Merge tag 'drm-fixes-2023-01-01' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Daniel Vetter:
 "I'm just back from the mountains, and Dave is out at the beach and
  should be back in a week again. Just i915 fixes and since Rodrigo
  bothered to make the pull last week I figured I should warm up gpg and
  forward this in a nice signed tag as a new years present!

   - i915 fixes for newer platforms

   - i915 locking rework to not give up in vm eviction fallback path too
     early"

* tag 'drm-fixes-2023-01-01' of git://anongit.freedesktop.org/drm/drm:
  drm/i915/dsi: fix MIPI_BKLT_EN_1 native GPIO index
  drm/i915/dsi: add support for ICL+ native MIPI GPIO sequence
  drm/i915/uc: Fix two issues with over-size firmware files
  drm/i915: improve the catch-all evict to handle lock contention
  drm/i915: Remove __maybe_unused from mtl_info
  drm/i915: fix TLB invalidation for Gen12.50 video and compute engines

drivers/gpu/drm/i915/display/intel_dsi_vbt.c

@@ -41,9 +41,11 @@
 
 #include "i915_drv.h"
 #include "i915_reg.h"
+#include "intel_de.h"
 #include "intel_display_types.h"
 #include "intel_dsi.h"
 #include "intel_dsi_vbt.h"
+#include "intel_gmbus_regs.h"
 #include "vlv_dsi.h"
 #include "vlv_dsi_regs.h"
 #include "vlv_sideband.h"
@@ -377,6 +379,85 @@ static void icl_exec_gpio(struct intel_connector *connector,
 	drm_dbg_kms(&dev_priv->drm, "Skipping ICL GPIO element execution\n");
 }
 
+enum {
+	MIPI_RESET_1 = 0,
+	MIPI_AVDD_EN_1,
+	MIPI_BKLT_EN_1,
+	MIPI_AVEE_EN_1,
+	MIPI_VIO_EN_1,
+	MIPI_RESET_2,
+	MIPI_AVDD_EN_2,
+	MIPI_BKLT_EN_2,
+	MIPI_AVEE_EN_2,
+	MIPI_VIO_EN_2,
+};
+
+static void icl_native_gpio_set_value(struct drm_i915_private *dev_priv,
+				      int gpio, bool value)
+{
+	int index;
+
+	if (drm_WARN_ON(&dev_priv->drm, DISPLAY_VER(dev_priv) == 11 && gpio >= MIPI_RESET_2))
+		return;
+
+	switch (gpio) {
+	case MIPI_RESET_1:
+	case MIPI_RESET_2:
+		index = gpio == MIPI_RESET_1 ? HPD_PORT_A : HPD_PORT_B;
+
+		/*
+		 * Disable HPD to set the pin to output, and set output
+		 * value. The HPD pin should not be enabled for DSI anyway,
+		 * assuming the board design and VBT are sane, and the pin isn't
+		 * used by a non-DSI encoder.
+		 *
+		 * The locking protects against concurrent SHOTPLUG_CTL_DDI
+		 * modifications in irq setup and handling.
+		 */
+		spin_lock_irq(&dev_priv->irq_lock);
+		intel_de_rmw(dev_priv, SHOTPLUG_CTL_DDI,
+			     SHOTPLUG_CTL_DDI_HPD_ENABLE(index) |
+			     SHOTPLUG_CTL_DDI_HPD_OUTPUT_DATA(index),
+			     value ? SHOTPLUG_CTL_DDI_HPD_OUTPUT_DATA(index) : 0);
+		spin_unlock_irq(&dev_priv->irq_lock);
+		break;
+	case MIPI_AVDD_EN_1:
+	case MIPI_AVDD_EN_2:
+		index = gpio == MIPI_AVDD_EN_1 ? 0 : 1;
+
+		intel_de_rmw(dev_priv, PP_CONTROL(index), PANEL_POWER_ON,
+			     value ? PANEL_POWER_ON : 0);
+		break;
+	case MIPI_BKLT_EN_1:
+	case MIPI_BKLT_EN_2:
+		index = gpio == MIPI_BKLT_EN_1 ? 0 : 1;
+
+		intel_de_rmw(dev_priv, PP_CONTROL(index), EDP_BLC_ENABLE,
+			     value ? EDP_BLC_ENABLE : 0);
+		break;
+	case MIPI_AVEE_EN_1:
+	case MIPI_AVEE_EN_2:
+		index = gpio == MIPI_AVEE_EN_1 ? 1 : 2;
+
+		intel_de_rmw(dev_priv, GPIO(dev_priv, index),
+			     GPIO_CLOCK_VAL_OUT,
+			     GPIO_CLOCK_DIR_MASK | GPIO_CLOCK_DIR_OUT |
+			     GPIO_CLOCK_VAL_MASK | (value ? GPIO_CLOCK_VAL_OUT : 0));
+		break;
+	case MIPI_VIO_EN_1:
+	case MIPI_VIO_EN_2:
+		index = gpio == MIPI_VIO_EN_1 ? 1 : 2;
+
+		intel_de_rmw(dev_priv, GPIO(dev_priv, index),
+			     GPIO_DATA_VAL_OUT,
+			     GPIO_DATA_DIR_MASK | GPIO_DATA_DIR_OUT |
+			     GPIO_DATA_VAL_MASK | (value ? GPIO_DATA_VAL_OUT : 0));
+		break;
+	default:
+		MISSING_CASE(gpio);
+	}
+}
+
 static const u8 *mipi_exec_gpio(struct intel_dsi *intel_dsi, const u8 *data)
 {
 	struct drm_device *dev = intel_dsi->base.base.dev;
@@ -384,8 +465,7 @@ static const u8 *mipi_exec_gpio(struct intel_dsi *intel_dsi, const u8 *data)
 	struct intel_connector *connector = intel_dsi->attached_connector;
 	u8 gpio_source, gpio_index = 0, gpio_number;
 	bool value;
-
-	drm_dbg_kms(&dev_priv->drm, "\n");
+	bool native = DISPLAY_VER(dev_priv) >= 11;
 
 	if (connector->panel.vbt.dsi.seq_version >= 3)
 		gpio_index = *data++;
@@ -398,10 +478,18 @@ static const u8 *mipi_exec_gpio(struct intel_dsi *intel_dsi, const u8 *data)
 	else
 		gpio_source = 0;
 
+	if (connector->panel.vbt.dsi.seq_version >= 4 && *data & BIT(1))
+		native = false;
+
 	/* pull up/down */
 	value = *data++ & 1;
 
-	if (DISPLAY_VER(dev_priv) >= 11)
+	drm_dbg_kms(&dev_priv->drm, "GPIO index %u, number %u, source %u, native %s, set to %s\n",
+		    gpio_index, gpio_number, gpio_source, str_yes_no(native), str_on_off(value));
+
+	if (native)
+		icl_native_gpio_set_value(dev_priv, gpio_number, value);
+	else if (DISPLAY_VER(dev_priv) >= 11)
 		icl_exec_gpio(connector, gpio_source, gpio_index, value);
 	else if (IS_VALLEYVIEW(dev_priv))
 		vlv_exec_gpio(connector, gpio_source, gpio_number, value);

drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c

@@ -730,32 +730,69 @@ static int eb_reserve(struct i915_execbuffer *eb)
 	bool unpinned;
 
 	/*
-	 * Attempt to pin all of the buffers into the GTT.
-	 * This is done in 2 phases:
+	 * We have one more buffers that we couldn't bind, which could be due to
+	 * various reasons. To resolve this we have 4 passes, with every next
+	 * level turning the screws tighter:
 	 *
-	 * 1. Unbind all objects that do not match the GTT constraints for
-	 *    the execbuffer (fenceable, mappable, alignment etc).
-	 * 2. Bind new objects.
+	 * 0. Unbind all objects that do not match the GTT constraints for the
+	 * execbuffer (fenceable, mappable, alignment etc). Bind all new
+	 * objects.  This avoids unnecessary unbinding of later objects in order
+	 * to make room for the earlier objects *unless* we need to defragment.
 	 *
-	 * This avoid unnecessary unbinding of later objects in order to make
-	 * room for the earlier objects *unless* we need to defragment.
+	 * 1. Reorder the buffers, where objects with the most restrictive
+	 * placement requirements go first (ignoring fixed location buffers for
+	 * now).  For example, objects needing the mappable aperture (the first
+	 * 256M of GTT), should go first vs objects that can be placed just
+	 * about anywhere. Repeat the previous pass.
 	 *
-	 * Defragmenting is skipped if all objects are pinned at a fixed location.
+	 * 2. Consider buffers that are pinned at a fixed location. Also try to
+	 * evict the entire VM this time, leaving only objects that we were
+	 * unable to lock. Try again to bind the buffers. (still using the new
+	 * buffer order).
+	 *
+	 * 3. We likely have object lock contention for one or more stubborn
+	 * objects in the VM, for which we need to evict to make forward
+	 * progress (perhaps we are fighting the shrinker?). When evicting the
+	 * VM this time around, anything that we can't lock we now track using
+	 * the busy_bo, using the full lock (after dropping the vm->mutex to
+	 * prevent deadlocks), instead of trylock. We then continue to evict the
+	 * VM, this time with the stubborn object locked, which we can now
+	 * hopefully unbind (if still bound in the VM). Repeat until the VM is
+	 * evicted. Finally we should be able bind everything.
 	 */
-	for (pass = 0; pass <= 2; pass++) {
+	for (pass = 0; pass <= 3; pass++) {
 		int pin_flags = PIN_USER | PIN_VALIDATE;
 
 		if (pass == 0)
 			pin_flags |= PIN_NONBLOCK;
 
 		if (pass >= 1)
-			unpinned = eb_unbind(eb, pass == 2);
+			unpinned = eb_unbind(eb, pass >= 2);
 
 		if (pass == 2) {
 			err = mutex_lock_interruptible(&eb->context->vm->mutex);
 			if (!err) {
-				err = i915_gem_evict_vm(eb->context->vm, &eb->ww);
+				err = i915_gem_evict_vm(eb->context->vm, &eb->ww, NULL);
+				mutex_unlock(&eb->context->vm->mutex);
+			}
+			if (err)
+				return err;
+		}
+
+		if (pass == 3) {
+retry:
+			err = mutex_lock_interruptible(&eb->context->vm->mutex);
+			if (!err) {
+				struct drm_i915_gem_object *busy_bo = NULL;
+
+				err = i915_gem_evict_vm(eb->context->vm, &eb->ww, &busy_bo);
 				mutex_unlock(&eb->context->vm->mutex);
+				if (err && busy_bo) {
+					err = i915_gem_object_lock(busy_bo, &eb->ww);
+					i915_gem_object_put(busy_bo);
+					if (!err)
+						goto retry;
+				}
 			}
 			if (err)
 				return err;

drivers/gpu/drm/i915/gem/i915_gem_mman.c

@@ -369,7 +369,7 @@ static vm_fault_t vm_fault_gtt(struct vm_fault *vmf)
 		if (vma == ERR_PTR(-ENOSPC)) {
 			ret = mutex_lock_interruptible(&ggtt->vm.mutex);
 			if (!ret) {
-				ret = i915_gem_evict_vm(&ggtt->vm, &ww);
+				ret = i915_gem_evict_vm(&ggtt->vm, &ww, NULL);
 				mutex_unlock(&ggtt->vm.mutex);
 			}
 			if (ret)

drivers/gpu/drm/i915/gt/intel_gt.c

@@ -1109,9 +1109,15 @@ static void mmio_invalidate_full(struct intel_gt *gt)
 			continue;
 
 		if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
+			u32 val = BIT(engine->instance);
+
+			if (engine->class == VIDEO_DECODE_CLASS ||
+			    engine->class == VIDEO_ENHANCEMENT_CLASS ||
+			    engine->class == COMPUTE_CLASS)
+				val = _MASKED_BIT_ENABLE(val);
 			intel_gt_mcr_multicast_write_fw(gt,
 							xehp_regs[engine->class],
-							BIT(engine->instance));
+							val);
 		} else {
 			rb = get_reg_and_bit(engine, regs == gen8_regs, regs, num);
 			if (!i915_mmio_reg_offset(rb.reg))

drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c

@@ -545,6 +545,32 @@ static int check_ccs_header(struct intel_gt *gt,
 	return 0;
 }
 
+static int try_firmware_load(struct intel_uc_fw *uc_fw, const struct firmware **fw)
+{
+	struct intel_gt *gt = __uc_fw_to_gt(uc_fw);
+	struct device *dev = gt->i915->drm.dev;
+	int err;
+
+	err = firmware_request_nowarn(fw, uc_fw->file_selected.path, dev);
+
+	if (err)
+		return err;
+
+	if ((*fw)->size > INTEL_UC_RSVD_GGTT_PER_FW) {
+		drm_err(&gt->i915->drm,
+			"%s firmware %s: size (%zuKB) exceeds max supported size (%uKB)\n",
+			intel_uc_fw_type_repr(uc_fw->type), uc_fw->file_selected.path,
+			(*fw)->size / SZ_1K, INTEL_UC_RSVD_GGTT_PER_FW / SZ_1K);
+
+		/* try to find another blob to load */
+		release_firmware(*fw);
+		*fw = NULL;
+		return -ENOENT;
+	}
+
+	return 0;
+}
+
 /**
  * intel_uc_fw_fetch - fetch uC firmware
  * @uc_fw: uC firmware
@@ -558,7 +584,6 @@ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw)
 	struct intel_gt *gt = __uc_fw_to_gt(uc_fw);
 	struct drm_i915_private *i915 = gt->i915;
 	struct intel_uc_fw_file file_ideal;
-	struct device *dev = i915->drm.dev;
 	struct drm_i915_gem_object *obj;
 	const struct firmware *fw = NULL;
 	bool old_ver = false;
@@ -574,20 +599,9 @@ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw)
 	__force_fw_fetch_failures(uc_fw, -EINVAL);
 	__force_fw_fetch_failures(uc_fw, -ESTALE);
 
-	err = firmware_request_nowarn(&fw, uc_fw->file_selected.path, dev);
+	err = try_firmware_load(uc_fw, &fw);
 	memcpy(&file_ideal, &uc_fw->file_wanted, sizeof(file_ideal));
 
-	if (!err && fw->size > INTEL_UC_RSVD_GGTT_PER_FW) {
-		drm_err(&i915->drm,
-			"%s firmware %s: size (%zuKB) exceeds max supported size (%uKB)\n",
-			intel_uc_fw_type_repr(uc_fw->type), uc_fw->file_selected.path,
-			fw->size / SZ_1K, INTEL_UC_RSVD_GGTT_PER_FW / SZ_1K);
-
-		/* try to find another blob to load */
-		release_firmware(fw);
-		err = -ENOENT;
-	}
-
 	/* Any error is terminal if overriding. Don't bother searching for older versions */
 	if (err && intel_uc_fw_is_overridden(uc_fw))
 		goto fail;
@@ -608,7 +622,7 @@ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw)
 			break;
 		}
 
-		err = firmware_request_nowarn(&fw, uc_fw->file_selected.path, dev);
+		err = try_firmware_load(uc_fw, &fw);
 	}
 
 	if (err)

drivers/gpu/drm/i915/i915_gem_evict.c

@@ -416,6 +416,11 @@ int i915_gem_evict_for_node(struct i915_address_space *vm,
  * @vm: Address space to cleanse
  * @ww: An optional struct i915_gem_ww_ctx. If not NULL, i915_gem_evict_vm
  * will be able to evict vma's locked by the ww as well.
+ * @busy_bo: Optional pointer to struct drm_i915_gem_object. If not NULL, then
+ * in the event i915_gem_evict_vm() is unable to trylock an object for eviction,
+ * then @busy_bo will point to it. -EBUSY is also returned. The caller must drop
+ * the vm->mutex, before trying again to acquire the contended lock. The caller
+ * also owns a reference to the object.
  *
  * This function evicts all vmas from a vm.
  *
@@ -425,7 +430,8 @@ int i915_gem_evict_for_node(struct i915_address_space *vm,
  * To clarify: This is for freeing up virtual address space, not for freeing
  * memory in e.g. the shrinker.
  */
-int i915_gem_evict_vm(struct i915_address_space *vm, struct i915_gem_ww_ctx *ww)
+int i915_gem_evict_vm(struct i915_address_space *vm, struct i915_gem_ww_ctx *ww,
+		      struct drm_i915_gem_object **busy_bo)
 {
 	int ret = 0;
 
@@ -457,15 +463,22 @@ int i915_gem_evict_vm(struct i915_address_space *vm, struct i915_gem_ww_ctx *ww)
 			 * the resv is shared among multiple objects, we still
 			 * need the object ref.
 			 */
-			if (dying_vma(vma) ||
+			if (!i915_gem_object_get_rcu(vma->obj) ||
 			    (ww && (dma_resv_locking_ctx(vma->obj->base.resv) == &ww->ctx))) {
 				__i915_vma_pin(vma);
 				list_add(&vma->evict_link, &locked_eviction_list);
 				continue;
 			}
 
-			if (!i915_gem_object_trylock(vma->obj, ww))
+			if (!i915_gem_object_trylock(vma->obj, ww)) {
+				if (busy_bo) {
+					*busy_bo = vma->obj; /* holds ref */
+					ret = -EBUSY;
+					break;
+				}
+				i915_gem_object_put(vma->obj);
 				continue;
+			}
 
 			__i915_vma_pin(vma);
 			list_add(&vma->evict_link, &eviction_list);
@@ -473,25 +486,29 @@ int i915_gem_evict_vm(struct i915_address_space *vm, struct i915_gem_ww_ctx *ww)
 		if (list_empty(&eviction_list) && list_empty(&locked_eviction_list))
 			break;
 
-		ret = 0;
 		/* Unbind locked objects first, before unlocking the eviction_list */
 		list_for_each_entry_safe(vma, vn, &locked_eviction_list, evict_link) {
 			__i915_vma_unpin(vma);
 
-			if (ret == 0)
+			if (ret == 0) {
 				ret = __i915_vma_unbind(vma);
 				if (ret != -EINTR) /* "Get me out of here!" */
 					ret = 0;
 			}
+			if (!dying_vma(vma))
+				i915_gem_object_put(vma->obj);
+		}
 
 		list_for_each_entry_safe(vma, vn, &eviction_list, evict_link) {
 			__i915_vma_unpin(vma);
-			if (ret == 0)
+			if (ret == 0) {
 				ret = __i915_vma_unbind(vma);
 				if (ret != -EINTR) /* "Get me out of here!" */
 					ret = 0;
+			}
 
 			i915_gem_object_unlock(vma->obj);
+			i915_gem_object_put(vma->obj);
 		}
 	} while (ret == 0);
 

drivers/gpu/drm/i915/i915_gem_evict.h

@@ -11,6 +11,7 @@
 struct drm_mm_node;
 struct i915_address_space;
 struct i915_gem_ww_ctx;
+struct drm_i915_gem_object;
 
 int __must_check i915_gem_evict_something(struct i915_address_space *vm,
 					  struct i915_gem_ww_ctx *ww,
@@ -23,6 +24,7 @@ int __must_check i915_gem_evict_for_node(struct i915_address_space *vm,
 					 struct drm_mm_node *node,
 					 unsigned int flags);
 int i915_gem_evict_vm(struct i915_address_space *vm,
-		      struct i915_gem_ww_ctx *ww);
+		      struct i915_gem_ww_ctx *ww,
+		      struct drm_i915_gem_object **busy_bo);
 
 #endif /* __I915_GEM_EVICT_H__ */

drivers/gpu/drm/i915/i915_irq.c

@@ -1974,7 +1974,10 @@ static void icp_irq_handler(struct drm_i915_private *dev_priv, u32 pch_iir)
 	if (ddi_hotplug_trigger) {
 		u32 dig_hotplug_reg;
 
+		/* Locking due to DSI native GPIO sequences */
+		spin_lock(&dev_priv->irq_lock);
 		dig_hotplug_reg = intel_uncore_rmw(&dev_priv->uncore, SHOTPLUG_CTL_DDI, 0, 0);
+		spin_unlock(&dev_priv->irq_lock);
 
 		intel_get_hpd_pins(dev_priv, &pin_mask, &long_mask,
 				   ddi_hotplug_trigger, dig_hotplug_reg,

drivers/gpu/drm/i915/i915_pci.c

@@ -1129,7 +1129,6 @@ static const struct intel_gt_definition xelpmp_extra_gt[] = {
 	{}
 };
 
-__maybe_unused
 static const struct intel_device_info mtl_info = {
 	XE_HP_FEATURES,
 	XE_LPDP_FEATURES,

drivers/gpu/drm/i915/i915_reg.h

@@ -5988,6 +5988,7 @@
 
 #define SHOTPLUG_CTL_DDI				_MMIO(0xc4030)
 #define   SHOTPLUG_CTL_DDI_HPD_ENABLE(hpd_pin)			(0x8 << (_HPD_PIN_DDI(hpd_pin) * 4))
+#define   SHOTPLUG_CTL_DDI_HPD_OUTPUT_DATA(hpd_pin)		(0x4 << (_HPD_PIN_DDI(hpd_pin) * 4))
 #define   SHOTPLUG_CTL_DDI_HPD_STATUS_MASK(hpd_pin)		(0x3 << (_HPD_PIN_DDI(hpd_pin) * 4))
 #define   SHOTPLUG_CTL_DDI_HPD_NO_DETECT(hpd_pin)		(0x0 << (_HPD_PIN_DDI(hpd_pin) * 4))
 #define   SHOTPLUG_CTL_DDI_HPD_SHORT_DETECT(hpd_pin)		(0x1 << (_HPD_PIN_DDI(hpd_pin) * 4))

drivers/gpu/drm/i915/i915_vma.c

@@ -1566,7 +1566,7 @@ static int __i915_ggtt_pin(struct i915_vma *vma, struct i915_gem_ww_ctx *ww,
 			 * locked objects when called from execbuf when pinning
 			 * is removed. This would probably regress badly.
 			 */
-			i915_gem_evict_vm(vm, NULL);
+			i915_gem_evict_vm(vm, NULL, NULL);
 			mutex_unlock(&vm->mutex);
 		}
 	} while (1);

drivers/gpu/drm/i915/selftests/i915_gem_evict.c

@@ -344,7 +344,7 @@ static int igt_evict_vm(void *arg)
 
 	/* Everything is pinned, nothing should happen */
 	mutex_lock(&ggtt->vm.mutex);
-	err = i915_gem_evict_vm(&ggtt->vm, NULL);
+	err = i915_gem_evict_vm(&ggtt->vm, NULL, NULL);
 	mutex_unlock(&ggtt->vm.mutex);
 	if (err) {
 		pr_err("i915_gem_evict_vm on a full GGTT returned err=%d]\n",
@@ -356,7 +356,7 @@ static int igt_evict_vm(void *arg)
 
 	for_i915_gem_ww(&ww, err, false) {
 		mutex_lock(&ggtt->vm.mutex);
-		err = i915_gem_evict_vm(&ggtt->vm, &ww);
+		err = i915_gem_evict_vm(&ggtt->vm, &ww, NULL);
 		mutex_unlock(&ggtt->vm.mutex);
 	}