sob(struct hl_device *hdev, void *data)
 {
 	struct hl_hw_sob *hw_sob = (struct hl_hw_sob *) data;
+	int rc;
 
 	dev_dbg(hdev->dev, "reset SOB, q_idx: %d, sob_id: %d\n", hw_sob->q_idx,
 		hw_sob->sob_id);
 
-	WREG32(mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_SOB_OBJ_0 + hw_sob->sob_id * 4,
-		0);
+	rc = gaudi_schedule_register_memset(hdev, hw_sob->q_idx,
+			CFG_BASE + mmSYNC_MNGR_W_S_SYNC_MNGR_OBJS_SOB_OBJ_0 +
+			hw_sob->sob_id * 4, 1, 0);
+	if (rc)
+		dev_err(hdev->dev, "failed resetting sob %u", hw_sob->sob_id);
 
 	kref_init(&hw_sob->kref);
 }
diff --git a/drivers/misc/habanalabs/gaudi/gaudiP.h b/drivers/misc/habanalabs/gaudi/gaudiP.h
index a7ab2d7e57d4..78830443341d 100644
--- a/drivers/misc/habanalabs/gaudi/gaudiP.h
+++ b/drivers/misc/habanalabs/gaudi/gaudiP.h
@@ -251,11 +251,13 @@ enum gaudi_nic_mask {
  * @hdev: habanalabs device structure.
  * @kref: refcount of this SOB group. group will reset once refcount is zero.
  * @base_sob_id: base sob id of this SOB group.
+ * @queue_id: id of the queue that waits on this sob group
  */
 struct gaudi_hw_sob_group {
 	struct hl_device	*hdev;
 	struct kref		kref;
 	u32			base_sob_id;
+	u32			queue_id;
 };
 
 #define NUM_SOB_GROUPS (HL_RSVD_SOBS * QMAN_STREAMS)
diff --git a/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h b/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h
index 784b5bc8d0ba..6e097ace2e96 100644
--- a/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h
+++ b/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h
@@ -78,6 +78,9 @@ struct packet_wreg_bulk {
 	__le64 values[0]; /* data starts here */
 };
 
+#define GAUDI_PKT_LONG_CTL_OP_SHIFT		20
+#define GAUDI_PKT_LONG_CTL_OP_MASK		0x00300000
+
 struct packet_msg_long {
 	__le32 value;
 	__le32 ctl;
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-01-03 20:30 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-01-03 20:29 [PATCH 1/4] habanalabs: separate common code to dedicated folders Oded Gabbay
2021-01-03 20:29 ` [PATCH 2/4] habanalabs: increment ctx ref from within a cs allocation Oded Gabbay
2021-01-03 20:29 ` [PATCH 3/4] habanalabs: add driver support for internal cb scheduling Oded Gabbay
2021-01-03 20:29 ` [PATCH 4/4] habanalabs/gaudi: remove PCI access to SM block Oded Gabbay

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.