[deliverable/linux.git] / drivers / gpu / drm / i915 / intel_ringbuffer.h

#ifndef _INTEL_RINGBUFFER_H_
#define _INTEL_RINGBUFFER_H_

/*
 * Gen2 BSpec "1. Programming Environment" / 1.4.4.6 "Ring Buffer Use"
 * Gen3 BSpec "vol1c Memory Interface Functions" / 2.3.4.5 "Ring Buffer Use"
 * Gen4+ BSpec "vol1c Memory Interface and Command Stream" / 5.3.4.5 "Ring Buffer Use"
 *
 * "If the Ring Buffer Head Pointer and the Tail Pointer are on the same
 * cacheline, the Head Pointer must not be greater than the Tail
 * Pointer."
 */
#define I915_RING_FREE_SPACE 64

struct  intel_hw_status_page {
	u32		*page_addr;
	unsigned int	gfx_addr;
	struct		drm_i915_gem_object *obj;
};

#define I915_READ_TAIL(ring) I915_READ(RING_TAIL((ring)->mmio_base))
#define I915_WRITE_TAIL(ring, val) I915_WRITE(RING_TAIL((ring)->mmio_base), val)

#define I915_READ_START(ring) I915_READ(RING_START((ring)->mmio_base))
#define I915_WRITE_START(ring, val) I915_WRITE(RING_START((ring)->mmio_base), val)

#define I915_READ_HEAD(ring)  I915_READ(RING_HEAD((ring)->mmio_base))
#define I915_WRITE_HEAD(ring, val) I915_WRITE(RING_HEAD((ring)->mmio_base), val)

#define I915_READ_CTL(ring) I915_READ(RING_CTL((ring)->mmio_base))
#define I915_WRITE_CTL(ring, val) I915_WRITE(RING_CTL((ring)->mmio_base), val)

#define I915_READ_IMR(ring) I915_READ(RING_IMR((ring)->mmio_base))
#define I915_WRITE_IMR(ring, val) I915_WRITE(RING_IMR((ring)->mmio_base), val)

enum intel_ring_hangcheck_action {
	HANGCHECK_IDLE = 0,
	HANGCHECK_WAIT,
	HANGCHECK_ACTIVE,
	HANGCHECK_KICK,
	HANGCHECK_HUNG,
};

#define HANGCHECK_SCORE_RING_HUNG 31

struct intel_ring_hangcheck {
	bool deadlock;
	u32 seqno;
	u32 acthd;
	int score;
	enum intel_ring_hangcheck_action action;
};

struct  intel_ring_buffer {
	const char	*name;
	enum intel_ring_id {
		RCS = 0x0,
		VCS,
		BCS,
		VECS,
	} id;
#define I915_NUM_RINGS 4
	u32		mmio_base;
	void		__iomem *virtual_start;
	struct		drm_device *dev;
	struct		drm_i915_gem_object *obj;

	u32		head;
	u32		tail;
	int		space;
	int		size;
	int		effective_size;
	struct intel_hw_status_page status_page;

	/** We track the position of the requests in the ring buffer, and
	 * when each is retired we increment last_retired_head as the GPU
	 * must have finished processing the request and so we know we
	 * can advance the ringbuffer up to that position.
	 *
	 * last_retired_head is set to -1 after the value is consumed so
	 * we can detect new retirements.
	 */
	u32		last_retired_head;

	unsigned irq_refcount; /* protected by dev_priv->irq_lock */
	u32		irq_enable_mask;	/* bitmask to enable ring interrupt */
	u32		trace_irq_seqno;
	u32		sync_seqno[I915_NUM_RINGS-1];
	bool __must_check (*irq_get)(struct intel_ring_buffer *ring);
	void		(*irq_put)(struct intel_ring_buffer *ring);

	int		(*init)(struct intel_ring_buffer *ring);

	void		(*write_tail)(struct intel_ring_buffer *ring,
				      u32 value);
	int __must_check (*flush)(struct intel_ring_buffer *ring,
				  u32	invalidate_domains,
				  u32	flush_domains);
	int		(*add_request)(struct intel_ring_buffer *ring);
	/* Some chipsets are not quite as coherent as advertised and need
	 * an expensive kick to force a true read of the up-to-date seqno.
	 * However, the up-to-date seqno is not always required and the last
	 * seen value is good enough. Note that the seqno will always be
	 * monotonic, even if not coherent.
	 */
	u32		(*get_seqno)(struct intel_ring_buffer *ring,
				     bool lazy_coherency);
	void		(*set_seqno)(struct intel_ring_buffer *ring,
				     u32 seqno);
	int		(*dispatch_execbuffer)(struct intel_ring_buffer *ring,
					       u32 offset, u32 length,
					       unsigned flags);
#define I915_DISPATCH_SECURE 0x1
#define I915_DISPATCH_PINNED 0x2
	void		(*cleanup)(struct intel_ring_buffer *ring);
	int		(*sync_to)(struct intel_ring_buffer *ring,
				   struct intel_ring_buffer *to,
				   u32 seqno);

	/* our mbox written by others */
	u32		semaphore_register[I915_NUM_RINGS];
	/* mboxes this ring signals to */
	u32		signal_mbox[I915_NUM_RINGS];

	/**
	 * List of objects currently involved in rendering from the
	 * ringbuffer.
	 *
	 * Includes buffers having the contents of their GPU caches
	 * flushed, not necessarily primitives.  last_rendering_seqno
	 * represents when the rendering involved will be completed.
	 *
	 * A reference is held on the buffer while on this list.
	 */
	struct list_head active_list;

	/**
	 * List of breadcrumbs associated with GPU requests currently
	 * outstanding.
	 */
	struct list_head request_list;

	/**
	 * Do we have some not yet emitted requests outstanding?
	 */
	struct drm_i915_gem_request *preallocated_lazy_request;
	u32 outstanding_lazy_seqno;
	bool gpu_caches_dirty;
	bool fbc_dirty;

	wait_queue_head_t irq_queue;

	/**
	 * Do an explicit TLB flush before MI_SET_CONTEXT
	 */
	bool itlb_before_ctx_switch;
	struct i915_hw_context *default_context;
	struct i915_hw_context *last_context;

	struct intel_ring_hangcheck hangcheck;

	struct {
		struct drm_i915_gem_object *obj;
		u32 gtt_offset;
		volatile u32 *cpu_page;
	} scratch;
};

static inline bool
intel_ring_initialized(struct intel_ring_buffer *ring)
{
	return ring->obj != NULL;
}

static inline unsigned
intel_ring_flag(struct intel_ring_buffer *ring)
{
	return 1 << ring->id;
}

static inline u32
intel_ring_sync_index(struct intel_ring_buffer *ring,
		      struct intel_ring_buffer *other)
{
	int idx;

	/*
	 * cs -> 0 = vcs, 1 = bcs
	 * vcs -> 0 = bcs, 1 = cs,
	 * bcs -> 0 = cs, 1 = vcs.
	 */

	idx = (other - ring) - 1;
	if (idx < 0)
		idx += I915_NUM_RINGS;

	return idx;
}

static inline u32
intel_read_status_page(struct intel_ring_buffer *ring,
		       int reg)
{
	/* Ensure that the compiler doesn't optimize away the load. */
	barrier();
	return ring->status_page.page_addr[reg];
}

static inline void
intel_write_status_page(struct intel_ring_buffer *ring,
			int reg, u32 value)
{
	ring->status_page.page_addr[reg] = value;
}

/**
 * Reads a dword out of the status page, which is written to from the command
 * queue by automatic updates, MI_REPORT_HEAD, MI_STORE_DATA_INDEX, or
 * MI_STORE_DATA_IMM.
 *
 * The following dwords have a reserved meaning:
 * 0x00: ISR copy, updated when an ISR bit not set in the HWSTAM changes.
 * 0x04: ring 0 head pointer
 * 0x05: ring 1 head pointer (915-class)
 * 0x06: ring 2 head pointer (915-class)
 * 0x10-0x1b: Context status DWords (GM45)
 * 0x1f: Last written status offset. (GM45)
 *
 * The area from dword 0x20 to 0x3ff is available for driver usage.
 */
#define I915_GEM_HWS_INDEX		0x20
#define I915_GEM_HWS_SCRATCH_INDEX	0x30
#define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)

void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring);

int __must_check intel_ring_begin(struct intel_ring_buffer *ring, int n);
static inline void intel_ring_emit(struct intel_ring_buffer *ring,
				   u32 data)
{
	iowrite32(data, ring->virtual_start + ring->tail);
	ring->tail += 4;
}
static inline void intel_ring_advance(struct intel_ring_buffer *ring)
{
	ring->tail &= ring->size - 1;
}
void __intel_ring_advance(struct intel_ring_buffer *ring);

int __must_check intel_ring_idle(struct intel_ring_buffer *ring);
void intel_ring_init_seqno(struct intel_ring_buffer *ring, u32 seqno);
int intel_ring_flush_all_caches(struct intel_ring_buffer *ring);
int intel_ring_invalidate_all_caches(struct intel_ring_buffer *ring);

int intel_init_render_ring_buffer(struct drm_device *dev);
int intel_init_bsd_ring_buffer(struct drm_device *dev);
int intel_init_blt_ring_buffer(struct drm_device *dev);
int intel_init_vebox_ring_buffer(struct drm_device *dev);

u32 intel_ring_get_active_head(struct intel_ring_buffer *ring);
void intel_ring_setup_status_page(struct intel_ring_buffer *ring);

static inline u32 intel_ring_get_tail(struct intel_ring_buffer *ring)
{
	return ring->tail;
}

static inline u32 intel_ring_get_seqno(struct intel_ring_buffer *ring)
{
	BUG_ON(ring->outstanding_lazy_seqno == 0);
	return ring->outstanding_lazy_seqno;
}

static inline void i915_trace_irq_get(struct intel_ring_buffer *ring, u32 seqno)
{
	if (ring->trace_irq_seqno == 0 && ring->irq_get(ring))
		ring->trace_irq_seqno = seqno;
}

/* DRI warts */
int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size);

#endif /* _INTEL_RINGBUFFER_H_ */
Commit	Line	Data
8187a2b7 ZN	1	#ifndef _INTEL_RINGBUFFER_H_
	2	#define _INTEL_RINGBUFFER_H_
	3
633cf8f5 VS	4	/*
	5	* Gen2 BSpec "1. Programming Environment" / 1.4.4.6 "Ring Buffer Use"
	6	* Gen3 BSpec "vol1c Memory Interface Functions" / 2.3.4.5 "Ring Buffer Use"
	7	* Gen4+ BSpec "vol1c Memory Interface and Command Stream" / 5.3.4.5 "Ring Buffer Use"
	8	*
	9	* "If the Ring Buffer Head Pointer and the Tail Pointer are on the same
	10	* cacheline, the Head Pointer must not be greater than the Tail
	11	* Pointer."
	12	*/
	13	#define I915_RING_FREE_SPACE 64
	14
8187a2b7	15	struct intel_hw_status_page {
4225d0f2	16	u32 *page_addr;
8187a2b7	17	unsigned int gfx_addr;
05394f39	18	struct drm_i915_gem_object *obj;
8187a2b7 ZN	19	};
8187a2b7 ZN	20
b7287d80 BW	21	#define I915_READ_TAIL(ring) I915_READ(RING_TAIL((ring)->mmio_base))
b7287d80 BW	22	#define I915_WRITE_TAIL(ring, val) I915_WRITE(RING_TAIL((ring)->mmio_base), val)
cae5852d	23
b7287d80 BW	24	#define I915_READ_START(ring) I915_READ(RING_START((ring)->mmio_base))
b7287d80 BW	25	#define I915_WRITE_START(ring, val) I915_WRITE(RING_START((ring)->mmio_base), val)
cae5852d	26
b7287d80 BW	27	#define I915_READ_HEAD(ring) I915_READ(RING_HEAD((ring)->mmio_base))
b7287d80 BW	28	#define I915_WRITE_HEAD(ring, val) I915_WRITE(RING_HEAD((ring)->mmio_base), val)
cae5852d	29
b7287d80 BW	30	#define I915_READ_CTL(ring) I915_READ(RING_CTL((ring)->mmio_base))
b7287d80 BW	31	#define I915_WRITE_CTL(ring, val) I915_WRITE(RING_CTL((ring)->mmio_base), val)
cae5852d	32
b7287d80 BW	33	#define I915_READ_IMR(ring) I915_READ(RING_IMR((ring)->mmio_base))
b7287d80 BW	34	#define I915_WRITE_IMR(ring, val) I915_WRITE(RING_IMR((ring)->mmio_base), val)
870e86dd	35
f2f4d82f	36	enum intel_ring_hangcheck_action {
da661464	37	HANGCHECK_IDLE = 0,
f2f4d82f JN	38	HANGCHECK_WAIT,
	39	HANGCHECK_ACTIVE,
	40	HANGCHECK_KICK,
	41	HANGCHECK_HUNG,
	42	};
ad8beaea	43
b6b0fac0 MK	44	#define HANGCHECK_SCORE_RING_HUNG 31
b6b0fac0 MK	45
92cab734	46	struct intel_ring_hangcheck {
6274f212	47	bool deadlock;
92cab734	48	u32 seqno;
05407ff8 MK	49	u32 acthd;
05407ff8 MK	50	int score;
ad8beaea	51	enum intel_ring_hangcheck_action action;
92cab734 MK	52	};
92cab734 MK	53
8187a2b7 ZN	54	struct intel_ring_buffer {
8187a2b7 ZN	55	const char *name;
9220434a	56	enum intel_ring_id {
96154f2f DV	57	RCS = 0x0,
	58	VCS,
	59	BCS,
4a3dd19d	60	VECS,
9220434a	61	} id;
4a3dd19d	62	#define I915_NUM_RINGS 4
333e9fe9	63	u32 mmio_base;
311bd68e	64	void __iomem *virtual_start;
8187a2b7	65	struct drm_device *dev;
05394f39	66	struct drm_i915_gem_object *obj;
8187a2b7	67
8c0a6bfe CW	68	u32 head;
8c0a6bfe CW	69	u32 tail;
780f0ca3	70	int space;
c2c347a9	71	int size;
55249baa	72	int effective_size;
8187a2b7 ZN	73	struct intel_hw_status_page status_page;
8187a2b7 ZN	74
a71d8d94 CW	75	/** We track the position of the requests in the ring buffer, and
	76	* when each is retired we increment last_retired_head as the GPU
	77	* must have finished processing the request and so we know we
	78	* can advance the ringbuffer up to that position.
	79	*
	80	* last_retired_head is set to -1 after the value is consumed so
	81	* we can detect new retirements.
	82	*/
	83	u32 last_retired_head;
	84
c7113cc3	85	unsigned irq_refcount; /* protected by dev_priv->irq_lock */
6a848ccb	86	u32 irq_enable_mask; /* bitmask to enable ring interrupt */
db53a302	87	u32 trace_irq_seqno;
1ec14ad3	88	u32 sync_seqno[I915_NUM_RINGS-1];
b13c2b96	89	bool __must_check (irq_get)(struct intel_ring_buffer ring);
1ec14ad3	90	void (irq_put)(struct intel_ring_buffer ring);
8187a2b7	91
78501eac	92	int (init)(struct intel_ring_buffer ring);
8187a2b7	93
78501eac	94	void (write_tail)(struct intel_ring_buffer ring,
297b0c5b	95	u32 value);
b72f3acb CW	96	int __must_check (flush)(struct intel_ring_buffer ring,
	97	u32 invalidate_domains,
	98	u32 flush_domains);
9d773091	99	int (add_request)(struct intel_ring_buffer ring);
b2eadbc8 CW	100	/* Some chipsets are not quite as coherent as advertised and need
	101	* an expensive kick to force a true read of the up-to-date seqno.
	102	* However, the up-to-date seqno is not always required and the last
	103	* seen value is good enough. Note that the seqno will always be
	104	* monotonic, even if not coherent.
	105	*/
	106	u32 (get_seqno)(struct intel_ring_buffer ring,
	107	bool lazy_coherency);
b70ec5bf MK	108	void (set_seqno)(struct intel_ring_buffer ring,
b70ec5bf MK	109	u32 seqno);
78501eac	110	int (dispatch_execbuffer)(struct intel_ring_buffer ring,
d7d4eedd CW	111	u32 offset, u32 length,
	112	unsigned flags);
	113	#define I915_DISPATCH_SECURE 0x1
b45305fc	114	#define I915_DISPATCH_PINNED 0x2
8d19215b	115	void (cleanup)(struct intel_ring_buffer ring);
c8c99b0f BW	116	int (sync_to)(struct intel_ring_buffer ring,
	117	struct intel_ring_buffer *to,
	118	u32 seqno);
ad776f8b	119
5586181f BW	120	/* our mbox written by others */
5586181f BW	121	u32 semaphore_register[I915_NUM_RINGS];
ad776f8b BW	122	/* mboxes this ring signals to */
	123	u32 signal_mbox[I915_NUM_RINGS];
	124
8187a2b7 ZN	125	/**
	126	* List of objects currently involved in rendering from the
	127	* ringbuffer.
	128	*
	129	* Includes buffers having the contents of their GPU caches
	130	* flushed, not necessarily primitives. last_rendering_seqno
	131	* represents when the rendering involved will be completed.
	132	*
	133	* A reference is held on the buffer while on this list.
	134	*/
	135	struct list_head active_list;
	136
	137	/**
	138	* List of breadcrumbs associated with GPU requests currently
	139	* outstanding.
	140	*/
	141	struct list_head request_list;
	142
a56ba56c CW	143	/**
	144	* Do we have some not yet emitted requests outstanding?
	145	*/
3c0e234c	146	struct drm_i915_gem_request *preallocated_lazy_request;
1823521d	147	u32 outstanding_lazy_seqno;
cc889e0f	148	bool gpu_caches_dirty;
c65355bb	149	bool fbc_dirty;
a56ba56c	150
8187a2b7	151	wait_queue_head_t irq_queue;
8d19215b	152
12b0286f BW	153	/**
	154	* Do an explicit TLB flush before MI_SET_CONTEXT
	155	*/
	156	bool itlb_before_ctx_switch;
40521054	157	struct i915_hw_context *default_context;
112522f6	158	struct i915_hw_context *last_context;
40521054	159
92cab734 MK	160	struct intel_ring_hangcheck hangcheck;
92cab734 MK	161
0d1aacac CW	162	struct {
	163	struct drm_i915_gem_object *obj;
	164	u32 gtt_offset;
	165	volatile u32 *cpu_page;
	166	} scratch;
8187a2b7 ZN	167	};
8187a2b7 ZN	168
b4519513 CW	169	static inline bool
	170	intel_ring_initialized(struct intel_ring_buffer *ring)
	171	{
	172	return ring->obj != NULL;
	173	}
	174
96154f2f DV	175	static inline unsigned
	176	intel_ring_flag(struct intel_ring_buffer *ring)
	177	{
	178	return 1 << ring->id;
	179	}
	180
1ec14ad3 CW	181	static inline u32
	182	intel_ring_sync_index(struct intel_ring_buffer *ring,
	183	struct intel_ring_buffer *other)
	184	{
	185	int idx;
	186
	187	/*
	188	* cs -> 0 = vcs, 1 = bcs
	189	* vcs -> 0 = bcs, 1 = cs,
	190	* bcs -> 0 = cs, 1 = vcs.
	191	*/
	192
	193	idx = (other - ring) - 1;
	194	if (idx < 0)
	195	idx += I915_NUM_RINGS;
	196
	197	return idx;
	198	}
	199
8187a2b7 ZN	200	static inline u32
8187a2b7 ZN	201	intel_read_status_page(struct intel_ring_buffer *ring,
78501eac	202	int reg)
8187a2b7	203	{
4225d0f2 DV	204	/* Ensure that the compiler doesn't optimize away the load. */
	205	barrier();
	206	return ring->status_page.page_addr[reg];
8187a2b7 ZN	207	}
8187a2b7 ZN	208
b70ec5bf MK	209	static inline void
	210	intel_write_status_page(struct intel_ring_buffer *ring,
	211	int reg, u32 value)
	212	{
	213	ring->status_page.page_addr[reg] = value;
	214	}
	215
311bd68e CW	216	/**
	217	* Reads a dword out of the status page, which is written to from the command
	218	* queue by automatic updates, MI_REPORT_HEAD, MI_STORE_DATA_INDEX, or
	219	* MI_STORE_DATA_IMM.
	220	*
	221	* The following dwords have a reserved meaning:
	222	* 0x00: ISR copy, updated when an ISR bit not set in the HWSTAM changes.
	223	* 0x04: ring 0 head pointer
	224	* 0x05: ring 1 head pointer (915-class)
	225	* 0x06: ring 2 head pointer (915-class)
	226	* 0x10-0x1b: Context status DWords (GM45)
	227	* 0x1f: Last written status offset. (GM45)
	228	*
	229	* The area from dword 0x20 to 0x3ff is available for driver usage.
	230	*/
311bd68e	231	#define I915_GEM_HWS_INDEX 0x20
9a289771 JB	232	#define I915_GEM_HWS_SCRATCH_INDEX 0x30
9a289771 JB	233	#define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)
311bd68e	234
78501eac	235	void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring);
96f298aa	236
e1f99ce6	237	int __must_check intel_ring_begin(struct intel_ring_buffer *ring, int n);
78501eac CW	238	static inline void intel_ring_emit(struct intel_ring_buffer *ring,
78501eac CW	239	u32 data)
e898cd22	240	{
78501eac	241	iowrite32(data, ring->virtual_start + ring->tail);
e898cd22 CW	242	ring->tail += 4;
e898cd22 CW	243	}
09246732 CW	244	static inline void intel_ring_advance(struct intel_ring_buffer *ring)
	245	{
	246	ring->tail &= ring->size - 1;
	247	}
	248	void __intel_ring_advance(struct intel_ring_buffer *ring);
	249
3e960501	250	int __must_check intel_ring_idle(struct intel_ring_buffer *ring);
f7e98ad4	251	void intel_ring_init_seqno(struct intel_ring_buffer *ring, u32 seqno);
a7b9761d CW	252	int intel_ring_flush_all_caches(struct intel_ring_buffer *ring);
a7b9761d CW	253	int intel_ring_invalidate_all_caches(struct intel_ring_buffer *ring);
8187a2b7	254
5c1143bb XH	255	int intel_init_render_ring_buffer(struct drm_device *dev);
5c1143bb XH	256	int intel_init_bsd_ring_buffer(struct drm_device *dev);
549f7365	257	int intel_init_blt_ring_buffer(struct drm_device *dev);
9a8a2213	258	int intel_init_vebox_ring_buffer(struct drm_device *dev);
8187a2b7	259
78501eac CW	260	u32 intel_ring_get_active_head(struct intel_ring_buffer *ring);
78501eac CW	261	void intel_ring_setup_status_page(struct intel_ring_buffer *ring);
79f321b7	262
a71d8d94 CW	263	static inline u32 intel_ring_get_tail(struct intel_ring_buffer *ring)
	264	{
	265	return ring->tail;
	266	}
	267
9d773091 CW	268	static inline u32 intel_ring_get_seqno(struct intel_ring_buffer *ring)
9d773091 CW	269	{
1823521d CW	270	BUG_ON(ring->outstanding_lazy_seqno == 0);
1823521d CW	271	return ring->outstanding_lazy_seqno;
9d773091 CW	272	}
9d773091 CW	273
db53a302 CW	274	static inline void i915_trace_irq_get(struct intel_ring_buffer *ring, u32 seqno)
	275	{
	276	if (ring->trace_irq_seqno == 0 && ring->irq_get(ring))
	277	ring->trace_irq_seqno = seqno;
	278	}
	279
e8616b6c CW	280	/* DRI warts */
	281	int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size);
	282
8187a2b7	283	#endif /* _INTEL_RINGBUFFER_H_ */