Merge branch 'drm-intel-fixes' of git://people.freedesktop.org/~danve…

…t/drm-intel into drm-next

Daniel writes: (this pull is the one with the bad patch dropped)
First pile of fixes for 3.6 already, and I'm afraid it's a bit larger than
what I'd wish for. But I've moved all the feature-y stuff to -next, so
this really is all -fixes. Most of it is handling fallout from the hw
context stuff, discovered now that mesa git has started using them for
real. Otherwise all just small fixes:
- unbreak modeset=0 on gen6+ (regressed in next)
- const mismatch fix for ->mode_fixup
- simplify overly clever lvds modeset code (current code can totally
  confuse backlights, resulting in broken panels until a full power draw
  restores them).
- fix some fallout from the flushing_list disabling (regression only
  introduced in -next)
- DP link train improvements (this also kills the last 3.2 dp regression
  afaik)
- bugfix for the new ddc VGA detection on newer platforms
- minor backlight fixes (one of them a -next regression)
- only enable the required PM interrupts (to avoid waking up the cpu
  unnecessarily)
- some really minor bits (workaround clarification, make coverty happy,
  hsw init fix)
* 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel: (23 commits)
  drm/i915: unbreak lastclose for failed driver init
  drm/i915: Set the context before setting up regs for the context.
  drm/i915: constify mode in crtc_mode_fixup
  drm/i915/lvds: ditch ->prepare special case
  drm/i915: dereferencing an error pointer
  drm/i915: fix invalid reference handling of the default ctx obj
  drm/i915: Add -EIO to the list of known errors for __wait_seqno
  drm/i915: Flush the context object from the CPU caches upon switching
  drm/i915: Make the lock for pageflips interruptible
  drm/i915: don't forget the PCH backlight registers
  drm/i915: Insert a flush between batches if the breadcrumb was dropped
  drm/i915: missing error case in init status page
  drm/i915: mask tiled bit when updating ILK sprites
  drm/i915: try to train DP even harder
  drm/i915: kill intel_ddc_probe
  drm/i915: check whether we actually received an edid in detect_ddc
  drm/i915: fix up PCH backlight #define mixup
  drm/i915: Add comments to explain the BSD tail write workaround
  drm/i915: Disable the BLT on pre-production SNB hardware
  drm/i915: initialize power wells in modeset_init_hw
  ...

drivers/gpu/drm/i915/i915_debugfs.c

@@ -676,6 +676,7 @@ static void i915_ring_error_state(struct seq_file *m,
 	seq_printf(m, "  INSTPM: 0x%08x\n", error->instpm[ring]);
 	seq_printf(m, "  FADDR: 0x%08x\n", error->faddr[ring]);
 	if (INTEL_INFO(dev)->gen >= 6) {
+		seq_printf(m, "  RC PSMI: 0x%08x\n", error->rc_psmi[ring]);
 		seq_printf(m, "  FAULT_REG: 0x%08x\n", error->fault_reg[ring]);
 		seq_printf(m, "  SYNC_0: 0x%08x\n",
 			   error->semaphore_mboxes[ring][0]);

drivers/gpu/drm/i915/i915_dma.c

@@ -1781,7 +1781,13 @@ void i915_driver_lastclose(struct drm_device * dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 
-	if (!dev_priv || drm_core_check_feature(dev, DRIVER_MODESET)) {
+	/* On gen6+ we refuse to init without kms enabled, but then the drm core
+	 * goes right around and calls lastclose. Check for this and don't clean
+	 * up anything. */
+	if (!dev_priv)
+		return;
+
+	if (drm_core_check_feature(dev, DRIVER_MODESET)) {
 		intel_fb_restore_mode(dev);
 		vga_switcheroo_process_delayed_switch();
 		return;

drivers/gpu/drm/i915/i915_drv.h

@@ -190,6 +190,7 @@ struct drm_i915_error_state {
 	u32 instdone[I915_NUM_RINGS];
 	u32 acthd[I915_NUM_RINGS];
 	u32 semaphore_mboxes[I915_NUM_RINGS][I915_NUM_RINGS - 1];
+	u32 rc_psmi[I915_NUM_RINGS]; /* sleep state */
 	/* our own tracking of ring head and tail */
 	u32 cpu_ring_head[I915_NUM_RINGS];
 	u32 cpu_ring_tail[I915_NUM_RINGS];

drivers/gpu/drm/i915/i915_gem.c

@@ -2003,6 +2003,7 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
 	}
 
 	switch (end) {
+	case -EIO:
 	case -EAGAIN: /* Wedged */
 	case -ERESTARTSYS: /* Signal */
 		return (int)end;
@@ -3726,6 +3727,22 @@ void i915_gem_init_ppgtt(struct drm_device *dev)
 	}
 }
 
+static bool
+intel_enable_blt(struct drm_device *dev)
+{
+	if (!HAS_BLT(dev))
+		return false;
+
+	/* The blitter was dysfunctional on early prototypes */
+	if (IS_GEN6(dev) && dev->pdev->revision < 8) {
+		DRM_INFO("BLT not supported on this pre-production hardware;"
+			 " graphics performance will be degraded.\n");
+		return false;
+	}
+
+	return true;
+}
+
 int
 i915_gem_init_hw(struct drm_device *dev)
 {
@@ -3749,7 +3766,7 @@ i915_gem_init_hw(struct drm_device *dev)
 			goto cleanup_render_ring;
 	}
 
-	if (HAS_BLT(dev)) {
+	if (intel_enable_blt(dev)) {
 		ret = intel_init_blt_ring_buffer(dev);
 		if (ret)
 			goto cleanup_bsd_ring;

drivers/gpu/drm/i915/i915_gem_context.c

@@ -112,8 +112,8 @@ static int get_context_size(struct drm_device *dev)
 		ret = GEN6_CXT_TOTAL_SIZE(reg) * 64;
 		break;
 	case 7:
-		reg = I915_READ(GEN7_CTX_SIZE);
-		ret = GEN7_CTX_TOTAL_SIZE(reg) * 64;
+		reg = I915_READ(GEN7_CXT_SIZE);
+		ret = GEN7_CXT_TOTAL_SIZE(reg) * 64;
 		break;
 	default:
 		BUG();
@@ -374,6 +374,17 @@ static int do_switch(struct drm_i915_gem_object *from_obj,
 	if (ret)
 		return ret;
 
+	/* Clear this page out of any CPU caches for coherent swap-in/out. Note
+	 * that thanks to write = false in this call and us not setting any gpu
+	 * write domains when putting a context object onto the active list
+	 * (when switching away from it), this won't block.
+	 * XXX: We need a real interface to do this instead of trickery. */
+	ret = i915_gem_object_set_to_gtt_domain(to->obj, false);
+	if (ret) {
+		i915_gem_object_unpin(to->obj);
+		return ret;
+	}
+
 	if (!to->obj->has_global_gtt_mapping)
 		i915_gem_gtt_bind_object(to->obj, to->obj->cache_level);
 
@@ -408,8 +419,11 @@ static int do_switch(struct drm_i915_gem_object *from_obj,
 		from_obj->dirty = 1;
 		BUG_ON(from_obj->ring != to->ring);
 		i915_gem_object_unpin(from_obj);
+
+		drm_gem_object_unreference(&from_obj->base);
 	}
 
+	drm_gem_object_reference(&to->obj->base);
 	ring->last_context_obj = to->obj;
 	to->is_initialized = true;
 
@@ -459,20 +473,7 @@ int i915_switch_context(struct intel_ring_buffer *ring,
 	if (from_obj == to->obj)
 		return 0;
 
-	ret = do_switch(from_obj, to, i915_gem_next_request_seqno(to->ring));
-	if (ret)
-		return ret;
-
-	/* Just to make the code a little cleaner we take the object reference
-	 * after the switch was successful. It would be more intuitive to ref
-	 * the 'to' object before the switch but we know the refcount must be >0
-	 * if context_get() succeeded, and we hold struct mutex. So it's safe to
-	 * do this here/now
-	 */
-	drm_gem_object_reference(&to->obj->base);
-	if (from_obj != NULL)
-		drm_gem_object_unreference(&from_obj->base);
-	return ret;
+	return do_switch(from_obj, to, i915_gem_next_request_seqno(to->ring));
 }
 
 int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
@@ -496,11 +497,13 @@ int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
 
 	ctx = create_hw_context(dev, file_priv);
 	mutex_unlock(&dev->struct_mutex);
+	if (IS_ERR(ctx))
+		return PTR_ERR(ctx);
 
 	args->ctx_id = ctx->id;
 	DRM_DEBUG_DRIVER("HW context %d created\n", args->ctx_id);
 
-	return PTR_RET(ctx);
+	return 0;
 }
 
 int i915_gem_context_destroy_ioctl(struct drm_device *dev, void *data,

drivers/gpu/drm/i915/i915_gem_execbuffer.c

@@ -885,11 +885,16 @@ i915_gem_execbuffer_move_to_gpu(struct intel_ring_buffer *ring,
 			return ret;
 	}
 
-	/* Unconditionally invalidate gpu caches. */
-	ret = i915_gem_flush_ring(ring, I915_GEM_GPU_DOMAINS, 0);
+	/* Unconditionally invalidate gpu caches and ensure that we do flush
+	 * any residual writes from the previous batch.
+	 */
+	ret = i915_gem_flush_ring(ring,
+				  I915_GEM_GPU_DOMAINS,
+				  ring->gpu_caches_dirty ? I915_GEM_GPU_DOMAINS : 0);
 	if (ret)
 		return ret;
 
+	ring->gpu_caches_dirty = false;
 	return 0;
 }
 
@@ -1223,6 +1228,10 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 		}
 	}
 
+	ret = i915_switch_context(ring, file, ctx_id);
+	if (ret)
+		goto err;
+
 	if (ring == &dev_priv->ring[RCS] &&
 	    mode != dev_priv->relative_constants_mode) {
 		ret = intel_ring_begin(ring, 4);
@@ -1244,10 +1253,6 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 			goto err;
 	}
 
-	ret = i915_switch_context(ring, file, ctx_id);
-	if (ret)
-		goto err;
-
 	trace_i915_gem_ring_dispatch(ring, seqno);
 
 	exec_start = batch_obj->gtt_offset + args->batch_start_offset;

drivers/gpu/drm/i915/i915_irq.c

@@ -1066,6 +1066,7 @@ static void i915_record_ring_state(struct drm_device *dev,
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
 	if (INTEL_INFO(dev)->gen >= 6) {
+		error->rc_psmi[ring->id] = I915_READ(ring->mmio_base + 0x50);
 		error->fault_reg[ring->id] = I915_READ(RING_FAULT_REG(ring));
 		error->semaphore_mboxes[ring->id][0]
 			= I915_READ(RING_SYNC_0(ring->mmio_base));

drivers/gpu/drm/i915/i915_reg.h

@@ -690,10 +690,10 @@
 #define   GEN6_BLITTER_FBC_NOTIFY			(1<<3)
 
 #define GEN6_BSD_SLEEP_PSMI_CONTROL	0x12050
-#define   GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_MODIFY_MASK	(1 << 16)
-#define   GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_DISABLE		(1 << 0)
-#define   GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_ENABLE		0
-#define   GEN6_BSD_SLEEP_PSMI_CONTROL_IDLE_INDICATOR			(1 << 3)
+#define   GEN6_BSD_SLEEP_MSG_DISABLE	(1 << 0)
+#define   GEN6_BSD_SLEEP_FLUSH_DISABLE	(1 << 2)
+#define   GEN6_BSD_SLEEP_INDICATOR	(1 << 3)
+#define   GEN6_BSD_GO_INDICATOR		(1 << 4)
 
 #define GEN6_BSD_HWSTAM			0x12098
 #define GEN6_BSD_IMR			0x120a8
@@ -1482,15 +1482,19 @@
 					GEN6_CXT_RENDER_SIZE(cxt_reg) + \
 					GEN6_CXT_EXTENDED_SIZE(cxt_reg) + \
 					GEN6_CXT_PIPELINE_SIZE(cxt_reg))
-#define GEN7_CTX_SIZE		0x21a8
-#define GEN7_CTX_RENDER_SIZE(ctx_reg)	((ctx_reg >> 16) & 0x3f)
-#define GEN7_CTX_EXTENDED_SIZE(ctx_reg)	((ctx_reg >> 9) & 0x7f)
-#define GEN7_CTX_GT1_SIZE(ctx_reg)	((ctx_reg >> 6) & 0x7)
-#define GEN7_CTX_VFSTATE_SIZE(ctx_reg)	((ctx_reg >> 0) & 0x3f)
-#define GEN7_CTX_TOTAL_SIZE(ctx_reg)	(GEN7_CTX_RENDER_SIZE(ctx_reg) + \
-					 GEN7_CTX_EXTENDED_SIZE(ctx_reg) + \
-					 GEN7_CTX_GT1_SIZE(ctx_reg) + \
-					 GEN7_CTX_VFSTATE_SIZE(ctx_reg))
+#define GEN7_CXT_SIZE		0x21a8
+#define GEN7_CXT_POWER_SIZE(ctx_reg)	((ctx_reg >> 25) & 0x7f)
+#define GEN7_CXT_RING_SIZE(ctx_reg)	((ctx_reg >> 22) & 0x7)
+#define GEN7_CXT_RENDER_SIZE(ctx_reg)	((ctx_reg >> 16) & 0x3f)
+#define GEN7_CXT_EXTENDED_SIZE(ctx_reg)	((ctx_reg >> 9) & 0x7f)
+#define GEN7_CXT_GT1_SIZE(ctx_reg)	((ctx_reg >> 6) & 0x7)
+#define GEN7_CXT_VFSTATE_SIZE(ctx_reg)	((ctx_reg >> 0) & 0x3f)
+#define GEN7_CXT_TOTAL_SIZE(ctx_reg)	(GEN7_CXT_POWER_SIZE(ctx_reg) + \
+					 GEN7_CXT_RING_SIZE(ctx_reg) + \
+					 GEN7_CXT_RENDER_SIZE(ctx_reg) + \
+					 GEN7_CXT_EXTENDED_SIZE(ctx_reg) + \
+					 GEN7_CXT_GT1_SIZE(ctx_reg) + \
+					 GEN7_CXT_VFSTATE_SIZE(ctx_reg))
 
 /*
  * Overlay regs
@@ -1896,7 +1900,7 @@
 /* PCH CTL1 is totally different, all but the below bits are reserved. CTL2 is
  * like the normal CTL from gen4 and earlier. Hooray for confusing naming. */
 #define BLC_PWM_PCH_CTL1	0xc8250
-#define   BLM_PCH_PWM_ENABLE			(1 << 30)
+#define   BLM_PCH_PWM_ENABLE			(1 << 31)
 #define   BLM_PCH_OVERRIDE_ENABLE		(1 << 30)
 #define   BLM_PCH_POLARITY			(1 << 29)
 #define BLC_PWM_PCH_CTL2	0xc8254

drivers/gpu/drm/i915/intel_crt.c

@@ -330,39 +330,34 @@ static bool intel_crt_detect_ddc(struct drm_connector *connector)
 {
 	struct intel_crt *crt = intel_attached_crt(connector);
 	struct drm_i915_private *dev_priv = crt->base.base.dev->dev_private;
+	struct edid *edid;
+	struct i2c_adapter *i2c;
+
+	BUG_ON(crt->base.type != INTEL_OUTPUT_ANALOG);
 
-	/* CRT should always be at 0, but check anyway */
-	if (crt->base.type != INTEL_OUTPUT_ANALOG)
-		return false;
+	i2c = intel_gmbus_get_adapter(dev_priv, dev_priv->crt_ddc_pin);
+	edid = drm_get_edid(connector, i2c);
 
-	if (intel_ddc_probe(&crt->base, dev_priv->crt_ddc_pin)) {
-		struct edid *edid;
-		bool is_digital = false;
-		struct i2c_adapter *i2c;
+	if (edid) {
+		bool is_digital = edid->input & DRM_EDID_INPUT_DIGITAL;
 
-		i2c = intel_gmbus_get_adapter(dev_priv, dev_priv->crt_ddc_pin);
-		edid = drm_get_edid(connector, i2c);
 		/*
 		 * This may be a DVI-I connector with a shared DDC
 		 * link between analog and digital outputs, so we
 		 * have to check the EDID input spec of the attached device.
-		 *
-		 * On the other hand, what should we do if it is a broken EDID?
 		 */
-		if (edid != NULL) {
-			is_digital = edid->input & DRM_EDID_INPUT_DIGITAL;
-			connector->display_info.raw_edid = NULL;
-			kfree(edid);
-		}
-
 		if (!is_digital) {
 			DRM_DEBUG_KMS("CRT detected via DDC:0x50 [EDID]\n");
 			return true;
-		} else {
-			DRM_DEBUG_KMS("CRT not detected via DDC:0x50 [EDID reports a digital panel]\n");
 		}
+
+		DRM_DEBUG_KMS("CRT not detected via DDC:0x50 [EDID reports a digital panel]\n");
+	} else {
+		DRM_DEBUG_KMS("CRT not detected via DDC:0x50 [no valid EDID found]\n");
 	}
 
+	kfree(edid);
+
 	return false;
 }
 

drivers/gpu/drm/i915/intel_display.c

@@ -3573,7 +3573,7 @@ void intel_encoder_destroy(struct drm_encoder *encoder)
 }
 
 static bool intel_crtc_mode_fixup(struct drm_crtc *crtc,
-				  struct drm_display_mode *mode,
+				  const struct drm_display_mode *mode,
 				  struct drm_display_mode *adjusted_mode)
 {
 	struct drm_device *dev = crtc->dev;
@@ -6486,7 +6486,9 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc,
 	intel_fb = to_intel_framebuffer(fb);
 	obj = intel_fb->obj;
 
-	mutex_lock(&dev->struct_mutex);
+	ret = i915_mutex_lock_interruptible(dev);
+	if (ret)
+		goto cleanup;
 
 	/* Reference the objects for the scheduled work. */
 	drm_gem_object_reference(&work->old_fb_obj->base);
@@ -6521,6 +6523,7 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc,
 	drm_gem_object_unreference(&obj->base);
 	mutex_unlock(&dev->struct_mutex);
 
+cleanup:
 	spin_lock_irqsave(&dev->event_lock, flags);
 	intel_crtc->unpin_work = NULL;
 	spin_unlock_irqrestore(&dev->event_lock, flags);
@@ -7174,6 +7177,11 @@ static void i915_disable_vga(struct drm_device *dev)
 
 void intel_modeset_init_hw(struct drm_device *dev)
 {
+	/* We attempt to init the necessary power wells early in the initialization
+	 * time, so the subsystems that expect power to be enabled can work.
+	 */
+	intel_init_power_wells(dev);
+
 	intel_prepare_ddi(dev);
 
 	intel_init_clock_gating(dev);

drivers/gpu/drm/i915/intel_dp.c

@@ -1771,7 +1771,7 @@ intel_dp_start_link_train(struct intel_dp *intel_dp)
 		for (i = 0; i < intel_dp->lane_count; i++)
 			if ((intel_dp->train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0)
 				break;
-		if (i == intel_dp->lane_count) {
+		if (i == intel_dp->lane_count && voltage_tries == 5) {
 			++loop_tries;
 			if (loop_tries == 5) {
 				DRM_DEBUG_KMS("too many full retries, give up\n");

drivers/gpu/drm/i915/intel_drv.h

@@ -342,7 +342,6 @@ struct intel_fbc_work {
 };
 
 int intel_ddc_get_modes(struct drm_connector *c, struct i2c_adapter *adapter);
-extern bool intel_ddc_probe(struct intel_encoder *intel_encoder, int ddc_bus);
 
 extern void intel_attach_force_audio_property(struct drm_connector *connector);
 extern void intel_attach_broadcast_rgb_property(struct drm_connector *connector);
@@ -496,6 +495,7 @@ extern void intel_update_fbc(struct drm_device *dev);
 extern void intel_gpu_ips_init(struct drm_i915_private *dev_priv);
 extern void intel_gpu_ips_teardown(void);
 
+extern void intel_init_power_wells(struct drm_device *dev);
 extern void intel_enable_gt_powersave(struct drm_device *dev);
 extern void intel_disable_gt_powersave(struct drm_device *dev);
 extern void gen6_gt_check_fifodbg(struct drm_i915_private *dev_priv);

drivers/gpu/drm/i915/intel_lvds.c

@@ -409,13 +409,7 @@ static void intel_lvds_prepare(struct drm_encoder *encoder)
 {
 	struct intel_lvds *intel_lvds = to_intel_lvds(encoder);
 
-	/*
-	 * Prior to Ironlake, we must disable the pipe if we want to adjust
-	 * the panel fitter. However at all other times we can just reset
-	 * the registers regardless.
-	 */
-	if (!HAS_PCH_SPLIT(encoder->dev) && intel_lvds->pfit_dirty)
-		intel_lvds_disable(intel_lvds);
+	intel_lvds_disable(intel_lvds);
 }
 
 static void intel_lvds_commit(struct drm_encoder *encoder)