[Intel-xe] [PATCH 02/21] drm/xe/oa: Add OA types
Umesh Nerlige Ramappa
umesh.nerlige.ramappa at intel.com
Fri Oct 13 17:05:28 UTC 2023
On Tue, Sep 19, 2023 at 09:10:30AM -0700, Ashutosh Dixit wrote:
>Add types and data structs used by OA. The data structs maintain device and
>gt level information, information about the open OA stream and OA buffer
>used internally to capture OA counters written by HW as well as capture
>configurations which can be selected for an OA stream.
>
>v2: Add linux includes to fix build
>v3: Change oa_unit_id to u16 (Umesh)
>
>Signed-off-by: Ashutosh Dixit <ashutosh.dixit at intel.com>
lgtm,
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa at intel.com>
Umesh
>---
> drivers/gpu/drm/xe/xe_oa_types.h | 295 +++++++++++++++++++++++++++++++
> 1 file changed, 295 insertions(+)
> create mode 100644 drivers/gpu/drm/xe/xe_oa_types.h
>
>diff --git a/drivers/gpu/drm/xe/xe_oa_types.h b/drivers/gpu/drm/xe/xe_oa_types.h
>new file mode 100644
>index 0000000000000..4063c81e353ff
>--- /dev/null
>+++ b/drivers/gpu/drm/xe/xe_oa_types.h
>@@ -0,0 +1,295 @@
>+/* SPDX-License-Identifier: MIT */
>+/*
>+ * Copyright © 2023 Intel Corporation
>+ */
>+
>+#ifndef _XE_OA_TYPES_H_
>+#define _XE_OA_TYPES_H__
>+
>+#include <linux/idr.h>
>+#include <linux/kobject.h>
>+#include <linux/poll.h>
>+#include <linux/sysfs.h>
>+#include <drm/xe_drm.h>
>+#include "regs/xe_reg_defs.h"
>+
>+struct drm_device;
>+struct drm_file;
>+
>+enum {
>+ OA_GROUP_OAG = 0,
>+ OA_GROUP_OAM_SAMEDIA_0 = 0,
>+
>+ OA_GROUP_MAX,
>+ OA_GROUP_INVALID = U32_MAX,
>+};
>+
>+enum oa_type {
>+ TYPE_OAG,
>+ TYPE_OAM,
>+};
>+
>+enum report_header {
>+ HDR_32_BIT = 0,
>+ HDR_64_BIT,
>+};
>+
>+struct xe_oa_format {
>+ u32 format;
>+ int size;
>+ int type;
>+ enum report_header header;
>+};
>+
>+struct xe_oa_reg {
>+ struct xe_reg addr;
>+ u32 value;
>+};
>+
>+struct xe_oa_config {
>+ struct xe_oa *oa;
>+
>+ char uuid[UUID_STRING_LEN + 1];
>+ int id;
>+
>+ const struct xe_oa_reg *mux_regs;
>+ u32 mux_regs_len;
>+ const struct xe_oa_reg *b_counter_regs;
>+ u32 b_counter_regs_len;
>+ const struct xe_oa_reg *flex_regs;
>+ u32 flex_regs_len;
>+
>+ struct attribute_group sysfs_metric;
>+ struct attribute *attrs[2];
>+ struct kobj_attribute sysfs_metric_id;
>+
>+ struct kref ref;
>+ struct rcu_head rcu;
>+};
>+
>+struct xe_oa_regs {
>+ u32 base;
>+ struct xe_reg oa_head_ptr;
>+ struct xe_reg oa_tail_ptr;
>+ struct xe_reg oa_buffer;
>+ struct xe_reg oa_ctx_ctrl;
>+ struct xe_reg oa_ctrl;
>+ struct xe_reg oa_debug;
>+ struct xe_reg oa_status;
>+ u32 oa_ctrl_counter_format_shift;
>+};
>+
>+/**
>+ * struct xe_oa_group - OA group representing one hardware OA unit
>+ */
>+struct xe_oa_group {
>+ /** @oa_unit_id: identifier for the OA unit */
>+ u16 oa_unit_id;
>+
>+ /**
>+ * @exclusive_stream: The stream currently using the OA unit. This is
>+ * sometimes accessed outside a syscall associated to its file
>+ * descriptor.
>+ */
>+ struct xe_oa_stream *exclusive_stream;
>+
>+ /** @num_engines: number of engines using this OA unit */
>+ u32 num_engines;
>+
>+ /** @regs: OA buffer register group for programming the OA unit */
>+ struct xe_oa_regs regs;
>+
>+ /** @type: Type of OA unit - OAM, OAG etc. */
>+ enum oa_type type;
>+};
>+
>+/**
>+ * struct xe_oa_gt - OA per-gt information
>+ */
>+struct xe_oa_gt {
>+ /** @lock: lock associated with anything below within this structure */
>+ struct mutex lock;
>+
>+ /** @num_oa_groups: number of oa groups per gt */
>+ u32 num_oa_groups;
>+
>+ /** @group: list of OA groups - one for each OA buffer */
>+ struct xe_oa_group *group;
>+};
>+
>+/**
>+ * struct xe_oa - OA device level information
>+ */
>+struct xe_oa {
>+ /** @xe: back pointer to xe device */
>+ struct xe_device *xe;
>+
>+ /** @metrics_kobj: kobj for metrics sysfs */
>+ struct kobject *metrics_kobj;
>+
>+ /**
>+ * @metrics_lock: lock associated with adding/modifying/removing OA
>+ * configs in oa->metrics_idr.
>+ */
>+ struct mutex metrics_lock;
>+
>+ /**
>+ * @metrics_idr: List of dynamic configurations (struct xe_oa_config)
>+ */
>+ struct idr metrics_idr;
>+
>+ /** @ctx_oactxctrl_offset: offset of OACTXCONTROL register in context image */
>+ u32 ctx_oactxctrl_offset;
>+
>+ /** @oa_formats: tracks all OA formats across platforms */
>+ const struct xe_oa_format *oa_formats;
>+
>+#define FORMAT_MASK_SIZE DIV_ROUND_UP(XE_OA_FORMAT_MAX - 1, BITS_PER_LONG)
>+
>+ /** @format_mask: tracks valid OA formats for a platform */
>+ unsigned long format_mask[FORMAT_MASK_SIZE];
>+
>+ /** @oa_unit_ids: tracks oa unit ids assigned across gt's */
>+ u16 oa_unit_ids;
>+};
>+
>+/**
>+ * struct xe_oa_stream - state for a single open stream FD
>+ */
>+struct xe_oa_stream {
>+ /** @oa: xe_oa backpointer */
>+ struct xe_oa *oa;
>+
>+ /** @gt: gt associated with the oa stream */
>+ struct xe_gt *gt;
>+
>+ /**
>+ * @hwe: hardware engine associated with this performance stream.
>+ */
>+ struct xe_hw_engine *hwe;
>+
>+ /** @lock: Lock associated with operations on stream */
>+ struct mutex lock;
>+
>+ /**
>+ * @sample: true when DRM_XE_OA_PROP_SAMPLE_OA is given when
>+ * opening a stream, representing the contents of a single sample
>+ * as read() by userspace.
>+ */
>+ bool sample;
>+
>+ /**
>+ * @sample_size: Considering the configured contents of a sample
>+ * combined with the required header size, this is the total size
>+ * of a single sample record.
>+ */
>+ int sample_size;
>+
>+ /**
>+ * @exec_q: %NULL if measuring system-wide across all exec_q's or a
>+ * specific exec_q that is being monitored.
>+ */
>+ struct xe_exec_queue *exec_q;
>+
>+ /**
>+ * @enabled: Whether the stream is currently enabled, considering
>+ * whether the stream was opened in a disabled state and based
>+ * on `XE_OA_IOCTL_ENABLE` and `XE_OA_IOCTL_DISABLE` calls.
>+ */
>+ bool enabled;
>+
>+ /** @oa_config: The OA configuration used by the stream */
>+ struct xe_oa_config *oa_config;
>+
>+ /**
>+ * @oa_config_bos: A list of struct i915_oa_config_bo allocated lazily
>+ * each time @oa_config changes.
>+ */
>+ struct llist_head oa_config_bos;
>+
>+ /** @specific_ctx_id: id of the context used for filtering reports */
>+ u32 specific_ctx_id;
>+
>+ /** @specific_ctx_id_mask: The mask used to masking specific_ctx_id bits */
>+ u32 specific_ctx_id_mask;
>+
>+ /**
>+ * @poll_check_timer: High resolution timer that will periodically
>+ * check for data in the circular OA buffer for notifying userspace
>+ * (e.g. during a read() or poll()).
>+ */
>+ struct hrtimer poll_check_timer;
>+
>+ /**
>+ * @poll_wq: The wait queue that hrtimer callback wakes when it
>+ * sees data ready to read in the circular OA buffer.
>+ */
>+ wait_queue_head_t poll_wq;
>+
>+ /** @pollin: Whether there is data available to read */
>+ bool pollin;
>+
>+ /** @periodic: Whether periodic sampling is currently enabled */
>+ bool periodic;
>+
>+ /** @period_exponent: The OA unit sampling frequency is derived from this */
>+ int period_exponent;
>+
>+ /** @oa_buffer: State of the OA buffer */
>+ struct {
>+ /** @format: data format */
>+ const struct xe_oa_format *format;
>+
>+ /** @format: xe_bo backing the OA buffer */
>+ struct xe_bo *bo;
>+
>+ /** @vaddr: mapped vaddr of the OA buffer */
>+ u8 *vaddr;
>+
>+ /** @last_ctx_id: last context id for OA data added */
>+ u32 last_ctx_id;
>+
>+ /**
>+ * @ptr_lock: Locks reads and writes to all head/tail state
>+ *
>+ * Consider: the head and tail pointer state needs to be read
>+ * consistently from a hrtimer callback (atomic context) and
>+ * read() fop (user context) with tail pointer updates happening
>+ * in atomic context and head updates in user context and the
>+ * (unlikely) possibility of read() errors needing to reset all
>+ * head/tail state.
>+ *
>+ * Note: Contention/performance aren't currently a significant
>+ * concern here considering the relatively low frequency of
>+ * hrtimer callbacks (5ms period) and that reads typically only
>+ * happen in response to a hrtimer event and likely complete
>+ * before the next callback.
>+ *
>+ * Note: This lock is not held *while* reading and copying data
>+ * to userspace so the value of head observed in htrimer
>+ * callbacks won't represent any partial consumption of data.
>+ */
>+ spinlock_t ptr_lock;
>+
>+ /**
>+ * @head: Although we can always read back the head pointer register,
>+ * we prefer to avoid trusting the HW state, just to avoid any
>+ * risk that some hardware condition could somehow bump the
>+ * head pointer unpredictably and cause us to forward the wrong
>+ * OA buffer data to userspace.
>+ */
>+ u32 head;
>+
>+ /**
>+ * @tail: The last verified tail that can be read by userspace.
>+ */
>+ u32 tail;
>+ } oa_buffer;
>+
>+ /**
>+ * @poll_oa_period: The period in nanoseconds at which the OA
>+ * buffer should be checked for available data.
>+ */
>+ u64 poll_oa_period;
>+};
>+#endif
>--
>2.41.0
>
More information about the Intel-xe
mailing list