summaryrefslogtreecommitdiff
path: root/drivers/misc
diff options
context:
space:
mode:
authorOfir Bitton <obitton@habana.ai>2020-06-18 09:51:16 +0300
committerOded Gabbay <oded.gabbay@gmail.com>2020-07-24 20:31:36 +0300
commitdb491e4f08a9fd84ebb1ebd22a6b0b988a81a0d8 (patch)
treec1581ca6328266d01a986c62e9af2bc6c6838517 /drivers/misc
parentc8f9b49d2db7e98df534c6e666b26c64da4cb667 (diff)
downloadlwn-db491e4f08a9fd84ebb1ebd22a6b0b988a81a0d8.tar.gz
lwn-db491e4f08a9fd84ebb1ebd22a6b0b988a81a0d8.zip
habanalabs: Add dropped cs statistics info struct
Add command submission statistics structure which can be obtained through the info ioctl. Each drop counter describes the reason for which the command submission was dropped. This information is needed for the user to be aware of the specific reason for which the submitted work was dropped. The user can then utilize the driver more efficiently. Signed-off-by: Ofir Bitton <obitton@habana.ai> Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com> Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Diffstat (limited to 'drivers/misc')
-rw-r--r--drivers/misc/habanalabs/command_submission.c24
-rw-r--r--drivers/misc/habanalabs/habanalabs.h5
-rw-r--r--drivers/misc/habanalabs/habanalabs_ioctl.c24
-rw-r--r--drivers/misc/habanalabs/hw_queue.c5
4 files changed, 56 insertions, 2 deletions
diff --git a/drivers/misc/habanalabs/command_submission.c b/drivers/misc/habanalabs/command_submission.c
index f81d6685e011..777f88d25acd 100644
--- a/drivers/misc/habanalabs/command_submission.c
+++ b/drivers/misc/habanalabs/command_submission.c
@@ -246,6 +246,18 @@ static void free_job(struct hl_device *hdev, struct hl_cs_job *job)
kfree(job);
}
+static void cs_counters_aggregate(struct hl_device *hdev, struct hl_ctx *ctx)
+{
+ hdev->aggregated_cs_counters.device_in_reset_drop_cnt +=
+ ctx->cs_counters.device_in_reset_drop_cnt;
+ hdev->aggregated_cs_counters.out_of_mem_drop_cnt +=
+ ctx->cs_counters.out_of_mem_drop_cnt;
+ hdev->aggregated_cs_counters.parsing_drop_cnt +=
+ ctx->cs_counters.parsing_drop_cnt;
+ hdev->aggregated_cs_counters.queue_full_drop_cnt +=
+ ctx->cs_counters.queue_full_drop_cnt;
+}
+
static void cs_do_release(struct kref *ref)
{
struct hl_cs *cs = container_of(ref, struct hl_cs,
@@ -349,6 +361,8 @@ static void cs_do_release(struct kref *ref)
dma_fence_signal(cs->fence);
dma_fence_put(cs->fence);
+ cs_counters_aggregate(hdev, cs->ctx);
+
kfree(cs);
}
@@ -632,12 +646,15 @@ static int cs_ioctl_default(struct hl_fpriv *hpriv, void __user *chunks,
rc = validate_queue_index(hdev, chunk, &queue_type,
&is_kernel_allocated_cb);
- if (rc)
+ if (rc) {
+ hpriv->ctx->cs_counters.parsing_drop_cnt++;
goto free_cs_object;
+ }
if (is_kernel_allocated_cb) {
cb = get_cb_from_cs_chunk(hdev, &hpriv->cb_mgr, chunk);
if (!cb) {
+ hpriv->ctx->cs_counters.parsing_drop_cnt++;
rc = -EINVAL;
goto free_cs_object;
}
@@ -651,6 +668,7 @@ static int cs_ioctl_default(struct hl_fpriv *hpriv, void __user *chunks,
job = hl_cs_allocate_job(hdev, queue_type,
is_kernel_allocated_cb);
if (!job) {
+ hpriv->ctx->cs_counters.out_of_mem_drop_cnt++;
dev_err(hdev->dev, "Failed to allocate a new job\n");
rc = -ENOMEM;
if (is_kernel_allocated_cb)
@@ -683,6 +701,7 @@ static int cs_ioctl_default(struct hl_fpriv *hpriv, void __user *chunks,
rc = cs_parser(hpriv, job);
if (rc) {
+ hpriv->ctx->cs_counters.parsing_drop_cnt++;
dev_err(hdev->dev,
"Failed to parse JOB %d.%llu.%d, err %d, rejecting the CS\n",
cs->ctx->asid, cs->sequence, job->id, rc);
@@ -691,6 +710,7 @@ static int cs_ioctl_default(struct hl_fpriv *hpriv, void __user *chunks,
}
if (int_queues_only) {
+ hpriv->ctx->cs_counters.parsing_drop_cnt++;
dev_err(hdev->dev,
"Reject CS %d.%llu because only internal queues jobs are present\n",
cs->ctx->asid, cs->sequence);
@@ -875,6 +895,7 @@ static int cs_ioctl_signal_wait(struct hl_fpriv *hpriv, enum hl_cs_type cs_type,
job = hl_cs_allocate_job(hdev, q_type, true);
if (!job) {
+ ctx->cs_counters.out_of_mem_drop_cnt++;
dev_err(hdev->dev, "Failed to allocate a new job\n");
rc = -ENOMEM;
goto put_cs;
@@ -882,6 +903,7 @@ static int cs_ioctl_signal_wait(struct hl_fpriv *hpriv, enum hl_cs_type cs_type,
cb = hl_cb_kernel_create(hdev, PAGE_SIZE);
if (!cb) {
+ ctx->cs_counters.out_of_mem_drop_cnt++;
kfree(job);
rc = -EFAULT;
goto put_cs;
diff --git a/drivers/misc/habanalabs/habanalabs.h b/drivers/misc/habanalabs/habanalabs.h
index e4d6f7c91194..ae781453a509 100644
--- a/drivers/misc/habanalabs/habanalabs.h
+++ b/drivers/misc/habanalabs/habanalabs.h
@@ -10,6 +10,7 @@
#include "include/armcp_if.h"
#include "include/qman_if.h"
+#include <uapi/misc/habanalabs.h>
#include <linux/cdev.h>
#include <linux/iopoll.h>
@@ -787,6 +788,7 @@ struct hl_ctx {
struct mutex mem_hash_lock;
struct mutex mmu_lock;
struct list_head debugfs_list;
+ struct hl_cs_counters cs_counters;
u64 cs_sequence;
u64 *dram_default_hops;
spinlock_t cs_lock;
@@ -1391,6 +1393,7 @@ struct hl_device_idle_busy_ts {
* @compute_ctx: current compute context executing.
* @idle_busy_ts_arr: array to hold time stamps of transitions from idle to busy
* and vice-versa
+ * @aggregated_cs_counters: aggregated cs counters among all contexts
* @dram_used_mem: current DRAM memory consumption.
* @timeout_jiffies: device CS timeout value.
* @max_power: the max power of the device, as configured by the sysadmin. This
@@ -1489,6 +1492,8 @@ struct hl_device {
struct hl_device_idle_busy_ts *idle_busy_ts_arr;
+ struct hl_cs_counters aggregated_cs_counters;
+
atomic64_t dram_used_mem;
u64 timeout_jiffies;
u64 max_power;
diff --git a/drivers/misc/habanalabs/habanalabs_ioctl.c b/drivers/misc/habanalabs/habanalabs_ioctl.c
index 52eedd3a6c3a..5af1c03da473 100644
--- a/drivers/misc/habanalabs/habanalabs_ioctl.c
+++ b/drivers/misc/habanalabs/habanalabs_ioctl.c
@@ -276,6 +276,27 @@ static int time_sync_info(struct hl_device *hdev, struct hl_info_args *args)
min((size_t) max_size, sizeof(time_sync))) ? -EFAULT : 0;
}
+static int cs_counters_info(struct hl_fpriv *hpriv, struct hl_info_args *args)
+{
+ struct hl_device *hdev = hpriv->hdev;
+ struct hl_info_cs_counters cs_counters = {0};
+ u32 max_size = args->return_size;
+ void __user *out = (void __user *) (uintptr_t) args->return_pointer;
+
+ if ((!max_size) || (!out))
+ return -EINVAL;
+
+ memcpy(&cs_counters.cs_counters, &hdev->aggregated_cs_counters,
+ sizeof(struct hl_cs_counters));
+
+ if (hpriv->ctx)
+ memcpy(&cs_counters.ctx_cs_counters, &hpriv->ctx->cs_counters,
+ sizeof(struct hl_cs_counters));
+
+ return copy_to_user(out, &cs_counters,
+ min((size_t) max_size, sizeof(cs_counters))) ? -EFAULT : 0;
+}
+
static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
struct device *dev)
{
@@ -336,6 +357,9 @@ static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
case HL_INFO_TIME_SYNC:
return time_sync_info(hdev, args);
+ case HL_INFO_CS_COUNTERS:
+ return cs_counters_info(hpriv, args);
+
default:
dev_err(dev, "Invalid request %d\n", args->op);
rc = -ENOTTY;
diff --git a/drivers/misc/habanalabs/hw_queue.c b/drivers/misc/habanalabs/hw_queue.c
index f5a10a5ac300..da66ffb528f8 100644
--- a/drivers/misc/habanalabs/hw_queue.c
+++ b/drivers/misc/habanalabs/hw_queue.c
@@ -514,6 +514,7 @@ int hl_hw_queue_schedule_cs(struct hl_cs *cs)
hdev->asic_funcs->hw_queues_lock(hdev);
if (hl_device_disabled_or_in_reset(hdev)) {
+ ctx->cs_counters.device_in_reset_drop_cnt++;
dev_err(hdev->dev,
"device is disabled or in reset, CS rejected!\n");
rc = -EPERM;
@@ -543,8 +544,10 @@ int hl_hw_queue_schedule_cs(struct hl_cs *cs)
break;
}
- if (rc)
+ if (rc) {
+ ctx->cs_counters.queue_full_drop_cnt++;
goto unroll_cq_resv;
+ }
if (q->queue_type == QUEUE_TYPE_EXT ||
q->queue_type == QUEUE_TYPE_HW)