drm/i915: take a reference to uncore in the engine and use it

A few advantages:

- Prepares us for the planned split of display uncore from GT uncore

- Improves our engine-centric view of the world in the engine code
  and allows us to avoid jumping back to dev_priv.

- Allows us to wrap accesses to engine register in nice macros that
  automatically pick the right mmio base.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190325214940.23632-10-daniele.ceraolospurio@intel.com

drivers/gpu/drm/i915/gvt/handlers.c

@@ -1848,7 +1848,7 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
 	MMIO_DH(GEN7_SC_INSTDONE, D_BDW_PLUS, mmio_read_from_hw, NULL);
 
 	MMIO_GM_RDR(_MMIO(0x2148), D_ALL, NULL, NULL);
-	MMIO_GM_RDR(CCID, D_ALL, NULL, NULL);
+	MMIO_GM_RDR(CCID(RENDER_RING_BASE), D_ALL, NULL, NULL);
 	MMIO_GM_RDR(_MMIO(0x12198), D_ALL, NULL, NULL);
 	MMIO_D(GEN7_CXT_SIZE, D_ALL);
 

drivers/gpu/drm/i915/i915_debugfs.c

@@ -880,7 +880,7 @@ static int i915_interrupt_info(struct seq_file *m, void *data)
 		for_each_engine(engine, dev_priv, id) {
 			seq_printf(m,
 				   "Graphics Interrupt mask (%s):	%08x\n",
-				   engine->name, I915_READ_IMR(engine));
+				   engine->name, ENGINE_READ(engine, RING_IMR));
 		}
 	}
 

drivers/gpu/drm/i915/i915_gpu_error.c

@@ -1136,7 +1136,7 @@ static void error_record_engine_registers(struct i915_gpu_state *error,
 	struct drm_i915_private *dev_priv = engine->i915;
 
 	if (INTEL_GEN(dev_priv) >= 6) {
-		ee->rc_psmi = I915_READ(RING_PSMI_CTL(engine->mmio_base));
+		ee->rc_psmi = ENGINE_READ(engine, RING_PSMI_CTL);
 		if (INTEL_GEN(dev_priv) >= 8)
 			ee->fault_reg = I915_READ(GEN8_RING_FAULT_REG);
 		else
@@ -1144,32 +1144,32 @@ static void error_record_engine_registers(struct i915_gpu_state *error,
 	}
 
 	if (INTEL_GEN(dev_priv) >= 4) {
-		ee->faddr = I915_READ(RING_DMA_FADD(engine->mmio_base));
-		ee->ipeir = I915_READ(RING_IPEIR(engine->mmio_base));
-		ee->ipehr = I915_READ(RING_IPEHR(engine->mmio_base));
-		ee->instps = I915_READ(RING_INSTPS(engine->mmio_base));
-		ee->bbaddr = I915_READ(RING_BBADDR(engine->mmio_base));
+		ee->faddr = ENGINE_READ(engine, RING_DMA_FADD);
+		ee->ipeir = ENGINE_READ(engine, RING_IPEIR);
+		ee->ipehr = ENGINE_READ(engine, RING_IPEHR);
+		ee->instps = ENGINE_READ(engine, RING_INSTPS);
+		ee->bbaddr = ENGINE_READ(engine, RING_BBADDR);
 		if (INTEL_GEN(dev_priv) >= 8) {
-			ee->faddr |= (u64) I915_READ(RING_DMA_FADD_UDW(engine->mmio_base)) << 32;
-			ee->bbaddr |= (u64) I915_READ(RING_BBADDR_UDW(engine->mmio_base)) << 32;
+			ee->faddr |= (u64)ENGINE_READ(engine, RING_DMA_FADD_UDW) << 32;
+			ee->bbaddr |= (u64)ENGINE_READ(engine, RING_BBADDR_UDW) << 32;
 		}
-		ee->bbstate = I915_READ(RING_BBSTATE(engine->mmio_base));
+		ee->bbstate = ENGINE_READ(engine, RING_BBSTATE);
 	} else {
-		ee->faddr = I915_READ(DMA_FADD_I8XX);
-		ee->ipeir = I915_READ(IPEIR);
-		ee->ipehr = I915_READ(IPEHR);
+		ee->faddr = ENGINE_READ(engine, DMA_FADD_I8XX);
+		ee->ipeir = ENGINE_READ(engine, IPEIR);
+		ee->ipehr = ENGINE_READ(engine, IPEHR);
 	}
 
 	intel_engine_get_instdone(engine, &ee->instdone);
 
-	ee->instpm = I915_READ(RING_INSTPM(engine->mmio_base));
+	ee->instpm = ENGINE_READ(engine, RING_INSTPM);
 	ee->acthd = intel_engine_get_active_head(engine);
-	ee->start = I915_READ_START(engine);
-	ee->head = I915_READ_HEAD(engine);
-	ee->tail = I915_READ_TAIL(engine);
-	ee->ctl = I915_READ_CTL(engine);
+	ee->start = ENGINE_READ(engine, RING_START);
+	ee->head = ENGINE_READ(engine, RING_HEAD);
+	ee->tail = ENGINE_READ(engine, RING_TAIL);
+	ee->ctl = ENGINE_READ(engine, RING_CTL);
 	if (INTEL_GEN(dev_priv) > 2)
-		ee->mode = I915_READ_MODE(engine);
+		ee->mode = ENGINE_READ(engine, RING_MI_MODE);
 
 	if (!HWS_NEEDS_PHYSICAL(dev_priv)) {
 		i915_reg_t mmio;
@@ -1214,10 +1214,10 @@ static void error_record_engine_registers(struct i915_gpu_state *error,
 
 		if (IS_GEN(dev_priv, 6))
 			ee->vm_info.pp_dir_base =
-				I915_READ(RING_PP_DIR_BASE_READ(engine));
+				ENGINE_READ(engine, RING_PP_DIR_BASE_READ);
 		else if (IS_GEN(dev_priv, 7))
 			ee->vm_info.pp_dir_base =
-				I915_READ(RING_PP_DIR_BASE(engine));
+					ENGINE_READ(engine, RING_PP_DIR_BASE);
 		else if (INTEL_GEN(dev_priv) >= 8)
 			for (i = 0; i < 4; i++) {
 				ee->vm_info.pdp[i] =
@@ -1601,7 +1601,7 @@ static void capture_reg_state(struct i915_gpu_state *error)
 	}
 
 	if (INTEL_GEN(dev_priv) >= 5)
-		error->ccid = I915_READ(CCID);
+		error->ccid = I915_READ(CCID(RENDER_RING_BASE));
 
 	/* 3: Feature specific registers */
 	if (IS_GEN_RANGE(dev_priv, 6, 7)) {

drivers/gpu/drm/i915/i915_reg.h

@@ -434,9 +434,9 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define GEN11_VECS_SFC_USAGE(engine)		_MMIO((engine)->mmio_base + 0x2014)
 #define   GEN11_VECS_SFC_USAGE_BIT		(1 << 0)
 
-#define RING_PP_DIR_BASE(engine)	_MMIO((engine)->mmio_base + 0x228)
-#define RING_PP_DIR_BASE_READ(engine)	_MMIO((engine)->mmio_base + 0x518)
-#define RING_PP_DIR_DCLV(engine)	_MMIO((engine)->mmio_base + 0x220)
+#define RING_PP_DIR_BASE(base)		_MMIO((base) + 0x228)
+#define RING_PP_DIR_BASE_READ(base)	_MMIO((base) + 0x518)
+#define RING_PP_DIR_DCLV(base)		_MMIO((base) + 0x220)
 #define   PP_DIR_DCLV_2G		0xffffffff
 
 #define GEN8_RING_PDP_UDW(engine, n)	_MMIO((engine)->mmio_base + 0x270 + (n) * 8 + 4)
@@ -2568,12 +2568,12 @@ enum i915_power_well_id {
 #define HWS_START_ADDRESS_SHIFT	4
 #define PWRCTXA		_MMIO(0x2088) /* 965GM+ only */
 #define   PWRCTX_EN	(1 << 0)
-#define IPEIR		_MMIO(0x2088)
-#define IPEHR		_MMIO(0x208c)
+#define IPEIR(base)	_MMIO((base) + 0x88)
+#define IPEHR(base)	_MMIO((base) + 0x8c)
 #define GEN2_INSTDONE	_MMIO(0x2090)
 #define NOPID		_MMIO(0x2094)
 #define HWSTAM		_MMIO(0x2098)
-#define DMA_FADD_I8XX	_MMIO(0x20d0)
+#define DMA_FADD_I8XX(base)	_MMIO((base) + 0xd0)
 #define RING_BBSTATE(base)	_MMIO((base) + 0x110)
 #define   RING_BB_PPGTT		(1 << 5)
 #define RING_SBBADDR(base)	_MMIO((base) + 0x114) /* hsw+ */
@@ -2747,7 +2747,7 @@ enum i915_power_well_id {
 #define   INSTPM_FORCE_ORDERING				(1 << 7) /* GEN6+ */
 #define   INSTPM_TLB_INVALIDATE	(1 << 9)
 #define   INSTPM_SYNC_FLUSH	(1 << 5)
-#define ACTHD	        _MMIO(0x20c8)
+#define ACTHD(base)	_MMIO((base) + 0xc8)
 #define MEM_MODE	_MMIO(0x20cc)
 #define   MEM_DISPLAY_B_TRICKLE_FEED_DISABLE (1 << 3) /* 830 only */
 #define   MEM_DISPLAY_A_TRICKLE_FEED_DISABLE (1 << 2) /* 830/845 only */
@@ -3947,7 +3947,7 @@ enum i915_power_well_id {
 /*
  * Logical Context regs
  */
-#define CCID				_MMIO(0x2180)
+#define CCID(base)			_MMIO((base) + 0x180)
 #define   CCID_EN			BIT(0)
 #define   CCID_EXTENDED_STATE_RESTORE	BIT(2)
 #define   CCID_EXTENDED_STATE_SAVE	BIT(3)

drivers/gpu/drm/i915/i915_reset.c

@@ -1173,19 +1173,24 @@ static void i915_reset_device(struct drm_i915_private *i915,
 		kobject_uevent_env(kobj, KOBJ_CHANGE, reset_done_event);
 }
 
+static void clear_register(struct drm_i915_private *dev_priv, i915_reg_t reg)
+{
+	I915_WRITE(reg, I915_READ(reg));
+}
+
 void i915_clear_error_registers(struct drm_i915_private *dev_priv)
 {
 	u32 eir;
 
 	if (!IS_GEN(dev_priv, 2))
-		I915_WRITE(PGTBL_ER, I915_READ(PGTBL_ER));
+		clear_register(dev_priv, PGTBL_ER);
 
 	if (INTEL_GEN(dev_priv) < 4)
-		I915_WRITE(IPEIR, I915_READ(IPEIR));
+		clear_register(dev_priv, IPEIR(RENDER_RING_BASE));
 	else
-		I915_WRITE(IPEIR_I965, I915_READ(IPEIR_I965));
+		clear_register(dev_priv, IPEIR_I965);
 
-	I915_WRITE(EIR, I915_READ(EIR));
+	clear_register(dev_priv, EIR);
 	eir = I915_READ(EIR);
 	if (eir) {
 		/*

drivers/gpu/drm/i915/intel_engine_types.h

@@ -29,6 +29,7 @@ struct drm_i915_reg_table;
 struct i915_gem_context;
 struct i915_request;
 struct i915_sched_attr;
+struct intel_uncore;
 
 struct intel_hw_status_page {
 	struct i915_vma *vma;
@@ -250,6 +251,7 @@ struct intel_engine_execlists {
 
 struct intel_engine_cs {
 	struct drm_i915_private *i915;
+	struct intel_uncore *uncore;
 	char name[INTEL_ENGINE_CS_MAX_NAME];
 
 	enum intel_engine_id id;

drivers/gpu/drm/i915/intel_hangcheck.c

@@ -118,11 +118,11 @@ engine_stuck(struct intel_engine_cs *engine, u64 acthd)
 	 * and break the hang. This should work on
 	 * all but the second generation chipsets.
 	 */
-	tmp = I915_READ_CTL(engine);
+	tmp = ENGINE_READ(engine, RING_CTL);
 	if (tmp & RING_WAIT) {
 		i915_handle_error(dev_priv, engine->mask, 0,
 				  "stuck wait on %s", engine->name);
-		I915_WRITE_CTL(engine, tmp);
+		ENGINE_WRITE(engine, RING_CTL, tmp);
 		return ENGINE_WAIT_KICK;
 	}
 

drivers/gpu/drm/i915/intel_lrc.c

@@ -2074,16 +2074,14 @@ static int gen8_emit_bb_start(struct i915_request *rq,
 
 static void gen8_logical_ring_enable_irq(struct intel_engine_cs *engine)
 {
-	struct drm_i915_private *dev_priv = engine->i915;
-	I915_WRITE_IMR(engine,
-		       ~(engine->irq_enable_mask | engine->irq_keep_mask));
-	POSTING_READ_FW(RING_IMR(engine->mmio_base));
+	ENGINE_WRITE(engine, RING_IMR,
+		     ~(engine->irq_enable_mask | engine->irq_keep_mask));
+	ENGINE_POSTING_READ(engine, RING_IMR);
 }
 
 static void gen8_logical_ring_disable_irq(struct intel_engine_cs *engine)
 {
-	struct drm_i915_private *dev_priv = engine->i915;
-	I915_WRITE_IMR(engine, ~engine->irq_keep_mask);
+	ENGINE_WRITE(engine, RING_IMR, ~engine->irq_keep_mask);
 }
 
 static int gen8_emit_flush(struct i915_request *request, u32 mode)
@@ -2288,7 +2286,7 @@ void intel_logical_ring_cleanup(struct intel_engine_cs *engine)
 	dev_priv = engine->i915;
 
 	if (engine->buffer) {
-		WARN_ON((I915_READ_MODE(engine) & MODE_IDLE) == 0);
+		WARN_ON((ENGINE_READ(engine, RING_MI_MODE) & MODE_IDLE) == 0);
 	}
 
 	if (engine->cleanup)
@@ -2400,6 +2398,7 @@ static int logical_ring_init(struct intel_engine_cs *engine)
 {
 	struct drm_i915_private *i915 = engine->i915;
 	struct intel_engine_execlists * const execlists = &engine->execlists;
+	u32 base = engine->mmio_base;
 	int ret;
 
 	ret = intel_engine_init_common(engine);
@@ -2410,12 +2409,12 @@ static int logical_ring_init(struct intel_engine_cs *engine)
 
 	if (HAS_LOGICAL_RING_ELSQ(i915)) {
 		execlists->submit_reg = i915->uncore.regs +
-			i915_mmio_reg_offset(RING_EXECLIST_SQ_CONTENTS(engine));
+			i915_mmio_reg_offset(RING_EXECLIST_SQ_CONTENTS(base));
 		execlists->ctrl_reg = i915->uncore.regs +
-			i915_mmio_reg_offset(RING_EXECLIST_CONTROL(engine));
+			i915_mmio_reg_offset(RING_EXECLIST_CONTROL(base));
 	} else {
 		execlists->submit_reg = i915->uncore.regs +
-			i915_mmio_reg_offset(RING_ELSP(engine));
+			i915_mmio_reg_offset(RING_ELSP(base));
 	}
 
 	execlists->preempt_complete_status = ~0u;
@@ -2658,7 +2657,7 @@ static void execlists_init_reg_state(u32 *regs,
 	regs[CTX_LRI_HEADER_0] = MI_LOAD_REGISTER_IMM(rcs ? 14 : 11) |
 				 MI_LRI_FORCE_POSTED;
 
-	CTX_REG(regs, CTX_CONTEXT_CONTROL, RING_CONTEXT_CONTROL(engine),
+	CTX_REG(regs, CTX_CONTEXT_CONTROL, RING_CONTEXT_CONTROL(base),
 		_MASKED_BIT_DISABLE(CTX_CTRL_ENGINE_CTX_RESTORE_INHIBIT) |
 		_MASKED_BIT_ENABLE(CTX_CTRL_INHIBIT_SYN_CTX_SWITCH));
 	if (INTEL_GEN(engine->i915) < 11) {

drivers/gpu/drm/i915/intel_lrc.h

@@ -28,20 +28,20 @@
 #include "i915_gem_context.h"
 
 /* Execlists regs */
-#define RING_ELSP(engine)			_MMIO((engine)->mmio_base + 0x230)
-#define RING_EXECLIST_STATUS_LO(engine)		_MMIO((engine)->mmio_base + 0x234)
-#define RING_EXECLIST_STATUS_HI(engine)		_MMIO((engine)->mmio_base + 0x234 + 4)
-#define RING_CONTEXT_CONTROL(engine)		_MMIO((engine)->mmio_base + 0x244)
+#define RING_ELSP(base)				_MMIO((base) + 0x230)
+#define RING_EXECLIST_STATUS_LO(base)		_MMIO((base) + 0x234)
+#define RING_EXECLIST_STATUS_HI(base)		_MMIO((base) + 0x234 + 4)
+#define RING_CONTEXT_CONTROL(base)		_MMIO((base) + 0x244)
 #define	  CTX_CTRL_INHIBIT_SYN_CTX_SWITCH	(1 << 3)
 #define	  CTX_CTRL_ENGINE_CTX_RESTORE_INHIBIT	(1 << 0)
-#define   CTX_CTRL_RS_CTX_ENABLE                (1 << 1)
+#define   CTX_CTRL_RS_CTX_ENABLE		(1 << 1)
 #define	  CTX_CTRL_ENGINE_CTX_SAVE_INHIBIT	(1 << 2)
-#define RING_CONTEXT_STATUS_BUF_BASE(engine)	_MMIO((engine)->mmio_base + 0x370)
-#define RING_CONTEXT_STATUS_BUF_LO(engine, i)	_MMIO((engine)->mmio_base + 0x370 + (i) * 8)
-#define RING_CONTEXT_STATUS_BUF_HI(engine, i)	_MMIO((engine)->mmio_base + 0x370 + (i) * 8 + 4)
-#define RING_CONTEXT_STATUS_PTR(engine)		_MMIO((engine)->mmio_base + 0x3a0)
-#define RING_EXECLIST_SQ_CONTENTS(engine)	_MMIO((engine)->mmio_base + 0x510)
-#define RING_EXECLIST_CONTROL(engine)		_MMIO((engine)->mmio_base + 0x550)
+#define RING_CONTEXT_STATUS_BUF_BASE(base)	_MMIO((base) + 0x370)
+#define RING_CONTEXT_STATUS_BUF_LO(base, i)	_MMIO((base) + 0x370 + (i) * 8)
+#define RING_CONTEXT_STATUS_BUF_HI(base, i)	_MMIO((base) + 0x370 + (i) * 8 + 4)
+#define RING_CONTEXT_STATUS_PTR(base)		_MMIO((base) + 0x3a0)
+#define RING_EXECLIST_SQ_CONTENTS(base)		_MMIO((base) + 0x510)
+#define RING_EXECLIST_CONTROL(base)		_MMIO((base) + 0x550)
 #define	  EL_CTRL_LOAD				(1 << 0)
 
 /* The docs specify that the write pointer wraps around after 5h, "After status

drivers/gpu/drm/i915/intel_ringbuffer.h

@@ -29,23 +29,44 @@ struct drm_printer;
 #define CACHELINE_BYTES 64
 #define CACHELINE_DWORDS (CACHELINE_BYTES / sizeof(u32))
 
-#define I915_READ_TAIL(engine) I915_READ(RING_TAIL((engine)->mmio_base))
-#define I915_WRITE_TAIL(engine, val) I915_WRITE(RING_TAIL((engine)->mmio_base), val)
+/*
+ * The register defines to be used with the following macros need to accept a
+ * base param, e.g:
+ *
+ * REG_FOO(base) _MMIO((base) + <relative offset>)
+ * ENGINE_READ(engine, REG_FOO);
+ *
+ * register arrays are to be defined and accessed as follows:
+ *
+ * REG_BAR(base, i) _MMIO((base) + <relative offset> + (i) * <shift>)
+ * ENGINE_READ_IDX(engine, REG_BAR, i)
+ */
+
+#define __ENGINE_REG_OP(op__, engine__, ...) \
+	intel_uncore_##op__((engine__)->uncore, __VA_ARGS__)
+
+#define __ENGINE_READ_OP(op__, engine__, reg__) \
+	__ENGINE_REG_OP(op__, (engine__), reg__((engine__)->mmio_base))
 
-#define I915_READ_START(engine) I915_READ(RING_START((engine)->mmio_base))
-#define I915_WRITE_START(engine, val) I915_WRITE(RING_START((engine)->mmio_base), val)
+#define ENGINE_READ16(...)	__ENGINE_READ_OP(read16, __VA_ARGS__)
+#define ENGINE_READ(...)	__ENGINE_READ_OP(read, __VA_ARGS__)
+#define ENGINE_READ_FW(...)	__ENGINE_READ_OP(read_fw, __VA_ARGS__)
+#define ENGINE_POSTING_READ(...) __ENGINE_READ_OP(posting_read, __VA_ARGS__)
 
-#define I915_READ_HEAD(engine)  I915_READ(RING_HEAD((engine)->mmio_base))
-#define I915_WRITE_HEAD(engine, val) I915_WRITE(RING_HEAD((engine)->mmio_base), val)
+#define ENGINE_READ64(engine__, lower_reg__, upper_reg__) \
+	__ENGINE_REG_OP(read64_2x32, (engine__), \
+			lower_reg__((engine__)->mmio_base), \
+			upper_reg__((engine__)->mmio_base))
 
-#define I915_READ_CTL(engine) I915_READ(RING_CTL((engine)->mmio_base))
-#define I915_WRITE_CTL(engine, val) I915_WRITE(RING_CTL((engine)->mmio_base), val)
+#define ENGINE_READ_IDX(engine__, reg__, idx__) \
+	__ENGINE_REG_OP(read, (engine__), reg__((engine__)->mmio_base, (idx__)))
 
-#define I915_READ_IMR(engine) I915_READ(RING_IMR((engine)->mmio_base))
-#define I915_WRITE_IMR(engine, val) I915_WRITE(RING_IMR((engine)->mmio_base), val)
+#define __ENGINE_WRITE_OP(op__, engine__, reg__, val__) \
+	__ENGINE_REG_OP(op__, (engine__), reg__((engine__)->mmio_base), (val__))
 
-#define I915_READ_MODE(engine) I915_READ(RING_MI_MODE((engine)->mmio_base))
-#define I915_WRITE_MODE(engine, val) I915_WRITE(RING_MI_MODE((engine)->mmio_base), val)
+#define ENGINE_WRITE16(...)	__ENGINE_WRITE_OP(write16, __VA_ARGS__)
+#define ENGINE_WRITE(...)	__ENGINE_WRITE_OP(write, __VA_ARGS__)
+#define ENGINE_WRITE_FW(...)	__ENGINE_WRITE_OP(write_fw, __VA_ARGS__)
 
 /* seqno size is actually only a uint32, but since we plan to use MI_FLUSH_DW to
  * do the writes, and that must have qw aligned offsets, simply pretend it's 8b.

drivers/gpu/drm/i915/intel_uncore.c

@@ -1814,7 +1814,7 @@ int __intel_wait_for_register_fw(struct intel_uncore *uncore,
 
 /**
  * __intel_wait_for_register - wait until register matches expected state
- * @dev_priv: the i915 device
+ * @uncore: the struct intel_uncore
  * @reg: the register to read
  * @mask: mask to apply to register value
  * @value: expected value