From 29105ccc43ead5a0179d04b1404611981e92e278 Mon Sep 17 00:00:00 2001
From: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu, 7 Jan 2010 10:39:13 +0000
Subject: drm/i915: Replace open-coded eviction in i915_gem_idle()

With the introduction of the hang-check, we can safely expect that
i915_wait_request() will always return even when the GPU hangs, and so
do not need to open code the wait in order to manually check for the
hang. Also we do not need to always evict all buffers, so only flush
the GPU (and wait for it to idle) for KMS, but continue to evict for UMS.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_gem.c | 144 ++++++++++++----------------------------
 1 file changed, 44 insertions(+), 100 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index ec8a0d7ffa3..129ac36ddc5 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -4441,129 +4441,73 @@ i915_gem_evict_from_inactive_list(struct drm_device *dev)
 	return 0;
 }
 
-int
-i915_gem_idle(struct drm_device *dev)
+static int
+i915_gpu_idle(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
-	uint32_t seqno, cur_seqno, last_seqno;
-	int stuck, ret;
+	bool lists_empty;
+	uint32_t seqno;
 
-	mutex_lock(&dev->struct_mutex);
+	spin_lock(&dev_priv->mm.active_list_lock);
+	lists_empty = list_empty(&dev_priv->mm.flushing_list) &&
+		      list_empty(&dev_priv->mm.active_list);
+	spin_unlock(&dev_priv->mm.active_list_lock);
 
-	if (dev_priv->mm.suspended || dev_priv->ring.ring_obj == NULL) {
-		mutex_unlock(&dev->struct_mutex);
+	if (lists_empty)
 		return 0;
-	}
-
-	/* Hack!  Don't let anybody do execbuf while we don't control the chip.
-	 * We need to replace this with a semaphore, or something.
-	 */
-	dev_priv->mm.suspended = 1;
-	del_timer(&dev_priv->hangcheck_timer);
-
-	/* Cancel the retire work handler, wait for it to finish if running
-	 */
-	mutex_unlock(&dev->struct_mutex);
-	cancel_delayed_work_sync(&dev_priv->mm.retire_work);
-	mutex_lock(&dev->struct_mutex);
 
-	i915_kernel_lost_context(dev);
-
-	/* Flush the GPU along with all non-CPU write domains
-	 */
+	/* Flush everything onto the inactive list. */
 	i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS);
 	seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS);
-
-	if (seqno == 0) {
-		mutex_unlock(&dev->struct_mutex);
+	if (seqno == 0)
 		return -ENOMEM;
-	}
-
-	dev_priv->mm.waiting_gem_seqno = seqno;
-	last_seqno = 0;
-	stuck = 0;
-	for (;;) {
-		cur_seqno = i915_get_gem_seqno(dev);
-		if (i915_seqno_passed(cur_seqno, seqno))
-			break;
-		if (last_seqno == cur_seqno) {
-			if (stuck++ > 100) {
-				DRM_ERROR("hardware wedged\n");
-				atomic_set(&dev_priv->mm.wedged, 1);
-				DRM_WAKEUP(&dev_priv->irq_queue);
-				break;
-			}
-		}
-		msleep(10);
-		last_seqno = cur_seqno;
-	}
-	dev_priv->mm.waiting_gem_seqno = 0;
-
-	i915_gem_retire_requests(dev);
-
-	spin_lock(&dev_priv->mm.active_list_lock);
-	if (!atomic_read(&dev_priv->mm.wedged)) {
-		/* Active and flushing should now be empty as we've
-		 * waited for a sequence higher than any pending execbuffer
-		 */
-		WARN_ON(!list_empty(&dev_priv->mm.active_list));
-		WARN_ON(!list_empty(&dev_priv->mm.flushing_list));
-		/* Request should now be empty as we've also waited
-		 * for the last request in the list
-		 */
-		WARN_ON(!list_empty(&dev_priv->mm.request_list));
-	}
-
-	/* Empty the active and flushing lists to inactive.  If there's
-	 * anything left at this point, it means that we're wedged and
-	 * nothing good's going to happen by leaving them there.  So strip
-	 * the GPU domains and just stuff them onto inactive.
-	 */
-	while (!list_empty(&dev_priv->mm.active_list)) {
-		struct drm_gem_object *obj;
-		uint32_t old_write_domain;
-
-		obj = list_first_entry(&dev_priv->mm.active_list,
-				       struct drm_i915_gem_object,
-				       list)->obj;
-		old_write_domain = obj->write_domain;
-		obj->write_domain &= ~I915_GEM_GPU_DOMAINS;
-		i915_gem_object_move_to_inactive(obj);
 
-		trace_i915_gem_object_change_domain(obj,
-						    obj->read_domains,
-						    old_write_domain);
-	}
-	spin_unlock(&dev_priv->mm.active_list_lock);
+	return i915_wait_request(dev, seqno);
+}
 
-	while (!list_empty(&dev_priv->mm.flushing_list)) {
-		struct drm_gem_object *obj;
-		uint32_t old_write_domain;
+int
+i915_gem_idle(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	int ret;
 
-		obj = list_first_entry(&dev_priv->mm.flushing_list,
-				       struct drm_i915_gem_object,
-				       list)->obj;
-		old_write_domain = obj->write_domain;
-		obj->write_domain &= ~I915_GEM_GPU_DOMAINS;
-		i915_gem_object_move_to_inactive(obj);
+	mutex_lock(&dev->struct_mutex);
 
-		trace_i915_gem_object_change_domain(obj,
-						    obj->read_domains,
-						    old_write_domain);
+	if (dev_priv->mm.suspended || dev_priv->ring.ring_obj == NULL) {
+		mutex_unlock(&dev->struct_mutex);
+		return 0;
 	}
 
-
-	/* Move all inactive buffers out of the GTT. */
-	ret = i915_gem_evict_from_inactive_list(dev);
-	WARN_ON(!list_empty(&dev_priv->mm.inactive_list));
+	ret = i915_gpu_idle(dev);
 	if (ret) {
 		mutex_unlock(&dev->struct_mutex);
 		return ret;
 	}
 
+	/* Under UMS, be paranoid and evict. */
+	if (!drm_core_check_feature(dev, DRIVER_MODESET)) {
+		ret = i915_gem_evict_from_inactive_list(dev);
+		if (ret) {
+			mutex_unlock(&dev->struct_mutex);
+			return ret;
+		}
+	}
+
+	/* Hack!  Don't let anybody do execbuf while we don't control the chip.
+	 * We need to replace this with a semaphore, or something.
+	 * And not confound mm.suspended!
+	 */
+	dev_priv->mm.suspended = 1;
+	del_timer(&dev_priv->hangcheck_timer);
+
+	i915_kernel_lost_context(dev);
 	i915_gem_cleanup_ringbuffer(dev);
+
 	mutex_unlock(&dev->struct_mutex);
 
+	/* Cancel the retire work handler, which should be idle now. */
+	cancel_delayed_work_sync(&dev_priv->mm.retire_work);
+
 	return 0;
 }
 
-- 
cgit v1.2.3


From c4804411691bdd7d8a57e942cbb502fd52a90074 Mon Sep 17 00:00:00 2001
From: Zhenyu Wang <zhenyuw@linux.intel.com>
Date: Thu, 17 Dec 2009 14:48:43 +0800
Subject: drm/i915: Keep MCHBAR always enabled

As we need more and more controls within MCHBAR for memory config and
power management, this trys to keep MCHBAR enabled from driver load and
only tear down in driver unload.

Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_dma.c        | 121 ++++++++++++++++++++++++++++++++
 drivers/gpu/drm/i915/i915_drv.h        |   1 +
 drivers/gpu/drm/i915/i915_gem_tiling.c | 122 ---------------------------------
 3 files changed, 122 insertions(+), 122 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 2307f98349f..3afe361ec55 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -35,6 +35,8 @@
 #include "i915_drv.h"
 #include "i915_trace.h"
 #include <linux/vgaarb.h>
+#include <linux/acpi.h>
+#include <linux/pnp.h>
 
 /* Really want an OS-independent resettable timer.  Would like to have
  * this loop run for (eg) 3 sec, but have the timer reset every time
@@ -933,6 +935,120 @@ static int i915_get_bridge_dev(struct drm_device *dev)
 	return 0;
 }
 
+#define MCHBAR_I915 0x44
+#define MCHBAR_I965 0x48
+#define MCHBAR_SIZE (4*4096)
+
+#define DEVEN_REG 0x54
+#define   DEVEN_MCHBAR_EN (1 << 28)
+
+/* Allocate space for the MCH regs if needed, return nonzero on error */
+static int
+intel_alloc_mchbar_resource(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	int reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915;
+	u32 temp_lo, temp_hi = 0;
+	u64 mchbar_addr;
+	int ret = 0;
+
+	if (IS_I965G(dev))
+		pci_read_config_dword(dev_priv->bridge_dev, reg + 4, &temp_hi);
+	pci_read_config_dword(dev_priv->bridge_dev, reg, &temp_lo);
+	mchbar_addr = ((u64)temp_hi << 32) | temp_lo;
+
+	/* If ACPI doesn't have it, assume we need to allocate it ourselves */
+#ifdef CONFIG_PNP
+	if (mchbar_addr &&
+	    pnp_range_reserved(mchbar_addr, mchbar_addr + MCHBAR_SIZE)) {
+		ret = 0;
+		goto out;
+	}
+#endif
+
+	/* Get some space for it */
+	ret = pci_bus_alloc_resource(dev_priv->bridge_dev->bus, &dev_priv->mch_res,
+				     MCHBAR_SIZE, MCHBAR_SIZE,
+				     PCIBIOS_MIN_MEM,
+				     0,   pcibios_align_resource,
+				     dev_priv->bridge_dev);
+	if (ret) {
+		DRM_DEBUG_DRIVER("failed bus alloc: %d\n", ret);
+		dev_priv->mch_res.start = 0;
+		goto out;
+	}
+
+	if (IS_I965G(dev))
+		pci_write_config_dword(dev_priv->bridge_dev, reg + 4,
+				       upper_32_bits(dev_priv->mch_res.start));
+
+	pci_write_config_dword(dev_priv->bridge_dev, reg,
+			       lower_32_bits(dev_priv->mch_res.start));
+out:
+	return ret;
+}
+
+/* Setup MCHBAR if possible, return true if we should disable it again */
+static void
+intel_setup_mchbar(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915;
+	u32 temp;
+	bool enabled;
+
+	dev_priv->mchbar_need_disable = false;
+
+	if (IS_I915G(dev) || IS_I915GM(dev)) {
+		pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp);
+		enabled = !!(temp & DEVEN_MCHBAR_EN);
+	} else {
+		pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp);
+		enabled = temp & 1;
+	}
+
+	/* If it's already enabled, don't have to do anything */
+	if (enabled)
+		return;
+
+	if (intel_alloc_mchbar_resource(dev))
+		return;
+
+	dev_priv->mchbar_need_disable = true;
+
+	/* Space is allocated or reserved, so enable it. */
+	if (IS_I915G(dev) || IS_I915GM(dev)) {
+		pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG,
+				       temp | DEVEN_MCHBAR_EN);
+	} else {
+		pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp);
+		pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp | 1);
+	}
+}
+
+static void
+intel_teardown_mchbar(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915;
+	u32 temp;
+
+	if (dev_priv->mchbar_need_disable) {
+		if (IS_I915G(dev) || IS_I915GM(dev)) {
+			pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp);
+			temp &= ~DEVEN_MCHBAR_EN;
+			pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, temp);
+		} else {
+			pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp);
+			temp &= ~1;
+			pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp);
+		}
+	}
+
+	if (dev_priv->mch_res.start)
+		release_resource(&dev_priv->mch_res);
+}
+
 /**
  * i915_probe_agp - get AGP bootup configuration
  * @pdev: PCI device
@@ -1450,6 +1566,9 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 		dev->driver->get_vblank_counter = gm45_get_vblank_counter;
 	}
 
+	/* Try to make sure MCHBAR is enabled before poking at it */
+	intel_setup_mchbar(dev);
+
 	i915_gem_load(dev);
 
 	/* Init HWS */
@@ -1569,6 +1688,8 @@ int i915_driver_unload(struct drm_device *dev)
 		intel_cleanup_overlay(dev);
 	}
 
+	intel_teardown_mchbar(dev);
+
 	pci_dev_put(dev_priv->bridge_dev);
 	kfree(dev->dev_private);
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index b99b6a841d9..f2742d6d3f5 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -590,6 +590,7 @@ typedef struct drm_i915_private {
 	int child_dev_num;
 	struct child_device_config *child_dev;
 	struct drm_connector *int_lvds_connector;
+	bool mchbar_need_disable;
 } drm_i915_private_t;
 
 /** driver private structure attached to each drm_gem_object */
diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c
index df278b2685b..f308fdf4372 100644
--- a/drivers/gpu/drm/i915/i915_gem_tiling.c
+++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
@@ -25,8 +25,6 @@
  *
  */
 
-#include <linux/acpi.h>
-#include <linux/pnp.h>
 #include "linux/string.h"
 #include "linux/bitops.h"
 #include "drmP.h"
@@ -83,120 +81,6 @@
  * to match what the GPU expects.
  */
 
-#define MCHBAR_I915 0x44
-#define MCHBAR_I965 0x48
-#define MCHBAR_SIZE (4*4096)
-
-#define DEVEN_REG 0x54
-#define   DEVEN_MCHBAR_EN (1 << 28)
-
-/* Allocate space for the MCH regs if needed, return nonzero on error */
-static int
-intel_alloc_mchbar_resource(struct drm_device *dev)
-{
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	int reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915;
-	u32 temp_lo, temp_hi = 0;
-	u64 mchbar_addr;
-	int ret = 0;
-
-	if (IS_I965G(dev))
-		pci_read_config_dword(dev_priv->bridge_dev, reg + 4, &temp_hi);
-	pci_read_config_dword(dev_priv->bridge_dev, reg, &temp_lo);
-	mchbar_addr = ((u64)temp_hi << 32) | temp_lo;
-
-	/* If ACPI doesn't have it, assume we need to allocate it ourselves */
-#ifdef CONFIG_PNP
-	if (mchbar_addr &&
-	    pnp_range_reserved(mchbar_addr, mchbar_addr + MCHBAR_SIZE)) {
-		ret = 0;
-		goto out;
-	}
-#endif
-
-	/* Get some space for it */
-	ret = pci_bus_alloc_resource(dev_priv->bridge_dev->bus, &dev_priv->mch_res,
-				     MCHBAR_SIZE, MCHBAR_SIZE,
-				     PCIBIOS_MIN_MEM,
-				     0,   pcibios_align_resource,
-				     dev_priv->bridge_dev);
-	if (ret) {
-		DRM_DEBUG_DRIVER("failed bus alloc: %d\n", ret);
-		dev_priv->mch_res.start = 0;
-		goto out;
-	}
-
-	if (IS_I965G(dev))
-		pci_write_config_dword(dev_priv->bridge_dev, reg + 4,
-				       upper_32_bits(dev_priv->mch_res.start));
-
-	pci_write_config_dword(dev_priv->bridge_dev, reg,
-			       lower_32_bits(dev_priv->mch_res.start));
-out:
-	return ret;
-}
-
-/* Setup MCHBAR if possible, return true if we should disable it again */
-static bool
-intel_setup_mchbar(struct drm_device *dev)
-{
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915;
-	u32 temp;
-	bool need_disable = false, enabled;
-
-	if (IS_I915G(dev) || IS_I915GM(dev)) {
-		pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp);
-		enabled = !!(temp & DEVEN_MCHBAR_EN);
-	} else {
-		pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp);
-		enabled = temp & 1;
-	}
-
-	/* If it's already enabled, don't have to do anything */
-	if (enabled)
-		goto out;
-
-	if (intel_alloc_mchbar_resource(dev))
-		goto out;
-
-	need_disable = true;
-
-	/* Space is allocated or reserved, so enable it. */
-	if (IS_I915G(dev) || IS_I915GM(dev)) {
-		pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG,
-				       temp | DEVEN_MCHBAR_EN);
-	} else {
-		pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp);
-		pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp | 1);
-	}
-out:
-	return need_disable;
-}
-
-static void
-intel_teardown_mchbar(struct drm_device *dev, bool disable)
-{
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	int mchbar_reg = IS_I965G(dev) ? MCHBAR_I965 : MCHBAR_I915;
-	u32 temp;
-
-	if (disable) {
-		if (IS_I915G(dev) || IS_I915GM(dev)) {
-			pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp);
-			temp &= ~DEVEN_MCHBAR_EN;
-			pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, temp);
-		} else {
-			pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp);
-			temp &= ~1;
-			pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp);
-		}
-	}
-
-	if (dev_priv->mch_res.start)
-		release_resource(&dev_priv->mch_res);
-}
-
 /**
  * Detects bit 6 swizzling of address lookup between IGD access and CPU
  * access through main memory.
@@ -207,7 +91,6 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	uint32_t swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN;
 	uint32_t swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN;
-	bool need_disable;
 
 	if (IS_IRONLAKE(dev)) {
 		/* On Ironlake whatever DRAM config, GPU always do
@@ -224,9 +107,6 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev)
 	} else if (IS_MOBILE(dev)) {
 		uint32_t dcc;
 
-		/* Try to make sure MCHBAR is enabled before poking at it */
-		need_disable = intel_setup_mchbar(dev);
-
 		/* On mobile 9xx chipsets, channel interleave by the CPU is
 		 * determined by DCC.  For single-channel, neither the CPU
 		 * nor the GPU do swizzling.  For dual channel interleaved,
@@ -266,8 +146,6 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev)
 			swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN;
 			swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN;
 		}
-
-		intel_teardown_mchbar(dev, need_disable);
 	} else {
 		/* The 965, G33, and newer, have a very flexible memory
 		 * configuration.  It will enable dual-channel mode
-- 
cgit v1.2.3


From b397c836eff58cd9a43f7bd8b853a51b3ecc3420 Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Tue, 26 Jan 2010 09:43:10 -0800
Subject: drm/i915: Don't reserve compatibility fence regs in KMS mode.

The fence start is for compatibility with UMS X Servers before fence
management.  KMS X Servers only started doing tiling after fence
management appeared.

Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_gem.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 129ac36ddc5..4a141993a37 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -4794,7 +4794,8 @@ i915_gem_load(struct drm_device *dev)
 	spin_unlock(&shrink_list_lock);
 
 	/* Old X drivers will take 0-2 for front, back, depth buffers */
-	dev_priv->fence_reg_start = 3;
+	if (!drm_core_check_feature(dev, DRIVER_MODESET))
+		dev_priv->fence_reg_start = 3;
 
 	if (IS_I965G(dev) || IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev))
 		dev_priv->num_fence_regs = 16;
-- 
cgit v1.2.3


From ee980b8003a25fbfed50c3367f2b426c870eaf90 Mon Sep 17 00:00:00 2001
From: Li Peng <peng.li@linux.intel.com>
Date: Wed, 27 Jan 2010 19:01:11 +0800
Subject: drm/i915: enable memory self refresh on 9xx

Enabling memory self refresh (SR) on 9xx needs to set additional
register bits. On 945, we need bit 31 of FW_BLC_SELF to enable the
write to self refresh bit and bit 16 to enable the write of self
refresh watermark. On 915, bit 12 of INSTPM is used to enable SR.

SR will take effect when CPU enters C3+ state and its entry/exit
should be automatically controlled by H/W, driver only needs to set
SR enable bits in wm update. But this isn't safe in my test on 945
because GPU is hung. So this patch explicitly enables SR when GPU
is idle, and disables SR when it is busy. In my test on a netbook of
945GSE chipset, it saves about 0.8W idle power.

Signed-off-by: Li Peng <peng.li@intel.com>
[anholt: rebased against 33c5fd121eabbccc9103daf6cda36941eb3c349f
by adding disable of INSTPM SR bit on 915GM for two pipe setup]
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_reg.h      |  5 ++++-
 drivers/gpu/drm/i915/intel_display.c | 31 ++++++++++++++++++++++++++++---
 2 files changed, 32 insertions(+), 4 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index ab1bd2d3d3b..6defb7f4734 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -306,11 +306,14 @@
 #define   I915_ERROR_MEMORY_REFRESH			(1<<1)
 #define   I915_ERROR_INSTRUCTION			(1<<0)
 #define INSTPM	        0x020c0
+#define   INSTPM_SELF_EN (1<<12) /* 915GM only */
 #define ACTHD	        0x020c8
 #define FW_BLC		0x020d8
 #define FW_BLC2	 	0x020dc
 #define FW_BLC_SELF	0x020e0 /* 915+ only */
-#define   FW_BLC_SELF_EN (1<<15)
+#define   FW_BLC_SELF_EN_MASK      (1<<31)
+#define   FW_BLC_SELF_FIFO_MASK    (1<<16) /* 945 only */
+#define   FW_BLC_SELF_EN           (1<<15) /* 945 only */
 #define MM_BURST_LENGTH     0x00700000
 #define MM_FIFO_WATERMARK   0x0001F000
 #define LM_BURST_LENGTH     0x00000700
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index b27202d23eb..af9ec217cd1 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -2757,11 +2757,22 @@ static void i9xx_update_wm(struct drm_device *dev, int planea_clock,
 		srwm = total_size - sr_entries;
 		if (srwm < 0)
 			srwm = 1;
-		I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN | (srwm & 0x3f));
+
+		if (IS_I945G(dev) || IS_I945GM(dev))
+			I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_FIFO_MASK | (srwm & 0xff));
+		else if (IS_I915GM(dev)) {
+			/* 915M has a smaller SRWM field */
+			I915_WRITE(FW_BLC_SELF, srwm & 0x3f);
+			I915_WRITE(INSTPM, I915_READ(INSTPM) | INSTPM_SELF_EN);
+		}
 	} else {
 		/* Turn off self refresh if both pipes are enabled */
-		I915_WRITE(FW_BLC_SELF, I915_READ(FW_BLC_SELF)
-					& ~FW_BLC_SELF_EN);
+		if (IS_I945G(dev) || IS_I945GM(dev)) {
+			I915_WRITE(FW_BLC_SELF, I915_READ(FW_BLC_SELF)
+				   & ~FW_BLC_SELF_EN);
+		} else if (IS_I915GM(dev)) {
+			I915_WRITE(INSTPM, I915_READ(INSTPM) & ~INSTPM_SELF_EN);
+		}
 	}
 
 	DRM_DEBUG_KMS("Setting FIFO watermarks - A: %d, B: %d, C: %d, SR %d\n",
@@ -4011,6 +4022,11 @@ static void intel_idle_update(struct work_struct *work)
 
 	mutex_lock(&dev->struct_mutex);
 
+	if (IS_I945G(dev) || IS_I945GM(dev)) {
+		DRM_DEBUG_DRIVER("enable memory self refresh on 945\n");
+		I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN_MASK | FW_BLC_SELF_EN);
+	}
+
 	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
 		/* Skip inactive CRTCs */
 		if (!crtc->fb)
@@ -4044,6 +4060,15 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj)
 	if (!drm_core_check_feature(dev, DRIVER_MODESET))
 		return;
 
+	if (IS_I945G(dev) || IS_I945GM(dev)) {
+		u32 fw_blc_self;
+
+		DRM_DEBUG_DRIVER("disable memory self refresh on 945\n");
+		fw_blc_self = I915_READ(FW_BLC_SELF);
+		fw_blc_self &= ~FW_BLC_SELF_EN;
+		I915_WRITE(FW_BLC_SELF, fw_blc_self | FW_BLC_SELF_EN_MASK);
+	}
+
 	if (!dev_priv->busy)
 		dev_priv->busy = true;
 	else
-- 
cgit v1.2.3


From f97108d1d0facc7902134ebc453b226bbd4d1cdb Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Fri, 29 Jan 2010 11:27:07 -0800
Subject: drm/i915: add dynamic performance control support for Ironlake

Ironlake (and 965GM, which this patch doesn't support) supports a
hardware performance and power management feature that allows it to
adjust to changes in GPU load over time with software help.  The goal
if this is to maximize performance/power for a given workload.

This patch enables that feature, which is also a requirement for
supporting Intelligent Power Sharing, a feature which allows for
dynamic budgeting of power between the CPU and GPU in Arrandale
platforms.

Tested-by: ykzhao <yakui.zhao@intel.com>
[anholt: Resolved against the irq handler loop removal]
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c  |  96 ++++++++++++++++++++++++
 drivers/gpu/drm/i915/i915_drv.c      |   5 ++
 drivers/gpu/drm/i915/i915_drv.h      |   6 ++
 drivers/gpu/drm/i915/i915_irq.c      |  63 ++++++++++++++++
 drivers/gpu/drm/i915/i915_reg.h      | 141 ++++++++++++++++++++++++++++++++++-
 drivers/gpu/drm/i915/i915_suspend.c  |   8 ++
 drivers/gpu/drm/i915/intel_display.c | 100 +++++++++++++++++++++++--
 drivers/gpu/drm/i915/intel_drv.h     |   2 +
 8 files changed, 411 insertions(+), 10 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index a894ade0309..55340de618e 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -386,6 +386,97 @@ out:
 	return 0;
 }
 
+static int i915_rstdby_delays(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u16 crstanddelay = I915_READ16(CRSTANDVID);
+
+	seq_printf(m, "w/ctx: %d, w/o ctx: %d\n", (crstanddelay >> 8) & 0x3f, (crstanddelay & 0x3f));
+
+	return 0;
+}
+
+static int i915_cur_delayinfo(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u16 rgvswctl = I915_READ16(MEMSWCTL);
+
+	seq_printf(m, "Last command: 0x%01x\n", (rgvswctl >> 13) & 0x3);
+	seq_printf(m, "Command status: %d\n", (rgvswctl >> 12) & 1);
+	seq_printf(m, "P%d DELAY 0x%02x\n", (rgvswctl >> 8) & 0xf,
+		   rgvswctl & 0x3f);
+
+	return 0;
+}
+
+static int i915_delayfreq_table(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u32 delayfreq;
+	int i;
+
+	for (i = 0; i < 16; i++) {
+		delayfreq = I915_READ(PXVFREQ_BASE + i * 4);
+		seq_printf(m, "P%02dVIDFREQ: 0x%08x\n", i, delayfreq);
+	}
+
+	return 0;
+}
+
+static inline int MAP_TO_MV(int map)
+{
+	return 1250 - (map * 25);
+}
+
+static int i915_inttoext_table(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u32 inttoext;
+	int i;
+
+	for (i = 1; i <= 32; i++) {
+		inttoext = I915_READ(INTTOEXT_BASE_ILK + i * 4);
+		seq_printf(m, "INTTOEXT%02d: 0x%08x\n", i, inttoext);
+	}
+
+	return 0;
+}
+
+static int i915_drpc_info(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u32 rgvmodectl = I915_READ(MEMMODECTL);
+
+	seq_printf(m, "HD boost: %s\n", (rgvmodectl & MEMMODE_BOOST_EN) ?
+		   "yes" : "no");
+	seq_printf(m, "Boost freq: %d\n",
+		   (rgvmodectl & MEMMODE_BOOST_FREQ_MASK) >>
+		   MEMMODE_BOOST_FREQ_SHIFT);
+	seq_printf(m, "HW control enabled: %s\n",
+		   rgvmodectl & MEMMODE_HWIDLE_EN ? "yes" : "no");
+	seq_printf(m, "SW control enabled: %s\n",
+		   rgvmodectl & MEMMODE_SWMODE_EN ? "yes" : "no");
+	seq_printf(m, "Gated voltage change: %s\n",
+		   rgvmodectl & MEMMODE_RCLK_GATE ? "yes" : "no");
+	seq_printf(m, "Starting frequency: P%d\n",
+		   (rgvmodectl & MEMMODE_FSTART_MASK) >> MEMMODE_FSTART_SHIFT);
+	seq_printf(m, "Max frequency: P%d\n",
+		   (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT);
+	seq_printf(m, "Min frequency: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK));
+
+	return 0;
+}
+
 static int
 i915_wedged_open(struct inode *inode,
 		 struct file *filp)
@@ -503,6 +594,11 @@ static struct drm_info_list i915_debugfs_list[] = {
 	{"i915_ringbuffer_info", i915_ringbuffer_info, 0},
 	{"i915_batchbuffers", i915_batchbuffer_info, 0},
 	{"i915_error_state", i915_error_state, 0},
+	{"i915_rstdby_delays", i915_rstdby_delays, 0},
+	{"i915_cur_delayinfo", i915_cur_delayinfo, 0},
+	{"i915_delayfreq_table", i915_delayfreq_table, 0},
+	{"i915_inttoext_table", i915_inttoext_table, 0},
+	{"i915_drpc_info", i915_drpc_info, 0},
 };
 #define I915_DEBUGFS_ENTRIES ARRAY_SIZE(i915_debugfs_list)
 
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 79beffcf593..89f1cb86c32 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -583,6 +583,11 @@ static int __init i915_init(void)
 		driver.driver_features &= ~DRIVER_MODESET;
 #endif
 
+	if (!(driver.driver_features & DRIVER_MODESET)) {
+		driver.suspend = i915_suspend;
+		driver.resume = i915_resume;
+	}
+
 	return drm_init(&driver);
 }
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index f2742d6d3f5..2e493ec1042 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -452,6 +452,7 @@ typedef struct drm_i915_private {
 	u32 savePIPEB_DATA_N1;
 	u32 savePIPEB_LINK_M1;
 	u32 savePIPEB_LINK_N1;
+	u32 saveRSTDBYCTL;
 
 	struct {
 		struct drm_mm gtt_space;
@@ -590,7 +591,12 @@ typedef struct drm_i915_private {
 	int child_dev_num;
 	struct child_device_config *child_dev;
 	struct drm_connector *int_lvds_connector;
+
 	bool mchbar_need_disable;
+
+	u8 cur_delay;
+	u8 min_delay;
+	u8 max_delay;
 } drm_i915_private_t;
 
 /** driver private structure attached to each drm_gem_object */
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index a17d6bdfe63..8b35f5e1c51 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -269,6 +269,57 @@ static void i915_hotplug_work_func(struct work_struct *work)
 	drm_sysfs_hotplug_event(dev);
 }
 
+static void i915_handle_rps_change(struct drm_device *dev)
+{
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	u32 slow_up, slow_down, max_avg, min_avg;
+	u16 rgvswctl;
+	u8 new_delay = dev_priv->cur_delay;
+
+	I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS) & ~MEMINT_EVAL_CHG);
+	slow_up = I915_READ(RCPREVBSYTUPAVG);
+	slow_down = I915_READ(RCPREVBSYTDNAVG);
+	max_avg = I915_READ(RCBMAXAVG);
+	min_avg = I915_READ(RCBMINAVG);
+
+	/* Handle RCS change request from hw */
+	if (slow_up > max_avg) {
+		if (dev_priv->cur_delay != dev_priv->max_delay)
+			new_delay = dev_priv->cur_delay - 1;
+		if (new_delay < dev_priv->max_delay)
+			new_delay = dev_priv->max_delay;
+	} else if (slow_down < min_avg) {
+		if (dev_priv->cur_delay != dev_priv->min_delay)
+			new_delay = dev_priv->cur_delay + 1;
+		if (new_delay > dev_priv->min_delay)
+			new_delay = dev_priv->min_delay;
+	}
+
+	DRM_DEBUG("rps change requested: %d -> %d\n",
+		  dev_priv->cur_delay, new_delay);
+
+	rgvswctl = I915_READ(MEMSWCTL);
+	if (rgvswctl & MEMCTL_CMD_STS) {
+		DRM_ERROR("gpu slow, RCS change rejected\n");
+		return; /* still slow with another command */
+	}
+
+	/* Program the new state */
+	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
+		(new_delay << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
+	I915_WRITE(MEMSWCTL, rgvswctl);
+	POSTING_READ(MEMSWCTL);
+
+	rgvswctl |= MEMCTL_CMD_STS;
+	I915_WRITE(MEMSWCTL, rgvswctl);
+
+	dev_priv->cur_delay = new_delay;
+
+	DRM_DEBUG("rps changed\n");
+
+	return;
+}
+
 irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
@@ -331,6 +382,11 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 		queue_work(dev_priv->wq, &dev_priv->hotplug_work);
 	}
 
+	if (de_iir & DE_PCU_EVENT) {
+		I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS));
+		i915_handle_rps_change(dev);
+	}
+
 	/* should clear PCH hotplug event before clear CPU irq */
 	I915_WRITE(SDEIIR, pch_iir);
 	I915_WRITE(GTIIR, gt_iir);
@@ -1064,6 +1120,13 @@ static int ironlake_irq_postinstall(struct drm_device *dev)
 	I915_WRITE(SDEIER, dev_priv->pch_irq_enable_reg);
 	(void) I915_READ(SDEIER);
 
+	if (IS_IRONLAKE_M(dev)) {
+		/* Clear & enable PCU event interrupts */
+		I915_WRITE(DEIIR, DE_PCU_EVENT);
+		I915_WRITE(DEIER, I915_READ(DEIER) | DE_PCU_EVENT);
+		ironlake_enable_display_irq(dev_priv, DE_PCU_EVENT);
+	}
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 6defb7f4734..c3948ee37c1 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -61,6 +61,7 @@
 #define   GC_CLOCK_100_200		(1 << 0)
 #define   GC_CLOCK_100_133		(2 << 0)
 #define   GC_CLOCK_166_250		(3 << 0)
+#define GCFGC2	0xda
 #define GCFGC	0xf0 /* 915+ only */
 #define   GC_LOW_FREQUENCY_ENABLE	(1 << 7)
 #define   GC_DISPLAY_CLOCK_190_200_MHZ	(0 << 4)
@@ -282,7 +283,7 @@
 #define   I915_PIPE_CONTROL_NOTIFY_INTERRUPT		(1<<18)
 #define   I915_DISPLAY_PORT_INTERRUPT			(1<<17)
 #define   I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT	(1<<15)
-#define   I915_GMCH_THERMAL_SENSOR_EVENT_INTERRUPT	(1<<14)
+#define   I915_GMCH_THERMAL_SENSOR_EVENT_INTERRUPT	(1<<14) /* p-state */
 #define   I915_HWB_OOM_INTERRUPT			(1<<13)
 #define   I915_SYNC_STATUS_INTERRUPT			(1<<12)
 #define   I915_DISPLAY_PLANE_A_FLIP_PENDING_INTERRUPT	(1<<11)
@@ -787,10 +788,144 @@
 #define CLKCFG_MEM_800					(3 << 4)
 #define CLKCFG_MEM_MASK					(7 << 4)
 
-/** GM965 GM45 render standby register */
-#define MCHBAR_RENDER_STANDBY	0x111B8
+#define CRSTANDVID		0x11100
+#define PXVFREQ_BASE		0x11110 /* P[0-15]VIDFREQ (0x1114c) (Ironlake) */
+#define   PXVFREQ_PX_MASK	0x7f000000
+#define   PXVFREQ_PX_SHIFT	24
+#define VIDFREQ_BASE		0x11110
+#define VIDFREQ1		0x11110 /* VIDFREQ1-4 (0x1111c) (Cantiga) */
+#define VIDFREQ2		0x11114
+#define VIDFREQ3		0x11118
+#define VIDFREQ4		0x1111c
+#define   VIDFREQ_P0_MASK	0x1f000000
+#define   VIDFREQ_P0_SHIFT	24
+#define   VIDFREQ_P0_CSCLK_MASK	0x00f00000
+#define   VIDFREQ_P0_CSCLK_SHIFT 20
+#define   VIDFREQ_P0_CRCLK_MASK	0x000f0000
+#define   VIDFREQ_P0_CRCLK_SHIFT 16
+#define   VIDFREQ_P1_MASK	0x00001f00
+#define   VIDFREQ_P1_SHIFT	8
+#define   VIDFREQ_P1_CSCLK_MASK	0x000000f0
+#define   VIDFREQ_P1_CSCLK_SHIFT 4
+#define   VIDFREQ_P1_CRCLK_MASK	0x0000000f
+#define INTTOEXT_BASE_ILK	0x11300
+#define INTTOEXT_BASE		0x11120 /* INTTOEXT1-8 (0x1113c) */
+#define   INTTOEXT_MAP3_SHIFT	24
+#define   INTTOEXT_MAP3_MASK	(0x1f << INTTOEXT_MAP3_SHIFT)
+#define   INTTOEXT_MAP2_SHIFT	16
+#define   INTTOEXT_MAP2_MASK	(0x1f << INTTOEXT_MAP2_SHIFT)
+#define   INTTOEXT_MAP1_SHIFT	8
+#define   INTTOEXT_MAP1_MASK	(0x1f << INTTOEXT_MAP1_SHIFT)
+#define   INTTOEXT_MAP0_SHIFT	0
+#define   INTTOEXT_MAP0_MASK	(0x1f << INTTOEXT_MAP0_SHIFT)
+#define MEMSWCTL		0x11170 /* Ironlake only */
+#define   MEMCTL_CMD_MASK	0xe000
+#define   MEMCTL_CMD_SHIFT	13
+#define   MEMCTL_CMD_RCLK_OFF	0
+#define   MEMCTL_CMD_RCLK_ON	1
+#define   MEMCTL_CMD_CHFREQ	2
+#define   MEMCTL_CMD_CHVID	3
+#define   MEMCTL_CMD_VMMOFF	4
+#define   MEMCTL_CMD_VMMON	5
+#define   MEMCTL_CMD_STS	(1<<12) /* write 1 triggers command, clears
+					   when command complete */
+#define   MEMCTL_FREQ_MASK	0x0f00 /* jitter, from 0-15 */
+#define   MEMCTL_FREQ_SHIFT	8
+#define   MEMCTL_SFCAVM		(1<<7)
+#define   MEMCTL_TGT_VID_MASK	0x007f
+#define MEMIHYST		0x1117c
+#define MEMINTREN		0x11180 /* 16 bits */
+#define   MEMINT_RSEXIT_EN	(1<<8)
+#define   MEMINT_CX_SUPR_EN	(1<<7)
+#define   MEMINT_CONT_BUSY_EN	(1<<6)
+#define   MEMINT_AVG_BUSY_EN	(1<<5)
+#define   MEMINT_EVAL_CHG_EN	(1<<4)
+#define   MEMINT_MON_IDLE_EN	(1<<3)
+#define   MEMINT_UP_EVAL_EN	(1<<2)
+#define   MEMINT_DOWN_EVAL_EN	(1<<1)
+#define   MEMINT_SW_CMD_EN	(1<<0)
+#define MEMINTRSTR		0x11182 /* 16 bits */
+#define   MEM_RSEXIT_MASK	0xc000
+#define   MEM_RSEXIT_SHIFT	14
+#define   MEM_CONT_BUSY_MASK	0x3000
+#define   MEM_CONT_BUSY_SHIFT	12
+#define   MEM_AVG_BUSY_MASK	0x0c00
+#define   MEM_AVG_BUSY_SHIFT	10
+#define   MEM_EVAL_CHG_MASK	0x0300
+#define   MEM_EVAL_BUSY_SHIFT	8
+#define   MEM_MON_IDLE_MASK	0x00c0
+#define   MEM_MON_IDLE_SHIFT	6
+#define   MEM_UP_EVAL_MASK	0x0030
+#define   MEM_UP_EVAL_SHIFT	4
+#define   MEM_DOWN_EVAL_MASK	0x000c
+#define   MEM_DOWN_EVAL_SHIFT	2
+#define   MEM_SW_CMD_MASK	0x0003
+#define   MEM_INT_STEER_GFX	0
+#define   MEM_INT_STEER_CMR	1
+#define   MEM_INT_STEER_SMI	2
+#define   MEM_INT_STEER_SCI	3
+#define MEMINTRSTS		0x11184
+#define   MEMINT_RSEXIT		(1<<7)
+#define   MEMINT_CONT_BUSY	(1<<6)
+#define   MEMINT_AVG_BUSY	(1<<5)
+#define   MEMINT_EVAL_CHG	(1<<4)
+#define   MEMINT_MON_IDLE	(1<<3)
+#define   MEMINT_UP_EVAL	(1<<2)
+#define   MEMINT_DOWN_EVAL	(1<<1)
+#define   MEMINT_SW_CMD		(1<<0)
+#define MEMMODECTL		0x11190
+#define   MEMMODE_BOOST_EN	(1<<31)
+#define   MEMMODE_BOOST_FREQ_MASK 0x0f000000 /* jitter for boost, 0-15 */
+#define   MEMMODE_BOOST_FREQ_SHIFT 24
+#define   MEMMODE_IDLE_MODE_MASK 0x00030000
+#define   MEMMODE_IDLE_MODE_SHIFT 16
+#define   MEMMODE_IDLE_MODE_EVAL 0
+#define   MEMMODE_IDLE_MODE_CONT 1
+#define   MEMMODE_HWIDLE_EN	(1<<15)
+#define   MEMMODE_SWMODE_EN	(1<<14)
+#define   MEMMODE_RCLK_GATE	(1<<13)
+#define   MEMMODE_HW_UPDATE	(1<<12)
+#define   MEMMODE_FSTART_MASK	0x00000f00 /* starting jitter, 0-15 */
+#define   MEMMODE_FSTART_SHIFT	8
+#define   MEMMODE_FMAX_MASK	0x000000f0 /* max jitter, 0-15 */
+#define   MEMMODE_FMAX_SHIFT	4
+#define   MEMMODE_FMIN_MASK	0x0000000f /* min jitter, 0-15 */
+#define RCBMAXAVG		0x1119c
+#define MEMSWCTL2		0x1119e /* Cantiga only */
+#define   SWMEMCMD_RENDER_OFF	(0 << 13)
+#define   SWMEMCMD_RENDER_ON	(1 << 13)
+#define   SWMEMCMD_SWFREQ	(2 << 13)
+#define   SWMEMCMD_TARVID	(3 << 13)
+#define   SWMEMCMD_VRM_OFF	(4 << 13)
+#define   SWMEMCMD_VRM_ON	(5 << 13)
+#define   CMDSTS		(1<<12)
+#define   SFCAVM		(1<<11)
+#define   SWFREQ_MASK		0x0380 /* P0-7 */
+#define   SWFREQ_SHIFT		7
+#define   TARVID_MASK		0x001f
+#define MEMSTAT_CTG		0x111a0
+#define RCBMINAVG		0x111a0
+#define RCUPEI			0x111b0
+#define RCDNEI			0x111b4
+#define RSTDBYCTL		0x111b8
 #define   RCX_SW_EXIT		(1<<23)
 #define   RSX_STATUS_MASK	0x00700000
+#define VIDCTL			0x111c0
+#define VIDSTS			0x111c8
+#define VIDSTART		0x111cc /* 8 bits */
+#define MEMSTAT_ILK			0x111f8
+#define   MEMSTAT_VID_MASK	0x7f00
+#define   MEMSTAT_VID_SHIFT	8
+#define   MEMSTAT_PSTATE_MASK	0x00f8
+#define   MEMSTAT_PSTATE_SHIFT  3
+#define   MEMSTAT_MON_ACTV	(1<<2)
+#define   MEMSTAT_SRC_CTL_MASK	0x0003
+#define   MEMSTAT_SRC_CTL_CORE	0
+#define   MEMSTAT_SRC_CTL_TRB	1
+#define   MEMSTAT_SRC_CTL_THM	2
+#define   MEMSTAT_SRC_CTL_STDBY 3
+#define RCPREVBSYTUPAVG		0x113b8
+#define RCPREVBSYTDNAVG		0x113bc
 #define PEG_BAND_GAP_DATA	0x14d68
 
 /*
diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c
index a3b90c9561d..2c346645acf 100644
--- a/drivers/gpu/drm/i915/i915_suspend.c
+++ b/drivers/gpu/drm/i915/i915_suspend.c
@@ -682,6 +682,7 @@ void i915_restore_display(struct drm_device *dev)
 		I915_WRITE(PCH_PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS);
 		I915_WRITE(PCH_PP_DIVISOR, dev_priv->savePP_DIVISOR);
 		I915_WRITE(PCH_PP_CONTROL, dev_priv->savePP_CONTROL);
+		I915_WRITE(RSTDBYCTL, dev_priv->saveRSTDBYCTL);
 	} else {
 		I915_WRITE(PFIT_PGM_RATIOS, dev_priv->savePFIT_PGM_RATIOS);
 		I915_WRITE(BLC_PWM_CTL, dev_priv->saveBLC_PWM_CTL);
@@ -745,11 +746,15 @@ int i915_save_state(struct drm_device *dev)
 		dev_priv->saveGTIMR = I915_READ(GTIMR);
 		dev_priv->saveFDI_RXA_IMR = I915_READ(FDI_RXA_IMR);
 		dev_priv->saveFDI_RXB_IMR = I915_READ(FDI_RXB_IMR);
+		dev_priv->saveRSTDBYCTL = I915_READ(RSTDBYCTL);
 	} else {
 		dev_priv->saveIER = I915_READ(IER);
 		dev_priv->saveIMR = I915_READ(IMR);
 	}
 
+	if (IS_IRONLAKE_M(dev))
+		ironlake_disable_drps(dev);
+
 	/* Cache mode state */
 	dev_priv->saveCACHE_MODE_0 = I915_READ(CACHE_MODE_0);
 
@@ -820,6 +825,9 @@ int i915_restore_state(struct drm_device *dev)
 	/* Clock gating state */
 	intel_init_clock_gating(dev);
 
+	if (IS_IRONLAKE_M(dev))
+		ironlake_enable_drps(dev);
+
 	/* Cache mode state */
 	I915_WRITE (CACHE_MODE_0, dev_priv->saveCACHE_MODE_0 | 0xffff0000);
 
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index af9ec217cd1..4a93f7a0f58 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4616,6 +4616,91 @@ err_unref:
 	return NULL;
 }
 
+void ironlake_enable_drps(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 rgvmodectl = I915_READ(MEMMODECTL), rgvswctl;
+	u8 fmax, fmin, fstart, vstart;
+	int i = 0;
+
+	/* 100ms RC evaluation intervals */
+	I915_WRITE(RCUPEI, 100000);
+	I915_WRITE(RCDNEI, 100000);
+
+	/* Set max/min thresholds to 90ms and 80ms respectively */
+	I915_WRITE(RCBMAXAVG, 90000);
+	I915_WRITE(RCBMINAVG, 80000);
+
+	I915_WRITE(MEMIHYST, 1);
+
+	/* Set up min, max, and cur for interrupt handling */
+	fmax = (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT;
+	fmin = (rgvmodectl & MEMMODE_FMIN_MASK);
+	fstart = (rgvmodectl & MEMMODE_FSTART_MASK) >>
+		MEMMODE_FSTART_SHIFT;
+	vstart = (I915_READ(PXVFREQ_BASE + (fstart * 4)) & PXVFREQ_PX_MASK) >>
+		PXVFREQ_PX_SHIFT;
+
+	dev_priv->max_delay = fstart; /* can't go to fmax w/o IPS */
+	dev_priv->min_delay = fmin;
+	dev_priv->cur_delay = fstart;
+
+	I915_WRITE(MEMINTREN, MEMINT_CX_SUPR_EN | MEMINT_EVAL_CHG_EN);
+
+	/*
+	 * Interrupts will be enabled in ironlake_irq_postinstall
+	 */
+
+	I915_WRITE(VIDSTART, vstart);
+	POSTING_READ(VIDSTART);
+
+	rgvmodectl |= MEMMODE_SWMODE_EN;
+	I915_WRITE(MEMMODECTL, rgvmodectl);
+
+	while (I915_READ(MEMSWCTL) & MEMCTL_CMD_STS) {
+		if (i++ > 100) {
+			DRM_ERROR("stuck trying to change perf mode\n");
+			break;
+		}
+		msleep(1);
+	}
+	msleep(1);
+
+	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
+		(fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
+	I915_WRITE(MEMSWCTL, rgvswctl);
+	POSTING_READ(MEMSWCTL);
+
+	rgvswctl |= MEMCTL_CMD_STS;
+	I915_WRITE(MEMSWCTL, rgvswctl);
+}
+
+void ironlake_disable_drps(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 rgvswctl;
+	u8 fstart;
+
+	/* Ack interrupts, disable EFC interrupt */
+	I915_WRITE(MEMINTREN, I915_READ(MEMINTREN) & ~MEMINT_EVAL_CHG_EN);
+	I915_WRITE(MEMINTRSTS, MEMINT_EVAL_CHG);
+	I915_WRITE(DEIER, I915_READ(DEIER) & ~DE_PCU_EVENT);
+	I915_WRITE(DEIIR, DE_PCU_EVENT);
+	I915_WRITE(DEIMR, I915_READ(DEIMR) | DE_PCU_EVENT);
+
+	/* Go back to the starting frequency */
+	fstart = (I915_READ(MEMMODECTL) & MEMMODE_FSTART_MASK) >>
+		MEMMODE_FSTART_SHIFT;
+	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
+		(fstart << MEMCTL_FREQ_SHIFT);
+	I915_WRITE(MEMSWCTL, rgvswctl);
+	msleep(1);
+	rgvswctl |= MEMCTL_CMD_STS;
+	I915_WRITE(MEMSWCTL, rgvswctl);
+	msleep(1);
+
+}
+
 void intel_init_clock_gating(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
@@ -4685,8 +4770,8 @@ void intel_init_clock_gating(struct drm_device *dev)
 
 		if (obj_priv) {
 			I915_WRITE(PWRCTXA, obj_priv->gtt_offset | PWRCTX_EN);
-			I915_WRITE(MCHBAR_RENDER_STANDBY,
-				   I915_READ(MCHBAR_RENDER_STANDBY) & ~RCX_SW_EXIT);
+			I915_WRITE(RSTDBYCTL, I915_READ(RSTDBYCTL) &
+				   ~RCX_SW_EXIT);
 		}
 	}
 }
@@ -4799,11 +4884,6 @@ void intel_modeset_init(struct drm_device *dev)
 	DRM_DEBUG_KMS("%d display pipe%s available.\n",
 		  num_pipe, num_pipe > 1 ? "s" : "");
 
-	if (IS_I85X(dev))
-		pci_read_config_word(dev->pdev, HPLLCC, &dev_priv->orig_clock);
-	else if (IS_I9XX(dev) || IS_G4X(dev))
-		pci_read_config_word(dev->pdev, GCFGC, &dev_priv->orig_clock);
-
 	for (i = 0; i < num_pipe; i++) {
 		intel_crtc_init(dev, i);
 	}
@@ -4812,6 +4892,9 @@ void intel_modeset_init(struct drm_device *dev)
 
 	intel_init_clock_gating(dev);
 
+	if (IS_IRONLAKE_M(dev))
+		ironlake_enable_drps(dev);
+
 	INIT_WORK(&dev_priv->idle_work, intel_idle_update);
 	setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer,
 		    (unsigned long)dev);
@@ -4859,6 +4942,9 @@ void intel_modeset_cleanup(struct drm_device *dev)
 		drm_gem_object_unreference(dev_priv->pwrctx);
 	}
 
+	if (IS_IRONLAKE_M(dev))
+		ironlake_disable_drps(dev);
+
 	mutex_unlock(&dev->struct_mutex);
 
 	drm_mode_config_cleanup(dev);
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index a51573da1ff..3a467ca5785 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -209,6 +209,8 @@ extern void intel_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green,
 extern void intel_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green,
 				    u16 *blue, int regno);
 extern void intel_init_clock_gating(struct drm_device *dev);
+extern void ironlake_enable_drps(struct drm_device *dev);
+extern void ironlake_disable_drps(struct drm_device *dev);
 
 extern int intel_framebuffer_create(struct drm_device *dev,
 				    struct drm_mode_fb_cmd *mode_cmd,
-- 
cgit v1.2.3


From b5b72e891a5a6056c849ef8eaf259f126090f88b Mon Sep 17 00:00:00 2001
From: Matthew Garrett <mjg59@srcf.ucam.org>
Date: Tue, 2 Feb 2010 18:30:47 +0000
Subject: drm/i915: Deobfuscate the render p-state obfuscation

The ironlake render p-state support includes some rather odd variable
names. Clean them up in order to improve the readability of the code.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_drv.h      |  2 +-
 drivers/gpu/drm/i915/i915_irq.c      | 14 +++++++-------
 drivers/gpu/drm/i915/i915_reg.h      |  2 +-
 drivers/gpu/drm/i915/i915_suspend.c  |  6 ++++--
 drivers/gpu/drm/i915/intel_display.c |  4 ++--
 5 files changed, 15 insertions(+), 13 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 2e493ec1042..7df89ae8172 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -452,7 +452,7 @@ typedef struct drm_i915_private {
 	u32 savePIPEB_DATA_N1;
 	u32 savePIPEB_LINK_M1;
 	u32 savePIPEB_LINK_N1;
-	u32 saveRSTDBYCTL;
+	u32 saveMCHBAR_RENDER_STANDBY;
 
 	struct {
 		struct drm_mm gtt_space;
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 8b35f5e1c51..1a56ae7b5a7 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -272,23 +272,23 @@ static void i915_hotplug_work_func(struct work_struct *work)
 static void i915_handle_rps_change(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
-	u32 slow_up, slow_down, max_avg, min_avg;
+	u32 busy_up, busy_down, max_avg, min_avg;
 	u16 rgvswctl;
 	u8 new_delay = dev_priv->cur_delay;
 
 	I915_WRITE(MEMINTRSTS, I915_READ(MEMINTRSTS) & ~MEMINT_EVAL_CHG);
-	slow_up = I915_READ(RCPREVBSYTUPAVG);
-	slow_down = I915_READ(RCPREVBSYTDNAVG);
+	busy_up = I915_READ(RCPREVBSYTUPAVG);
+	busy_down = I915_READ(RCPREVBSYTDNAVG);
 	max_avg = I915_READ(RCBMAXAVG);
 	min_avg = I915_READ(RCBMINAVG);
 
 	/* Handle RCS change request from hw */
-	if (slow_up > max_avg) {
+	if (busy_up > max_avg) {
 		if (dev_priv->cur_delay != dev_priv->max_delay)
 			new_delay = dev_priv->cur_delay - 1;
 		if (new_delay < dev_priv->max_delay)
 			new_delay = dev_priv->max_delay;
-	} else if (slow_down < min_avg) {
+	} else if (busy_down < min_avg) {
 		if (dev_priv->cur_delay != dev_priv->min_delay)
 			new_delay = dev_priv->cur_delay + 1;
 		if (new_delay > dev_priv->min_delay)
@@ -300,8 +300,8 @@ static void i915_handle_rps_change(struct drm_device *dev)
 
 	rgvswctl = I915_READ(MEMSWCTL);
 	if (rgvswctl & MEMCTL_CMD_STS) {
-		DRM_ERROR("gpu slow, RCS change rejected\n");
-		return; /* still slow with another command */
+		DRM_ERROR("gpu busy, RCS change rejected\n");
+		return; /* still busy with another command */
 	}
 
 	/* Program the new state */
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index c3948ee37c1..d344c031f18 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -907,7 +907,7 @@
 #define RCBMINAVG		0x111a0
 #define RCUPEI			0x111b0
 #define RCDNEI			0x111b4
-#define RSTDBYCTL		0x111b8
+#define MCHBAR_RENDER_STANDBY		0x111b8
 #define   RCX_SW_EXIT		(1<<23)
 #define   RSX_STATUS_MASK	0x00700000
 #define VIDCTL			0x111c0
diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c
index 2c346645acf..ac0d1a73ac2 100644
--- a/drivers/gpu/drm/i915/i915_suspend.c
+++ b/drivers/gpu/drm/i915/i915_suspend.c
@@ -682,7 +682,8 @@ void i915_restore_display(struct drm_device *dev)
 		I915_WRITE(PCH_PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS);
 		I915_WRITE(PCH_PP_DIVISOR, dev_priv->savePP_DIVISOR);
 		I915_WRITE(PCH_PP_CONTROL, dev_priv->savePP_CONTROL);
-		I915_WRITE(RSTDBYCTL, dev_priv->saveRSTDBYCTL);
+		I915_WRITE(MCHBAR_RENDER_STANDBY,
+			   dev_priv->saveMCHBAR_RENDER_STANDBY);
 	} else {
 		I915_WRITE(PFIT_PGM_RATIOS, dev_priv->savePFIT_PGM_RATIOS);
 		I915_WRITE(BLC_PWM_CTL, dev_priv->saveBLC_PWM_CTL);
@@ -746,7 +747,8 @@ int i915_save_state(struct drm_device *dev)
 		dev_priv->saveGTIMR = I915_READ(GTIMR);
 		dev_priv->saveFDI_RXA_IMR = I915_READ(FDI_RXA_IMR);
 		dev_priv->saveFDI_RXB_IMR = I915_READ(FDI_RXB_IMR);
-		dev_priv->saveRSTDBYCTL = I915_READ(RSTDBYCTL);
+		dev_priv->saveMCHBAR_RENDER_STANDBY =
+			I915_READ(MCHBAR_RENDER_STANDBY);
 	} else {
 		dev_priv->saveIER = I915_READ(IER);
 		dev_priv->saveIMR = I915_READ(IMR);
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 4a93f7a0f58..9e80020ae2f 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4770,8 +4770,8 @@ void intel_init_clock_gating(struct drm_device *dev)
 
 		if (obj_priv) {
 			I915_WRITE(PWRCTXA, obj_priv->gtt_offset | PWRCTX_EN);
-			I915_WRITE(RSTDBYCTL, I915_READ(RSTDBYCTL) &
-				   ~RCX_SW_EXIT);
+			I915_WRITE(MCHBAR_RENDER_STANDBY,
+				   I915_READ(MCHBAR_RENDER_STANDBY) & ~RCX_SW_EXIT);
 		}
 	}
 }
-- 
cgit v1.2.3


From 060e645ac0181f499478a7be0ccb8a5600edc71a Mon Sep 17 00:00:00 2001
From: Li Peng <peng.li@linux.intel.com>
Date: Wed, 10 Feb 2010 01:54:24 +0800
Subject: drm/i915: Fix OGLC performance regression on 945

He Shuang reported an OGLC performance regression introduced in the patch
"enable memory self refresh on 9xx", In that patch, SR on 945 is disabled
everytime when calling intel_mark_busy(), while too much of such operation
will impact performance. Actually disable SR is necessary only when GPU and
Crtc changing from idle to busy. This patch make such optimization.

It fixes upstream bug
http://bugs.freedesktop.org/show_bug.cgi?id=26422

Signed-off-by: Li Peng <peng.li@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/intel_display.c | 27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 9e80020ae2f..cba51c2b512 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4060,18 +4060,17 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj)
 	if (!drm_core_check_feature(dev, DRIVER_MODESET))
 		return;
 
-	if (IS_I945G(dev) || IS_I945GM(dev)) {
-		u32 fw_blc_self;
-
-		DRM_DEBUG_DRIVER("disable memory self refresh on 945\n");
-		fw_blc_self = I915_READ(FW_BLC_SELF);
-		fw_blc_self &= ~FW_BLC_SELF_EN;
-		I915_WRITE(FW_BLC_SELF, fw_blc_self | FW_BLC_SELF_EN_MASK);
-	}
+	if (!dev_priv->busy) {
+		if (IS_I945G(dev) || IS_I945GM(dev)) {
+			u32 fw_blc_self;
 
-	if (!dev_priv->busy)
+			DRM_DEBUG_DRIVER("disable memory self refresh on 945\n");
+			fw_blc_self = I915_READ(FW_BLC_SELF);
+			fw_blc_self &= ~FW_BLC_SELF_EN;
+			I915_WRITE(FW_BLC_SELF, fw_blc_self | FW_BLC_SELF_EN_MASK);
+		}
 		dev_priv->busy = true;
-	else
+	} else
 		mod_timer(&dev_priv->idle_timer, jiffies +
 			  msecs_to_jiffies(GPU_IDLE_TIMEOUT));
 
@@ -4083,6 +4082,14 @@ void intel_mark_busy(struct drm_device *dev, struct drm_gem_object *obj)
 		intel_fb = to_intel_framebuffer(crtc->fb);
 		if (intel_fb->obj == obj) {
 			if (!intel_crtc->busy) {
+				if (IS_I945G(dev) || IS_I945GM(dev)) {
+					u32 fw_blc_self;
+
+					DRM_DEBUG_DRIVER("disable memory self refresh on 945\n");
+					fw_blc_self = I915_READ(FW_BLC_SELF);
+					fw_blc_self &= ~FW_BLC_SELF_EN;
+					I915_WRITE(FW_BLC_SELF, fw_blc_self | FW_BLC_SELF_EN_MASK);
+				}
 				/* Non-busy -> busy, upclock */
 				intel_increase_pllclock(crtc, true);
 				intel_crtc->busy = true;
-- 
cgit v1.2.3


From 357b13c3e498bb658f511f91a9e4f09c9553be6e Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Thu, 4 Feb 2010 14:17:47 -0800
Subject: drm/i915: fix drps disable so unload & re-load works

At unload time, we need to disable DRPS, but we need to do it correctly
or the GPU will hang and we won't be able to load the module again.  So
set the SFCAVM bit so we can properly restore the DRPS config at unload.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/intel_display.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index cba51c2b512..e879885f475 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4699,7 +4699,7 @@ void ironlake_disable_drps(struct drm_device *dev)
 	fstart = (I915_READ(MEMMODECTL) & MEMMODE_FSTART_MASK) >>
 		MEMMODE_FSTART_SHIFT;
 	rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) |
-		(fstart << MEMCTL_FREQ_SHIFT);
+		(fstart << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM;
 	I915_WRITE(MEMSWCTL, rgvswctl);
 	msleep(1);
 	rgvswctl |= MEMCTL_CMD_STS;
-- 
cgit v1.2.3


From b5e50c3f56ee4aa0d0168eab5ece413ac5df76aa Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Fri, 5 Feb 2010 12:42:41 -0800
Subject: drm/i915: provide FBC status in debugfs

Tools like powertop want to check the current FBC status and report it
to the user.  So add a debugfs file indicating whether FBC is enabled,
and if not, why.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c  | 49 ++++++++++++++++++++++++++++++++++++
 drivers/gpu/drm/i915/i915_dma.c      |  2 ++
 drivers/gpu/drm/i915/i915_drv.h      | 10 ++++++++
 drivers/gpu/drm/i915/intel_display.c |  5 ++++
 4 files changed, 66 insertions(+)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 55340de618e..96e8b9b2cb3 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -477,6 +477,54 @@ static int i915_drpc_info(struct seq_file *m, void *unused)
 	return 0;
 }
 
+static int i915_fbc_status(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct drm_crtc *crtc;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	bool fbc_enabled = false;
+
+	if (!dev_priv->display.fbc_enabled) {
+		seq_printf(m, "FBC unsupported on this chipset\n");
+		return 0;
+	}
+
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		if (!crtc->enabled)
+			continue;
+		if (dev_priv->display.fbc_enabled(crtc))
+			fbc_enabled = true;
+	}
+
+	if (fbc_enabled) {
+		seq_printf(m, "FBC enabled\n");
+	} else {
+		seq_printf(m, "FBC disabled: ");
+		switch (dev_priv->no_fbc_reason) {
+		case FBC_STOLEN_TOO_SMALL:
+			seq_printf(m, "not enough stolen memory");
+			break;
+		case FBC_UNSUPPORTED_MODE:
+			seq_printf(m, "mode not supported");
+			break;
+		case FBC_MODE_TOO_LARGE:
+			seq_printf(m, "mode too large");
+			break;
+		case FBC_BAD_PLANE:
+			seq_printf(m, "FBC unsupported on plane");
+			break;
+		case FBC_NOT_TILED:
+			seq_printf(m, "scanout buffer not tiled");
+			break;
+		default:
+			seq_printf(m, "unknown reason");
+		}
+		seq_printf(m, "\n");
+	}
+	return 0;
+}
+
 static int
 i915_wedged_open(struct inode *inode,
 		 struct file *filp)
@@ -599,6 +647,7 @@ static struct drm_info_list i915_debugfs_list[] = {
 	{"i915_delayfreq_table", i915_delayfreq_table, 0},
 	{"i915_inttoext_table", i915_inttoext_table, 0},
 	{"i915_drpc_info", i915_drpc_info, 0},
+	{"i915_fbc_status", i915_fbc_status, 0},
 };
 #define I915_DEBUGFS_ENTRIES ARRAY_SIZE(i915_debugfs_list)
 
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 3afe361ec55..47805a41395 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1249,6 +1249,7 @@ static void i915_setup_compression(struct drm_device *dev, int size)
 	/* Leave 1M for line length buffer & misc. */
 	compressed_fb = drm_mm_search_free(&dev_priv->vram, size, 4096, 0);
 	if (!compressed_fb) {
+		dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL;
 		i915_warn_stolen(dev);
 		return;
 	}
@@ -1256,6 +1257,7 @@ static void i915_setup_compression(struct drm_device *dev, int size)
 	compressed_fb = drm_mm_get_block(compressed_fb, size, 4096);
 	if (!compressed_fb) {
 		i915_warn_stolen(dev);
+		dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL;
 		return;
 	}
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 7df89ae8172..953ad64e055 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -192,6 +192,14 @@ struct intel_device_info {
 	u8 cursor_needs_physical : 1;
 };
 
+enum no_fbc_reason {
+	FBC_STOLEN_TOO_SMALL, /* not enough space to hold compressed buffers */
+	FBC_UNSUPPORTED_MODE, /* interlace or doublescanned mode */
+	FBC_MODE_TOO_LARGE, /* mode too large for compression */
+	FBC_BAD_PLANE, /* fbc not supported on plane */
+	FBC_NOT_TILED, /* buffer not tiled */
+};
+
 typedef struct drm_i915_private {
 	struct drm_device *dev;
 
@@ -597,6 +605,8 @@ typedef struct drm_i915_private {
 	u8 cur_delay;
 	u8 min_delay;
 	u8 max_delay;
+
+	enum no_fbc_reason no_fbc_reason;
 } drm_i915_private_t;
 
 /** driver private structure attached to each drm_gem_object */
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index e879885f475..a483f41e9f2 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -1188,25 +1188,30 @@ static void intel_update_fbc(struct drm_crtc *crtc,
 	if (intel_fb->obj->size > dev_priv->cfb_size) {
 		DRM_DEBUG_KMS("framebuffer too large, disabling "
 				"compression\n");
+		dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL;
 		goto out_disable;
 	}
 	if ((mode->flags & DRM_MODE_FLAG_INTERLACE) ||
 	    (mode->flags & DRM_MODE_FLAG_DBLSCAN)) {
 		DRM_DEBUG_KMS("mode incompatible with compression, "
 				"disabling\n");
+		dev_priv->no_fbc_reason = FBC_UNSUPPORTED_MODE;
 		goto out_disable;
 	}
 	if ((mode->hdisplay > 2048) ||
 	    (mode->vdisplay > 1536)) {
 		DRM_DEBUG_KMS("mode too large for compression, disabling\n");
+		dev_priv->no_fbc_reason = FBC_MODE_TOO_LARGE;
 		goto out_disable;
 	}
 	if ((IS_I915GM(dev) || IS_I945GM(dev)) && plane != 0) {
 		DRM_DEBUG_KMS("plane not 0, disabling compression\n");
+		dev_priv->no_fbc_reason = FBC_BAD_PLANE;
 		goto out_disable;
 	}
 	if (obj_priv->tiling_mode != I915_TILING_X) {
 		DRM_DEBUG_KMS("framebuffer not tiled, disabling compression\n");
+		dev_priv->no_fbc_reason = FBC_NOT_TILED;
 		goto out_disable;
 	}
 
-- 
cgit v1.2.3


From 4a9bef37e5cf2c73b61ecc9ead52cad2eb4a372b Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Fri, 5 Feb 2010 12:47:35 -0800
Subject: drm/i915: provide self-refresh status in debugfs

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 96e8b9b2cb3..56095b3d28c 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -525,6 +525,26 @@ static int i915_fbc_status(struct seq_file *m, void *unused)
 	return 0;
 }
 
+static int i915_sr_status(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
+	bool sr_enabled = false;
+
+	if (IS_I965G(dev) || IS_I945G(dev) || IS_I945GM(dev))
+		sr_enabled = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN;
+	else if (IS_I915GM(dev))
+		sr_enabled = I915_READ(INSTPM) & INSTPM_SELF_EN;
+	else if (IS_PINEVIEW(dev))
+		sr_enabled = I915_READ(DSPFW3) & PINEVIEW_SELF_REFRESH_EN;
+
+	seq_printf(m, "self-refresh: %s\n", sr_enabled ? "enabled" :
+		   "disabled");
+
+	return 0;
+}
+
 static int
 i915_wedged_open(struct inode *inode,
 		 struct file *filp)
@@ -648,6 +668,7 @@ static struct drm_info_list i915_debugfs_list[] = {
 	{"i915_inttoext_table", i915_inttoext_table, 0},
 	{"i915_drpc_info", i915_drpc_info, 0},
 	{"i915_fbc_status", i915_fbc_status, 0},
+	{"i915_sr_status", i915_sr_status, 0},
 };
 #define I915_DEBUGFS_ENTRIES ARRAY_SIZE(i915_debugfs_list)
 
-- 
cgit v1.2.3


From 4667a4a52ffc3789c45f46096cdffb053591a8e1 Mon Sep 17 00:00:00 2001
From: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu, 11 Feb 2010 14:14:42 +0100
Subject: drm/i915: overlay: nuke readback to flush wc caches

I retested this and whatever this papered over, the problem doesn't seem
to exist anymore.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Eric Anholt <eric@anholt.net>
[anholt: fixed up compile warning]
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/intel_overlay.c | 5 -----
 1 file changed, 5 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 2639591c72e..1257dc3cb47 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -199,16 +199,11 @@ static struct overlay_registers *intel_overlay_map_regs_atomic(struct intel_over
 
 static void intel_overlay_unmap_regs_atomic(struct intel_overlay *overlay)
 {
-	struct drm_device *dev = overlay->dev;
-        drm_i915_private_t *dev_priv = dev->dev_private;
-
 	if (OVERLAY_NONPHYSICAL(overlay->dev))
 		io_mapping_unmap_atomic(overlay->virt_addr);
 
 	overlay->virt_addr = NULL;
 
-	I915_READ(OVADD); /* flush wc cashes */
-
 	return;
 }
 
-- 
cgit v1.2.3


From 4f8a567c4666c73284832240db89fdadb2c50bd5 Mon Sep 17 00:00:00 2001
From: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu, 11 Feb 2010 14:14:43 +0100
Subject: drm/i915: overlay: drop superflous gpu flushes

Cache-coherency is maintained by gem. Drop these leftover MI_FLUSH
commands from the userspace code.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/intel_overlay.c | 20 +++++---------------
 1 file changed, 5 insertions(+), 15 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 1257dc3cb47..f3086fd4773 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -220,9 +220,7 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 	overlay->active = 1;
 	overlay->hw_wedged = NEEDS_WAIT_FOR_FLIP;
 
-	BEGIN_LP_RING(6);
-	OUT_RING(MI_FLUSH);
-	OUT_RING(MI_NOOP);
+	BEGIN_LP_RING(4);
 	OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_ON);
 	OUT_RING(overlay->flip_addr | OFC_UPDATE);
 	OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
@@ -262,9 +260,7 @@ static void intel_overlay_continue(struct intel_overlay *overlay,
 	if (tmp & (1 << 17))
 		DRM_DEBUG("overlay underrun, DOVSTA: %x\n", tmp);
 
-	BEGIN_LP_RING(4);
-	OUT_RING(MI_FLUSH);
-	OUT_RING(MI_NOOP);
+	BEGIN_LP_RING(2);
 	OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE);
 	OUT_RING(flip_addr);
         ADVANCE_LP_RING();
@@ -333,9 +329,7 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 	/* wait for overlay to go idle */
 	overlay->hw_wedged = SWITCH_OFF_STAGE_1;
 
-	BEGIN_LP_RING(6);
-	OUT_RING(MI_FLUSH);
-	OUT_RING(MI_NOOP);
+	BEGIN_LP_RING(4);
 	OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE);
 	OUT_RING(flip_addr);
         OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
@@ -353,9 +347,7 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 	/* turn overlay off */
 	overlay->hw_wedged = SWITCH_OFF_STAGE_2;
 
-	BEGIN_LP_RING(6);
-        OUT_RING(MI_FLUSH);
-        OUT_RING(MI_NOOP);
+	BEGIN_LP_RING(4);
         OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_OFF);
 	OUT_RING(flip_addr);
         OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
@@ -430,9 +422,7 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 
 			overlay->hw_wedged = SWITCH_OFF_STAGE_2;
 
-			BEGIN_LP_RING(6);
-			OUT_RING(MI_FLUSH);
-			OUT_RING(MI_NOOP);
+			BEGIN_LP_RING(4);
 			OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_OFF);
 			OUT_RING(flip_addr);
 			OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP);
-- 
cgit v1.2.3


From 4a7266123fce399f695b62b7f87b467b317f1487 Mon Sep 17 00:00:00 2001
From: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Mon, 1 Feb 2010 13:59:16 +0100
Subject: drm/i915: move a gtt flush to the correct place

No functional change, because gtt flushing is a no-op. Still, try
to keep the bookkeeping accurate. The if is still slightly wrong
for with execbuf2 even i915-class hw doesn't always need a fence
reg for gpu access. But that's for somewhen lateron.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_gem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 4a141993a37..a236bfb3084 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2552,12 +2552,12 @@ i915_gem_object_put_fence_reg(struct drm_gem_object *obj)
 		int ret;
 
 		i915_gem_object_flush_gpu_write_domain(obj);
-		i915_gem_object_flush_gtt_write_domain(obj);
 		ret = i915_gem_object_wait_rendering(obj);
 		if (ret != 0)
 			return ret;
 	}
 
+	i915_gem_object_flush_gtt_write_domain(obj);
 	i915_gem_clear_fence_reg (obj);
 
 	return 0;
-- 
cgit v1.2.3


From 10ae9bd25acf394c8fa2f9d795dfa9cec4d19ed6 Mon Sep 17 00:00:00 2001
From: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Mon, 1 Feb 2010 13:59:17 +0100
Subject: drm/i915: blow away userspace mappings before fence change

This aligns it with the other user of i915_gem_clear_fence_reg,
which blows away the mapping before changing the fence reg.

Only affects userspace if it races against itself when changing
tiling parameters, i.e. behaviour is undefined, anyway.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_gem.c        | 6 ++++++
 drivers/gpu/drm/i915/i915_gem_tiling.c | 6 ------
 2 files changed, 6 insertions(+), 6 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a236bfb3084..7b12604a9eb 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2544,6 +2544,12 @@ i915_gem_object_put_fence_reg(struct drm_gem_object *obj)
 	if (obj_priv->fence_reg == I915_FENCE_REG_NONE)
 		return 0;
 
+	/* If we've changed tiling, GTT-mappings of the object
+	 * need to re-fault to ensure that the correct fence register
+	 * setup is in place.
+	 */
+	i915_gem_release_mmap(obj);
+
 	/* On the i915, GPU access to tiled buffers is via a fence,
 	 * therefore we must wait for any outstanding access to complete
 	 * before clearing the fence.
diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c
index f308fdf4372..b0cbe3a62f8 100644
--- a/drivers/gpu/drm/i915/i915_gem_tiling.c
+++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
@@ -371,12 +371,6 @@ i915_gem_set_tiling(struct drm_device *dev, void *data,
 			goto err;
 		}
 
-		/* If we've changed tiling, GTT-mappings of the object
-		 * need to re-fault to ensure that the correct fence register
-		 * setup is in place.
-		 */
-		i915_gem_release_mmap(obj);
-
 		obj_priv->tiling_mode = args->tiling_mode;
 		obj_priv->stride = args->stride;
 	}
-- 
cgit v1.2.3


From f590d279eb4978352af163a88b001f156c7147d2 Mon Sep 17 00:00:00 2001
From: Owain Ainsworth <zerooa@googlemail.com>
Date: Thu, 18 Feb 2010 15:33:00 +0000
Subject: drm/i915: reduce some of the duplication of tiling checking

i915_gem_object_fenceable was mostly just a repeat of the
i915_gem_object_fence_offset_ok, but also checking the size (which was
checkecd when we allowed that BO to be tiled in the first place). So
instead, export the latter function and use it in place.

Signed-Off-By: Owain G. Ainsworth <oga@openbsd.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_drv.h        |  3 ++-
 drivers/gpu/drm/i915/i915_gem.c        |  3 ++-
 drivers/gpu/drm/i915/i915_gem_tiling.c | 35 +---------------------------------
 3 files changed, 5 insertions(+), 36 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 953ad64e055..40b0da37b1f 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -914,7 +914,8 @@ void i915_gem_object_do_bit_17_swizzle(struct drm_gem_object *obj);
 void i915_gem_object_save_bit_17_swizzle(struct drm_gem_object *obj);
 bool i915_tiling_ok(struct drm_device *dev, int stride, int size,
 		    int tiling_mode);
-bool i915_obj_fenceable(struct drm_device *dev, struct drm_gem_object *obj);
+bool i915_gem_object_fence_offset_ok(struct drm_gem_object *obj,
+				     int tiling_mode);
 
 /* i915_gem_debug.c */
 void i915_gem_dump_object(struct drm_gem_object *obj, int len,
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 7b12604a9eb..59dcce054d1 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3253,7 +3253,8 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj,
 	             obj_priv->tiling_mode != I915_TILING_NONE;
 
 	/* Check fence reg constraints and rebind if necessary */
-	if (need_fence && !i915_obj_fenceable(dev, obj))
+	if (need_fence && !i915_gem_object_fence_offset_ok(obj,
+	    obj_priv->tiling_mode))
 		i915_gem_object_unbind(obj);
 
 	/* Choose the GTT offset for our buffer and put it there. */
diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c
index b0cbe3a62f8..ba247d1f9bc 100644
--- a/drivers/gpu/drm/i915/i915_gem_tiling.c
+++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
@@ -180,39 +180,6 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev)
 	dev_priv->mm.bit_6_swizzle_y = swizzle_y;
 }
 
-
-/**
- * Returns whether an object is currently fenceable.  If not, it may need
- * to be unbound and have its pitch adjusted.
- */
-bool
-i915_obj_fenceable(struct drm_device *dev, struct drm_gem_object *obj)
-{
-	struct drm_i915_gem_object *obj_priv = obj->driver_private;
-
-	if (IS_I965G(dev)) {
-		/* The 965 can have fences at any page boundary. */
-		if (obj->size & 4095)
-			return false;
-		return true;
-	} else if (IS_I9XX(dev)) {
-		if (obj_priv->gtt_offset & ~I915_FENCE_START_MASK)
-			return false;
-	} else {
-		if (obj_priv->gtt_offset & ~I830_FENCE_START_MASK)
-			return false;
-	}
-
-	/* Power of two sized... */
-	if (obj->size & (obj->size - 1))
-		return false;
-
-	/* Objects must be size aligned as well */
-	if (obj_priv->gtt_offset & (obj->size - 1))
-		return false;
-	return true;
-}
-
 /* Check pitch constriants for all chips & tiling formats */
 bool
 i915_tiling_ok(struct drm_device *dev, int stride, int size, int tiling_mode)
@@ -269,7 +236,7 @@ i915_tiling_ok(struct drm_device *dev, int stride, int size, int tiling_mode)
 	return true;
 }
 
-static bool
+bool
 i915_gem_object_fence_offset_ok(struct drm_gem_object *obj, int tiling_mode)
 {
 	struct drm_device *dev = obj->dev;
-- 
cgit v1.2.3


From 7b9c5abee98c54f85bcc04bd4d7ec8d5094c73f4 Mon Sep 17 00:00:00 2001
From: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Fri, 12 Feb 2010 09:30:00 -0800
Subject: drm/i915: give up on 8xx lid status

These old machines more often than not lie about their lid state.  So
don't use it to detect LVDS presence, but leave the event handler to
deal with lid open/close, when we might need to reset the mode.

Fixes kernel bug #15248

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/intel_lvds.c | 7 +++++++
 1 file changed, 7 insertions(+)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c
index b1d0acbae4e..1740577c617 100644
--- a/drivers/gpu/drm/i915/intel_lvds.c
+++ b/drivers/gpu/drm/i915/intel_lvds.c
@@ -648,8 +648,15 @@ static const struct dmi_system_id bad_lid_status[] = {
  */
 static enum drm_connector_status intel_lvds_detect(struct drm_connector *connector)
 {
+	struct drm_device *dev = connector->dev;
 	enum drm_connector_status status = connector_status_connected;
 
+	/* ACPI lid methods were generally unreliable in this generation, so
+	 * don't even bother.
+	 */
+	if (IS_I8XX(dev))
+		return connector_status_connected;
+
 	if (!dmi_check_system(bad_lid_status) && !acpi_lid_open())
 		status = connector_status_disconnected;
 
-- 
cgit v1.2.3


From 9df30794f609d9412f14cfd0eb7b45dd64d0b14e Mon Sep 17 00:00:00 2001
From: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu, 18 Feb 2010 10:24:56 +0000
Subject: drm/i915: Record batch buffer following GPU error

In order to improve our diagnostic capabilities following a GPU hang
and subsequent reset, we need to record the batch buffer that triggered
the error. We assume that the current batch buffer, plus a few details
about what else is on the active list, will be sufficient -- at the very
least an improvement over nothing.

The extra information is stored in /debug/dri/.../i915_error_state
following an error, and may be decoded using
intel_gpu_tools/tools/intel_error_decode.

v2: Avoid excessive work under spinlocks.
v3: Include ringbuffer for later analysis.
v4: Use kunmap correctly and record more buffer state.
v5: Search ringbuffer for current batch buffer
v6: Use a work fn for the impossible IRQ error case.
v7: Avoid non-atomic paths whilst in IRQ context.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/i915/i915_debugfs.c |  85 ++++++++++++++
 drivers/gpu/drm/i915/i915_dma.c     |   2 +
 drivers/gpu/drm/i915/i915_drv.h     |  21 ++++
 drivers/gpu/drm/i915/i915_irq.c     | 224 ++++++++++++++++++++++++++++++++++--
 drivers/gpu/drm/i915/i915_reg.h     |   1 +
 5 files changed, 326 insertions(+), 7 deletions(-)

(limited to 'drivers/gpu')

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 56095b3d28c..5eed4631244 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -350,6 +350,36 @@ static int i915_ringbuffer_info(struct seq_file *m, void *data)
 	return 0;
 }
 
+static const char *pin_flag(int pinned)
+{
+	if (pinned > 0)
+		return " P";
+	else if (pinned < 0)
+		return " p";
+	else
+		return "";
+}
+
+static const char *tiling_flag(int tiling)
+{
+	switch (tiling) {
+	default:
+	case I915_TILING_NONE: return "";
+	case I915_TILING_X: return " X";
+	case I915_TILING_Y: return " Y";
+	}
+}
+
+static const char *dirty_flag(int dirty)
+{
+	return dirty ? " dirty" : "";
+}
+
+static const char *purgeable_flag(int purgeable)
+{
+	return purgeable ? " purgeable" : "";
+}
+
 static int i915_error_state(struct seq_file *m, void *unused)
 {
 	struct drm_info_node *node = (struct drm_info_node *) m->private;
@@ -357,6 +387,7 @@ static int i915_error_state(struct seq_file *m, void *unused)
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct drm_i915_error_state *error;
 	unsigned long flags;
+	int i, page, offset, elt;
 
 	spin_lock_irqsave(&dev_priv->error_lock, flags);
 	if (!dev_priv->first_error) {
@@ -368,6 +399,7 @@ static int i915_error_state(struct seq_file *m, void *unused)
 
 	seq_printf(m, "Time: %ld s %ld us\n", error->time.tv_sec,
 		   error->time.tv_usec);
+	seq_printf(m, "PCI ID: 0x%04x\n", dev->pci_device);
 	seq_printf(m, "EIR: 0x%08x\n", error->eir);
 	seq_printf(m, "  PGTBL_ER: 0x%08x\n", error->pgtbl_er);
 	seq_printf(m, "  INSTPM: 0x%08x\n", error->instpm);
@@ -379,6 +411,59 @@ static int i915_error_state(struct seq_file *m, void *unused)
 		seq_printf(m, "  INSTPS: 0x%08x\n", error->instps);
 		seq_printf(m, "  INSTDONE1: 0x%08x\n", error->instdone1);
 	}
+	seq_printf(m, "seqno: 0x%08x\n", error->seqno);
+
+	if (error->active_bo_count) {
+		seq_printf(m, "Buffers [%d]:\n", error->active_bo_count);
+
+		for (i = 0; i < error->active_bo_count; i++) {
+			seq_printf(m, "  %08x %8zd %08x %08x %08x%s%s%s%s",
+				   error->active_bo[i].gtt_offset,
+				   error->active_bo[i].size,
+				   error->active_bo[i].read_domains,
+				   error->active_bo[i].write_domain,
+				   error->active_bo[i].seqno,
+				   pin_flag(error->active_bo[i].pinned),
+				   tiling_flag(error->active_bo[i].tiling),
+				   dirty_flag(error->active_bo[i].dirty),
+				   purgeable_flag(error->active_bo[i].purgeable));
+
+			if (error->active_bo[i].name)
+				seq_printf(m, " (name: %d)", error->active_bo[i].name);
+			if (error->active_bo[i].fence_reg != I915_FENCE_REG_NONE)
+				seq_printf(m, " (fence: %d)", error->active_bo[i].fence_reg);
+
+			seq_printf(m, "\n");
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(error->batchbuffer); i++) {
+		if (error->batchbuffer[i]) {
+			struct drm_i915_error_object *obj = error->batchbuffer[i];
+
+			seq_printf(m, "--- gtt_offset = 0x%08x\n", obj->gtt_offset);
+			offset = 0;
+			for (page = 0; page < obj->page_count; page++) {
+				for (elt = 0; elt < PAGE_SIZE/4; elt++) {
+					seq_printf(m, "%08x :  %08x\n", offset, obj->pages[page][elt]);
+					offset += 4;
+				}
+			}
+		}
+	}
+
+	if (error->ringbuffer) {
+		struct drm_i915_error_object *obj = error->ringbuffer;
+
+		seq_printf(m, "--- ringbuffer = 0x%08x\n", obj->gtt_offset);
+		offset = 0;
+		for (page = 0; page < obj->page_count; page++) {
+			for (elt = 0; elt < PAGE_SIZE/4; elt++) {
+				seq_printf(m, "%08x :  %08x\n", offset, obj->pages[page][elt]);
+				offset += 4;
+			}
+		}
+	}
 
 out:
 	spin_unlock_irqrestore(&dev_priv->error_lock, flags);
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 47805a41395..dbfe07c90cb 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1644,6 +1644,8 @@ int i915_driver_unload(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
+	i915_destroy_error_state(dev);
+
 	destroy_workqueue(dev_priv->wq);
 	del_timer_sync(&dev_priv->hangcheck_timer);
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 40b0da37b1f..ec06d4865a5 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -150,7 +150,27 @@ struct drm_i915_error_state {
 	u32 instps;
 	u32 instdone1;
 	u32 seqno;
+	u64 bbaddr;
 	struct timeval time;
+	struct drm_i915_error_object {
+		int page_count;
+		u32 gtt_offset;
+		u32 *pages[0];
+	} *ringbuffer, *batchbuffer[2];
+	struct drm_i915_error_buffer {
+		size_t size;
+		u32 name;
+		u32 seqno;
+		u32 gtt_offset;
+		u32 read_domains;
+		u32 write_domain;
+		u32 fence_reg;
+		s32 pinned:2;
+		u32 tiling:2;
+		u32 dirty:1;
+		u32 purgeable:1;
+	} *active_bo;
+	u32 active_bo_count;
 };
 
 struct drm_i915_display_funcs {
@@ -778,6 +798,7 @@ extern int i965_reset(struct drm_device *dev, u8 flags);
 
 /* i915_irq.c */
 void i915_hangcheck_elapsed(unsigned long data);
+void i915_destroy_error_state(struct drm_device *dev);
 extern int i915_irq_emit(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
 extern int i915_irq_wait(struct drm_device *dev, void *data,
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 1a56ae7b5a7..ba1d8314c1c 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -432,6 +432,121 @@ static void i915_error_work_func(struct work_struct *work)
 	}
 }
 
+static struct drm_i915_error_object *
+i915_error_object_create(struct drm_device *dev,
+			 struct drm_gem_object *src)
+{
+	struct drm_i915_error_object *dst;
+	struct drm_i915_gem_object *src_priv;
+	int page, page_count;
+
+	if (src == NULL)
+		return NULL;
+
+	src_priv = src->driver_private;
+	if (src_priv->pages == NULL)
+		return NULL;
+
+	page_count = src->size / PAGE_SIZE;
+
+	dst = kmalloc(sizeof(*dst) + page_count * sizeof (u32 *), GFP_ATOMIC);
+	if (dst == NULL)
+		return NULL;
+
+	for (page = 0; page < page_count; page++) {
+		void *s, *d = kmalloc(PAGE_SIZE, GFP_ATOMIC);
+		if (d == NULL)
+			goto unwind;
+		s = kmap_atomic(src_priv->pages[page], KM_USER0);
+		memcpy(d, s, PAGE_SIZE);
+		kunmap_atomic(s, KM_USER0);
+		dst->pages[page] = d;
+	}
+	dst->page_count = page_count;
+	dst->gtt_offset = src_priv->gtt_offset;
+
+	return dst;
+
+unwind:
+	while (page--)
+		kfree(dst->pages[page]);
+	kfree(dst);
+	return NULL;
+}
+
+static void
+i915_error_object_free(struct drm_i915_error_object *obj)
+{
+	int page;
+
+	if (obj == NULL)
+		return;
+
+	for (page = 0; page < obj->page_count; page++)
+		kfree(obj->pages[page]);
+
+	kfree(obj);
+}
+
+static void
+i915_error_state_free(struct drm_device *dev,
+		      struct drm_i915_error_state *error)
+{
+	i915_error_object_free(error->batchbuffer[0]);
+	i915_error_object_free(error->batchbuffer[1]);
+	i915_error_object_free(error->ringbuffer);
+	kfree(error->active_bo);
+	kfree(error);
+}
+
+static u32
+i915_get_bbaddr(struct drm_device *dev, u32 *ring)
+{
+	u32 cmd;
+
+	if (IS_I830(dev) || IS_845G(dev))
+		cmd = MI_BATCH_BUFFER;
+	else if (IS_I965G(dev))
+		cmd = (MI_BATCH_BUFFER_START | (2 << 6) |
+		       MI_BATCH_NON_SECURE_I965);
+	else
+		cmd = (MI_BATCH_BUFFER_START | (2 << 6));
+
+	return ring[0] == cmd ? ring[1] : 0;
+}
+
+static u32
+i915_ringbuffer_last_batch(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 head, bbaddr;
+	u32 *ring;
+
+	/* Locate the current position in the ringbuffer and walk back
+	 * to find the most recently dispatched batch buffer.
+	 */
+	bbaddr = 0;
+	head = I915_READ(PRB0_HEAD) & HEAD_ADDR;
+	ring = (u32 *)(dev_priv->ring.virtual_start + head);
+
+	while (--ring >= (u32 *)dev_priv->ring.virtual_start) {
+		bbaddr = i915_get_bbaddr(dev, ring);
+		if (bbaddr)
+			break;
+	}
+
+	if (bbaddr == 0) {
+		ring = (u32 *)(dev_priv->ring.virtual_start + dev_priv->ring.Size);
+		while (--ring >= (u32 *)dev_priv->ring.virtual_start) {
+			bbaddr = i915_get_bbaddr(dev, ring);
+			if (bbaddr)
+				break;
+		}
+	}
+
+	return bbaddr;
+}
+
 /**
  * i915_capture_error_state - capture an error record for later analysis
  * @dev: drm device
@@ -444,19 +559,26 @@ static void i915_error_work_func(struct work_struct *work)
 static void i915_capture_error_state(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_i915_gem_object *obj_priv;
 	struct drm_i915_error_state *error;
+	struct drm_gem_object *batchbuffer[2];
 	unsigned long flags;
+	u32 bbaddr;
+	int count;
 
 	spin_lock_irqsave(&dev_priv->error_lock, flags);
-	if (dev_priv->first_error)
-		goto out;
+	error = dev_priv->first_error;
+	spin_unlock_irqrestore(&dev_priv->error_lock, flags);
+	if (error)
+		return;
 
 	error = kmalloc(sizeof(*error), GFP_ATOMIC);
 	if (!error) {
-		DRM_DEBUG_DRIVER("out ot memory, not capturing error state\n");
-		goto out;
+		DRM_DEBUG_DRIVER("out of memory, not capturing error state\n");
+		return;
 	}
 
+	error->seqno = i915_get_gem_seqno(dev);
 	error->eir = I915_READ(EIR);
 	error->pgtbl_er = I915_READ(PGTBL_ER);
 	error->pipeastat = I915_READ(PIPEASTAT);
@@ -467,6 +589,7 @@ static void i915_capture_error_state(struct drm_device *dev)
 		error->ipehr = I915_READ(IPEHR);
 		error->instdone = I915_READ(INSTDONE);
 		error->acthd = I915_READ(ACTHD);
+		error->bbaddr = 0;
 	} else {
 		error->ipeir = I915_READ(IPEIR_I965);
 		error->ipehr = I915_READ(IPEHR_I965);
@@ -474,14 +597,101 @@ static void i915_capture_error_state(struct drm_device *dev)
 		error->instps = I915_READ(INSTPS);
 		error->instdone1 = I915_READ(INSTDONE1);
 		error->acthd = I915_READ(ACTHD_I965);
+		error->bbaddr = I915_READ64(BB_ADDR);
 	}
 
-	do_gettimeofday(&error->time);
+	bbaddr = i915_ringbuffer_last_batch(dev);
+
+	/* Grab the current batchbuffer, most likely to have crashed. */
+	batchbuffer[0] = NULL;
+	batchbuffer[1] = NULL;
+	count = 0;
+	list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+		struct drm_gem_object *obj = obj_priv->obj;
+
+		if (batchbuffer[0] == NULL &&
+		    bbaddr >= obj_priv->gtt_offset &&
+		    bbaddr < obj_priv->gtt_offset + obj->size)
+			batchbuffer[0] = obj;
+
+		if (batchbuffer[1] == NULL &&
+		    error->acthd >= obj_priv->gtt_offset &&
+		    error->acthd < obj_priv->gtt_offset + obj->size &&
+		    batchbuffer[0] != obj)
+			batchbuffer[1] = obj;
+
+		count++;
+	}
 
-	dev_priv->first_error = error;
+	/* We need to copy these to an anonymous buffer as the simplest
+	 * method to avoid being overwritten by userpace.
+	 */
+	error->batchbuffer[0] = i915_error_object_create(dev, batchbuffer[0]);
+	error->batchbuffer[1] = i915_error_object_create(dev, batchbuffer[1]);
+
+	/* Record the ringbuffer */
+	error->ringbuffer = i915_error_object_create(dev, dev_priv->ring.ring_obj);
+
+	/* Record buffers on the active list. */
+	error->active_bo = NULL;
+	error->active_bo_count = 0;
+
+	if (count)
+		error->active_bo = kmalloc(sizeof(*error->active_bo)*count,
+					   GFP_ATOMIC);
+
+	if (error->active_bo) {
+		int i = 0;
+		list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+			struct drm_gem_object *obj = obj_priv->obj;
+
+			error->active_bo[i].size = obj->size;
+			error->active_bo[i].name = obj->name;
+			error->active_bo[i].seqno = obj_priv->last_rendering_seqno;
+			error->active_bo[i].gtt_offset = obj_priv->gtt_offset;
+			error->active_bo[i].read_domains = obj->read_domains;
+			error->active_bo[i].write_domain = obj->write_domain;
+			error->active_bo[i].fence_reg = obj_priv->fence_reg;
+			error->active_bo[i].pinned = 0;
+			if (obj_priv->pin_count > 0)
+				error->active_bo[i].pinned = 1;
+			if (obj_priv->user_pin_count > 0)
+				error->active_bo[i].pinned = -1;
+			error->active_bo[i].tiling = obj_priv->tiling_mode;
+			error->active_bo[i].dirty = obj_priv->dirty;
+			error->active_bo[i].purgeable = obj_priv->madv != I915_MADV_WILLNEED;
+
+			if (++i == count)
+				break;
+		}
+		error->active_bo_count = i;
+	}
+
+	do_gettimeofday(&error->time);
 
-out:
+	spin_lock_irqsave(&dev_priv->error_lock, flags);
+	if (dev_priv->first_error == NULL) {
+		dev_priv->first_error = error;
+		error = NULL;
+	}
 	spin_unlock_irqrestore(&dev_priv->error_lock, flags);
+
+	if (error)
+		i915_error_state_free(dev, error);
+}
+
+void i915_destroy_error_state(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_i915_error_state *error;
+
+	spin_lock(&dev_priv->error_lock);
+	error = dev_priv->first_error;
+	dev_priv->first_error = NULL;
+	spin_unlock(&dev_priv->error_lock);
+
+	if (error)
+		i915_error_state_free(dev, error);
 }
 
 /**
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index d344c031f18..eff8d850a75 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -328,6 +328,7 @@
 #define   CM0_COLOR_EVICT_DISABLE (1<<3)
 #define   CM0_DEPTH_WRITE_DISABLE (1<<1)
 #define   CM0_RC_OP_FLUSH_DISABLE (1<<0)
+#define BB_ADDR		0x02140 /* 8 bytes */
 #define GFX_FLSH_CNTL	0x02170 /* 915+ only */
 
 
-- 
cgit v1.2.3