Merge airlied/drm-next into drm-misc-fixes

Fast forwarding -fixes for 4.17.

Signed-off-by: Sean Paul <seanpaul@chromium.org>

Documentation/ABI/testing/sysfs-ata

@@ -1,110 +1,139 @@
 What:		/sys/class/ata_...
-Date:		August 2008
-Contact:	Gwendal Grignou<gwendal@google.com>
 Description:
-
-Provide a place in sysfs for storing the ATA topology of the system.  This allows
-retrieving various information about ATA objects.
+		Provide a place in sysfs for storing the ATA topology of the
+		system. This allows retrieving various information about ATA
+		objects.
 
 Files under /sys/class/ata_port
 -------------------------------
 
-	For each port, a directory ataX is created where X is the ata_port_id of
-	the port. The device parent is the ata host device.
+For each port, a directory ataX is created where X is the ata_port_id of the
+port. The device parent is the ata host device.
 
-idle_irq (read)
 
-	Number of IRQ received by the port while idle [some ata HBA only].
+What:		/sys/class/ata_port/ataX/nr_pmp_links
+What:		/sys/class/ata_port/ataX/idle_irq
+Date:		May, 2010
+KernelVersion:	v2.6.37
+Contact:	Gwendal Grignou <gwendal@chromium.org>
+Description:
+		nr_pmp_links:	(RO) If a SATA Port Multiplier (PM) is
+				connected, the number of links behind it.
 
-nr_pmp_links (read)
+		idle_irq:	(RO) Number of IRQ received by the port while
+				idle [some ata HBA only].
 
-	If a SATA Port Multiplier (PM) is connected, number of link behind it.
+
+What:		/sys/class/ata_port/ataX/port_no
+Date:		May, 2013
+KernelVersion:	v3.11
+Contact:	Gwendal Grignou <gwendal@chromium.org>
+Description:
+		(RO) Host local port number. While registering host controller,
+		port numbers are tracked based upon number of ports available on
+		the controller. This attribute is needed by udev for composing
+		persistent links in /dev/disk/by-path.
 
 Files under /sys/class/ata_link
 -------------------------------
 
-	Behind each port, there is a ata_link. If there is a SATA PM in the
-	topology, 15 ata_link objects are created.
-
-	If a link is behind a port, the directory name is linkX, where X is
-	ata_port_id of the port.
-	If a link is behind a PM, its name is linkX.Y where X is ata_port_id
-	of the parent port and Y the PM port.
+Behind each port, there is a ata_link. If there is a SATA PM in the topology, 15
+ata_link objects are created.
 
-hw_sata_spd_limit
+If a link is behind a port, the directory name is linkX, where X is ata_port_id
+of the port. If a link is behind a PM, its name is linkX.Y where X is
+ata_port_id of the parent port and Y the PM port.
 
-	Maximum speed supported by the connected SATA device.
 
-sata_spd_limit
+What:		/sys/class/ata_link/linkX[.Y]/hw_sata_spd_limit
+What:		/sys/class/ata_link/linkX[.Y]/sata_spd_limit
+What:		/sys/class/ata_link/linkX[.Y]/sata_spd
+Date:		May, 2010
+KernelVersion:	v2.6.37
+Contact:	Gwendal Grignou <gwendal@chromium.org>
+Description:
+		hw_sata_spd_limit:	(RO) Maximum speed supported by the
+					connected SATA device.
 
-	Maximum speed imposed by libata.
+		sata_spd_limit:		(RO) Maximum speed imposed by libata.
 
-sata_spd
+		sata_spd:		(RO) Current speed of the link
+					eg. 1.5, 3 Gbps etc.
 
-	Current speed of the link [1.5, 3Gps,...].
 
 Files under /sys/class/ata_device
 ---------------------------------
 
-	Behind each link, up to two ata device are created.
-	The name of the directory is devX[.Y].Z where:
-	- X is ata_port_id of the port where the device is connected,
-	- Y the port of the PM if any, and
-	- Z the device id: for PATA, there is usually 2 devices [0,1],
-	only 1 for SATA.
-
-class
-	Device class. Can be "ata" for disk, "atapi" for packet device,
-	"pmp" for PM, or "none" if no device was found behind the link.
-
-dma_mode
+Behind each link, up to two ata devices are created.
+The name of the directory is devX[.Y].Z where:
+- X is ata_port_id of the port where the device is connected,
+- Y the port of the PM if any, and
+- Z the device id: for PATA, there is usually 2 devices [0,1], only 1 for SATA.
+
+
+What:		/sys/class/ata_device/devX[.Y].Z/spdn_cnt
+What:		/sys/class/ata_device/devX[.Y].Z/gscr
+What:		/sys/class/ata_device/devX[.Y].Z/ering
+What:		/sys/class/ata_device/devX[.Y].Z/id
+What:		/sys/class/ata_device/devX[.Y].Z/pio_mode
+What:		/sys/class/ata_device/devX[.Y].Z/xfer_mode
+What:		/sys/class/ata_device/devX[.Y].Z/dma_mode
+What:		/sys/class/ata_device/devX[.Y].Z/class
+Date:		May, 2010
+KernelVersion:	v2.6.37
+Contact:	Gwendal Grignou <gwendal@chromium.org>
+Description:
+		spdn_cnt:	(RO) Number of times libata decided to lower the
+				speed of link due to errors.
 
-	Transfer modes supported by the device when in DMA mode.
-	Mostly used by PATA device.
+		gscr:		(RO) Cached result of the dump of PM GSCR
+				register. Valid registers are:
 
-pio_mode
+				0:      SATA_PMP_GSCR_PROD_ID,
+				1:	SATA_PMP_GSCR_REV,
+				2:      SATA_PMP_GSCR_PORT_INFO,
+				32:     SATA_PMP_GSCR_ERROR,
+				33:     SATA_PMP_GSCR_ERROR_EN,
+				64:     SATA_PMP_GSCR_FEAT,
+				96:     SATA_PMP_GSCR_FEAT_EN,
+				130:    SATA_PMP_GSCR_SII_GPIO
 
-	Transfer modes supported by the device when in PIO mode.
-	Mostly used by PATA device.
+				Only valid if the device is a PM.
 
-xfer_mode
+		ering:		(RO) Formatted output of the error ring of the
+				device.
 
-	Current transfer mode.
+		id:		(RO) Cached result of IDENTIFY command, as
+				described in ATA8 7.16 and 7.17. Only valid if
+				the device is not a PM.
 
-id
+		pio_mode:	(RO) Transfer modes supported by the device when
+				in PIO mode. Mostly used by PATA device.
 
-	Cached result of IDENTIFY command, as described in ATA8 7.16 and 7.17.
-	Only valid if the device is not a PM.
+		xfer_mode:	(RO) Current transfer mode
 
-gscr
+		dma_mode:	(RO) Transfer modes supported by the device when
+				in DMA mode. Mostly used by PATA device.
 
-	Cached result of the dump of PM GSCR register.
-	Valid registers are:
-	0: 	SATA_PMP_GSCR_PROD_ID,
-	1: 	SATA_PMP_GSCR_REV,
-	2: 	SATA_PMP_GSCR_PORT_INFO,
-	32:	SATA_PMP_GSCR_ERROR,
-	33:	SATA_PMP_GSCR_ERROR_EN,
-	64:	SATA_PMP_GSCR_FEAT,
-	96:	SATA_PMP_GSCR_FEAT_EN,
-	130:	SATA_PMP_GSCR_SII_GPIO
-	Only valid if the device is a PM.
+		class:		(RO) Device class. Can be "ata" for disk,
+				"atapi" for packet device, "pmp" for PM, or
+				"none" if no device was found behind the link.
 
-trim
 
-	Shows the DSM TRIM mode currently used by the device. Valid
-	values are:
-	unsupported:		Drive does not support DSM TRIM
-	unqueued:		Drive supports unqueued DSM TRIM only
-	queued:			Drive supports queued DSM TRIM
-	forced_unqueued:	Drive's queued DSM support is known to be
-				buggy and only unqueued TRIM commands
-				are sent
+What:		/sys/class/ata_device/devX[.Y].Z/trim
+Date:		May, 2015
+KernelVersion:	v4.10
+Contact:	Gwendal Grignou <gwendal@chromium.org>
+Description:
+		(RO) Shows the DSM TRIM mode currently used by the device. Valid
+		values are:
 
-spdn_cnt
+		unsupported:		Drive does not support DSM TRIM
 
-	Number of time libata decided to lower the speed of link due to errors.
+		unqueued:               Drive supports unqueued DSM TRIM only
 
-ering
+		queued:                 Drive supports queued DSM TRIM
 
-	Formatted output of the error ring of the device.
+		forced_unqueued:	Drive's queued DSM support is known to
+					be buggy and only unqueued TRIM commands
+					are sent

Documentation/ABI/testing/sysfs-block-device

@@ -0,0 +1,58 @@
+What:		/sys/block/*/device/sw_activity
+Date:		Jun, 2008
+KernelVersion:	v2.6.27
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RW) Used by drivers which support software controlled activity
+		LEDs.
+
+		It has the following valid values:
+
+		0	OFF - the LED is not activated on activity
+		1	BLINK_ON - the LED blinks on every 10ms when activity is
+			detected.
+		2	BLINK_OFF - the LED is on when idle, and blinks off
+			every 10ms when activity is detected.
+
+		Note that the user must turn sw_activity OFF it they wish to
+		control the activity LED via the em_message file.
+
+
+What:		/sys/block/*/device/unload_heads
+Date:		Sep, 2008
+KernelVersion:	v2.6.28
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RW) Hard disk shock protection
+
+		Writing an integer value to this file will take the heads of the
+		respective drive off the platter and block all I/O operations
+		for the specified number of milliseconds.
+
+		- If the device does not support the unload heads feature,
+		  access is denied with -EOPNOTSUPP.
+		- The maximal value accepted for a timeout is 30000
+		  milliseconds.
+		- A previously set timeout can be cancelled and disk can resume
+		  normal operation immediately by specifying a timeout of 0.
+		- Some hard drives only comply with an earlier version of the
+		  ATA standard, but support the unload feature nonetheless.
+		  There is no safe way Linux can detect these devices, so this
+		  is not enabled by default. If it is known that your device
+		  does support the unload feature, then you can tell the kernel
+		  to enable it by writing -1. It can be disabled again by
+		  writing -2.
+		- Values below -2 are rejected with -EINVAL
+
+		For more information, see
+		Documentation/laptops/disk-shock-protection.txt
+
+
+What:		/sys/block/*/device/ncq_prio_enable
+Date:		Oct, 2016
+KernelVersion:	v4.10
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RW) Write to the file to turn on or off the SATA ncq (native
+		command queueing) support. By default this feature is turned
+		off.

Documentation/ABI/testing/sysfs-class-scsi_host

@@ -27,3 +27,92 @@ Description:	This file contains the current status of the "SSD Smart Path"
 		the direct i/o path to physical devices.  This setting is
 		controller wide, affecting all configured logical drives on the
 		controller.  This file is readable and writable.
+
+What:		/sys/class/scsi_host/hostX/link_power_management_policy
+Date:		Oct, 2007
+KernelVersion:	v2.6.24
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RW) This parameter allows the user to read and set the link
+		(interface) power management.
+
+		There are four possible options:
+
+		min_power: Tell the controller to try to make the link use the
+		least possible power when possible. This may sacrifice some
+		performance due to increased latency when coming out of lower
+		power states.
+
+		max_performance: Generally, this means no power management.
+		Tell the controller to have performance be a priority over power
+		management.
+
+		medium_power: Tell the controller to enter a lower power state
+		when possible, but do not enter the lowest power state, thus
+		improving latency over min_power setting.
+
+		med_power_with_dipm: Identical to the existing medium_power
+		setting except that it enables dipm (device initiated power
+		management) on top, which makes it match the Windows IRST (Intel
+		Rapid Storage Technology) driver settings. This setting is also
+		close to min_power, except that:
+		a) It does not use host-initiated slumber mode, but it does
+		allow device-initiated slumber
+		b) It does not enable low power device sleep mode (DevSlp).
+
+What:		/sys/class/scsi_host/hostX/em_message
+What:		/sys/class/scsi_host/hostX/em_message_type
+Date:		Jun, 2008
+KernelVersion:	v2.6.27
+Contact:	linux-ide@vger.kernel.org
+Description:
+		em_message: (RW) Enclosure management support. For the LED
+		protocol, writes and reads correspond to the LED message format
+		as defined in the AHCI spec.
+
+		The user must turn sw_activity (under /sys/block/*/device/) OFF
+		it they wish to control the activity LED via the em_message
+		file.
+
+		em_message_type: (RO) Displays the current enclosure management
+		protocol that is being used by the driver (for eg. LED, SAF-TE,
+		SES-2, SGPIO etc).
+
+What:		/sys/class/scsi_host/hostX/ahci_port_cmd
+What:		/sys/class/scsi_host/hostX/ahci_host_caps
+What:		/sys/class/scsi_host/hostX/ahci_host_cap2
+Date:		Mar, 2010
+KernelVersion:	v2.6.35
+Contact:	linux-ide@vger.kernel.org
+Description:
+		[to be documented]
+
+What:		/sys/class/scsi_host/hostX/ahci_host_version
+Date:		Mar, 2010
+KernelVersion:	v2.6.35
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RO) Display the version of the AHCI spec implemented by the
+		host.
+
+What:		/sys/class/scsi_host/hostX/em_buffer
+Date:		Apr, 2010
+KernelVersion:	v2.6.35
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RW) Allows access to AHCI EM (enclosure management) buffer
+		directly if the host supports EM.
+
+		For eg. the AHCI driver supports SGPIO EM messages but the
+		SATA/AHCI specs do not define the SGPIO message format of the EM
+		buffer. Different hardware(HW) vendors may have different
+		definitions. With the em_buffer attribute, this issue can be
+		solved by allowing HW vendors to provide userland drivers and
+		tools for their SGPIO initiators.
+
+What:		/sys/class/scsi_host/hostX/em_message_supported
+Date:		Oct, 2009
+KernelVersion:	v2.6.39
+Contact:	linux-ide@vger.kernel.org
+Description:
+		(RO) Displays supported enclosure management message types.

Documentation/devicetree/bindings/display/bridge/renesas,lvds.txt

@@ -0,0 +1,58 @@
+Renesas R-Car LVDS Encoder
+==========================
+
+These DT bindings describe the LVDS encoder embedded in the Renesas R-Car
+Gen2, R-Car Gen3 and RZ/G SoCs.
+
+Required properties:
+
+- compatible : Shall contain one of
+  - "renesas,r8a7743-lvds" for R8A7743 (RZ/G1M) compatible LVDS encoders
+  - "renesas,r8a7790-lvds" for R8A7790 (R-Car H2) compatible LVDS encoders
+  - "renesas,r8a7791-lvds" for R8A7791 (R-Car M2-W) compatible LVDS encoders
+  - "renesas,r8a7793-lvds" for R8A7793 (R-Car M2-N) compatible LVDS encoders
+  - "renesas,r8a7795-lvds" for R8A7795 (R-Car H3) compatible LVDS encoders
+  - "renesas,r8a7796-lvds" for R8A7796 (R-Car M3-W) compatible LVDS encoders
+  - "renesas,r8a77970-lvds" for R8A77970 (R-Car V3M) compatible LVDS encoders
+  - "renesas,r8a77995-lvds" for R8A77995 (R-Car D3) compatible LVDS encoders
+
+- reg: Base address and length for the memory-mapped registers
+- clocks: A phandle + clock-specifier pair for the functional clock
+- resets: A phandle + reset specifier for the module reset
+
+Required nodes:
+
+The LVDS encoder has two video ports. Their connections are modelled using the
+OF graph bindings specified in Documentation/devicetree/bindings/graph.txt.
+
+- Video port 0 corresponds to the parallel RGB input
+- Video port 1 corresponds to the LVDS output
+
+Each port shall have a single endpoint.
+
+
+Example:
+
+	lvds0: lvds@feb90000 {
+		compatible = "renesas,r8a7790-lvds";
+		reg = <0 0xfeb90000 0 0x1c>;
+		clocks = <&cpg CPG_MOD 726>;
+		resets = <&cpg 726>;
+
+		ports {
+			#address-cells = <1>;
+			#size-cells = <0>;
+
+			port@0 {
+				reg = <0>;
+				lvds0_in: endpoint {
+					remote-endpoint = <&du_out_lvds0>;
+				};
+			};
+			port@1 {
+				reg = <1>;
+				lvds0_out: endpoint {
+				};
+			};
+		};
+	};

Documentation/devicetree/bindings/display/bridge/ti,ths8135.txt → Documentation/devicetree/bindings/display/bridge/ti,ths813x.txt

@@ -1,11 +1,16 @@
-THS8135 Video DAC
------------------
+THS8134 and THS8135 Video DAC
+-----------------------------
 
-This is the binding for Texas Instruments THS8135 Video DAC bridge.
+This is the binding for Texas Instruments THS8134, THS8134A, THS8134B and
+THS8135 Video DAC bridges.
 
 Required properties:
 
-- compatible: Must be "ti,ths8135"
+- compatible: Must be one of
+  "ti,ths8134"
+  "ti,ths8134a," "ti,ths8134"
+  "ti,ths8134b", "ti,ths8134"
+  "ti,ths8135"
 
 Required nodes:
 

Documentation/devicetree/bindings/display/connector/dvi-connector.txt

@@ -10,6 +10,7 @@ Optional properties:
 - analog: the connector has DVI analog pins
 - digital: the connector has DVI digital pins
 - dual-link: the connector has pins for DVI dual-link
+- hpd-gpios: HPD GPIO number
 
 Required nodes:
 - Video port for DVI input

Documentation/devicetree/bindings/display/etnaviv/etnaviv-drm.txt

@@ -1,23 +1,3 @@
-Etnaviv DRM master device
-=========================
-
-The Etnaviv DRM master device is a virtual device needed to list all
-Vivante GPU cores that comprise the GPU subsystem.
-
-Required properties:
-- compatible: Should be one of
-    "fsl,imx-gpu-subsystem"
-    "marvell,dove-gpu-subsystem"
-- cores: Should contain a list of phandles pointing to Vivante GPU devices
-
-example:
-
-gpu-subsystem {
-	compatible = "fsl,imx-gpu-subsystem";
-	cores = <&gpu_2d>, <&gpu_3d>;
-};
-
-
 Vivante GPU core devices
 ========================
 
@@ -32,7 +12,9 @@ Required properties:
 - clocks: should contain one clock for entry in clock-names
   see Documentation/devicetree/bindings/clock/clock-bindings.txt
 - clock-names:
-   - "bus":    AXI/register clock
+   - "bus":    AXI/master interface clock
+   - "reg":    AHB/slave interface clock
+               (only required if GPU can gate slave interface independently)
    - "core":   GPU core clock
    - "shader": Shader clock (only required if GPU has feature PIPE_3D)
 

Documentation/devicetree/bindings/display/msm/dsi.txt

@@ -7,8 +7,6 @@ Required properties:
 - reg: Physical base address and length of the registers of controller
 - reg-names: The names of register regions. The following regions are required:
   * "dsi_ctrl"
-- qcom,dsi-host-index: The ID of DSI controller hardware instance. This should
-  be 0 or 1, since we have 2 DSI controllers at most for now.
 - interrupts: The interrupt signal from the DSI block.
 - power-domains: Should be <&mmcc MDSS_GDSC>.
 - clocks: Phandles to device clocks.
@@ -22,6 +20,8 @@ Required properties:
   * "core"
   For DSIv2, we need an additional clock:
    * "src"
+  For DSI6G v2.0 onwards, we need also need the clock:
+   * "byte_intf"
 - assigned-clocks: Parents of "byte" and "pixel" for the given platform.
 - assigned-clock-parents: The Byte clock and Pixel clock PLL outputs provided
   by a DSI PHY block. See [1] for details on clock bindings.
@@ -88,21 +88,35 @@ Required properties:
   * "qcom,dsi-phy-28nm-lp"
   * "qcom,dsi-phy-20nm"
   * "qcom,dsi-phy-28nm-8960"
-- reg: Physical base address and length of the registers of PLL, PHY and PHY
-  regulator
+  * "qcom,dsi-phy-14nm"
+  * "qcom,dsi-phy-10nm"
+- reg: Physical base address and length of the registers of PLL, PHY. Some
+  revisions require the PHY regulator base address, whereas others require the
+  PHY lane base address. See below for each PHY revision.
 - reg-names: The names of register regions. The following regions are required:
+  For DSI 28nm HPM/LP/8960 PHYs and 20nm PHY:
   * "dsi_pll"
   * "dsi_phy"
   * "dsi_phy_regulator"
+  For DSI 14nm and 10nm PHYs:
+  * "dsi_pll"
+  * "dsi_phy"
+  * "dsi_phy_lane"
 - clock-cells: Must be 1. The DSI PHY block acts as a clock provider, creating
   2 clocks: A byte clock (index 0), and a pixel clock (index 1).
-- qcom,dsi-phy-index: The ID of DSI PHY hardware instance. This should
-  be 0 or 1, since we have 2 DSI PHYs at most for now.
 - power-domains: Should be <&mmcc MDSS_GDSC>.
 - clocks: Phandles to device clocks. See [1] for details on clock bindings.
 - clock-names: the following clocks are required:
   * "iface"
+  For 28nm HPM/LP, 28nm 8960 PHYs:
+- vddio-supply: phandle to vdd-io regulator device node
+  For 20nm PHY:
 - vddio-supply: phandle to vdd-io regulator device node
+- vcca-supply: phandle to vcca regulator device node
+  For 14nm PHY:
+- vcca-supply: phandle to vcca regulator device node
+  For 10nm PHY:
+- vdds-supply: phandle to vdds regulator device node
 
 Optional properties:
 - qcom,dsi-phy-regulator-ldo-mode: Boolean value indicating if the LDO mode PHY

Documentation/devicetree/bindings/display/panel/arm,versatile-tft-panel.txt

@@ -0,0 +1,31 @@
+ARM Versatile TFT Panels
+
+These panels are connected to the daughterboards found on the
+ARM Versatile reference designs.
+
+This device node must appear as a child to a "syscon"-compatible
+node.
+
+Required properties:
+- compatible: should be "arm,versatile-tft-panel"
+
+Required subnodes:
+- port: see display/panel/panel-common.txt, graph.txt
+
+
+Example:
+
+sysreg@0 {
+	compatible = "arm,versatile-sysreg", "syscon", "simple-mfd";
+	reg = <0x00000 0x1000>;
+
+	panel: display@0 {
+		compatible = "arm,versatile-tft-panel";
+
+		port {
+			panel_in: endpoint {
+				remote-endpoint = <&foo>;
+			};
+		};
+	};
+};

Documentation/devicetree/bindings/display/panel/auo,g104sn02.txt

@@ -0,0 +1,12 @@
+AU Optronics Corporation 10.4" (800x600) color TFT LCD panel
+
+Required properties:
+- compatible: should be "auo,g104sn02"
+- power-supply: as specified in the base binding
+
+Optional properties:
+- backlight: as specified in the base binding
+- enable-gpios: as specified in the base binding
+
+This binding is compatible with the simple-panel binding, which is specified
+in simple-panel.txt in this directory.

Documentation/devicetree/bindings/display/panel/display-timing.txt

@@ -80,6 +80,11 @@ The parameters are defined as:
   |          |        v                            |          |       |
   +----------+-------------------------------------+----------+-------+
 
+Note: In addition to being used as subnode(s) of display-timings, the timing
+      subnode may also be used on its own. This is appropriate if only one mode
+      need be conveyed. In this case, the node should be named 'panel-timing'.
+
+
 Example:
 
 	display-timings {

Documentation/devicetree/bindings/display/panel/koe,tx31d200vm0baa.txt

@@ -0,0 +1,25 @@
+Kaohsiung Opto-Electronics. TX31D200VM0BAA 12.3" HSXGA LVDS panel
+
+This binding is compatible with the simple-panel binding, which is specified
+in simple-panel.txt in this directory.
+
+Required properties:
+- compatible: should be "koe,tx31d200vm0baa"
+
+Optional properties:
+- backlight: phandle of the backlight device attached to the panel
+
+Optional nodes:
+- Video port for LVDS panel input.
+
+Example:
+	panel {
+		compatible = "koe,tx31d200vm0baa";
+		backlight = <&backlight_lvds>;
+
+		port {
+			panel_in: endpoint {
+				remote-endpoint = <&lvds0_out>;
+			};
+		};
+	};

Documentation/devicetree/bindings/display/panel/orisetech,otm8009a.txt

@@ -9,6 +9,7 @@ Required properties:
 
 Optional properties:
   - reset-gpios: a GPIO spec for the reset pin (active low).
+  - power-supply: phandle of the regulator that provides the supply voltage.
 
 Example:
 &dsi {
@@ -17,5 +18,6 @@ Example:
 		compatible = "orisetech,otm8009a";
 		reg = <0>;
 		reset-gpios = <&gpioh 7 GPIO_ACTIVE_LOW>;
+		power-supply = <&v1v8>;
 	};
 };

Documentation/devicetree/bindings/display/panel/raydium,rm68200.txt

@@ -0,0 +1,25 @@
+Raydium Semiconductor Corporation RM68200 5.5" 720p MIPI-DSI TFT LCD panel
+
+The Raydium Semiconductor Corporation RM68200 is a 5.5" 720x1280 TFT LCD
+panel connected using a MIPI-DSI video interface.
+
+Required properties:
+  - compatible: "raydium,rm68200"
+  - reg: the virtual channel number of a DSI peripheral
+
+Optional properties:
+  - reset-gpios: a GPIO spec for the reset pin (active low).
+  - power-supply: phandle of the regulator that provides the supply voltage.
+  - backlight: phandle of the backlight device attached to the panel.
+
+Example:
+&dsi {
+	...
+	panel@0 {
+		compatible = "raydium,rm68200";
+		reg = <0>;
+		reset-gpios = <&gpiof 15 GPIO_ACTIVE_LOW>;
+		power-supply = <&v1v8>;
+		backlight = <&pwm_backlight>;
+	};
+};

Documentation/devicetree/bindings/display/panel/simple-panel.txt

@@ -1,4 +1,8 @@
 Simple display panel
+====================
+
+panel node
+----------
 
 Required properties:
 - power-supply: See panel-common.txt

Documentation/devicetree/bindings/display/renesas,du.txt

@@ -13,13 +13,10 @@ Required Properties:
     - "renesas,du-r8a7794" for R8A7794 (R-Car E2) compatible DU
     - "renesas,du-r8a7795" for R8A7795 (R-Car H3) compatible DU
     - "renesas,du-r8a7796" for R8A7796 (R-Car M3-W) compatible DU
+    - "renesas,du-r8a77970" for R8A77970 (R-Car V3M) compatible DU
+    - "renesas,du-r8a77995" for R8A77995 (R-Car D3) compatible DU
 
-  - reg: A list of base address and length of each memory resource, one for
-    each entry in the reg-names property.
-  - reg-names: Name of the memory resources. The DU requires one memory
-    resource for the DU core (named "du") and one memory resource for each
-    LVDS encoder (named "lvds.x" with "x" being the LVDS controller numerical
-    index).
+  - reg: the memory-mapped I/O registers base address and length
 
   - interrupt-parent: phandle of the parent interrupt controller.
   - interrupts: Interrupt specifiers for the DU interrupts.
@@ -29,14 +26,13 @@ Required Properties:
   - clock-names: Name of the clocks. This property is model-dependent.
     - R8A7779 uses a single functional clock. The clock doesn't need to be
       named.
-    - All other DU instances use one functional clock per channel and one
-      clock per LVDS encoder (if available). The functional clocks must be
-      named "du.x" with "x" being the channel numerical index. The LVDS clocks
-      must be named "lvds.x" with "x" being the LVDS encoder numerical index.
-    - In addition to the functional and encoder clocks, all DU versions also
-      support externally supplied pixel clocks. Those clocks are optional.
-      When supplied they must be named "dclkin.x" with "x" being the input
-      clock numerical index.
+    - All other DU instances use one functional clock per channel The
+      functional clocks must be named "du.x" with "x" being the channel
+      numerical index.
+    - In addition to the functional clocks, all DU versions also support
+      externally supplied pixel clocks. Those clocks are optional. When
+      supplied they must be named "dclkin.x" with "x" being the input clock
+      numerical index.
 
   - vsps: A list of phandle and channel index tuples to the VSPs that handle
     the memory interfaces for the DU channels. The phandle identifies the VSP
@@ -63,15 +59,15 @@ corresponding to each DU output.
  R8A7794 (R-Car E2)   DPAD 0         DPAD 1         -              -
  R8A7795 (R-Car H3)   DPAD 0         HDMI 0         HDMI 1         LVDS 0
  R8A7796 (R-Car M3-W) DPAD 0         HDMI 0         LVDS 0         -
+ R8A77970 (R-Car V3M) DPAD 0         LVDS 0         -              -
+ R8A77995 (R-Car D3)  DPAD 0         LVDS 0         LVDS 1         -
 
 
 Example: R8A7795 (R-Car H3) ES2.0 DU
 
 	du: display@feb00000 {
 		compatible = "renesas,du-r8a7795";
-		reg = <0 0xfeb00000 0 0x80000>,
-		      <0 0xfeb90000 0 0x14>;
-		reg-names = "du", "lvds.0";
+		reg = <0 0xfeb00000 0 0x80000>;
 		interrupts = <GIC_SPI 256 IRQ_TYPE_LEVEL_HIGH>,
 			     <GIC_SPI 268 IRQ_TYPE_LEVEL_HIGH>,
 			     <GIC_SPI 269 IRQ_TYPE_LEVEL_HIGH>,
@@ -79,9 +75,8 @@ Example: R8A7795 (R-Car H3) ES2.0 DU
 		clocks = <&cpg CPG_MOD 724>,
 			 <&cpg CPG_MOD 723>,
 			 <&cpg CPG_MOD 722>,
-			 <&cpg CPG_MOD 721>,
-			 <&cpg CPG_MOD 727>;
-		clock-names = "du.0", "du.1", "du.2", "du.3", "lvds.0";
+			 <&cpg CPG_MOD 721>;
+		clock-names = "du.0", "du.1", "du.2", "du.3";
 		vsps = <&vspd0 0>, <&vspd1 0>, <&vspd2 0>, <&vspd0 1>;
 
 		ports {

Documentation/devicetree/bindings/display/rockchip/cdn-dp-rockchip.txt

@@ -0,0 +1,74 @@
+Rockchip RK3399 specific extensions to the cdn Display Port
+================================
+
+Required properties:
+- compatible: must be "rockchip,rk3399-cdn-dp"
+
+- reg: physical base address of the controller and length
+
+- clocks: from common clock binding: handle to dp clock.
+
+- clock-names: from common clock binding:
+	       Required elements: "core-clk" "pclk" "spdif" "grf"
+
+- resets : a list of phandle + reset specifier pairs
+- reset-names : string of reset names
+		Required elements: "apb", "core", "dptx", "spdif"
+- power-domains : power-domain property defined with a phandle
+		  to respective power domain.
+- assigned-clocks: main clock, should be <&cru SCLK_DP_CORE>
+- assigned-clock-rates : the DP core clk frequency, shall be: 100000000
+
+- rockchip,grf: this soc should set GRF regs, so need get grf here.
+
+- ports: contain a port nodes with endpoint definitions as defined in
+	 Documentation/devicetree/bindings/media/video-interfaces.txt.
+	 contained 2 endpoints, connecting to the output of vop.
+
+- phys: from general PHY binding: the phandle for the PHY device.
+
+- extcon: extcon specifier for the Power Delivery
+
+- #sound-dai-cells = it must be 1 if your system is using 2 DAIs: I2S, SPDIF
+
+-------------------------------------------------------------------------------
+
+Example:
+	cdn_dp: dp@fec00000 {
+		compatible = "rockchip,rk3399-cdn-dp";
+		reg = <0x0 0xfec00000 0x0 0x100000>;
+		interrupts = <GIC_SPI 9 IRQ_TYPE_LEVEL_HIGH>;
+		clocks = <&cru SCLK_DP_CORE>, <&cru PCLK_DP_CTRL>,
+			 <&cru SCLK_SPDIF_REC_DPTX>, <&cru PCLK_VIO_GRF>;
+		clock-names = "core-clk", "pclk", "spdif", "grf";
+		assigned-clocks = <&cru SCLK_DP_CORE>;
+		assigned-clock-rates = <100000000>;
+		power-domains = <&power RK3399_PD_HDCP>;
+		phys = <&tcphy0_dp>, <&tcphy1_dp>;
+		resets = <&cru SRST_DPTX_SPDIF_REC>;
+		reset-names = "spdif";
+		extcon = <&fusb0>, <&fusb1>;
+		rockchip,grf = <&grf>;
+		#address-cells = <1>;
+		#size-cells = <0>;
+		#sound-dai-cells = <1>;
+
+		ports {
+			#address-cells = <1>;
+			#size-cells = <0>;
+
+			dp_in: port {
+				#address-cells = <1>;
+				#size-cells = <0>;
+				dp_in_vopb: endpoint@0 {
+					reg = <0>;
+					remote-endpoint = <&vopb_out_dp>;
+				};
+
+				dp_in_vopl: endpoint@1 {
+					reg = <1>;
+					remote-endpoint = <&vopl_out_dp>;
+				};
+			};
+		};
+	};

Documentation/devicetree/bindings/display/st,stm32-ltdc.txt

@@ -98,7 +98,7 @@ Example 2: DSI panel
 			compatible = "st,stm32-dsi";
 			reg = <0x40016c00 0x800>;
 			clocks = <&rcc 1 CLK_F469_DSI>, <&clk_hse>;
-			clock-names = "ref", "pclk";
+			clock-names = "pclk", "ref";
 			resets = <&rcc STM32F4_APB2_RESET(DSI)>;
 			reset-names = "apb";
 

Documentation/devicetree/bindings/display/sunxi/sun4i-drm.txt

@@ -64,6 +64,56 @@ Required properties:
     first port should be the input endpoint. The second should be the
     output, usually to an HDMI connector.
 
+DWC HDMI TX Encoder
+-------------------
+
+The HDMI transmitter is a Synopsys DesignWare HDMI 1.4 TX controller IP
+with Allwinner's own PHY IP. It supports audio and video outputs and CEC.
+
+These DT bindings follow the Synopsys DWC HDMI TX bindings defined in
+Documentation/devicetree/bindings/display/bridge/dw_hdmi.txt with the
+following device-specific properties.
+
+Required properties:
+
+  - compatible: value must be one of:
+    * "allwinner,sun8i-a83t-dw-hdmi"
+  - reg: base address and size of memory-mapped region
+  - reg-io-width: See dw_hdmi.txt. Shall be 1.
+  - interrupts: HDMI interrupt number
+  - clocks: phandles to the clocks feeding the HDMI encoder
+    * iahb: the HDMI bus clock
+    * isfr: the HDMI register clock
+    * tmds: TMDS clock
+  - clock-names: the clock names mentioned above
+  - resets: phandle to the reset controller
+  - reset-names: must be "ctrl"
+  - phys: phandle to the DWC HDMI PHY
+  - phy-names: must be "phy"
+
+  - ports: A ports node with endpoint definitions as defined in
+    Documentation/devicetree/bindings/media/video-interfaces.txt. The
+    first port should be the input endpoint. The second should be the
+    output, usually to an HDMI connector.
+
+DWC HDMI PHY
+------------
+
+Required properties:
+  - compatible: value must be one of:
+    * allwinner,sun8i-a83t-hdmi-phy
+    * allwinner,sun8i-h3-hdmi-phy
+  - reg: base address and size of memory-mapped region
+  - clocks: phandles to the clocks feeding the HDMI PHY
+    * bus: the HDMI PHY interface clock
+    * mod: the HDMI PHY module clock
+  - clock-names: the clock names mentioned above
+  - resets: phandle to the reset controller driving the PHY
+  - reset-names: must be "phy"
+
+H3 HDMI PHY requires additional clock:
+  - pll-0: parent of phy clock
+
 TV Encoder
 ----------
 
@@ -94,24 +144,29 @@ Required properties:
    * allwinner,sun7i-a20-tcon
    * allwinner,sun8i-a33-tcon
    * allwinner,sun8i-a83t-tcon-lcd
+   * allwinner,sun8i-a83t-tcon-tv
    * allwinner,sun8i-v3s-tcon
+   * allwinner,sun9i-a80-tcon-lcd
+   * allwinner,sun9i-a80-tcon-tv
  - reg: base address and size of memory-mapped region
  - interrupts: interrupt associated to this IP
- - clocks: phandles to the clocks feeding the TCON. Three are needed:
+ - clocks: phandles to the clocks feeding the TCON.
    - 'ahb': the interface clocks
-   - 'tcon-ch0': The clock driving the TCON channel 0
+   - 'tcon-ch0': The clock driving the TCON channel 0, if supported
  - resets: phandles to the reset controllers driving the encoder
-   - "lcd": the reset line for the TCON channel 0
+   - "lcd": the reset line for the TCON
+   - "edp": the reset line for the eDP block (A80 only)
 
  - clock-names: the clock names mentioned above
  - reset-names: the reset names mentioned above
- - clock-output-names: Name of the pixel clock created
+ - clock-output-names: Name of the pixel clock created, if TCON supports
+   channel 0.
 
 - ports: A ports node with endpoint definitions as defined in
   Documentation/devicetree/bindings/media/video-interfaces.txt. The
   first port should be the input endpoint, the second one the output
 
-  The output may have multiple endpoints. The TCON has two channels,
+  The output may have multiple endpoints. TCON can have 1 or 2 channels,
   usually with the first channel being used for the panels interfaces
   (RGB, LVDS, etc.), and the second being used for the outputs that
   require another controller (TV Encoder, HDMI, etc.). The endpoints
@@ -119,11 +174,13 @@ Required properties:
   channel the endpoint is associated to. If that property is not
   present, the endpoint number will be used as the channel number.
 
-On SoCs other than the A33 and V3s, there is one more clock required:
+For TCONs with channel 0, there is one more clock required:
+   - 'tcon-ch0': The clock driving the TCON channel 0
+For TCONs with channel 1, there is one more clock required:
    - 'tcon-ch1': The clock driving the TCON channel 1
 
-On SoCs that support LVDS (all SoCs but the A13, H3, H5 and V3s), you
-need one more reset line:
+When TCON support LVDS (all TCONs except TV TCON on A83T and those found
+in A13, H3, H5 and V3s SoCs), you need one more reset line:
    - 'lvds': The reset line driving the LVDS logic
 
 And on the A23, A31, A31s and A33, you need one more clock line:
@@ -134,7 +191,7 @@ DRC
 ---
 
 The DRC (Dynamic Range Controller), found in the latest Allwinner SoCs
-(A31, A23, A33), allows to dynamically adjust pixel
+(A31, A23, A33, A80), allows to dynamically adjust pixel
 brightness/contrast based on histogram measurements for LCD content
 adaptive backlight control.
 
@@ -144,6 +201,7 @@ Required properties:
     * allwinner,sun6i-a31-drc
     * allwinner,sun6i-a31s-drc
     * allwinner,sun8i-a33-drc
+    * allwinner,sun9i-a80-drc
   - reg: base address and size of the memory-mapped region.
   - interrupts: interrupt associated to this IP
   - clocks: phandles to the clocks feeding the DRC
@@ -170,6 +228,7 @@ Required properties:
     * allwinner,sun6i-a31-display-backend
     * allwinner,sun7i-a20-display-backend
     * allwinner,sun8i-a33-display-backend
+    * allwinner,sun9i-a80-display-backend
   - reg: base address and size of the memory-mapped region.
   - interrupts: interrupt associated to this IP
   - clocks: phandles to the clocks feeding the frontend and backend
@@ -191,6 +250,28 @@ On the A33, some additional properties are required:
   - resets and reset-names need to have a phandle to the SAT bus
     resets, whose name will be "sat"
 
+DEU
+---
+
+The DEU (Detail Enhancement Unit), found in the Allwinner A80 SoC,
+can sharpen the display content in both luma and chroma channels.
+
+Required properties:
+  - compatible: value must be one of:
+    * allwinner,sun9i-a80-deu
+  - reg: base address and size of the memory-mapped region.
+  - interrupts: interrupt associated to this IP
+  - clocks: phandles to the clocks feeding the DEU
+    * ahb: the DEU interface clock
+    * mod: the DEU module clock
+    * ram: the DEU DRAM clock
+  - clock-names: the clock names mentioned above
+  - resets: phandles to the reset line driving the DEU
+
+- ports: A ports node with endpoint definitions as defined in
+  Documentation/devicetree/bindings/media/video-interfaces.txt. The
+  first port should be the input endpoints, the second one the outputs
+
 Display Engine Frontend
 -----------------------
 
@@ -204,6 +285,7 @@ Required properties:
     * allwinner,sun6i-a31-display-frontend
     * allwinner,sun7i-a20-display-frontend
     * allwinner,sun8i-a33-display-frontend
+    * allwinner,sun9i-a80-display-frontend
   - reg: base address and size of the memory-mapped region.
   - interrupts: interrupt associated to this IP
   - clocks: phandles to the clocks feeding the frontend and backend
@@ -226,6 +308,8 @@ supported.
 Required properties:
   - compatible: value must be one of:
     * allwinner,sun8i-a83t-de2-mixer-0
+    * allwinner,sun8i-a83t-de2-mixer-1
+    * allwinner,sun8i-h3-de2-mixer-0
     * allwinner,sun8i-v3s-de2-mixer
   - reg: base address and size of the memory-mapped region.
   - clocks: phandles to the clocks feeding the mixer
@@ -256,7 +340,9 @@ Required properties:
     * allwinner,sun7i-a20-display-engine
     * allwinner,sun8i-a33-display-engine
     * allwinner,sun8i-a83t-display-engine
+    * allwinner,sun8i-h3-display-engine
     * allwinner,sun8i-v3s-display-engine
+    * allwinner,sun9i-a80-display-engine
 
   - allwinner,pipelines: list of phandle to the display engine
     frontends (DE 1.0) or mixers (DE 2.0) available.

Documentation/devicetree/bindings/iio/adc/sigma-delta-modulator.txt

@@ -3,11 +3,11 @@ Device-Tree bindings for sigma delta modulator
 Required properties:
 - compatible: should be "ads1201", "sd-modulator". "sd-modulator" can be use
 	as a generic SD modulator if modulator not specified in compatible list.
-- #io-channel-cells = <1>: See the IIO bindings section "IIO consumers".
+- #io-channel-cells = <0>: See the IIO bindings section "IIO consumers".
 
 Example node:
 
 	ads1202: adc@0 {
 		compatible = "sd-modulator";
-		#io-channel-cells = <1>;
+		#io-channel-cells = <0>;
 	};

Documentation/devicetree/bindings/net/dsa/marvell.txt

@@ -50,14 +50,15 @@ Example:
 			compatible = "marvell,mv88e6085";
 			reg = <0>;
 			reset-gpios = <&gpio5 1 GPIO_ACTIVE_LOW>;
-		};
-		mdio {
-			#address-cells = <1>;
-			#size-cells = <0>;
-			switch1phy0: switch1phy0@0 {
-				reg = <0>;
-				interrupt-parent = <&switch0>;
-				interrupts = <0 IRQ_TYPE_LEVEL_HIGH>;
+
+			mdio {
+				#address-cells = <1>;
+				#size-cells = <0>;
+				switch1phy0: switch1phy0@0 {
+					reg = <0>;
+					interrupt-parent = <&switch0>;
+					interrupts = <0 IRQ_TYPE_LEVEL_HIGH>;
+				};
 			};
 		};
 	};
@@ -74,23 +75,24 @@ Example:
 			compatible = "marvell,mv88e6390";
 			reg = <0>;
 			reset-gpios = <&gpio5 1 GPIO_ACTIVE_LOW>;
-		};
-		mdio {
-			#address-cells = <1>;
-			#size-cells = <0>;
-			switch1phy0: switch1phy0@0 {
-				reg = <0>;
-				interrupt-parent = <&switch0>;
-				interrupts = <0 IRQ_TYPE_LEVEL_HIGH>;
+
+			mdio {
+				#address-cells = <1>;
+				#size-cells = <0>;
+				switch1phy0: switch1phy0@0 {
+					reg = <0>;
+					interrupt-parent = <&switch0>;
+					interrupts = <0 IRQ_TYPE_LEVEL_HIGH>;
+				};
 			};
-		};
 
-		mdio1 {
-			compatible = "marvell,mv88e6xxx-mdio-external";
-			#address-cells = <1>;
-			#size-cells = <0>;
-			switch1phy9: switch1phy0@9 {
-				reg = <9>;
+			mdio1 {
+				compatible = "marvell,mv88e6xxx-mdio-external";
+				#address-cells = <1>;
+				#size-cells = <0>;
+				switch1phy9: switch1phy0@9 {
+					reg = <9>;
+				};
 			};
 		};
 	};

Documentation/devicetree/bindings/net/renesas,ravb.txt

@@ -27,7 +27,11 @@ Required properties:
 	SoC-specific version corresponding to the platform first followed by
 	the generic version.
 
-- reg: offset and length of (1) the register block and (2) the stream buffer.
+- reg: Offset and length of (1) the register block and (2) the stream buffer.
+       The region for the register block is mandatory.
+       The region for the stream buffer is optional, as it is only present on
+       R-Car Gen2 and RZ/G1 SoCs, and on R-Car H3 (R8A7795), M3-W (R8A7796),
+       and M3-N (R8A77965).
 - interrupts: A list of interrupt-specifiers, one for each entry in
 	      interrupt-names.
 	      If interrupt-names is not present, an interrupt specifier

Documentation/devicetree/bindings/vendor-prefixes.txt

@@ -104,6 +104,7 @@ eeti	eGalax_eMPIA Technology Inc
 elan	Elan Microelectronic Corp.
 embest	Shenzhen Embest Technology Co., Ltd.
 emmicro	EM Microelectronic
+emtrion	emtrion GmbH
 energymicro	Silicon Laboratories (formerly Energy Micro AS)
 engicam	Engicam S.r.l.
 epcos	EPCOS AG

Documentation/devicetree/overlay-notes.txt

@@ -87,8 +87,8 @@ Overlay in-kernel API
 
 The API is quite easy to use.
 
-1. Call of_overlay_apply() to create and apply an overlay changeset. The return
-value is an error or a cookie identifying this overlay.
+1. Call of_overlay_fdt_apply() to create and apply an overlay changeset. The
+return value is an error or a cookie identifying this overlay.
 
 2. Call of_overlay_remove() to remove and cleanup the overlay changeset
 previously created via the call to of_overlay_apply(). Removal of an overlay

Documentation/gpu/drivers.rst

@@ -0,0 +1,21 @@
+========================
+GPU Driver Documentation
+========================
+
+.. toctree::
+
+   i915
+   meson
+   pl111
+   tegra
+   tinydrm
+   tve200
+   vc4
+   bridge/dw-hdmi
+
+.. only::  subproject and html
+
+   Indices
+   =======
+
+   * :ref:`genindex`

Documentation/gpu/drm-kms.rst

@@ -286,6 +286,9 @@ Atomic Mode Setting Function Reference
 .. kernel-doc:: drivers/gpu/drm/drm_atomic.c
    :export:
 
+.. kernel-doc:: drivers/gpu/drm/drm_atomic.c
+   :internal:
+
 CRTC Abstraction
 ================
 
@@ -547,8 +550,9 @@ Explicit Fencing Properties
 Existing KMS Properties
 -----------------------
 
-The following table gives description of drm properties exposed by
-various modules/drivers.
+The following table gives description of drm properties exposed by various
+modules/drivers. Because this table is very unwieldy, do not add any new
+properties here. Instead document them in a section above.
 
 .. csv-table::
    :header-rows: 1

Documentation/gpu/index.rst

@@ -10,16 +10,9 @@ Linux GPU Driver Developer's Guide
    drm-kms
    drm-kms-helpers
    drm-uapi
-   i915
-   meson
-   pl111
-   tegra
-   tinydrm
-   tve200
-   vc4
+   drivers
    vga-switcheroo
    vgaarbiter
-   bridge/dw-hdmi
    todo
 
 .. only::  subproject and html

Documentation/gpu/kms-properties.csv

@@ -1,5 +1,4 @@
 Owner Module/Drivers,Group,Property Name,Type,Property Values,Object attached,Description/Restrictions
-,,“scaling mode”,ENUM,"{ ""None"", ""Full"", ""Center"", ""Full aspect"" }",Connector,"Supported by: amdgpu, gma500, i915, nouveau and radeon."
 ,DVI-I,“subconnector”,ENUM,"{ “Unknown”, “DVI-D”, “DVI-A” }",Connector,TBD
 ,,“select subconnector”,ENUM,"{ “Automatic”, “DVI-D”, “DVI-A” }",Connector,TBD
 ,TV,“subconnector”,ENUM,"{ ""Unknown"", ""Composite"", ""SVIDEO"", ""Component"", ""SCART"" }",Connector,TBD

Documentation/gpu/todo.rst

@@ -212,6 +212,16 @@ probably use drm_fb_helper_fbdev_teardown().
 
 Contact: Maintainer of the driver you plan to convert
 
+idr_init_base()
+---------------
+
+DRM core&drivers uses a lot of idr (integer lookup directories) for mapping
+userspace IDs to internal objects, and in most places ID=0 means NULL and hence
+is never used. Switching to idr_init_base() for these would make the idr more
+efficient.
+
+Contact: Daniel Vetter
+
 Core refactorings
 =================
 
@@ -440,5 +450,12 @@ See drivers/gpu/drm/amd/display/TODO for tasks.
 
 Contact: Harry Wentland, Alex Deucher
 
+i915
+----
+
+- Our early/late pm callbacks could be removed in favour of using
+  device_link_add to model the dependency between i915 and snd_had. See
+  https://dri.freedesktop.org/docs/drm/driver-api/device_link.html
+
 Outside DRM
 ===========

Documentation/networking/segmentation-offloads.txt

@@ -20,8 +20,8 @@ TCP Segmentation Offload
 
 TCP segmentation allows a device to segment a single frame into multiple
 frames with a data payload size specified in skb_shinfo()->gso_size.
-When TCP segmentation requested the bit for either SKB_GSO_TCP or
-SKB_GSO_TCP6 should be set in skb_shinfo()->gso_type and
+When TCP segmentation requested the bit for either SKB_GSO_TCPV4 or
+SKB_GSO_TCPV6 should be set in skb_shinfo()->gso_type and
 skb_shinfo()->gso_size should be set to a non-zero value.
 
 TCP segmentation is dependent on support for the use of partial checksum
@@ -153,8 +153,18 @@ To signal this, gso_size is set to the special value GSO_BY_FRAGS.
 
 Therefore, any code in the core networking stack must be aware of the
 possibility that gso_size will be GSO_BY_FRAGS and handle that case
-appropriately. (For size checks, the skb_gso_validate_*_len family of
-helpers do this automatically.)
+appropriately.
+
+There are some helpers to make this easier:
+
+ - skb_is_gso(skb) && skb_is_gso_sctp(skb) is the best way to see if
+   an skb is an SCTP GSO skb.
+
+ - For size checks, the skb_gso_validate_*_len family of helpers correctly
+   considers GSO_BY_FRAGS.
+
+ - For manipulating packets, skb_increase_gso_size and skb_decrease_gso_size
+   will check for GSO_BY_FRAGS and WARN if asked to manipulate these skbs.
 
 This also affects drivers with the NETIF_F_FRAGLIST & NETIF_F_GSO_SCTP bits
 set. Note also that NETIF_F_GSO_SCTP is included in NETIF_F_GSO_SOFTWARE.

MAINTAINERS

@@ -766,6 +766,8 @@ F:	drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
 F:	drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
 F:	drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
 F:	drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+F:	drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
+F:	drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
 F:	drivers/gpu/drm/amd/amdkfd/
 F:	drivers/gpu/drm/amd/include/cik_structs.h
 F:	drivers/gpu/drm/amd/include/kgd_kfd_interface.h
@@ -4458,6 +4460,13 @@ T:	git git://anongit.freedesktop.org/drm/drm-misc
 S:	Supported
 F:	drivers/gpu/drm/pl111/
 
+DRM DRIVER FOR ARM VERSATILE TFT PANELS
+M:	Linus Walleij <linus.walleij@linaro.org>
+T:	git git://anongit.freedesktop.org/drm/drm-misc
+S:	Maintained
+F:	drivers/gpu/drm/panel/panel-arm-versatile.c
+F:	Documentation/devicetree/bindings/display/panel/arm,versatile-tft-panel.txt
+
 DRM DRIVER FOR AST SERVER GRAPHICS CHIPS
 M:	Dave Airlie <airlied@redhat.com>
 S:	Odd Fixes
@@ -4612,8 +4621,8 @@ F:	include/uapi/drm/
 F:	include/linux/vga*
 
 DRM DRIVERS AND MISC GPU PATCHES
-M:	Daniel Vetter <daniel.vetter@intel.com>
 M:	Gustavo Padovan <gustavo@padovan.org>
+M:	Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
 M:	Sean Paul <seanpaul@chromium.org>
 W:	https://01.org/linuxgraphics/gfx-docs/maintainer-tools/drm-misc.html
 S:	Maintained
@@ -4739,6 +4748,7 @@ F:	drivers/gpu/drm/rcar-du/
 F:	drivers/gpu/drm/shmobile/
 F:	include/linux/platform_data/shmob_drm.h
 F:	Documentation/devicetree/bindings/display/bridge/renesas,dw-hdmi.txt
+F:	Documentation/devicetree/bindings/display/bridge/renesas,lvds.txt
 F:	Documentation/devicetree/bindings/display/renesas,du.txt
 
 DRM DRIVERS FOR ROCKCHIP
@@ -10334,7 +10344,7 @@ F:	drivers/oprofile/
 F:	include/linux/oprofile.h
 
 ORACLE CLUSTER FILESYSTEM 2 (OCFS2)
-M:	Mark Fasheh <mfasheh@versity.com>
+M:	Mark Fasheh <mark@fasheh.com>
 M:	Joel Becker <jlbec@evilplan.org>
 L:	ocfs2-devel@oss.oracle.com (moderated for non-subscribers)
 W:	http://ocfs2.wiki.kernel.org
@@ -10844,6 +10854,7 @@ F:	drivers/platform/x86/peaq-wmi.c
 PER-CPU MEMORY ALLOCATOR
 M:	Tejun Heo <tj@kernel.org>
 M:	Christoph Lameter <cl@linux.com>
+M:	Dennis Zhou <dennisszhou@gmail.com>
 T:	git git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu.git
 S:	Maintained
 F:	include/linux/percpu*.h

Makefile

@@ -2,7 +2,7 @@
 VERSION = 4
 PATCHLEVEL = 16
 SUBLEVEL = 0
-EXTRAVERSION = -rc6
+EXTRAVERSION = -rc7
 NAME = Fearless Coyote
 
 # *DOCUMENTATION*
@@ -826,6 +826,15 @@ KBUILD_CFLAGS += $(call cc-disable-warning, pointer-sign)
 # disable invalid "can't wrap" optimizations for signed / pointers
 KBUILD_CFLAGS	+= $(call cc-option,-fno-strict-overflow)
 
+# clang sets -fmerge-all-constants by default as optimization, but this
+# is non-conforming behavior for C and in fact breaks the kernel, so we
+# need to disable it here generally.
+KBUILD_CFLAGS	+= $(call cc-option,-fno-merge-all-constants)
+
+# for gcc -fno-merge-all-constants disables everything, but it is fine
+# to have actual conforming behavior enabled.
+KBUILD_CFLAGS	+= $(call cc-option,-fmerge-constants)
+
 # Make sure -fstack-check isn't enabled (like gentoo apparently did)
 KBUILD_CFLAGS  += $(call cc-option,-fno-stack-check,)
 

arch/arm64/mm/mmu.c

@@ -972,3 +972,13 @@ int pmd_clear_huge(pmd_t *pmdp)
 	pmd_clear(pmdp);
 	return 1;
 }
+
+int pud_free_pmd_page(pud_t *pud)
+{
+	return pud_none(*pud);
+}
+
+int pmd_free_pte_page(pmd_t *pmd)
+{
+	return pmd_none(*pmd);
+}

arch/h8300/include/asm/byteorder.h

@@ -2,7 +2,6 @@
 #ifndef __H8300_BYTEORDER_H__
 #define __H8300_BYTEORDER_H__
 
-#define __BIG_ENDIAN __ORDER_BIG_ENDIAN__
 #include <linux/byteorder/big_endian.h>
 
 #endif

arch/mips/lantiq/Kconfig

@@ -13,6 +13,8 @@ choice
 config SOC_AMAZON_SE
 	bool "Amazon SE"
 	select SOC_TYPE_XWAY
+	select MFD_SYSCON
+	select MFD_CORE
 
 config SOC_XWAY
 	bool "XWAY"

arch/mips/lantiq/xway/sysctrl.c

@@ -549,9 +549,9 @@ void __init ltq_soc_init(void)
 		clkdev_add_static(ltq_ar9_cpu_hz(), ltq_ar9_fpi_hz(),
 				ltq_ar9_fpi_hz(), CLOCK_250M);
 		clkdev_add_pmu("1f203018.usb2-phy", "phy", 1, 0, PMU_USB0_P);
-		clkdev_add_pmu("1e101000.usb", "otg", 1, 0, PMU_USB0);
+		clkdev_add_pmu("1e101000.usb", "otg", 1, 0, PMU_USB0 | PMU_AHBM);
 		clkdev_add_pmu("1f203034.usb2-phy", "phy", 1, 0, PMU_USB1_P);
-		clkdev_add_pmu("1e106000.usb", "otg", 1, 0, PMU_USB1);
+		clkdev_add_pmu("1e106000.usb", "otg", 1, 0, PMU_USB1 | PMU_AHBM);
 		clkdev_add_pmu("1e180000.etop", "switch", 1, 0, PMU_SWITCH);
 		clkdev_add_pmu("1e103000.sdio", NULL, 1, 0, PMU_SDIO);
 		clkdev_add_pmu("1e103100.deu", NULL, 1, 0, PMU_DEU);
@@ -560,7 +560,7 @@ void __init ltq_soc_init(void)
 	} else {
 		clkdev_add_static(ltq_danube_cpu_hz(), ltq_danube_fpi_hz(),
 				ltq_danube_fpi_hz(), ltq_danube_pp32_hz());
-		clkdev_add_pmu("1f203018.usb2-phy", "ctrl", 1, 0, PMU_USB0);
+		clkdev_add_pmu("1e101000.usb", "otg", 1, 0, PMU_USB0 | PMU_AHBM);
 		clkdev_add_pmu("1f203018.usb2-phy", "phy", 1, 0, PMU_USB0_P);
 		clkdev_add_pmu("1e103000.sdio", NULL, 1, 0, PMU_SDIO);
 		clkdev_add_pmu("1e103100.deu", NULL, 1, 0, PMU_DEU);

arch/mips/ralink/mt7621.c

@@ -170,6 +170,28 @@ void prom_soc_init(struct ralink_soc_info *soc_info)
 	u32 n1;
 	u32 rev;
 
+	/* Early detection of CMP support */
+	mips_cm_probe();
+	mips_cpc_probe();
+
+	if (mips_cps_numiocu(0)) {
+		/*
+		 * mips_cm_probe() wipes out bootloader
+		 * config for CM regions and we have to configure them
+		 * again. This SoC cannot talk to pamlbus devices
+		 * witout proper iocu region set up.
+		 *
+		 * FIXME: it would be better to do this with values
+		 * from DT, but we need this very early because
+		 * without this we cannot talk to pretty much anything
+		 * including serial.
+		 */
+		write_gcr_reg0_base(MT7621_PALMBUS_BASE);
+		write_gcr_reg0_mask(~MT7621_PALMBUS_SIZE |
+				    CM_GCR_REGn_MASK_CMTGT_IOCU0);
+		__sync();
+	}
+
 	n0 = __raw_readl(sysc + SYSC_REG_CHIP_NAME0);
 	n1 = __raw_readl(sysc + SYSC_REG_CHIP_NAME1);
 
@@ -194,26 +216,6 @@ void prom_soc_init(struct ralink_soc_info *soc_info)
 
 	rt2880_pinmux_data = mt7621_pinmux_data;
 
-	/* Early detection of CMP support */
-	mips_cm_probe();
-	mips_cpc_probe();
-
-	if (mips_cps_numiocu(0)) {
-		/*
-		 * mips_cm_probe() wipes out bootloader
-		 * config for CM regions and we have to configure them
-		 * again. This SoC cannot talk to pamlbus devices
-		 * witout proper iocu region set up.
-		 *
-		 * FIXME: it would be better to do this with values
-		 * from DT, but we need this very early because
-		 * without this we cannot talk to pretty much anything
-		 * including serial.
-		 */
-		write_gcr_reg0_base(MT7621_PALMBUS_BASE);
-		write_gcr_reg0_mask(~MT7621_PALMBUS_SIZE |
-				    CM_GCR_REGn_MASK_CMTGT_IOCU0);
-	}
 
 	if (!register_cps_smp_ops())
 		return;

arch/mips/ralink/reset.c

@@ -96,16 +96,9 @@ static void ralink_restart(char *command)
 	unreachable();
 }
 
-static void ralink_halt(void)
-{
-	local_irq_disable();
-	unreachable();
-}
-
 static int __init mips_reboot_setup(void)
 {
 	_machine_restart = ralink_restart;
-	_machine_halt = ralink_halt;
 
 	return 0;
 }

arch/x86/Kconfig.cpu

@@ -315,19 +315,6 @@ config X86_L1_CACHE_SHIFT
 	default "4" if MELAN || M486 || MGEODEGX1
 	default "5" if MWINCHIP3D || MWINCHIPC6 || MCRUSOE || MEFFICEON || MCYRIXIII || MK6 || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || M586 || MVIAC3_2 || MGEODE_LX
 
-config X86_PPRO_FENCE
-	bool "PentiumPro memory ordering errata workaround"
-	depends on M686 || M586MMX || M586TSC || M586 || M486 || MGEODEGX1
-	---help---
-	  Old PentiumPro multiprocessor systems had errata that could cause
-	  memory operations to violate the x86 ordering standard in rare cases.
-	  Enabling this option will attempt to work around some (but not all)
-	  occurrences of this problem, at the cost of much heavier spinlock and
-	  memory barrier operations.
-
-	  If unsure, say n here. Even distro kernels should think twice before
-	  enabling this: there are few systems, and an unlikely bug.
-
 config X86_F00F_BUG
 	def_bool y
 	depends on M586MMX || M586TSC || M586 || M486

arch/x86/Makefile

@@ -223,6 +223,15 @@ KBUILD_CFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr) $(avx_instr)
 
 LDFLAGS := -m elf_$(UTS_MACHINE)
 
+#
+# The 64-bit kernel must be aligned to 2MB.  Pass -z max-page-size=0x200000 to
+# the linker to force 2MB page size regardless of the default page size used
+# by the linker.
+#
+ifdef CONFIG_X86_64
+LDFLAGS += $(call ld-option, -z max-page-size=0x200000)
+endif
+
 # Speed up the build
 KBUILD_CFLAGS += -pipe
 # Workaround for a gcc prelease that unfortunately was shipped in a suse release

arch/x86/boot/compressed/misc.c

@@ -309,6 +309,10 @@ static void parse_elf(void *output)
 
 		switch (phdr->p_type) {
 		case PT_LOAD:
+#ifdef CONFIG_X86_64
+			if ((phdr->p_align % 0x200000) != 0)
+				error("Alignment of LOAD segment isn't multiple of 2MB");
+#endif
 #ifdef CONFIG_RELOCATABLE
 			dest = output;
 			dest += (phdr->p_paddr - LOAD_PHYSICAL_ADDR);

arch/x86/entry/entry_64.S

@@ -1138,7 +1138,7 @@ apicinterrupt3 HYPERV_REENLIGHTENMENT_VECTOR \
 #endif /* CONFIG_HYPERV */
 
 idtentry debug			do_debug		has_error_code=0	paranoid=1 shift_ist=DEBUG_STACK
-idtentry int3			do_int3			has_error_code=0	paranoid=1 shift_ist=DEBUG_STACK
+idtentry int3			do_int3			has_error_code=0
 idtentry stack_segment		do_stack_segment	has_error_code=1
 
 #ifdef CONFIG_XEN

arch/x86/entry/vdso/vdso32/vclock_gettime.c

@@ -5,8 +5,6 @@
 #undef CONFIG_OPTIMIZE_INLINING
 #endif
 
-#undef CONFIG_X86_PPRO_FENCE
-
 #ifdef CONFIG_X86_64
 
 /*

arch/x86/entry/vsyscall/vsyscall_64.c

@@ -347,7 +347,7 @@ void __init set_vsyscall_pgtable_user_bits(pgd_t *root)
 	set_pgd(pgd, __pgd(pgd_val(*pgd) | _PAGE_USER));
 	p4d = p4d_offset(pgd, VSYSCALL_ADDR);
 #if CONFIG_PGTABLE_LEVELS >= 5
-	p4d->p4d |= _PAGE_USER;
+	set_p4d(p4d, __p4d(p4d_val(*p4d) | _PAGE_USER));
 #endif
 	pud = pud_offset(p4d, VSYSCALL_ADDR);
 	set_pud(pud, __pud(pud_val(*pud) | _PAGE_USER));

arch/x86/events/core.c

@@ -2118,7 +2118,8 @@ static int x86_pmu_event_init(struct perf_event *event)
 			event->destroy(event);
 	}
 
-	if (READ_ONCE(x86_pmu.attr_rdpmc))
+	if (READ_ONCE(x86_pmu.attr_rdpmc) &&
+	    !(event->hw.flags & PERF_X86_EVENT_LARGE_PEBS))
 		event->hw.flags |= PERF_X86_EVENT_RDPMC_ALLOWED;
 
 	return err;

arch/x86/events/intel/core.c

@@ -2952,9 +2952,9 @@ static void intel_pebs_aliases_skl(struct perf_event *event)
 	return intel_pebs_aliases_precdist(event);
 }
 
-static unsigned long intel_pmu_free_running_flags(struct perf_event *event)
+static unsigned long intel_pmu_large_pebs_flags(struct perf_event *event)
 {
-	unsigned long flags = x86_pmu.free_running_flags;
+	unsigned long flags = x86_pmu.large_pebs_flags;
 
 	if (event->attr.use_clockid)
 		flags &= ~PERF_SAMPLE_TIME;
@@ -2976,8 +2976,8 @@ static int intel_pmu_hw_config(struct perf_event *event)
 		if (!event->attr.freq) {
 			event->hw.flags |= PERF_X86_EVENT_AUTO_RELOAD;
 			if (!(event->attr.sample_type &
-			      ~intel_pmu_free_running_flags(event)))
-				event->hw.flags |= PERF_X86_EVENT_FREERUNNING;
+			      ~intel_pmu_large_pebs_flags(event)))
+				event->hw.flags |= PERF_X86_EVENT_LARGE_PEBS;
 		}
 		if (x86_pmu.pebs_aliases)
 			x86_pmu.pebs_aliases(event);
@@ -3194,7 +3194,7 @@ static unsigned bdw_limit_period(struct perf_event *event, unsigned left)
 			X86_CONFIG(.event=0xc0, .umask=0x01)) {
 		if (left < 128)
 			left = 128;
-		left &= ~0x3fu;
+		left &= ~0x3fULL;
 	}
 	return left;
 }
@@ -3460,7 +3460,7 @@ static __initconst const struct x86_pmu core_pmu = {
 	.event_map		= intel_pmu_event_map,
 	.max_events		= ARRAY_SIZE(intel_perfmon_event_map),
 	.apic			= 1,
-	.free_running_flags	= PEBS_FREERUNNING_FLAGS,
+	.large_pebs_flags	= LARGE_PEBS_FLAGS,
 
 	/*
 	 * Intel PMCs cannot be accessed sanely above 32-bit width,
@@ -3502,7 +3502,7 @@ static __initconst const struct x86_pmu intel_pmu = {
 	.event_map		= intel_pmu_event_map,
 	.max_events		= ARRAY_SIZE(intel_perfmon_event_map),
 	.apic			= 1,
-	.free_running_flags	= PEBS_FREERUNNING_FLAGS,
+	.large_pebs_flags	= LARGE_PEBS_FLAGS,
 	/*
 	 * Intel PMCs cannot be accessed sanely above 32 bit width,
 	 * so we install an artificial 1<<31 period regardless of

arch/x86/events/intel/ds.c

@@ -935,7 +935,7 @@ void intel_pmu_pebs_add(struct perf_event *event)
 	bool needed_cb = pebs_needs_sched_cb(cpuc);
 
 	cpuc->n_pebs++;
-	if (hwc->flags & PERF_X86_EVENT_FREERUNNING)
+	if (hwc->flags & PERF_X86_EVENT_LARGE_PEBS)
 		cpuc->n_large_pebs++;
 
 	pebs_update_state(needed_cb, cpuc, event->ctx->pmu);
@@ -975,7 +975,7 @@ void intel_pmu_pebs_del(struct perf_event *event)
 	bool needed_cb = pebs_needs_sched_cb(cpuc);
 
 	cpuc->n_pebs--;
-	if (hwc->flags & PERF_X86_EVENT_FREERUNNING)
+	if (hwc->flags & PERF_X86_EVENT_LARGE_PEBS)
 		cpuc->n_large_pebs--;
 
 	pebs_update_state(needed_cb, cpuc, event->ctx->pmu);
@@ -1530,7 +1530,7 @@ void __init intel_ds_init(void)
 			x86_pmu.pebs_record_size =
 						sizeof(struct pebs_record_skl);
 			x86_pmu.drain_pebs = intel_pmu_drain_pebs_nhm;
-			x86_pmu.free_running_flags |= PERF_SAMPLE_TIME;
+			x86_pmu.large_pebs_flags |= PERF_SAMPLE_TIME;
 			break;
 
 		default:

arch/x86/events/intel/uncore_snbep.c

@@ -3343,6 +3343,7 @@ static struct extra_reg skx_uncore_cha_extra_regs[] = {
 	SNBEP_CBO_EVENT_EXTRA_REG(0x9134, 0xffff, 0x4),
 	SNBEP_CBO_EVENT_EXTRA_REG(0x35, 0xff, 0x8),
 	SNBEP_CBO_EVENT_EXTRA_REG(0x36, 0xff, 0x8),
+	SNBEP_CBO_EVENT_EXTRA_REG(0x38, 0xff, 0x3),
 	EVENT_EXTRA_END
 };
 
@@ -3562,24 +3563,27 @@ static struct intel_uncore_type *skx_msr_uncores[] = {
 	NULL,
 };
 
+/*
+ * To determine the number of CHAs, it should read bits 27:0 in the CAPID6
+ * register which located at Device 30, Function 3, Offset 0x9C. PCI ID 0x2083.
+ */
+#define SKX_CAPID6		0x9c
+#define SKX_CHA_BIT_MASK	GENMASK(27, 0)
+
 static int skx_count_chabox(void)
 {
-	struct pci_dev *chabox_dev = NULL;
-	int bus, count = 0;
+	struct pci_dev *dev = NULL;
+	u32 val = 0;
 
-	while (1) {
-		chabox_dev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x208d, chabox_dev);
-		if (!chabox_dev)
-			break;
-		if (count == 0)
-			bus = chabox_dev->bus->number;
-		if (bus != chabox_dev->bus->number)
-			break;
-		count++;
-	}
+	dev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x2083, dev);
+	if (!dev)
+		goto out;
 
-	pci_dev_put(chabox_dev);
-	return count;
+	pci_read_config_dword(dev, SKX_CAPID6, &val);
+	val &= SKX_CHA_BIT_MASK;
+out:
+	pci_dev_put(dev);
+	return hweight32(val);
 }
 
 void skx_uncore_cpu_init(void)

arch/x86/events/perf_event.h

@@ -69,7 +69,7 @@ struct event_constraint {
 #define PERF_X86_EVENT_RDPMC_ALLOWED	0x0100 /* grant rdpmc permission */
 #define PERF_X86_EVENT_EXCL_ACCT	0x0200 /* accounted EXCL event */
 #define PERF_X86_EVENT_AUTO_RELOAD	0x0400 /* use PEBS auto-reload */
-#define PERF_X86_EVENT_FREERUNNING	0x0800 /* use freerunning PEBS */
+#define PERF_X86_EVENT_LARGE_PEBS	0x0800 /* use large PEBS */
 
 
 struct amd_nb {
@@ -88,7 +88,7 @@ struct amd_nb {
  * REGS_USER can be handled for events limited to ring 3.
  *
  */
-#define PEBS_FREERUNNING_FLAGS \
+#define LARGE_PEBS_FLAGS \
 	(PERF_SAMPLE_IP | PERF_SAMPLE_TID | PERF_SAMPLE_ADDR | \
 	PERF_SAMPLE_ID | PERF_SAMPLE_CPU | PERF_SAMPLE_STREAM_ID | \
 	PERF_SAMPLE_DATA_SRC | PERF_SAMPLE_IDENTIFIER | \
@@ -608,7 +608,7 @@ struct x86_pmu {
 	struct event_constraint *pebs_constraints;
 	void		(*pebs_aliases)(struct perf_event *event);
 	int 		max_pebs_events;
-	unsigned long	free_running_flags;
+	unsigned long	large_pebs_flags;
 
 	/*
 	 * Intel LBR

arch/x86/include/asm/barrier.h

@@ -52,11 +52,7 @@ static inline unsigned long array_index_mask_nospec(unsigned long index,
 #define barrier_nospec() alternative_2("", "mfence", X86_FEATURE_MFENCE_RDTSC, \
 					   "lfence", X86_FEATURE_LFENCE_RDTSC)
 
-#ifdef CONFIG_X86_PPRO_FENCE
-#define dma_rmb()	rmb()
-#else
 #define dma_rmb()	barrier()
-#endif
 #define dma_wmb()	barrier()
 
 #ifdef CONFIG_X86_32
@@ -68,30 +64,6 @@ static inline unsigned long array_index_mask_nospec(unsigned long index,
 #define __smp_wmb()	barrier()
 #define __smp_store_mb(var, value) do { (void)xchg(&var, value); } while (0)
 
-#if defined(CONFIG_X86_PPRO_FENCE)
-
-/*
- * For this option x86 doesn't have a strong TSO memory
- * model and we should fall back to full barriers.
- */
-
-#define __smp_store_release(p, v)					\
-do {									\
-	compiletime_assert_atomic_type(*p);				\
-	__smp_mb();							\
-	WRITE_ONCE(*p, v);						\
-} while (0)
-
-#define __smp_load_acquire(p)						\
-({									\
-	typeof(*p) ___p1 = READ_ONCE(*p);				\
-	compiletime_assert_atomic_type(*p);				\
-	__smp_mb();							\
-	___p1;								\
-})
-
-#else /* regular x86 TSO memory ordering */
-
 #define __smp_store_release(p, v)					\
 do {									\
 	compiletime_assert_atomic_type(*p);				\
@@ -107,8 +79,6 @@ do {									\
 	___p1;								\
 })
 
-#endif
-
 /* Atomic operations are already serializing on x86 */
 #define __smp_mb__before_atomic()	barrier()
 #define __smp_mb__after_atomic()	barrier()

arch/x86/include/asm/io.h

@@ -232,21 +232,6 @@ extern void set_iounmap_nonlazy(void);
  */
 #define __ISA_IO_base ((char __iomem *)(PAGE_OFFSET))
 
-/*
- *	Cache management
- *
- *	This needed for two cases
- *	1. Out of order aware processors
- *	2. Accidentally out of order processors (PPro errata #51)
- */
-
-static inline void flush_write_buffers(void)
-{
-#if defined(CONFIG_X86_PPRO_FENCE)
-	asm volatile("lock; addl $0,0(%%esp)": : :"memory");
-#endif
-}
-
 #endif /* __KERNEL__ */
 
 extern void native_io_delay(void);

arch/x86/include/asm/vmx.h

@@ -352,6 +352,7 @@ enum vmcs_field {
 #define INTR_TYPE_NMI_INTR		(2 << 8) /* NMI */
 #define INTR_TYPE_HARD_EXCEPTION	(3 << 8) /* processor exception */
 #define INTR_TYPE_SOFT_INTR             (4 << 8) /* software interrupt */
+#define INTR_TYPE_PRIV_SW_EXCEPTION	(5 << 8) /* ICE breakpoint - undocumented */
 #define INTR_TYPE_SOFT_EXCEPTION	(6 << 8) /* software exception */
 
 /* GUEST_INTERRUPTIBILITY_INFO flags. */

arch/x86/kernel/devicetree.c

@@ -259,7 +259,7 @@ static void __init dtb_apic_setup(void)
 	dtb_ioapic_setup();
 }
 
-#ifdef CONFIG_OF_FLATTREE
+#ifdef CONFIG_OF_EARLY_FLATTREE
 static void __init x86_flattree_get_config(void)
 {
 	u32 size, map_len;

arch/x86/kernel/idt.c

@@ -160,7 +160,6 @@ static const __initconst struct idt_data early_pf_idts[] = {
  */
 static const __initconst struct idt_data dbg_idts[] = {
 	INTG(X86_TRAP_DB,	debug),
-	INTG(X86_TRAP_BP,	int3),
 };
 #endif
 
@@ -183,7 +182,6 @@ gate_desc debug_idt_table[IDT_ENTRIES] __page_aligned_bss;
 static const __initconst struct idt_data ist_idts[] = {
 	ISTG(X86_TRAP_DB,	debug,		DEBUG_STACK),
 	ISTG(X86_TRAP_NMI,	nmi,		NMI_STACK),
-	SISTG(X86_TRAP_BP,	int3,		DEBUG_STACK),
 	ISTG(X86_TRAP_DF,	double_fault,	DOUBLEFAULT_STACK),
 #ifdef CONFIG_X86_MCE
 	ISTG(X86_TRAP_MC,	&machine_check,	MCE_STACK),

arch/x86/kernel/pci-nommu.c

@@ -37,7 +37,6 @@ static dma_addr_t nommu_map_page(struct device *dev, struct page *page,
 	WARN_ON(size == 0);
 	if (!check_addr("map_single", dev, bus, size))
 		return NOMMU_MAPPING_ERROR;
-	flush_write_buffers();
 	return bus;
 }
 
@@ -72,25 +71,9 @@ static int nommu_map_sg(struct device *hwdev, struct scatterlist *sg,
 			return 0;
 		s->dma_length = s->length;
 	}
-	flush_write_buffers();
 	return nents;
 }
 
-static void nommu_sync_single_for_device(struct device *dev,
-			dma_addr_t addr, size_t size,
-			enum dma_data_direction dir)
-{
-	flush_write_buffers();
-}
-
-
-static void nommu_sync_sg_for_device(struct device *dev,
-			struct scatterlist *sg, int nelems,
-			enum dma_data_direction dir)
-{
-	flush_write_buffers();
-}
-
 static int nommu_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
 	return dma_addr == NOMMU_MAPPING_ERROR;
@@ -101,8 +84,6 @@ const struct dma_map_ops nommu_dma_ops = {
 	.free			= dma_generic_free_coherent,
 	.map_sg			= nommu_map_sg,
 	.map_page		= nommu_map_page,
-	.sync_single_for_device = nommu_sync_single_for_device,
-	.sync_sg_for_device	= nommu_sync_sg_for_device,
 	.is_phys		= 1,
 	.mapping_error		= nommu_mapping_error,
 	.dma_supported		= x86_dma_supported,

arch/x86/kernel/traps.c

@@ -577,7 +577,6 @@ do_general_protection(struct pt_regs *regs, long error_code)
 }
 NOKPROBE_SYMBOL(do_general_protection);
 
-/* May run on IST stack. */
 dotraplinkage void notrace do_int3(struct pt_regs *regs, long error_code)
 {
 #ifdef CONFIG_DYNAMIC_FTRACE
@@ -592,6 +591,13 @@ dotraplinkage void notrace do_int3(struct pt_regs *regs, long error_code)
 	if (poke_int3_handler(regs))
 		return;
 
+	/*
+	 * Use ist_enter despite the fact that we don't use an IST stack.
+	 * We can be called from a kprobe in non-CONTEXT_KERNEL kernel
+	 * mode or even during context tracking state changes.
+	 *
+	 * This means that we can't schedule.  That's okay.
+	 */
 	ist_enter(regs);
 	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
 #ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
@@ -609,15 +615,10 @@ dotraplinkage void notrace do_int3(struct pt_regs *regs, long error_code)
 			SIGTRAP) == NOTIFY_STOP)
 		goto exit;
 
-	/*
-	 * Let others (NMI) know that the debug stack is in use
-	 * as we may switch to the interrupt stack.
-	 */
-	debug_stack_usage_inc();
 	cond_local_irq_enable(regs);
 	do_trap(X86_TRAP_BP, SIGTRAP, "int3", regs, error_code, NULL);
 	cond_local_irq_disable(regs);
-	debug_stack_usage_dec();
+
 exit:
 	ist_exit(regs);
 }

arch/x86/kvm/vmx.c

@@ -1045,6 +1045,13 @@ static inline bool is_machine_check(u32 intr_info)
 		(INTR_TYPE_HARD_EXCEPTION | MC_VECTOR | INTR_INFO_VALID_MASK);
 }
 
+/* Undocumented: icebp/int1 */
+static inline bool is_icebp(u32 intr_info)
+{
+	return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VALID_MASK))
+		== (INTR_TYPE_PRIV_SW_EXCEPTION | INTR_INFO_VALID_MASK);
+}
+
 static inline bool cpu_has_vmx_msr_bitmap(void)
 {
 	return vmcs_config.cpu_based_exec_ctrl & CPU_BASED_USE_MSR_BITMAPS;
@@ -6179,7 +6186,7 @@ static int handle_exception(struct kvm_vcpu *vcpu)
 		      (KVM_GUESTDBG_SINGLESTEP | KVM_GUESTDBG_USE_HW_BP))) {
 			vcpu->arch.dr6 &= ~15;
 			vcpu->arch.dr6 |= dr6 | DR6_RTM;
-			if (!(dr6 & ~DR6_RESERVED)) /* icebp */
+			if (is_icebp(intr_info))
 				skip_emulated_instruction(vcpu);
 
 			kvm_queue_exception(vcpu, DB_VECTOR);

arch/x86/mm/init_64.c

@@ -800,17 +800,11 @@ int arch_add_memory(int nid, u64 start, u64 size, struct vmem_altmap *altmap,
 
 #define PAGE_INUSE 0xFD
 
-static void __meminit free_pagetable(struct page *page, int order,
-		struct vmem_altmap *altmap)
+static void __meminit free_pagetable(struct page *page, int order)
 {
 	unsigned long magic;
 	unsigned int nr_pages = 1 << order;
 
-	if (altmap) {
-		vmem_altmap_free(altmap, nr_pages);
-		return;
-	}
-
 	/* bootmem page has reserved flag */
 	if (PageReserved(page)) {
 		__ClearPageReserved(page);
@@ -826,8 +820,16 @@ static void __meminit free_pagetable(struct page *page, int order,
 		free_pages((unsigned long)page_address(page), order);
 }
 
-static void __meminit free_pte_table(pte_t *pte_start, pmd_t *pmd,
+static void __meminit free_hugepage_table(struct page *page,
 		struct vmem_altmap *altmap)
+{
+	if (altmap)
+		vmem_altmap_free(altmap, PMD_SIZE / PAGE_SIZE);
+	else
+		free_pagetable(page, get_order(PMD_SIZE));
+}
+
+static void __meminit free_pte_table(pte_t *pte_start, pmd_t *pmd)
 {
 	pte_t *pte;
 	int i;
@@ -839,14 +841,13 @@ static void __meminit free_pte_table(pte_t *pte_start, pmd_t *pmd,
 	}
 
 	/* free a pte talbe */
-	free_pagetable(pmd_page(*pmd), 0, altmap);
+	free_pagetable(pmd_page(*pmd), 0);
 	spin_lock(&init_mm.page_table_lock);
 	pmd_clear(pmd);
 	spin_unlock(&init_mm.page_table_lock);
 }
 
-static void __meminit free_pmd_table(pmd_t *pmd_start, pud_t *pud,
-		struct vmem_altmap *altmap)
+static void __meminit free_pmd_table(pmd_t *pmd_start, pud_t *pud)
 {
 	pmd_t *pmd;
 	int i;
@@ -858,14 +859,13 @@ static void __meminit free_pmd_table(pmd_t *pmd_start, pud_t *pud,
 	}
 
 	/* free a pmd talbe */
-	free_pagetable(pud_page(*pud), 0, altmap);
+	free_pagetable(pud_page(*pud), 0);
 	spin_lock(&init_mm.page_table_lock);
 	pud_clear(pud);
 	spin_unlock(&init_mm.page_table_lock);
 }
 
-static void __meminit free_pud_table(pud_t *pud_start, p4d_t *p4d,
-		struct vmem_altmap *altmap)
+static void __meminit free_pud_table(pud_t *pud_start, p4d_t *p4d)
 {
 	pud_t *pud;
 	int i;
@@ -877,7 +877,7 @@ static void __meminit free_pud_table(pud_t *pud_start, p4d_t *p4d,
 	}
 
 	/* free a pud talbe */
-	free_pagetable(p4d_page(*p4d), 0, altmap);
+	free_pagetable(p4d_page(*p4d), 0);
 	spin_lock(&init_mm.page_table_lock);
 	p4d_clear(p4d);
 	spin_unlock(&init_mm.page_table_lock);
@@ -885,7 +885,7 @@ static void __meminit free_pud_table(pud_t *pud_start, p4d_t *p4d,
 
 static void __meminit
 remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
-		 struct vmem_altmap *altmap, bool direct)
+		 bool direct)
 {
 	unsigned long next, pages = 0;
 	pte_t *pte;
@@ -916,7 +916,7 @@ remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
 			 * freed when offlining, or simplely not in use.
 			 */
 			if (!direct)
-				free_pagetable(pte_page(*pte), 0, altmap);
+				free_pagetable(pte_page(*pte), 0);
 
 			spin_lock(&init_mm.page_table_lock);
 			pte_clear(&init_mm, addr, pte);
@@ -939,7 +939,7 @@ remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
 
 			page_addr = page_address(pte_page(*pte));
 			if (!memchr_inv(page_addr, PAGE_INUSE, PAGE_SIZE)) {
-				free_pagetable(pte_page(*pte), 0, altmap);
+				free_pagetable(pte_page(*pte), 0);
 
 				spin_lock(&init_mm.page_table_lock);
 				pte_clear(&init_mm, addr, pte);
@@ -974,9 +974,8 @@ remove_pmd_table(pmd_t *pmd_start, unsigned long addr, unsigned long end,
 			if (IS_ALIGNED(addr, PMD_SIZE) &&
 			    IS_ALIGNED(next, PMD_SIZE)) {
 				if (!direct)
-					free_pagetable(pmd_page(*pmd),
-						       get_order(PMD_SIZE),
-						       altmap);
+					free_hugepage_table(pmd_page(*pmd),
+							    altmap);
 
 				spin_lock(&init_mm.page_table_lock);
 				pmd_clear(pmd);
@@ -989,9 +988,8 @@ remove_pmd_table(pmd_t *pmd_start, unsigned long addr, unsigned long end,
 				page_addr = page_address(pmd_page(*pmd));
 				if (!memchr_inv(page_addr, PAGE_INUSE,
 						PMD_SIZE)) {
-					free_pagetable(pmd_page(*pmd),
-						       get_order(PMD_SIZE),
-						       altmap);
+					free_hugepage_table(pmd_page(*pmd),
+							    altmap);
 
 					spin_lock(&init_mm.page_table_lock);
 					pmd_clear(pmd);
@@ -1003,8 +1001,8 @@ remove_pmd_table(pmd_t *pmd_start, unsigned long addr, unsigned long end,
 		}
 
 		pte_base = (pte_t *)pmd_page_vaddr(*pmd);
-		remove_pte_table(pte_base, addr, next, altmap, direct);
-		free_pte_table(pte_base, pmd, altmap);
+		remove_pte_table(pte_base, addr, next, direct);
+		free_pte_table(pte_base, pmd);
 	}
 
 	/* Call free_pmd_table() in remove_pud_table(). */
@@ -1033,8 +1031,7 @@ remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
 			    IS_ALIGNED(next, PUD_SIZE)) {
 				if (!direct)
 					free_pagetable(pud_page(*pud),
-						       get_order(PUD_SIZE),
-						       altmap);
+						       get_order(PUD_SIZE));
 
 				spin_lock(&init_mm.page_table_lock);
 				pud_clear(pud);
@@ -1048,8 +1045,7 @@ remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
 				if (!memchr_inv(page_addr, PAGE_INUSE,
 						PUD_SIZE)) {
 					free_pagetable(pud_page(*pud),
-						       get_order(PUD_SIZE),
-						       altmap);
+						       get_order(PUD_SIZE));
 
 					spin_lock(&init_mm.page_table_lock);
 					pud_clear(pud);
@@ -1062,7 +1058,7 @@ remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
 
 		pmd_base = pmd_offset(pud, 0);
 		remove_pmd_table(pmd_base, addr, next, direct, altmap);
-		free_pmd_table(pmd_base, pud, altmap);
+		free_pmd_table(pmd_base, pud);
 	}
 
 	if (direct)
@@ -1094,7 +1090,7 @@ remove_p4d_table(p4d_t *p4d_start, unsigned long addr, unsigned long end,
 		 * to adapt for boot-time switching between 4 and 5 level page tables.
 		 */
 		if (CONFIG_PGTABLE_LEVELS == 5)
-			free_pud_table(pud_base, p4d, altmap);
+			free_pud_table(pud_base, p4d);
 	}
 
 	if (direct)

arch/x86/mm/pgtable.c

@@ -702,4 +702,52 @@ int pmd_clear_huge(pmd_t *pmd)
 
 	return 0;
 }
+
+/**
+ * pud_free_pmd_page - Clear pud entry and free pmd page.
+ * @pud: Pointer to a PUD.
+ *
+ * Context: The pud range has been unmaped and TLB purged.
+ * Return: 1 if clearing the entry succeeded. 0 otherwise.
+ */
+int pud_free_pmd_page(pud_t *pud)
+{
+	pmd_t *pmd;
+	int i;
+
+	if (pud_none(*pud))
+		return 1;
+
+	pmd = (pmd_t *)pud_page_vaddr(*pud);
+
+	for (i = 0; i < PTRS_PER_PMD; i++)
+		if (!pmd_free_pte_page(&pmd[i]))
+			return 0;
+
+	pud_clear(pud);
+	free_page((unsigned long)pmd);
+
+	return 1;
+}
+
+/**
+ * pmd_free_pte_page - Clear pmd entry and free pte page.
+ * @pmd: Pointer to a PMD.
+ *
+ * Context: The pmd range has been unmaped and TLB purged.
+ * Return: 1 if clearing the entry succeeded. 0 otherwise.
+ */
+int pmd_free_pte_page(pmd_t *pmd)
+{
+	pte_t *pte;
+
+	if (pmd_none(*pmd))
+		return 1;
+
+	pte = (pte_t *)pmd_page_vaddr(*pmd);
+	pmd_clear(pmd);
+	free_page((unsigned long)pte);
+
+	return 1;
+}
 #endif	/* CONFIG_HAVE_ARCH_HUGE_VMAP */

arch/x86/net/bpf_jit_comp.c

@@ -1188,7 +1188,7 @@ skip_init_addrs:
 	 * may converge on the last pass. In such case do one more
 	 * pass to emit the final image
 	 */
-	for (pass = 0; pass < 10 || image; pass++) {
+	for (pass = 0; pass < 20 || image; pass++) {
 		proglen = do_jit(prog, addrs, image, oldproglen, &ctx);
 		if (proglen <= 0) {
 			image = NULL;
@@ -1215,6 +1215,7 @@ skip_init_addrs:
 			}
 		}
 		oldproglen = proglen;
+		cond_resched();
 	}
 
 	if (bpf_jit_enable > 1)

arch/x86/platform/efi/efi_64.c

@@ -227,7 +227,7 @@ int __init efi_alloc_page_tables(void)
 	if (!pud) {
 		if (CONFIG_PGTABLE_LEVELS > 4)
 			free_page((unsigned long) pgd_page_vaddr(*pgd));
-		free_page((unsigned long)efi_pgd);
+		free_pages((unsigned long)efi_pgd, PGD_ALLOCATION_ORDER);
 		return -ENOMEM;
 	}
 

arch/x86/um/asm/barrier.h

@@ -30,11 +30,7 @@
 
 #endif /* CONFIG_X86_32 */
 
-#ifdef CONFIG_X86_PPRO_FENCE
-#define dma_rmb()	rmb()
-#else /* CONFIG_X86_PPRO_FENCE */
 #define dma_rmb()	barrier()
-#endif /* CONFIG_X86_PPRO_FENCE */
 #define dma_wmb()	barrier()
 
 #include <asm-generic/barrier.h>

drivers/acpi/acpi_watchdog.c

@@ -74,10 +74,10 @@ void __init acpi_watchdog_init(void)
 		res.start = gas->address;
 		if (gas->space_id == ACPI_ADR_SPACE_SYSTEM_MEMORY) {
 			res.flags = IORESOURCE_MEM;
-			res.end = res.start + ALIGN(gas->access_width, 4);
+			res.end = res.start + ALIGN(gas->access_width, 4) - 1;
 		} else if (gas->space_id == ACPI_ADR_SPACE_SYSTEM_IO) {
 			res.flags = IORESOURCE_IO;
-			res.end = res.start + gas->access_width;
+			res.end = res.start + gas->access_width - 1;
 		} else {
 			pr_warn("Unsupported address space: %u\n",
 				gas->space_id);

drivers/acpi/battery.c

@@ -70,7 +70,6 @@ static async_cookie_t async_cookie;
 static bool battery_driver_registered;
 static int battery_bix_broken_package;
 static int battery_notification_delay_ms;
-static int battery_full_discharging;
 static unsigned int cache_time = 1000;
 module_param(cache_time, uint, 0644);
 MODULE_PARM_DESC(cache_time, "cache time in milliseconds");
@@ -215,12 +214,9 @@ static int acpi_battery_get_property(struct power_supply *psy,
 		return -ENODEV;
 	switch (psp) {
 	case POWER_SUPPLY_PROP_STATUS:
-		if (battery->state & ACPI_BATTERY_STATE_DISCHARGING) {
-			if (battery_full_discharging && battery->rate_now == 0)
-				val->intval = POWER_SUPPLY_STATUS_FULL;
-			else
-				val->intval = POWER_SUPPLY_STATUS_DISCHARGING;
-		} else if (battery->state & ACPI_BATTERY_STATE_CHARGING)
+		if (battery->state & ACPI_BATTERY_STATE_DISCHARGING)
+			val->intval = POWER_SUPPLY_STATUS_DISCHARGING;
+		else if (battery->state & ACPI_BATTERY_STATE_CHARGING)
 			val->intval = POWER_SUPPLY_STATUS_CHARGING;
 		else if (acpi_battery_is_charged(battery))
 			val->intval = POWER_SUPPLY_STATUS_FULL;
@@ -1170,12 +1166,6 @@ battery_notification_delay_quirk(const struct dmi_system_id *d)
 	return 0;
 }
 
-static int __init battery_full_discharging_quirk(const struct dmi_system_id *d)
-{
-	battery_full_discharging = 1;
-	return 0;
-}
-
 static const struct dmi_system_id bat_dmi_table[] __initconst = {
 	{
 		.callback = battery_bix_broken_package_quirk,
@@ -1193,38 +1183,6 @@ static const struct dmi_system_id bat_dmi_table[] __initconst = {
 			DMI_MATCH(DMI_PRODUCT_NAME, "Aspire V5-573G"),
 		},
 	},
-	{
-		.callback = battery_full_discharging_quirk,
-		.ident = "ASUS GL502VSK",
-		.matches = {
-			DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."),
-			DMI_MATCH(DMI_PRODUCT_NAME, "GL502VSK"),
-		},
-	},
-	{
-		.callback = battery_full_discharging_quirk,
-		.ident = "ASUS UX305LA",
-		.matches = {
-			DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."),
-			DMI_MATCH(DMI_PRODUCT_NAME, "UX305LA"),
-		},
-	},
-	{
-		.callback = battery_full_discharging_quirk,
-		.ident = "ASUS UX360UA",
-		.matches = {
-			DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."),
-			DMI_MATCH(DMI_PRODUCT_NAME, "UX360UA"),
-		},
-	},
-	{
-		.callback = battery_full_discharging_quirk,
-		.ident = "ASUS UX410UAK",
-		.matches = {
-			DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."),
-			DMI_MATCH(DMI_PRODUCT_NAME, "UX410UAK"),
-		},
-	},
 	{},
 };
 

drivers/acpi/nfit/core.c

@@ -2675,10 +2675,14 @@ static int acpi_nfit_register_region(struct acpi_nfit_desc *acpi_desc,
 	else
 		ndr_desc->numa_node = NUMA_NO_NODE;
 
-	if(acpi_desc->platform_cap & ACPI_NFIT_CAPABILITY_CACHE_FLUSH)
+	/*
+	 * Persistence domain bits are hierarchical, if
+	 * ACPI_NFIT_CAPABILITY_CACHE_FLUSH is set then
+	 * ACPI_NFIT_CAPABILITY_MEM_FLUSH is implied.
+	 */
+	if (acpi_desc->platform_cap & ACPI_NFIT_CAPABILITY_CACHE_FLUSH)
 		set_bit(ND_REGION_PERSIST_CACHE, &ndr_desc->flags);
-
-	if (acpi_desc->platform_cap & ACPI_NFIT_CAPABILITY_MEM_FLUSH)
+	else if (acpi_desc->platform_cap & ACPI_NFIT_CAPABILITY_MEM_FLUSH)
 		set_bit(ND_REGION_PERSIST_MEMCTRL, &ndr_desc->flags);
 
 	list_for_each_entry(nfit_memdev, &acpi_desc->memdevs, list) {

drivers/acpi/numa.c

@@ -103,25 +103,27 @@ int acpi_map_pxm_to_node(int pxm)
  */
 int acpi_map_pxm_to_online_node(int pxm)
 {
-	int node, n, dist, min_dist;
+	int node, min_node;
 
 	node = acpi_map_pxm_to_node(pxm);
 
 	if (node == NUMA_NO_NODE)
 		node = 0;
 
+	min_node = node;
 	if (!node_online(node)) {
-		min_dist = INT_MAX;
+		int min_dist = INT_MAX, dist, n;
+
 		for_each_online_node(n) {
 			dist = node_distance(node, n);
 			if (dist < min_dist) {
 				min_dist = dist;
-				node = n;
+				min_node = n;
 			}
 		}
 	}
 
-	return node;
+	return min_node;
 }
 EXPORT_SYMBOL(acpi_map_pxm_to_online_node);
 

drivers/ata/ahci.c

@@ -550,7 +550,9 @@ static const struct pci_device_id ahci_pci_tbl[] = {
 	  .driver_data = board_ahci_yes_fbs },
 	{ PCI_DEVICE(PCI_VENDOR_ID_MARVELL_EXT, 0x9230),
 	  .driver_data = board_ahci_yes_fbs },
-	{ PCI_DEVICE(PCI_VENDOR_ID_TTI, 0x0642),
+	{ PCI_DEVICE(PCI_VENDOR_ID_TTI, 0x0642), /* highpoint rocketraid 642L */
+	  .driver_data = board_ahci_yes_fbs },
+	{ PCI_DEVICE(PCI_VENDOR_ID_TTI, 0x0645), /* highpoint rocketraid 644L */
 	  .driver_data = board_ahci_yes_fbs },
 
 	/* Promise */

drivers/ata/libahci.c

@@ -665,6 +665,16 @@ int ahci_stop_engine(struct ata_port *ap)
 	if ((tmp & (PORT_CMD_START | PORT_CMD_LIST_ON)) == 0)
 		return 0;
 
+	/*
+	 * Don't try to issue commands but return with ENODEV if the
+	 * AHCI controller not available anymore (e.g. due to PCIe hot
+	 * unplugging). Otherwise a 500ms delay for each port is added.
+	 */
+	if (tmp == 0xffffffff) {
+		dev_err(ap->host->dev, "AHCI controller unavailable!\n");
+		return -ENODEV;
+	}
+
 	/* setting HBA to idle */
 	tmp &= ~PORT_CMD_START;
 	writel(tmp, port_mmio + PORT_CMD);

drivers/ata/libahci_platform.c

@@ -340,7 +340,7 @@ static int ahci_platform_get_regulator(struct ahci_host_priv *hpriv, u32 port,
  * 2) regulator for controlling the targets power (optional)
  * 3) 0 - AHCI_MAX_CLKS clocks, as specified in the devs devicetree node,
  *    or for non devicetree enabled platforms a single clock
- *	4) phys (optional)
+ * 4) phys (optional)
  *
  * RETURNS:
  * The allocated ahci_host_priv on success, otherwise an ERR_PTR value

drivers/ata/libata-core.c

@@ -4530,6 +4530,25 @@ static const struct ata_blacklist_entry ata_device_blacklist [] = {
 	{ "PIONEER DVD-RW  DVR-212D",	NULL,	ATA_HORKAGE_NOSETXFER },
 	{ "PIONEER DVD-RW  DVR-216D",	NULL,	ATA_HORKAGE_NOSETXFER },
 
+	/* Crucial BX100 SSD 500GB has broken LPM support */
+	{ "CT500BX100SSD1",		NULL,	ATA_HORKAGE_NOLPM },
+
+	/* 512GB MX100 with MU01 firmware has both queued TRIM and LPM issues */
+	{ "Crucial_CT512MX100*",	"MU01",	ATA_HORKAGE_NO_NCQ_TRIM |
+						ATA_HORKAGE_ZERO_AFTER_TRIM |
+						ATA_HORKAGE_NOLPM, },
+	/* 512GB MX100 with newer firmware has only LPM issues */
+	{ "Crucial_CT512MX100*",	NULL,	ATA_HORKAGE_ZERO_AFTER_TRIM |
+						ATA_HORKAGE_NOLPM, },
+
+	/* 480GB+ M500 SSDs have both queued TRIM and LPM issues */
+	{ "Crucial_CT480M500*",		NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
+						ATA_HORKAGE_ZERO_AFTER_TRIM |
+						ATA_HORKAGE_NOLPM, },
+	{ "Crucial_CT960M500*",		NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
+						ATA_HORKAGE_ZERO_AFTER_TRIM |
+						ATA_HORKAGE_NOLPM, },
+
 	/* devices that don't properly handle queued TRIM commands */
 	{ "Micron_M500_*",		NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
 						ATA_HORKAGE_ZERO_AFTER_TRIM, },
@@ -4541,7 +4560,9 @@ static const struct ata_blacklist_entry ata_device_blacklist [] = {
 						ATA_HORKAGE_ZERO_AFTER_TRIM, },
 	{ "Crucial_CT*MX100*",		"MU01",	ATA_HORKAGE_NO_NCQ_TRIM |
 						ATA_HORKAGE_ZERO_AFTER_TRIM, },
-	{ "Samsung SSD 8*",		NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
+	{ "Samsung SSD 840*",		NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
+						ATA_HORKAGE_ZERO_AFTER_TRIM, },
+	{ "Samsung SSD 850*",		NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
 						ATA_HORKAGE_ZERO_AFTER_TRIM, },
 	{ "FCCT*M500*",			NULL,	ATA_HORKAGE_NO_NCQ_TRIM |
 						ATA_HORKAGE_ZERO_AFTER_TRIM, },
@@ -5401,8 +5422,7 @@ void ata_qc_issue(struct ata_queued_cmd *qc)
 	 * We guarantee to LLDs that they will have at least one
 	 * non-zero sg if the command is a data command.
 	 */
-	if (WARN_ON_ONCE(ata_is_data(prot) &&
-			 (!qc->sg || !qc->n_elem || !qc->nbytes)))
+	if (ata_is_data(prot) && (!qc->sg || !qc->n_elem || !qc->nbytes))
 		goto sys_err;
 
 	if (ata_is_dma(prot) || (ata_is_pio(prot) &&

drivers/ata/libata-eh.c

@@ -815,7 +815,8 @@ void ata_scsi_port_error_handler(struct Scsi_Host *host, struct ata_port *ap)
 
 	if (ap->pflags & ATA_PFLAG_LOADING)
 		ap->pflags &= ~ATA_PFLAG_LOADING;
-	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
+	else if ((ap->pflags & ATA_PFLAG_SCSI_HOTPLUG) &&
+		!(ap->flags & ATA_FLAG_SAS_HOST))
 		schedule_delayed_work(&ap->hotplug_task, 0);
 
 	if (ap->pflags & ATA_PFLAG_RECOVERED)

drivers/ata/libata-scsi.c

@@ -3316,6 +3316,12 @@ static unsigned int ata_scsi_pass_thru(struct ata_queued_cmd *qc)
 		goto invalid_fld;
 	}
 
+	/* We may not issue NCQ commands to devices not supporting NCQ */
+	if (ata_is_ncq(tf->protocol) && !ata_ncq_enabled(dev)) {
+		fp = 1;
+		goto invalid_fld;
+	}
+
 	/* sanity check for pio multi commands */
 	if ((cdb[1] & 0xe0) && !is_multi_taskfile(tf)) {
 		fp = 1;
@@ -4282,7 +4288,7 @@ static inline void ata_scsi_dump_cdb(struct ata_port *ap,
 #ifdef ATA_DEBUG
 	struct scsi_device *scsidev = cmd->device;
 
-	DPRINTK("CDB (%u:%d,%d,%d) %9ph\n",
+	DPRINTK("CDB (%u:%d,%d,%lld) %9ph\n",
 		ap->print_id,
 		scsidev->channel, scsidev->id, scsidev->lun,
 		cmd->cmnd);
@@ -4309,7 +4315,9 @@ static inline int __ata_scsi_queuecmd(struct scsi_cmnd *scmd,
 		if (likely((scsi_op != ATA_16) || !atapi_passthru16)) {
 			/* relay SCSI command to ATAPI device */
 			int len = COMMAND_SIZE(scsi_op);
-			if (unlikely(len > scmd->cmd_len || len > dev->cdb_len))
+			if (unlikely(len > scmd->cmd_len ||
+				     len > dev->cdb_len ||
+				     scmd->cmd_len > ATAPI_CDB_LEN))
 				goto bad_cdb_len;
 
 			xlat_func = atapi_xlat;

drivers/ata/sata_rcar.c

@@ -146,6 +146,7 @@
 enum sata_rcar_type {
 	RCAR_GEN1_SATA,
 	RCAR_GEN2_SATA,
+	RCAR_GEN3_SATA,
 	RCAR_R8A7790_ES1_SATA,
 };
 
@@ -784,26 +785,11 @@ static void sata_rcar_setup_port(struct ata_host *host)
 	ioaddr->command_addr	= ioaddr->cmd_addr + (ATA_REG_CMD << 2);
 }
 
-static void sata_rcar_init_controller(struct ata_host *host)
+static void sata_rcar_init_module(struct sata_rcar_priv *priv)
 {
-	struct sata_rcar_priv *priv = host->private_data;
 	void __iomem *base = priv->base;
 	u32 val;
 
-	/* reset and setup phy */
-	switch (priv->type) {
-	case RCAR_GEN1_SATA:
-		sata_rcar_gen1_phy_init(priv);
-		break;
-	case RCAR_GEN2_SATA:
-	case RCAR_R8A7790_ES1_SATA:
-		sata_rcar_gen2_phy_init(priv);
-		break;
-	default:
-		dev_warn(host->dev, "SATA phy is not initialized\n");
-		break;
-	}
-
 	/* SATA-IP reset state */
 	val = ioread32(base + ATAPI_CONTROL1_REG);
 	val |= ATAPI_CONTROL1_RESET;
@@ -824,10 +810,33 @@ static void sata_rcar_init_controller(struct ata_host *host)
 	/* ack and mask */
 	iowrite32(0, base + SATAINTSTAT_REG);
 	iowrite32(0x7ff, base + SATAINTMASK_REG);
+
 	/* enable interrupts */
 	iowrite32(ATAPI_INT_ENABLE_SATAINT, base + ATAPI_INT_ENABLE_REG);
 }
 
+static void sata_rcar_init_controller(struct ata_host *host)
+{
+	struct sata_rcar_priv *priv = host->private_data;
+
+	/* reset and setup phy */
+	switch (priv->type) {
+	case RCAR_GEN1_SATA:
+		sata_rcar_gen1_phy_init(priv);
+		break;
+	case RCAR_GEN2_SATA:
+	case RCAR_GEN3_SATA:
+	case RCAR_R8A7790_ES1_SATA:
+		sata_rcar_gen2_phy_init(priv);
+		break;
+	default:
+		dev_warn(host->dev, "SATA phy is not initialized\n");
+		break;
+	}
+
+	sata_rcar_init_module(priv);
+}
+
 static const struct of_device_id sata_rcar_match[] = {
 	{
 		/* Deprecated by "renesas,sata-r8a7779" */
@@ -856,7 +865,7 @@ static const struct of_device_id sata_rcar_match[] = {
 	},
 	{
 		.compatible = "renesas,sata-r8a7795",
-		.data = (void *)RCAR_GEN2_SATA
+		.data = (void *)RCAR_GEN3_SATA
 	},
 	{
 		.compatible = "renesas,rcar-gen2-sata",
@@ -864,7 +873,7 @@ static const struct of_device_id sata_rcar_match[] = {
 	},
 	{
 		.compatible = "renesas,rcar-gen3-sata",
-		.data = (void *)RCAR_GEN2_SATA
+		.data = (void *)RCAR_GEN3_SATA
 	},
 	{ },
 };
@@ -982,11 +991,18 @@ static int sata_rcar_resume(struct device *dev)
 	if (ret)
 		return ret;
 
-	/* ack and mask */
-	iowrite32(0, base + SATAINTSTAT_REG);
-	iowrite32(0x7ff, base + SATAINTMASK_REG);
-	/* enable interrupts */
-	iowrite32(ATAPI_INT_ENABLE_SATAINT, base + ATAPI_INT_ENABLE_REG);
+	if (priv->type == RCAR_GEN3_SATA) {
+		sata_rcar_gen2_phy_init(priv);
+		sata_rcar_init_module(priv);
+	} else {
+		/* ack and mask */
+		iowrite32(0, base + SATAINTSTAT_REG);
+		iowrite32(0x7ff, base + SATAINTMASK_REG);
+
+		/* enable interrupts */
+		iowrite32(ATAPI_INT_ENABLE_SATAINT,
+			  base + ATAPI_INT_ENABLE_REG);
+	}
 
 	ata_host_resume(host);
 

drivers/bluetooth/btusb.c

@@ -231,7 +231,6 @@ static const struct usb_device_id blacklist_table[] = {
 	{ USB_DEVICE(0x0930, 0x0227), .driver_info = BTUSB_ATH3012 },
 	{ USB_DEVICE(0x0b05, 0x17d0), .driver_info = BTUSB_ATH3012 },
 	{ USB_DEVICE(0x0cf3, 0x0036), .driver_info = BTUSB_ATH3012 },
-	{ USB_DEVICE(0x0cf3, 0x3004), .driver_info = BTUSB_ATH3012 },
 	{ USB_DEVICE(0x0cf3, 0x3008), .driver_info = BTUSB_ATH3012 },
 	{ USB_DEVICE(0x0cf3, 0x311d), .driver_info = BTUSB_ATH3012 },
 	{ USB_DEVICE(0x0cf3, 0x311e), .driver_info = BTUSB_ATH3012 },
@@ -264,6 +263,7 @@ static const struct usb_device_id blacklist_table[] = {
 	{ USB_DEVICE(0x0489, 0xe03c), .driver_info = BTUSB_ATH3012 },
 
 	/* QCA ROME chipset */
+	{ USB_DEVICE(0x0cf3, 0x3004), .driver_info = BTUSB_QCA_ROME },
 	{ USB_DEVICE(0x0cf3, 0xe007), .driver_info = BTUSB_QCA_ROME },
 	{ USB_DEVICE(0x0cf3, 0xe009), .driver_info = BTUSB_QCA_ROME },
 	{ USB_DEVICE(0x0cf3, 0xe010), .driver_info = BTUSB_QCA_ROME },
@@ -386,10 +386,10 @@ static const struct usb_device_id blacklist_table[] = {
  */
 static const struct dmi_system_id btusb_needs_reset_resume_table[] = {
 	{
-		/* Lenovo Yoga 920 (QCA Rome device 0cf3:e300) */
+		/* Dell OptiPlex 3060 (QCA ROME device 0cf3:e007) */
 		.matches = {
-			DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
-			DMI_MATCH(DMI_PRODUCT_VERSION, "Lenovo YOGA 920"),
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex 3060"),
 		},
 	},
 	{}

drivers/bluetooth/hci_bcm.c

@@ -244,7 +244,9 @@ static irqreturn_t bcm_host_wake(int irq, void *data)
 
 	bt_dev_dbg(bdev, "Host wake IRQ");
 
-	pm_request_resume(bdev->dev);
+	pm_runtime_get(bdev->dev);
+	pm_runtime_mark_last_busy(bdev->dev);
+	pm_runtime_put_autosuspend(bdev->dev);
 
 	return IRQ_HANDLED;
 }
@@ -301,7 +303,7 @@ static const struct bcm_set_sleep_mode default_sleep_params = {
 	.usb_auto_sleep = 0,
 	.usb_resume_timeout = 0,
 	.break_to_host = 0,
-	.pulsed_host_wake = 0,
+	.pulsed_host_wake = 1,
 };
 
 static int bcm_setup_sleep(struct hci_uart *hu)
@@ -586,8 +588,11 @@ static int bcm_recv(struct hci_uart *hu, const void *data, int count)
 	} else if (!bcm->rx_skb) {
 		/* Delay auto-suspend when receiving completed packet */
 		mutex_lock(&bcm_device_lock);
-		if (bcm->dev && bcm_device_exists(bcm->dev))
-			pm_request_resume(bcm->dev->dev);
+		if (bcm->dev && bcm_device_exists(bcm->dev)) {
+			pm_runtime_get(bcm->dev->dev);
+			pm_runtime_mark_last_busy(bcm->dev->dev);
+			pm_runtime_put_autosuspend(bcm->dev->dev);
+		}
 		mutex_unlock(&bcm_device_lock);
 	}
 

drivers/clk/bcm/clk-bcm2835.c

@@ -449,17 +449,17 @@ struct bcm2835_pll_ana_bits {
 static const struct bcm2835_pll_ana_bits bcm2835_ana_default = {
 	.mask0 = 0,
 	.set0 = 0,
-	.mask1 = (u32)~(A2W_PLL_KI_MASK | A2W_PLL_KP_MASK),
+	.mask1 = A2W_PLL_KI_MASK | A2W_PLL_KP_MASK,
 	.set1 = (2 << A2W_PLL_KI_SHIFT) | (8 << A2W_PLL_KP_SHIFT),
-	.mask3 = (u32)~A2W_PLL_KA_MASK,
+	.mask3 = A2W_PLL_KA_MASK,
 	.set3 = (2 << A2W_PLL_KA_SHIFT),
 	.fb_prediv_mask = BIT(14),
 };
 
 static const struct bcm2835_pll_ana_bits bcm2835_ana_pllh = {
-	.mask0 = (u32)~(A2W_PLLH_KA_MASK | A2W_PLLH_KI_LOW_MASK),
+	.mask0 = A2W_PLLH_KA_MASK | A2W_PLLH_KI_LOW_MASK,
 	.set0 = (2 << A2W_PLLH_KA_SHIFT) | (2 << A2W_PLLH_KI_LOW_SHIFT),
-	.mask1 = (u32)~(A2W_PLLH_KI_HIGH_MASK | A2W_PLLH_KP_MASK),
+	.mask1 = A2W_PLLH_KI_HIGH_MASK | A2W_PLLH_KP_MASK,
 	.set1 = (6 << A2W_PLLH_KP_SHIFT),
 	.mask3 = 0,
 	.set3 = 0,
@@ -623,8 +623,10 @@ static int bcm2835_pll_on(struct clk_hw *hw)
 		     ~A2W_PLL_CTRL_PWRDN);
 
 	/* Take the PLL out of reset. */
+	spin_lock(&cprman->regs_lock);
 	cprman_write(cprman, data->cm_ctrl_reg,
 		     cprman_read(cprman, data->cm_ctrl_reg) & ~CM_PLL_ANARST);
+	spin_unlock(&cprman->regs_lock);
 
 	/* Wait for the PLL to lock. */
 	timeout = ktime_add_ns(ktime_get(), LOCK_TIMEOUT_NS);
@@ -701,9 +703,11 @@ static int bcm2835_pll_set_rate(struct clk_hw *hw,
 	}
 
 	/* Unmask the reference clock from the oscillator. */
+	spin_lock(&cprman->regs_lock);
 	cprman_write(cprman, A2W_XOSC_CTRL,
 		     cprman_read(cprman, A2W_XOSC_CTRL) |
 		     data->reference_enable_mask);
+	spin_unlock(&cprman->regs_lock);
 
 	if (do_ana_setup_first)
 		bcm2835_pll_write_ana(cprman, data->ana_reg_base, ana);

drivers/clk/clk-aspeed.c

@@ -205,6 +205,18 @@ static const struct aspeed_clk_soc_data ast2400_data = {
 	.calc_pll = aspeed_ast2400_calc_pll,
 };
 
+static int aspeed_clk_is_enabled(struct clk_hw *hw)
+{
+	struct aspeed_clk_gate *gate = to_aspeed_clk_gate(hw);
+	u32 clk = BIT(gate->clock_idx);
+	u32 enval = (gate->flags & CLK_GATE_SET_TO_DISABLE) ? 0 : clk;
+	u32 reg;
+
+	regmap_read(gate->map, ASPEED_CLK_STOP_CTRL, &reg);
+
+	return ((reg & clk) == enval) ? 1 : 0;
+}
+
 static int aspeed_clk_enable(struct clk_hw *hw)
 {
 	struct aspeed_clk_gate *gate = to_aspeed_clk_gate(hw);
@@ -215,6 +227,11 @@ static int aspeed_clk_enable(struct clk_hw *hw)
 
 	spin_lock_irqsave(gate->lock, flags);
 
+	if (aspeed_clk_is_enabled(hw)) {
+		spin_unlock_irqrestore(gate->lock, flags);
+		return 0;
+	}
+
 	if (gate->reset_idx >= 0) {
 		/* Put IP in reset */
 		regmap_update_bits(gate->map, ASPEED_RESET_CTRL, rst, rst);
@@ -255,17 +272,6 @@ static void aspeed_clk_disable(struct clk_hw *hw)
 	spin_unlock_irqrestore(gate->lock, flags);
 }
 
-static int aspeed_clk_is_enabled(struct clk_hw *hw)
-{
-	struct aspeed_clk_gate *gate = to_aspeed_clk_gate(hw);
-	u32 clk = BIT(gate->clock_idx);
-	u32 reg;
-
-	regmap_read(gate->map, ASPEED_CLK_STOP_CTRL, &reg);
-
-	return (reg & clk) ? 0 : 1;
-}
-
 static const struct clk_ops aspeed_clk_gate_ops = {
 	.enable = aspeed_clk_enable,
 	.disable = aspeed_clk_disable,

drivers/clk/clk.c

@@ -1125,8 +1125,10 @@ static int clk_core_round_rate_nolock(struct clk_core *core,
 {
 	lockdep_assert_held(&prepare_lock);
 
-	if (!core)
+	if (!core) {
+		req->rate = 0;
 		return 0;
+	}
 
 	clk_core_init_rate_req(core, req);
 
@@ -2309,8 +2311,11 @@ static int clk_core_set_phase_nolock(struct clk_core *core, int degrees)
 
 	trace_clk_set_phase(core, degrees);
 
-	if (core->ops->set_phase)
+	if (core->ops->set_phase) {
 		ret = core->ops->set_phase(core->hw, degrees);
+		if (!ret)
+			core->phase = degrees;
+	}
 
 	trace_clk_set_phase_complete(core, degrees);
 
@@ -2967,23 +2972,38 @@ static int __clk_core_init(struct clk_core *core)
 		rate = 0;
 	core->rate = core->req_rate = rate;
 
+	/*
+	 * Enable CLK_IS_CRITICAL clocks so newly added critical clocks
+	 * don't get accidentally disabled when walking the orphan tree and
+	 * reparenting clocks
+	 */
+	if (core->flags & CLK_IS_CRITICAL) {
+		unsigned long flags;
+
+		clk_core_prepare(core);
+
+		flags = clk_enable_lock();
+		clk_core_enable(core);
+		clk_enable_unlock(flags);
+	}
+
 	/*
 	 * walk the list of orphan clocks and reparent any that newly finds a
 	 * parent.
 	 */
 	hlist_for_each_entry_safe(orphan, tmp2, &clk_orphan_list, child_node) {
 		struct clk_core *parent = __clk_init_parent(orphan);
-		unsigned long flags;
 
 		/*
-		 * we could call __clk_set_parent, but that would result in a
-		 * redundant call to the .set_rate op, if it exists
+		 * We need to use __clk_set_parent_before() and _after() to
+		 * to properly migrate any prepare/enable count of the orphan
+		 * clock. This is important for CLK_IS_CRITICAL clocks, which
+		 * are enabled during init but might not have a parent yet.
 		 */
 		if (parent) {
 			/* update the clk tree topology */
-			flags = clk_enable_lock();
-			clk_reparent(orphan, parent);
-			clk_enable_unlock(flags);
+			__clk_set_parent_before(orphan, parent);
+			__clk_set_parent_after(orphan, parent, NULL);
 			__clk_recalc_accuracies(orphan);
 			__clk_recalc_rates(orphan, 0);
 		}
@@ -3000,16 +3020,6 @@ static int __clk_core_init(struct clk_core *core)
 	if (core->ops->init)
 		core->ops->init(core->hw);
 
-	if (core->flags & CLK_IS_CRITICAL) {
-		unsigned long flags;
-
-		clk_core_prepare(core);
-
-		flags = clk_enable_lock();
-		clk_core_enable(core);
-		clk_enable_unlock(flags);
-	}
-
 	kref_init(&core->ref);
 out:
 	clk_pm_runtime_put(core);

drivers/clk/hisilicon/clk-hi3660-stub.c

@@ -149,6 +149,8 @@ static int hi3660_stub_clk_probe(struct platform_device *pdev)
 		return PTR_ERR(stub_clk_chan.mbox);
 
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res)
+		return -EINVAL;
 	freq_reg = devm_ioremap(dev, res->start, resource_size(res));
 	if (!freq_reg)
 		return -ENOMEM;

drivers/clk/imx/clk-imx51-imx53.c

@@ -131,7 +131,17 @@ static const char *ieee1588_sels[] = { "pll3_sw", "pll4_sw", "dummy" /* usbphy2_
 static struct clk *clk[IMX5_CLK_END];
 static struct clk_onecell_data clk_data;
 
-static struct clk ** const uart_clks[] __initconst = {
+static struct clk ** const uart_clks_mx51[] __initconst = {
+	&clk[IMX5_CLK_UART1_IPG_GATE],
+	&clk[IMX5_CLK_UART1_PER_GATE],
+	&clk[IMX5_CLK_UART2_IPG_GATE],
+	&clk[IMX5_CLK_UART2_PER_GATE],
+	&clk[IMX5_CLK_UART3_IPG_GATE],
+	&clk[IMX5_CLK_UART3_PER_GATE],
+	NULL
+};
+
+static struct clk ** const uart_clks_mx50_mx53[] __initconst = {
 	&clk[IMX5_CLK_UART1_IPG_GATE],
 	&clk[IMX5_CLK_UART1_PER_GATE],
 	&clk[IMX5_CLK_UART2_IPG_GATE],
@@ -321,8 +331,6 @@ static void __init mx5_clocks_common_init(void __iomem *ccm_base)
 	clk_prepare_enable(clk[IMX5_CLK_TMAX1]);
 	clk_prepare_enable(clk[IMX5_CLK_TMAX2]); /* esdhc2, fec */
 	clk_prepare_enable(clk[IMX5_CLK_TMAX3]); /* esdhc1, esdhc4 */
-
-	imx_register_uart_clocks(uart_clks);
 }
 
 static void __init mx50_clocks_init(struct device_node *np)
@@ -388,6 +396,8 @@ static void __init mx50_clocks_init(struct device_node *np)
 
 	r = clk_round_rate(clk[IMX5_CLK_USBOH3_PER_GATE], 54000000);
 	clk_set_rate(clk[IMX5_CLK_USBOH3_PER_GATE], r);
+
+	imx_register_uart_clocks(uart_clks_mx50_mx53);
 }
 CLK_OF_DECLARE(imx50_ccm, "fsl,imx50-ccm", mx50_clocks_init);
 
@@ -477,6 +487,8 @@ static void __init mx51_clocks_init(struct device_node *np)
 	val = readl(MXC_CCM_CLPCR);
 	val |= 1 << 23;
 	writel(val, MXC_CCM_CLPCR);
+
+	imx_register_uart_clocks(uart_clks_mx51);
 }
 CLK_OF_DECLARE(imx51_ccm, "fsl,imx51-ccm", mx51_clocks_init);
 
@@ -606,5 +618,7 @@ static void __init mx53_clocks_init(struct device_node *np)
 
 	r = clk_round_rate(clk[IMX5_CLK_USBOH3_PER_GATE], 54000000);
 	clk_set_rate(clk[IMX5_CLK_USBOH3_PER_GATE], r);
+
+	imx_register_uart_clocks(uart_clks_mx50_mx53);
 }
 CLK_OF_DECLARE(imx53_ccm, "fsl,imx53-ccm", mx53_clocks_init);

drivers/clk/qcom/apcs-msm8916.c

@@ -49,11 +49,10 @@ static int qcom_apcs_msm8916_clk_probe(struct platform_device *pdev)
 	struct clk_regmap_mux_div *a53cc;
 	struct regmap *regmap;
 	struct clk_init_data init = { };
-	int ret;
+	int ret = -ENODEV;
 
 	regmap = dev_get_regmap(parent, NULL);
-	if (IS_ERR(regmap)) {
-		ret = PTR_ERR(regmap);
+	if (!regmap) {
 		dev_err(dev, "failed to get regmap: %d\n", ret);
 		return ret;
 	}

drivers/clk/sunxi-ng/ccu-sun6i-a31.c

@@ -762,7 +762,7 @@ static struct ccu_mp out_a_clk = {
 		.features	= CCU_FEATURE_FIXED_PREDIV,
 		.hw.init	= CLK_HW_INIT_PARENTS("out-a",
 						      clk_out_parents,
-						      &ccu_div_ops,
+						      &ccu_mp_ops,
 						      0),
 	},
 };
@@ -783,7 +783,7 @@ static struct ccu_mp out_b_clk = {
 		.features	= CCU_FEATURE_FIXED_PREDIV,
 		.hw.init	= CLK_HW_INIT_PARENTS("out-b",
 						      clk_out_parents,
-						      &ccu_div_ops,
+						      &ccu_mp_ops,
 						      0),
 	},
 };
@@ -804,7 +804,7 @@ static struct ccu_mp out_c_clk = {
 		.features	= CCU_FEATURE_FIXED_PREDIV,
 		.hw.init	= CLK_HW_INIT_PARENTS("out-c",
 						      clk_out_parents,
-						      &ccu_div_ops,
+						      &ccu_mp_ops,
 						      0),
 	},
 };

drivers/clk/ti/clk-33xx.c

@@ -45,7 +45,7 @@ static const struct omap_clkctrl_bit_data am3_gpio4_bit_data[] __initconst = {
 
 static const struct omap_clkctrl_reg_data am3_l4_per_clkctrl_regs[] __initconst = {
 	{ AM3_CPGMAC0_CLKCTRL, NULL, CLKF_SW_SUP, "cpsw_125mhz_gclk", "cpsw_125mhz_clkdm" },
-	{ AM3_LCDC_CLKCTRL, NULL, CLKF_SW_SUP, "lcd_gclk", "lcdc_clkdm" },
+	{ AM3_LCDC_CLKCTRL, NULL, CLKF_SW_SUP | CLKF_SET_RATE_PARENT, "lcd_gclk", "lcdc_clkdm" },
 	{ AM3_USB_OTG_HS_CLKCTRL, NULL, CLKF_SW_SUP, "usbotg_fck", "l3s_clkdm" },
 	{ AM3_TPTC0_CLKCTRL, NULL, CLKF_SW_SUP, "l3_gclk", "l3_clkdm" },
 	{ AM3_EMIF_CLKCTRL, NULL, CLKF_SW_SUP, "dpll_ddr_m2_div2_ck", "l3_clkdm" },

drivers/clk/ti/clk-43xx.c

@@ -187,7 +187,7 @@ static const struct omap_clkctrl_reg_data am4_l4_per_clkctrl_regs[] __initconst
 	{ AM4_OCP2SCP0_CLKCTRL, NULL, CLKF_SW_SUP, "l4ls_gclk" },
 	{ AM4_OCP2SCP1_CLKCTRL, NULL, CLKF_SW_SUP, "l4ls_gclk" },
 	{ AM4_EMIF_CLKCTRL, NULL, CLKF_SW_SUP, "dpll_ddr_m2_ck", "emif_clkdm" },
-	{ AM4_DSS_CORE_CLKCTRL, NULL, CLKF_SW_SUP, "disp_clk", "dss_clkdm" },
+	{ AM4_DSS_CORE_CLKCTRL, NULL, CLKF_SW_SUP | CLKF_SET_RATE_PARENT, "disp_clk", "dss_clkdm" },
 	{ AM4_CPGMAC0_CLKCTRL, NULL, CLKF_SW_SUP, "cpsw_125mhz_gclk", "cpsw_125mhz_clkdm" },
 	{ 0 },
 };

drivers/clk/ti/clkctrl.c

@@ -537,6 +537,8 @@ static void __init _ti_omap4_clkctrl_setup(struct device_node *node)
 		init.parent_names = &reg_data->parent;
 		init.num_parents = 1;
 		init.flags = 0;
+		if (reg_data->flags & CLKF_SET_RATE_PARENT)
+			init.flags |= CLK_SET_RATE_PARENT;
 		init.name = kasprintf(GFP_KERNEL, "%s:%s:%04x:%d",
 				      node->parent->name, node->name,
 				      reg_data->offset, 0);

drivers/dma-buf/dma-fence.c

@@ -171,6 +171,7 @@ void dma_fence_release(struct kref *kref)
 
 	trace_dma_fence_destroy(fence);
 
+	/* Failed to signal before release, could be a refcounting issue */
 	WARN_ON(!list_empty(&fence->cb_list));
 
 	if (fence->ops->release)

drivers/dma-buf/reservation.c

@@ -374,8 +374,9 @@ EXPORT_SYMBOL(reservation_object_copy_fences);
  * @pshared: the array of shared fence ptrs returned (array is krealloc'd to
  * the required size, and must be freed by caller)
  *
- * RETURNS
- * Zero or -errno
+ * Retrieve all fences from the reservation object. If the pointer for the
+ * exclusive fence is not specified the fence is put into the array of the
+ * shared fences as well. Returns either zero or -ENOMEM.
  */
 int reservation_object_get_fences_rcu(struct reservation_object *obj,
 				      struct dma_fence **pfence_excl,
@@ -389,8 +390,8 @@ int reservation_object_get_fences_rcu(struct reservation_object *obj,
 
 	do {
 		struct reservation_object_list *fobj;
-		unsigned seq;
-		unsigned int i;
+		unsigned int i, seq;
+		size_t sz = 0;
 
 		shared_count = i = 0;
 
@@ -402,9 +403,14 @@ int reservation_object_get_fences_rcu(struct reservation_object *obj,
 			goto unlock;
 
 		fobj = rcu_dereference(obj->fence);
-		if (fobj) {
+		if (fobj)
+			sz += sizeof(*shared) * fobj->shared_max;
+
+		if (!pfence_excl && fence_excl)
+			sz += sizeof(*shared);
+
+		if (sz) {
 			struct dma_fence **nshared;
-			size_t sz = sizeof(*shared) * fobj->shared_max;
 
 			nshared = krealloc(shared, sz,
 					   GFP_NOWAIT | __GFP_NOWARN);
@@ -420,13 +426,19 @@ int reservation_object_get_fences_rcu(struct reservation_object *obj,
 				break;
 			}
 			shared = nshared;
-			shared_count = fobj->shared_count;
-
+			shared_count = fobj ? fobj->shared_count : 0;
 			for (i = 0; i < shared_count; ++i) {
 				shared[i] = rcu_dereference(fobj->shared[i]);
 				if (!dma_fence_get_rcu(shared[i]))
 					break;
 			}
+
+			if (!pfence_excl && fence_excl) {
+				shared[i] = fence_excl;
+				fence_excl = NULL;
+				++i;
+				++shared_count;
+			}
 		}
 
 		if (i != shared_count || read_seqcount_retry(&obj->seq, seq)) {
@@ -448,7 +460,8 @@ unlock:
 
 	*pshared_count = shared_count;
 	*pshared = shared;
-	*pfence_excl = fence_excl;
+	if (pfence_excl)
+		*pfence_excl = fence_excl;
 
 	return ret;
 }

drivers/dma-buf/sw_sync.c

@@ -43,14 +43,14 @@
  * timelines.
  *
  * Fences can be created with SW_SYNC_IOC_CREATE_FENCE ioctl with struct
- * sw_sync_ioctl_create_fence as parameter.
+ * sw_sync_create_fence_data as parameter.
  *
  * To increment the timeline counter, SW_SYNC_IOC_INC ioctl should be used
  * with the increment as u32. This will update the last signaled value
  * from the timeline and signal any fence that has a seqno smaller or equal
  * to it.
  *
- * struct sw_sync_ioctl_create_fence
+ * struct sw_sync_create_fence_data
  * @value:	the seqno to initialise the fence with
  * @name:	the name of the new sync point
  * @fence:	return the fd of the new sync_file with the created fence
@@ -235,10 +235,10 @@ static void sync_timeline_signal(struct sync_timeline *obj, unsigned int inc)
 
 /**
  * sync_pt_create() - creates a sync pt
- * @parent:	fence's parent sync_timeline
- * @inc:	value of the fence
+ * @obj:	parent sync_timeline
+ * @value:	value of the fence
  *
- * Creates a new sync_pt as a child of @parent.  @size bytes will be
+ * Creates a new sync_pt (fence) as a child of @parent.  @size bytes will be
  * allocated allowing for implementation specific data to be kept after
  * the generic sync_timeline struct. Returns the sync_pt object or
  * NULL in case of error.

drivers/dma/stm32-dmamux.c

@@ -118,14 +118,15 @@ static void *stm32_dmamux_route_allocate(struct of_phandle_args *dma_spec,
 	spin_lock_irqsave(&dmamux->lock, flags);
 	mux->chan_id = find_first_zero_bit(dmamux->dma_inuse,
 					   dmamux->dma_requests);
-	set_bit(mux->chan_id, dmamux->dma_inuse);
-	spin_unlock_irqrestore(&dmamux->lock, flags);
 
 	if (mux->chan_id == dmamux->dma_requests) {
+		spin_unlock_irqrestore(&dmamux->lock, flags);
 		dev_err(&pdev->dev, "Run out of free DMA requests\n");
 		ret = -ENOMEM;
-		goto error;
+		goto error_chan_id;
 	}
+	set_bit(mux->chan_id, dmamux->dma_inuse);
+	spin_unlock_irqrestore(&dmamux->lock, flags);
 
 	/* Look for DMA Master */
 	for (i = 1, min = 0, max = dmamux->dma_reqs[i];
@@ -173,6 +174,8 @@ static void *stm32_dmamux_route_allocate(struct of_phandle_args *dma_spec,
 
 error:
 	clear_bit(mux->chan_id, dmamux->dma_inuse);
+
+error_chan_id:
 	kfree(mux);
 	return ERR_PTR(ret);
 }

drivers/gpu/drm/amd/acp/include/acp_gfx_if.h

@@ -25,7 +25,6 @@
 #define _ACP_GFX_IF_H
 
 #include <linux/types.h>
-#include "cgs_linux.h"
 #include "cgs_common.h"
 
 int amd_acp_hw_init(struct cgs_device *cgs_device,

drivers/gpu/drm/amd/amdgpu/Makefile

@@ -30,7 +30,6 @@ FULL_AMD_DISPLAY_PATH = $(FULL_AMD_PATH)/$(DISPLAY_FOLDER_NAME)
 ccflags-y := -I$(FULL_AMD_PATH)/include/asic_reg \
 	-I$(FULL_AMD_PATH)/include \
 	-I$(FULL_AMD_PATH)/amdgpu \
-	-I$(FULL_AMD_PATH)/scheduler \
 	-I$(FULL_AMD_PATH)/powerplay/inc \
 	-I$(FULL_AMD_PATH)/acp/include \
 	-I$(FULL_AMD_DISPLAY_PATH) \
@@ -63,7 +62,7 @@ amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
 amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o si_dpm.o si_smc.o
 
 amdgpu-y += \
-	vi.o mxgpu_vi.o nbio_v6_1.o soc15.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o
+	vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o
 
 # add GMC block
 amdgpu-y += \
@@ -88,8 +87,7 @@ amdgpu-y += \
 
 # add SMC block
 amdgpu-y += \
-	amdgpu_dpm.o \
-	amdgpu_powerplay.o
+	amdgpu_dpm.o
 
 # add DCE block
 amdgpu-y += \
@@ -130,6 +128,8 @@ amdgpu-y += \
 # add amdkfd interfaces
 amdgpu-y += \
 	 amdgpu_amdkfd.o \
+	 amdgpu_amdkfd_fence.o \
+	 amdgpu_amdkfd_gpuvm.o \
 	 amdgpu_amdkfd_gfx_v8.o
 
 # add cgs

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -68,6 +68,7 @@
 #include "amdgpu_vce.h"
 #include "amdgpu_vcn.h"
 #include "amdgpu_mn.h"
+#include "amdgpu_gmc.h"
 #include "amdgpu_dm.h"
 #include "amdgpu_virt.h"
 #include "amdgpu_gart.h"
@@ -127,6 +128,7 @@ extern int amdgpu_job_hang_limit;
 extern int amdgpu_lbpw;
 extern int amdgpu_compute_multipipe;
 extern int amdgpu_gpu_recovery;
+extern int amdgpu_emu_mode;
 
 #ifdef CONFIG_DRM_AMDGPU_SI
 extern int amdgpu_si_support;
@@ -179,10 +181,6 @@ extern int amdgpu_cik_support;
 #define CIK_CURSOR_WIDTH 128
 #define CIK_CURSOR_HEIGHT 128
 
-/* GPU RESET flags */
-#define AMDGPU_RESET_INFO_VRAM_LOST  (1 << 0)
-#define AMDGPU_RESET_INFO_FULLRESET  (1 << 1)
-
 struct amdgpu_device;
 struct amdgpu_ib;
 struct amdgpu_cs_parser;
@@ -318,13 +316,6 @@ struct amdgpu_vm_pte_funcs {
 	void (*write_pte)(struct amdgpu_ib *ib, uint64_t pe,
 			  uint64_t value, unsigned count,
 			  uint32_t incr);
-
-	/* maximum nums of PTEs/PDEs in a single operation */
-	uint32_t	set_max_nums_pte_pde;
-
-	/* number of dw to reserve per operation */
-	unsigned	set_pte_pde_num_dw;
-
 	/* for linear pte/pde updates without addr mapping */
 	void (*set_pte_pde)(struct amdgpu_ib *ib,
 			    uint64_t pe,
@@ -332,28 +323,6 @@ struct amdgpu_vm_pte_funcs {
 			    uint32_t incr, uint64_t flags);
 };
 
-/* provided by the gmc block */
-struct amdgpu_gart_funcs {
-	/* flush the vm tlb via mmio */
-	void (*flush_gpu_tlb)(struct amdgpu_device *adev,
-			      uint32_t vmid);
-	/* write pte/pde updates using the cpu */
-	int (*set_pte_pde)(struct amdgpu_device *adev,
-			   void *cpu_pt_addr, /* cpu addr of page table */
-			   uint32_t gpu_page_idx, /* pte/pde to update */
-			   uint64_t addr, /* addr to write into pte/pde */
-			   uint64_t flags); /* access flags */
-	/* enable/disable PRT support */
-	void (*set_prt)(struct amdgpu_device *adev, bool enable);
-	/* set pte flags based per asic */
-	uint64_t (*get_vm_pte_flags)(struct amdgpu_device *adev,
-				     uint32_t flags);
-	/* get the pde for a given mc addr */
-	void (*get_vm_pde)(struct amdgpu_device *adev, int level,
-			   u64 *dst, u64 *flags);
-	uint32_t (*get_invalidate_req)(unsigned int vmid);
-};
-
 /* provided by the ih block */
 struct amdgpu_ih_funcs {
 	/* ring read/write ptr handling, called from interrupt context */
@@ -370,14 +339,6 @@ struct amdgpu_ih_funcs {
 bool amdgpu_get_bios(struct amdgpu_device *adev);
 bool amdgpu_read_bios(struct amdgpu_device *adev);
 
-/*
- * Dummy page
- */
-struct amdgpu_dummy_page {
-	struct page	*page;
-	dma_addr_t	addr;
-};
-
 /*
  * Clocks
  */
@@ -418,8 +379,8 @@ amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
 struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
 					struct drm_gem_object *gobj,
 					int flags);
-int amdgpu_gem_prime_pin(struct drm_gem_object *obj);
-void amdgpu_gem_prime_unpin(struct drm_gem_object *obj);
+struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev,
+					    struct dma_buf *dma_buf);
 struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *);
 void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj);
 void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr);
@@ -480,7 +441,7 @@ struct amdgpu_sa_bo {
 void amdgpu_gem_force_release(struct amdgpu_device *adev);
 int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
 			     int alignment, u32 initial_domain,
-			     u64 flags, bool kernel,
+			     u64 flags, enum ttm_bo_type type,
 			     struct reservation_object *resv,
 			     struct drm_gem_object **obj);
 
@@ -493,56 +454,6 @@ int amdgpu_mode_dumb_mmap(struct drm_file *filp,
 int amdgpu_fence_slab_init(void);
 void amdgpu_fence_slab_fini(void);
 
-/*
- * VMHUB structures, functions & helpers
- */
-struct amdgpu_vmhub {
-	uint32_t	ctx0_ptb_addr_lo32;
-	uint32_t	ctx0_ptb_addr_hi32;
-	uint32_t	vm_inv_eng0_req;
-	uint32_t	vm_inv_eng0_ack;
-	uint32_t	vm_context0_cntl;
-	uint32_t	vm_l2_pro_fault_status;
-	uint32_t	vm_l2_pro_fault_cntl;
-};
-
-/*
- * GPU MC structures, functions & helpers
- */
-struct amdgpu_mc {
-	resource_size_t		aper_size;
-	resource_size_t		aper_base;
-	resource_size_t		agp_base;
-	/* for some chips with <= 32MB we need to lie
-	 * about vram size near mc fb location */
-	u64			mc_vram_size;
-	u64			visible_vram_size;
-	u64			gart_size;
-	u64			gart_start;
-	u64			gart_end;
-	u64			vram_start;
-	u64			vram_end;
-	unsigned		vram_width;
-	u64			real_vram_size;
-	int			vram_mtrr;
-	u64                     mc_mask;
-	const struct firmware   *fw;	/* MC firmware */
-	uint32_t                fw_version;
-	struct amdgpu_irq_src	vm_fault;
-	uint32_t		vram_type;
-	uint32_t                srbm_soft_reset;
-	bool			prt_warning;
-	uint64_t		stolen_size;
-	/* apertures */
-	u64					shared_aperture_start;
-	u64					shared_aperture_end;
-	u64					private_aperture_start;
-	u64					private_aperture_end;
-	/* protects concurrent invalidation */
-	spinlock_t		invalidate_lock;
-	bool			translate_further;
-};
-
 /*
  * GPU doorbell structures, functions & helpers
  */
@@ -979,6 +890,7 @@ struct amdgpu_gfx_funcs {
 	void (*read_wave_data)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t *dst, int *no_fields);
 	void (*read_wave_vgprs)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t thread, uint32_t start, uint32_t size, uint32_t *dst);
 	void (*read_wave_sgprs)(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t start, uint32_t size, uint32_t *dst);
+	void (*select_me_pipe_q)(struct amdgpu_device *adev, u32 me, u32 pipe, u32 queue);
 };
 
 struct amdgpu_ngg_buf {
@@ -1125,8 +1037,9 @@ struct amdgpu_job {
 	void			*owner;
 	uint64_t		fence_ctx; /* the fence_context this job uses */
 	bool                    vm_needs_flush;
-	unsigned		vmid;
 	uint64_t		vm_pd_addr;
+	unsigned		vmid;
+	unsigned		pasid;
 	uint32_t		gds_base, gds_size;
 	uint32_t		gws_base, gws_size;
 	uint32_t		oa_base, oa_size;
@@ -1169,8 +1082,6 @@ struct amdgpu_wb {
 int amdgpu_device_wb_get(struct amdgpu_device *adev, u32 *wb);
 void amdgpu_device_wb_free(struct amdgpu_device *adev, u32 wb);
 
-void amdgpu_device_get_pcie_info(struct amdgpu_device *adev);
-
 /*
  * SDMA
  */
@@ -1288,6 +1199,11 @@ struct amdgpu_asic_funcs {
 	void (*set_pcie_lanes)(struct amdgpu_device *adev, int lanes);
 	/* get config memsize register */
 	u32 (*get_config_memsize)(struct amdgpu_device *adev);
+	/* flush hdp write queue */
+	void (*flush_hdp)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+	/* invalidate hdp read cache */
+	void (*invalidate_hdp)(struct amdgpu_device *adev,
+			       struct amdgpu_ring *ring);
 };
 
 /*
@@ -1431,7 +1347,7 @@ struct amdgpu_nbio_funcs {
 	u32 (*get_pcie_data_offset)(struct amdgpu_device *adev);
 	u32 (*get_rev_id)(struct amdgpu_device *adev);
 	void (*mc_access_enable)(struct amdgpu_device *adev, bool enable);
-	void (*hdp_flush)(struct amdgpu_device *adev);
+	void (*hdp_flush)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
 	u32 (*get_memsize)(struct amdgpu_device *adev);
 	void (*sdma_doorbell_range)(struct amdgpu_device *adev, int instance,
 				    bool use_doorbell, int doorbell_index);
@@ -1463,6 +1379,7 @@ enum amd_hw_ip_block_type {
 	ATHUB_HWIP,
 	NBIO_HWIP,
 	MP0_HWIP,
+	MP1_HWIP,
 	UVD_HWIP,
 	VCN_HWIP = UVD_HWIP,
 	VCE_HWIP,
@@ -1472,15 +1389,14 @@ enum amd_hw_ip_block_type {
 	SMUIO_HWIP,
 	PWR_HWIP,
 	NBIF_HWIP,
+	THM_HWIP,
 	MAX_HWIP
 };
 
 #define HWIP_MAX_INSTANCE	6
 
 struct amd_powerplay {
-	struct cgs_device *cgs_device;
 	void *pp_handle;
-	const struct amd_ip_funcs *ip_funcs;
 	const struct amd_pm_funcs *pp_funcs;
 };
 
@@ -1504,6 +1420,7 @@ struct amdgpu_device {
 	const struct amdgpu_asic_funcs	*asic_funcs;
 	bool				shutdown;
 	bool				need_dma32;
+	bool				need_swiotlb;
 	bool				accel_working;
 	struct work_struct		reset_work;
 	struct notifier_block		acpi_nb;
@@ -1573,9 +1490,9 @@ struct amdgpu_device {
 	struct amdgpu_clock            clock;
 
 	/* MC */
-	struct amdgpu_mc		mc;
+	struct amdgpu_gmc		gmc;
 	struct amdgpu_gart		gart;
-	struct amdgpu_dummy_page	dummy_page;
+	dma_addr_t			dummy_page_addr;
 	struct amdgpu_vm_manager	vm_manager;
 	struct amdgpu_vmhub             vmhub[AMDGPU_MAX_VMHUBS];
 
@@ -1714,6 +1631,9 @@ uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
 			uint32_t acc_flags);
 void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
 		    uint32_t acc_flags);
+void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value);
+uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset);
+
 u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg);
 void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v);
 
@@ -1725,6 +1645,8 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
 bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type);
 bool amdgpu_device_has_dc_support(struct amdgpu_device *adev);
 
+int emu_soc_asic_init(struct amdgpu_device *adev);
+
 /*
  * Registers read & write functions.
  */
@@ -1735,6 +1657,9 @@ bool amdgpu_device_has_dc_support(struct amdgpu_device *adev);
 #define RREG32_NO_KIQ(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_NO_KIQ)
 #define WREG32_NO_KIQ(reg, v) amdgpu_mm_wreg(adev, (reg), (v), AMDGPU_REGS_NO_KIQ)
 
+#define RREG8(reg) amdgpu_mm_rreg8(adev, (reg))
+#define WREG8(reg, v) amdgpu_mm_wreg8(adev, (reg), (v))
+
 #define RREG32(reg) amdgpu_mm_rreg(adev, (reg), 0)
 #define RREG32_IDX(reg) amdgpu_mm_rreg(adev, (reg), AMDGPU_REGS_IDX)
 #define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_mm_rreg(adev, (reg), 0))
@@ -1837,13 +1762,17 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l))
 #define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v)))
 #define amdgpu_asic_get_config_memsize(adev) (adev)->asic_funcs->get_config_memsize((adev))
-#define amdgpu_gart_flush_gpu_tlb(adev, vmid) (adev)->gart.gart_funcs->flush_gpu_tlb((adev), (vmid))
-#define amdgpu_gart_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gart.gart_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags))
-#define amdgpu_gart_get_vm_pde(adev, level, dst, flags) (adev)->gart.gart_funcs->get_vm_pde((adev), (level), (dst), (flags))
+#define amdgpu_asic_flush_hdp(adev, r) (adev)->asic_funcs->flush_hdp((adev), (r))
+#define amdgpu_asic_invalidate_hdp(adev, r) (adev)->asic_funcs->invalidate_hdp((adev), (r))
+#define amdgpu_gmc_flush_gpu_tlb(adev, vmid) (adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid))
+#define amdgpu_gmc_emit_flush_gpu_tlb(r, vmid, addr) (r)->adev->gmc.gmc_funcs->emit_flush_gpu_tlb((r), (vmid), (addr))
+#define amdgpu_gmc_emit_pasid_mapping(r, vmid, pasid) (r)->adev->gmc.gmc_funcs->emit_pasid_mapping((r), (vmid), (pasid))
+#define amdgpu_gmc_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gmc.gmc_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags))
+#define amdgpu_gmc_get_vm_pde(adev, level, dst, flags) (adev)->gmc.gmc_funcs->get_vm_pde((adev), (level), (dst), (flags))
+#define amdgpu_gmc_get_pte_flags(adev, flags) (adev)->gmc.gmc_funcs->get_vm_pte_flags((adev),(flags))
 #define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count)))
 #define amdgpu_vm_write_pte(adev, ib, pe, value, count, incr) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (value), (count), (incr)))
 #define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
-#define amdgpu_vm_get_pte_flags(adev, flags) (adev)->gart.gart_funcs->get_vm_pte_flags((adev),(flags))
 #define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
 #define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
 #define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
@@ -1856,11 +1785,11 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
 #define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
 #define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
-#define amdgpu_ring_emit_hdp_invalidate(r) (r)->funcs->emit_hdp_invalidate((r))
 #define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
 #define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
 #define amdgpu_ring_emit_rreg(r, d) (r)->funcs->emit_rreg((r), (d))
 #define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v))
+#define amdgpu_ring_emit_reg_wait(r, d, v, m) (r)->funcs->emit_reg_wait((r), (d), (v), (m))
 #define amdgpu_ring_emit_tmz(r, b) (r)->funcs->emit_tmz((r), (b))
 #define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
 #define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r))
@@ -1870,7 +1799,6 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
 #define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
 #define amdgpu_display_vblank_get_counter(adev, crtc) (adev)->mode_info.funcs->vblank_get_counter((adev), (crtc))
-#define amdgpu_display_vblank_wait(adev, crtc) (adev)->mode_info.funcs->vblank_wait((adev), (crtc))
 #define amdgpu_display_backlight_set_level(adev, e, l) (adev)->mode_info.funcs->backlight_set_level((e), (l))
 #define amdgpu_display_backlight_get_level(adev, e) (adev)->mode_info.funcs->backlight_get_level((e))
 #define amdgpu_display_hpd_sense(adev, h) (adev)->mode_info.funcs->hpd_sense((adev), (h))
@@ -1887,26 +1815,24 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_gfx_select_se_sh(adev, se, sh, instance) (adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance))
 #define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a))
 #define amdgpu_psp_check_fw_loading_status(adev, i) (adev)->firmware.funcs->check_fw_loading_status((adev), (i))
+#define amdgpu_gfx_select_me_pipe_q(adev, me, pipe, q) (adev)->gfx.funcs->select_me_pipe_q((adev), (me), (pipe), (q))
 
 /* Common functions */
 int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 			      struct amdgpu_job* job, bool force);
 void amdgpu_device_pci_config_reset(struct amdgpu_device *adev);
 bool amdgpu_device_need_post(struct amdgpu_device *adev);
-void amdgpu_update_display_priority(struct amdgpu_device *adev);
+void amdgpu_display_update_priority(struct amdgpu_device *adev);
 
 void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
 				  u64 num_vis_bytes);
 void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *abo, u32 domain);
 bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
 void amdgpu_device_vram_location(struct amdgpu_device *adev,
-				 struct amdgpu_mc *mc, u64 base);
+				 struct amdgpu_gmc *mc, u64 base);
 void amdgpu_device_gart_location(struct amdgpu_device *adev,
-				 struct amdgpu_mc *mc);
+				 struct amdgpu_gmc *mc);
 int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev);
-void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size);
-int amdgpu_ttm_init(struct amdgpu_device *adev);
-void amdgpu_ttm_fini(struct amdgpu_device *adev);
 void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
 					     const u32 *registers,
 					     const u32 array_size);

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

@@ -30,6 +30,8 @@
 const struct kgd2kfd_calls *kgd2kfd;
 bool (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
 
+static const unsigned int compute_vmid_bitmap = 0xFF00;
+
 int amdgpu_amdkfd_init(void)
 {
 	int ret;
@@ -56,6 +58,7 @@ int amdgpu_amdkfd_init(void)
 #else
 	ret = -ENOENT;
 #endif
+	amdgpu_amdkfd_gpuvm_init_mem_limits();
 
 	return ret;
 }
@@ -78,10 +81,15 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_KAVERI:
+	case CHIP_HAWAII:
 		kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
 		break;
 #endif
 	case CHIP_CARRIZO:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
 		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
 		break;
 	default:
@@ -132,9 +140,13 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 	int last_valid_bit;
 	if (adev->kfd) {
 		struct kgd2kfd_shared_resources gpu_resources = {
-			.compute_vmid_bitmap = 0xFF00,
+			.compute_vmid_bitmap = compute_vmid_bitmap,
 			.num_pipe_per_mec = adev->gfx.mec.num_pipe_per_mec,
-			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe
+			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe,
+			.gpuvm_size = min(adev->vm_manager.max_pfn
+					  << AMDGPU_GPU_PAGE_SHIFT,
+					  AMDGPU_VA_HOLE_START),
+			.drm_render_minor = adev->ddev->render->index
 		};
 
 		/* this is going to have a few of the MSBs set that we need to
@@ -204,20 +216,14 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 			void **cpu_ptr)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
-	struct kgd_mem **mem = (struct kgd_mem **) mem_obj;
+	struct amdgpu_bo *bo = NULL;
 	int r;
+	uint64_t gpu_addr_tmp = 0;
+	void *cpu_ptr_tmp = NULL;
 
-	BUG_ON(kgd == NULL);
-	BUG_ON(gpu_addr == NULL);
-	BUG_ON(cpu_ptr == NULL);
-
-	*mem = kmalloc(sizeof(struct kgd_mem), GFP_KERNEL);
-	if ((*mem) == NULL)
-		return -ENOMEM;
-
-	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
-			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, 0,
-			     &(*mem)->bo);
+	r = amdgpu_bo_create(adev, size, PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, ttm_bo_type_kernel,
+			     NULL, &bo);
 	if (r) {
 		dev_err(adev->dev,
 			"failed to allocate BO for amdkfd (%d)\n", r);
@@ -225,54 +231,53 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 	}
 
 	/* map the buffer */
-	r = amdgpu_bo_reserve((*mem)->bo, true);
+	r = amdgpu_bo_reserve(bo, true);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
 		goto allocate_mem_reserve_bo_failed;
 	}
 
-	r = amdgpu_bo_pin((*mem)->bo, AMDGPU_GEM_DOMAIN_GTT,
-				&(*mem)->gpu_addr);
+	r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT,
+				&gpu_addr_tmp);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to pin bo for amdkfd\n", r);
 		goto allocate_mem_pin_bo_failed;
 	}
-	*gpu_addr = (*mem)->gpu_addr;
 
-	r = amdgpu_bo_kmap((*mem)->bo, &(*mem)->cpu_ptr);
+	r = amdgpu_bo_kmap(bo, &cpu_ptr_tmp);
 	if (r) {
 		dev_err(adev->dev,
 			"(%d) failed to map bo to kernel for amdkfd\n", r);
 		goto allocate_mem_kmap_bo_failed;
 	}
-	*cpu_ptr = (*mem)->cpu_ptr;
 
-	amdgpu_bo_unreserve((*mem)->bo);
+	*mem_obj = bo;
+	*gpu_addr = gpu_addr_tmp;
+	*cpu_ptr = cpu_ptr_tmp;
+
+	amdgpu_bo_unreserve(bo);
 
 	return 0;
 
 allocate_mem_kmap_bo_failed:
-	amdgpu_bo_unpin((*mem)->bo);
+	amdgpu_bo_unpin(bo);
 allocate_mem_pin_bo_failed:
-	amdgpu_bo_unreserve((*mem)->bo);
+	amdgpu_bo_unreserve(bo);
 allocate_mem_reserve_bo_failed:
-	amdgpu_bo_unref(&(*mem)->bo);
+	amdgpu_bo_unref(&bo);
 
 	return r;
 }
 
 void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
 {
-	struct kgd_mem *mem = (struct kgd_mem *) mem_obj;
-
-	BUG_ON(mem == NULL);
+	struct amdgpu_bo *bo = (struct amdgpu_bo *) mem_obj;
 
-	amdgpu_bo_reserve(mem->bo, true);
-	amdgpu_bo_kunmap(mem->bo);
-	amdgpu_bo_unpin(mem->bo);
-	amdgpu_bo_unreserve(mem->bo);
-	amdgpu_bo_unref(&(mem->bo));
-	kfree(mem);
+	amdgpu_bo_reserve(bo, true);
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&(bo));
 }
 
 void get_local_mem_info(struct kgd_dev *kgd,
@@ -281,24 +286,29 @@ void get_local_mem_info(struct kgd_dev *kgd,
 	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
 	uint64_t address_mask = adev->dev->dma_mask ? ~*adev->dev->dma_mask :
 					     ~((1ULL << 32) - 1);
-	resource_size_t aper_limit = adev->mc.aper_base + adev->mc.aper_size;
+	resource_size_t aper_limit = adev->gmc.aper_base + adev->gmc.aper_size;
 
 	memset(mem_info, 0, sizeof(*mem_info));
-	if (!(adev->mc.aper_base & address_mask || aper_limit & address_mask)) {
-		mem_info->local_mem_size_public = adev->mc.visible_vram_size;
-		mem_info->local_mem_size_private = adev->mc.real_vram_size -
-				adev->mc.visible_vram_size;
+	if (!(adev->gmc.aper_base & address_mask || aper_limit & address_mask)) {
+		mem_info->local_mem_size_public = adev->gmc.visible_vram_size;
+		mem_info->local_mem_size_private = adev->gmc.real_vram_size -
+				adev->gmc.visible_vram_size;
 	} else {
 		mem_info->local_mem_size_public = 0;
-		mem_info->local_mem_size_private = adev->mc.real_vram_size;
+		mem_info->local_mem_size_private = adev->gmc.real_vram_size;
 	}
-	mem_info->vram_width = adev->mc.vram_width;
+	mem_info->vram_width = adev->gmc.vram_width;
 
 	pr_debug("Address base: %pap limit %pap public 0x%llx private 0x%llx\n",
-			&adev->mc.aper_base, &aper_limit,
+			&adev->gmc.aper_base, &aper_limit,
 			mem_info->local_mem_size_public,
 			mem_info->local_mem_size_private);
 
+	if (amdgpu_emu_mode == 1) {
+		mem_info->mem_clk_max = 100;
+		return;
+	}
+
 	if (amdgpu_sriov_vf(adev))
 		mem_info->mem_clk_max = adev->clock.default_mclk / 100;
 	else
@@ -319,6 +329,9 @@ uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
 	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
 
 	/* the sclk is in quantas of 10kHz */
+	if (amdgpu_emu_mode == 1)
+		return 100;
+
 	if (amdgpu_sriov_vf(adev))
 		return adev->clock.default_sclk / 100;
 
@@ -354,3 +367,68 @@ uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd)
 
 	return amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 }
+
+int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct amdgpu_ring *ring;
+	struct dma_fence *f = NULL;
+	int ret;
+
+	switch (engine) {
+	case KGD_ENGINE_MEC1:
+		ring = &adev->gfx.compute_ring[0];
+		break;
+	case KGD_ENGINE_SDMA1:
+		ring = &adev->sdma.instance[0].ring;
+		break;
+	case KGD_ENGINE_SDMA2:
+		ring = &adev->sdma.instance[1].ring;
+		break;
+	default:
+		pr_err("Invalid engine in IB submission: %d\n", engine);
+		ret = -EINVAL;
+		goto err;
+	}
+
+	ret = amdgpu_job_alloc(adev, 1, &job, NULL);
+	if (ret)
+		goto err;
+
+	ib = &job->ibs[0];
+	memset(ib, 0, sizeof(struct amdgpu_ib));
+
+	ib->gpu_addr = gpu_addr;
+	ib->ptr = ib_cmd;
+	ib->length_dw = ib_len;
+	/* This works for NO_HWS. TODO: need to handle without knowing VMID */
+	job->vmid = vmid;
+
+	ret = amdgpu_ib_schedule(ring, 1, ib, job, &f);
+	if (ret) {
+		DRM_ERROR("amdgpu: failed to schedule IB.\n");
+		goto err_ib_sched;
+	}
+
+	ret = dma_fence_wait(f, false);
+
+err_ib_sched:
+	dma_fence_put(f);
+	amdgpu_job_free(job);
+err:
+	return ret;
+}
+
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid)
+{
+	if (adev->kfd) {
+		if ((1 << vmid) & compute_vmid_bitmap)
+			return true;
+	}
+
+	return false;
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h

@@ -26,15 +26,71 @@
 #define AMDGPU_AMDKFD_H_INCLUDED
 
 #include <linux/types.h>
+#include <linux/mm.h>
 #include <linux/mmu_context.h>
 #include <kgd_kfd_interface.h>
+#include <drm/ttm/ttm_execbuf_util.h>
+#include "amdgpu_sync.h"
+#include "amdgpu_vm.h"
+
+extern const struct kgd2kfd_calls *kgd2kfd;
 
 struct amdgpu_device;
 
+struct kfd_bo_va_list {
+	struct list_head bo_list;
+	struct amdgpu_bo_va *bo_va;
+	void *kgd_dev;
+	bool is_mapped;
+	uint64_t va;
+	uint64_t pte_flags;
+};
+
 struct kgd_mem {
+	struct mutex lock;
 	struct amdgpu_bo *bo;
-	uint64_t gpu_addr;
-	void *cpu_ptr;
+	struct list_head bo_va_list;
+	/* protected by amdkfd_process_info.lock */
+	struct ttm_validate_buffer validate_list;
+	struct ttm_validate_buffer resv_list;
+	uint32_t domain;
+	unsigned int mapped_to_gpu_memory;
+	uint64_t va;
+
+	uint32_t mapping_flags;
+
+	struct amdkfd_process_info *process_info;
+
+	struct amdgpu_sync sync;
+
+	bool aql_queue;
+};
+
+/* KFD Memory Eviction */
+struct amdgpu_amdkfd_fence {
+	struct dma_fence base;
+	struct mm_struct *mm;
+	spinlock_t lock;
+	char timeline_name[TASK_COMM_LEN];
+};
+
+struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
+						       struct mm_struct *mm);
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm);
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f);
+
+struct amdkfd_process_info {
+	/* List head of all VMs that belong to a KFD process */
+	struct list_head vm_list_head;
+	/* List head for all KFD BOs that belong to a KFD process. */
+	struct list_head kfd_bo_list;
+	/* Lock to protect kfd_bo_list */
+	struct mutex lock;
+
+	/* Number of VMs */
+	unsigned int n_vms;
+	/* Eviction Fence */
+	struct amdgpu_amdkfd_fence *eviction_fence;
 };
 
 int amdgpu_amdkfd_init(void);
@@ -48,9 +104,15 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev);
 void amdgpu_amdkfd_device_init(struct amdgpu_device *adev);
 void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev);
 
+int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len);
+
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void);
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void);
 
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid);
+
 /* Shared API */
 int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 			void **mem_obj, uint64_t *gpu_addr,
@@ -79,4 +141,36 @@ uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd);
 		valid;							\
 	})
 
+/* GPUVM API */
+int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, void **vm,
+					  void **process_info,
+					  struct dma_fence **ef);
+int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd,
+					   struct file *filp,
+					   void **vm, void **process_info,
+					   struct dma_fence **ef);
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm);
+void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm);
+uint32_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm);
+int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		struct kgd_dev *kgd, uint64_t va, uint64_t size,
+		void *vm, struct kgd_mem **mem,
+		uint64_t *offset, uint32_t flags);
+int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem);
+int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm);
+int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm);
+int amdgpu_amdkfd_gpuvm_sync_memory(
+		struct kgd_dev *kgd, struct kgd_mem *mem, bool intr);
+int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_dev *kgd,
+		struct kgd_mem *mem, void **kptr, uint64_t *size);
+int amdgpu_amdkfd_gpuvm_restore_process_bos(void *process_info,
+					    struct dma_fence **ef);
+
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void);
+void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo);
+
 #endif /* AMDGPU_AMDKFD_H_INCLUDED */

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c

@@ -0,0 +1,179 @@
+/*
+ * Copyright 2016-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/dma-fence.h>
+#include <linux/spinlock.h>
+#include <linux/atomic.h>
+#include <linux/stacktrace.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/sched/mm.h>
+#include "amdgpu_amdkfd.h"
+
+static const struct dma_fence_ops amdkfd_fence_ops;
+static atomic_t fence_seq = ATOMIC_INIT(0);
+
+/* Eviction Fence
+ * Fence helper functions to deal with KFD memory eviction.
+ * Big Idea - Since KFD submissions are done by user queues, a BO cannot be
+ *  evicted unless all the user queues for that process are evicted.
+ *
+ * All the BOs in a process share an eviction fence. When process X wants
+ * to map VRAM memory but TTM can't find enough space, TTM will attempt to
+ * evict BOs from its LRU list. TTM checks if the BO is valuable to evict
+ * by calling ttm_bo_driver->eviction_valuable().
+ *
+ * ttm_bo_driver->eviction_valuable() - will return false if the BO belongs
+ *  to process X. Otherwise, it will return true to indicate BO can be
+ *  evicted by TTM.
+ *
+ * If ttm_bo_driver->eviction_valuable returns true, then TTM will continue
+ * the evcition process for that BO by calling ttm_bo_evict --> amdgpu_bo_move
+ * --> amdgpu_copy_buffer(). This sets up job in GPU scheduler.
+ *
+ * GPU Scheduler (amd_sched_main) - sets up a cb (fence_add_callback) to
+ *  nofity when the BO is free to move. fence_add_callback --> enable_signaling
+ *  --> amdgpu_amdkfd_fence.enable_signaling
+ *
+ * amdgpu_amdkfd_fence.enable_signaling - Start a work item that will quiesce
+ * user queues and signal fence. The work item will also start another delayed
+ * work item to restore BOs
+ */
+
+struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
+						       struct mm_struct *mm)
+{
+	struct amdgpu_amdkfd_fence *fence;
+
+	fence = kzalloc(sizeof(*fence), GFP_KERNEL);
+	if (fence == NULL)
+		return NULL;
+
+	/* This reference gets released in amdkfd_fence_release */
+	mmgrab(mm);
+	fence->mm = mm;
+	get_task_comm(fence->timeline_name, current);
+	spin_lock_init(&fence->lock);
+
+	dma_fence_init(&fence->base, &amdkfd_fence_ops, &fence->lock,
+		   context, atomic_inc_return(&fence_seq));
+
+	return fence;
+}
+
+struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence;
+
+	if (!f)
+		return NULL;
+
+	fence = container_of(f, struct amdgpu_amdkfd_fence, base);
+	if (fence && f->ops == &amdkfd_fence_ops)
+		return fence;
+
+	return NULL;
+}
+
+static const char *amdkfd_fence_get_driver_name(struct dma_fence *f)
+{
+	return "amdgpu_amdkfd_fence";
+}
+
+static const char *amdkfd_fence_get_timeline_name(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	return fence->timeline_name;
+}
+
+/**
+ * amdkfd_fence_enable_signaling - This gets called when TTM wants to evict
+ *  a KFD BO and schedules a job to move the BO.
+ *  If fence is already signaled return true.
+ *  If fence is not signaled schedule a evict KFD process work item.
+ */
+static bool amdkfd_fence_enable_signaling(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	if (!fence)
+		return false;
+
+	if (dma_fence_is_signaled(f))
+		return true;
+
+	if (!kgd2kfd->schedule_evict_and_restore_process(fence->mm, f))
+		return true;
+
+	return false;
+}
+
+/**
+ * amdkfd_fence_release - callback that fence can be freed
+ *
+ * @fence: fence
+ *
+ * This function is called when the reference count becomes zero.
+ * Drops the mm_struct reference and RCU schedules freeing up the fence.
+ */
+static void amdkfd_fence_release(struct dma_fence *f)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	/* Unconditionally signal the fence. The process is getting
+	 * terminated.
+	 */
+	if (WARN_ON(!fence))
+		return; /* Not an amdgpu_amdkfd_fence */
+
+	mmdrop(fence->mm);
+	kfree_rcu(f, rcu);
+}
+
+/**
+ * amdkfd_fence_check_mm - Check if @mm is same as that of the fence @f
+ *  if same return TRUE else return FALSE.
+ *
+ * @f: [IN] fence
+ * @mm: [IN] mm that needs to be verified
+ */
+bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
+{
+	struct amdgpu_amdkfd_fence *fence = to_amdgpu_amdkfd_fence(f);
+
+	if (!fence)
+		return false;
+	else if (fence->mm == mm)
+		return true;
+
+	return false;
+}
+
+static const struct dma_fence_ops amdkfd_fence_ops = {
+	.get_driver_name = amdkfd_fence_get_driver_name,
+	.get_timeline_name = amdkfd_fence_get_timeline_name,
+	.enable_signaling = amdkfd_fence_enable_signaling,
+	.signaled = NULL,
+	.wait = dma_fence_default_wait,
+	.release = amdkfd_fence_release,
+};

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c

@@ -139,11 +139,14 @@ static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
 static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd, uint8_t vmid);
 static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 							uint8_t vmid);
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
 
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
 static void set_scratch_backing_va(struct kgd_dev *kgd,
 					uint64_t va, uint32_t vmid);
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base);
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
 
 /* Because of REG_GET_FIELD() being used, we put this function in the
  * asic specific file.
@@ -196,12 +199,26 @@ static const struct kfd2kgd_calls kfd2kgd = {
 	.address_watch_get_offset = kgd_address_watch_get_offset,
 	.get_atc_vmid_pasid_mapping_pasid = get_atc_vmid_pasid_mapping_pasid,
 	.get_atc_vmid_pasid_mapping_valid = get_atc_vmid_pasid_mapping_valid,
-	.write_vmid_invalidate_request = write_vmid_invalidate_request,
 	.get_fw_version = get_fw_version,
 	.set_scratch_backing_va = set_scratch_backing_va,
 	.get_tile_config = get_tile_config,
 	.get_cu_info = get_cu_info,
-	.get_vram_usage = amdgpu_amdkfd_get_vram_usage
+	.get_vram_usage = amdgpu_amdkfd_get_vram_usage,
+	.create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm,
+	.acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm,
+	.destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm,
+	.get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu,
+	.free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu,
+	.map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu,
+	.unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu,
+	.sync_memory = amdgpu_amdkfd_gpuvm_sync_memory,
+	.map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel,
+	.restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos,
+	.invalidate_tlbs = invalidate_tlbs,
+	.invalidate_tlbs_vmid = invalidate_tlbs_vmid,
+	.submit_ib = amdgpu_amdkfd_submit_ib,
 };
 
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void)
@@ -787,14 +804,7 @@ static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
 
 	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
-	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
-}
-
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-
-	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__PASID_MASK;
 }
 
 static void set_scratch_backing_va(struct kgd_dev *kgd,
@@ -812,8 +822,6 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
 	const union amdgpu_firmware_header *hdr;
 
-	BUG_ON(kgd == NULL);
-
 	switch (type) {
 	case KGD_ENGINE_PFP:
 		hdr = (const union amdgpu_firmware_header *)
@@ -866,3 +874,50 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 	return hdr->common.ucode_version;
 }
 
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+			uint32_t page_table_base)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID\n");
+		return;
+	}
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8, page_table_base);
+}
+
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	int vmid;
+	unsigned int tmp;
+
+	for (vmid = 0; vmid < 16; vmid++) {
+		if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid))
+			continue;
+
+		tmp = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+		if ((tmp & ATC_VMID0_PASID_MAPPING__VALID_MASK) &&
+			(tmp & ATC_VMID0_PASID_MAPPING__PASID_MASK) == pasid) {
+			WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+			RREG32(mmVM_INVALIDATE_RESPONSE);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("non kfd vmid\n");
+		return 0;
+	}
+
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	RREG32(mmVM_INVALIDATE_RESPONSE);
+	return 0;
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c

@@ -81,7 +81,6 @@ static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
 				uint32_t queue_id);
 static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
 				unsigned int utimeout);
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
 static int kgd_address_watch_disable(struct kgd_dev *kgd);
 static int kgd_address_watch_execute(struct kgd_dev *kgd,
 					unsigned int watch_point_id,
@@ -99,10 +98,13 @@ static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
 		uint8_t vmid);
 static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 		uint8_t vmid);
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
 static void set_scratch_backing_va(struct kgd_dev *kgd,
 					uint64_t va, uint32_t vmid);
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base);
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
 
 /* Because of REG_GET_FIELD() being used, we put this function in the
  * asic specific file.
@@ -157,12 +159,26 @@ static const struct kfd2kgd_calls kfd2kgd = {
 			get_atc_vmid_pasid_mapping_pasid,
 	.get_atc_vmid_pasid_mapping_valid =
 			get_atc_vmid_pasid_mapping_valid,
-	.write_vmid_invalidate_request = write_vmid_invalidate_request,
 	.get_fw_version = get_fw_version,
 	.set_scratch_backing_va = set_scratch_backing_va,
 	.get_tile_config = get_tile_config,
 	.get_cu_info = get_cu_info,
-	.get_vram_usage = amdgpu_amdkfd_get_vram_usage
+	.get_vram_usage = amdgpu_amdkfd_get_vram_usage,
+	.create_process_vm = amdgpu_amdkfd_gpuvm_create_process_vm,
+	.acquire_process_vm = amdgpu_amdkfd_gpuvm_acquire_process_vm,
+	.destroy_process_vm = amdgpu_amdkfd_gpuvm_destroy_process_vm,
+	.get_process_page_dir = amdgpu_amdkfd_gpuvm_get_process_page_dir,
+	.set_vm_context_page_table_base = set_vm_context_page_table_base,
+	.alloc_memory_of_gpu = amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu,
+	.free_memory_of_gpu = amdgpu_amdkfd_gpuvm_free_memory_of_gpu,
+	.map_memory_to_gpu = amdgpu_amdkfd_gpuvm_map_memory_to_gpu,
+	.unmap_memory_to_gpu = amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu,
+	.sync_memory = amdgpu_amdkfd_gpuvm_sync_memory,
+	.map_gtt_bo_to_kernel = amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel,
+	.restore_process_bos = amdgpu_amdkfd_gpuvm_restore_process_bos,
+	.invalidate_tlbs = invalidate_tlbs,
+	.invalidate_tlbs_vmid = invalidate_tlbs_vmid,
+	.submit_ib = amdgpu_amdkfd_submit_ib,
 };
 
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
@@ -704,14 +720,7 @@ static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
 
 	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
-	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
-}
-
-static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
-{
-	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
-
-	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__PASID_MASK;
 }
 
 static int kgd_address_watch_disable(struct kgd_dev *kgd)
@@ -775,8 +784,6 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
 	const union amdgpu_firmware_header *hdr;
 
-	BUG_ON(kgd == NULL);
-
 	switch (type) {
 	case KGD_ENGINE_PFP:
 		hdr = (const union amdgpu_firmware_header *)
@@ -828,3 +835,51 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 	/* Only 12 bit in use*/
 	return hdr->common.ucode_version;
 }
+
+static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t page_table_base)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("trying to set page table base for wrong VMID\n");
+		return;
+	}
+	WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vmid - 8, page_table_base);
+}
+
+static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	int vmid;
+	unsigned int tmp;
+
+	for (vmid = 0; vmid < 16; vmid++) {
+		if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid))
+			continue;
+
+		tmp = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+		if ((tmp & ATC_VMID0_PASID_MAPPING__VALID_MASK) &&
+			(tmp & ATC_VMID0_PASID_MAPPING__PASID_MASK) == pasid) {
+			WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+			RREG32(mmVM_INVALIDATE_RESPONSE);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
+		pr_err("non kfd vmid %d\n", vmid);
+		return -EINVAL;
+	}
+
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+	RREG32(mmVM_INVALIDATE_RESPONSE);
+	return 0;
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c

@@ -0,0 +1,1577 @@
+/*
+ * Copyright 2014-2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#define pr_fmt(fmt) "kfd2kgd: " fmt
+
+#include <linux/list.h>
+#include <drm/drmP.h>
+#include "amdgpu_object.h"
+#include "amdgpu_vm.h"
+#include "amdgpu_amdkfd.h"
+
+/* Special VM and GART address alignment needed for VI pre-Fiji due to
+ * a HW bug.
+ */
+#define VI_BO_SIZE_ALIGN (0x8000)
+
+/* Impose limit on how much memory KFD can use */
+static struct {
+	uint64_t max_system_mem_limit;
+	int64_t system_mem_used;
+	spinlock_t mem_limit_lock;
+} kfd_mem_limit;
+
+/* Struct used for amdgpu_amdkfd_bo_validate */
+struct amdgpu_vm_parser {
+	uint32_t        domain;
+	bool            wait;
+};
+
+static const char * const domain_bit_to_string[] = {
+		"CPU",
+		"GTT",
+		"VRAM",
+		"GDS",
+		"GWS",
+		"OA"
+};
+
+#define domain_string(domain) domain_bit_to_string[ffs(domain)-1]
+
+
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static bool check_if_add_bo_to_vm(struct amdgpu_vm *avm,
+		struct kgd_mem *mem)
+{
+	struct kfd_bo_va_list *entry;
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list)
+		if (entry->bo_va->base.vm == avm)
+			return false;
+
+	return true;
+}
+
+/* Set memory usage limits. Current, limits are
+ *  System (kernel) memory - 3/8th System RAM
+ */
+void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
+{
+	struct sysinfo si;
+	uint64_t mem;
+
+	si_meminfo(&si);
+	mem = si.totalram - si.totalhigh;
+	mem *= si.mem_unit;
+
+	spin_lock_init(&kfd_mem_limit.mem_limit_lock);
+	kfd_mem_limit.max_system_mem_limit = (mem >> 1) - (mem >> 3);
+	pr_debug("Kernel memory limit %lluM\n",
+		(kfd_mem_limit.max_system_mem_limit >> 20));
+}
+
+static int amdgpu_amdkfd_reserve_system_mem_limit(struct amdgpu_device *adev,
+					      uint64_t size, u32 domain)
+{
+	size_t acc_size;
+	int ret = 0;
+
+	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
+				       sizeof(struct amdgpu_bo));
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+	if (domain == AMDGPU_GEM_DOMAIN_GTT) {
+		if (kfd_mem_limit.system_mem_used + (acc_size + size) >
+			kfd_mem_limit.max_system_mem_limit) {
+			ret = -ENOMEM;
+			goto err_no_mem;
+		}
+		kfd_mem_limit.system_mem_used += (acc_size + size);
+	}
+err_no_mem:
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+	return ret;
+}
+
+static void unreserve_system_mem_limit(struct amdgpu_device *adev,
+				       uint64_t size, u32 domain)
+{
+	size_t acc_size;
+
+	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
+				       sizeof(struct amdgpu_bo));
+
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+	if (domain == AMDGPU_GEM_DOMAIN_GTT)
+		kfd_mem_limit.system_mem_used -= (acc_size + size);
+	WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
+		  "kfd system memory accounting unbalanced");
+
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+}
+
+void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo)
+{
+	spin_lock(&kfd_mem_limit.mem_limit_lock);
+
+	if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT) {
+		kfd_mem_limit.system_mem_used -=
+			(bo->tbo.acc_size + amdgpu_bo_size(bo));
+	}
+	WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
+		  "kfd system memory accounting unbalanced");
+
+	spin_unlock(&kfd_mem_limit.mem_limit_lock);
+}
+
+
+/* amdgpu_amdkfd_remove_eviction_fence - Removes eviction fence(s) from BO's
+ *  reservation object.
+ *
+ * @bo: [IN] Remove eviction fence(s) from this BO
+ * @ef: [IN] If ef is specified, then this eviction fence is removed if it
+ *  is present in the shared list.
+ * @ef_list: [OUT] Returns list of eviction fences. These fences are removed
+ *  from BO's reservation object shared list.
+ * @ef_count: [OUT] Number of fences in ef_list.
+ *
+ * NOTE: If called with ef_list, then amdgpu_amdkfd_add_eviction_fence must be
+ *  called to restore the eviction fences and to avoid memory leak. This is
+ *  useful for shared BOs.
+ * NOTE: Must be called with BO reserved i.e. bo->tbo.resv->lock held.
+ */
+static int amdgpu_amdkfd_remove_eviction_fence(struct amdgpu_bo *bo,
+					struct amdgpu_amdkfd_fence *ef,
+					struct amdgpu_amdkfd_fence ***ef_list,
+					unsigned int *ef_count)
+{
+	struct reservation_object_list *fobj;
+	struct reservation_object *resv;
+	unsigned int i = 0, j = 0, k = 0, shared_count;
+	unsigned int count = 0;
+	struct amdgpu_amdkfd_fence **fence_list;
+
+	if (!ef && !ef_list)
+		return -EINVAL;
+
+	if (ef_list) {
+		*ef_list = NULL;
+		*ef_count = 0;
+	}
+
+	resv = bo->tbo.resv;
+	fobj = reservation_object_get_list(resv);
+
+	if (!fobj)
+		return 0;
+
+	preempt_disable();
+	write_seqcount_begin(&resv->seq);
+
+	/* Go through all the shared fences in the resevation object. If
+	 * ef is specified and it exists in the list, remove it and reduce the
+	 * count. If ef is not specified, then get the count of eviction fences
+	 * present.
+	 */
+	shared_count = fobj->shared_count;
+	for (i = 0; i < shared_count; ++i) {
+		struct dma_fence *f;
+
+		f = rcu_dereference_protected(fobj->shared[i],
+					      reservation_object_held(resv));
+
+		if (ef) {
+			if (f->context == ef->base.context) {
+				dma_fence_put(f);
+				fobj->shared_count--;
+			} else {
+				RCU_INIT_POINTER(fobj->shared[j++], f);
+			}
+		} else if (to_amdgpu_amdkfd_fence(f))
+			count++;
+	}
+	write_seqcount_end(&resv->seq);
+	preempt_enable();
+
+	if (ef || !count)
+		return 0;
+
+	/* Alloc memory for count number of eviction fence pointers. Fill the
+	 * ef_list array and ef_count
+	 */
+	fence_list = kcalloc(count, sizeof(struct amdgpu_amdkfd_fence *),
+			     GFP_KERNEL);
+	if (!fence_list)
+		return -ENOMEM;
+
+	preempt_disable();
+	write_seqcount_begin(&resv->seq);
+
+	j = 0;
+	for (i = 0; i < shared_count; ++i) {
+		struct dma_fence *f;
+		struct amdgpu_amdkfd_fence *efence;
+
+		f = rcu_dereference_protected(fobj->shared[i],
+			reservation_object_held(resv));
+
+		efence = to_amdgpu_amdkfd_fence(f);
+		if (efence) {
+			fence_list[k++] = efence;
+			fobj->shared_count--;
+		} else {
+			RCU_INIT_POINTER(fobj->shared[j++], f);
+		}
+	}
+
+	write_seqcount_end(&resv->seq);
+	preempt_enable();
+
+	*ef_list = fence_list;
+	*ef_count = k;
+
+	return 0;
+}
+
+/* amdgpu_amdkfd_add_eviction_fence - Adds eviction fence(s) back into BO's
+ *  reservation object.
+ *
+ * @bo: [IN] Add eviction fences to this BO
+ * @ef_list: [IN] List of eviction fences to be added
+ * @ef_count: [IN] Number of fences in ef_list.
+ *
+ * NOTE: Must call amdgpu_amdkfd_remove_eviction_fence before calling this
+ *  function.
+ */
+static void amdgpu_amdkfd_add_eviction_fence(struct amdgpu_bo *bo,
+				struct amdgpu_amdkfd_fence **ef_list,
+				unsigned int ef_count)
+{
+	int i;
+
+	if (!ef_list || !ef_count)
+		return;
+
+	for (i = 0; i < ef_count; i++) {
+		amdgpu_bo_fence(bo, &ef_list[i]->base, true);
+		/* Re-adding the fence takes an additional reference. Drop that
+		 * reference.
+		 */
+		dma_fence_put(&ef_list[i]->base);
+	}
+
+	kfree(ef_list);
+}
+
+static int amdgpu_amdkfd_bo_validate(struct amdgpu_bo *bo, uint32_t domain,
+				     bool wait)
+{
+	struct ttm_operation_ctx ctx = { false, false };
+	int ret;
+
+	if (WARN(amdgpu_ttm_tt_get_usermm(bo->tbo.ttm),
+		 "Called with userptr BO"))
+		return -EINVAL;
+
+	amdgpu_ttm_placement_from_domain(bo, domain);
+
+	ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
+	if (ret)
+		goto validate_fail;
+	if (wait) {
+		struct amdgpu_amdkfd_fence **ef_list;
+		unsigned int ef_count;
+
+		ret = amdgpu_amdkfd_remove_eviction_fence(bo, NULL, &ef_list,
+							  &ef_count);
+		if (ret)
+			goto validate_fail;
+
+		ttm_bo_wait(&bo->tbo, false, false);
+		amdgpu_amdkfd_add_eviction_fence(bo, ef_list, ef_count);
+	}
+
+validate_fail:
+	return ret;
+}
+
+static int amdgpu_amdkfd_validate(void *param, struct amdgpu_bo *bo)
+{
+	struct amdgpu_vm_parser *p = param;
+
+	return amdgpu_amdkfd_bo_validate(bo, p->domain, p->wait);
+}
+
+/* vm_validate_pt_pd_bos - Validate page table and directory BOs
+ *
+ * Page directories are not updated here because huge page handling
+ * during page table updates can invalidate page directory entries
+ * again. Page directories are only updated after updating page
+ * tables.
+ */
+static int vm_validate_pt_pd_bos(struct amdgpu_vm *vm)
+{
+	struct amdgpu_bo *pd = vm->root.base.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+	struct amdgpu_vm_parser param;
+	uint64_t addr, flags = AMDGPU_PTE_VALID;
+	int ret;
+
+	param.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	param.wait = false;
+
+	ret = amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_amdkfd_validate,
+					&param);
+	if (ret) {
+		pr_err("amdgpu: failed to validate PT BOs\n");
+		return ret;
+	}
+
+	ret = amdgpu_amdkfd_validate(&param, pd);
+	if (ret) {
+		pr_err("amdgpu: failed to validate PD\n");
+		return ret;
+	}
+
+	addr = amdgpu_bo_gpu_offset(vm->root.base.bo);
+	amdgpu_gmc_get_vm_pde(adev, -1, &addr, &flags);
+	vm->pd_phys_addr = addr;
+
+	if (vm->use_cpu_for_update) {
+		ret = amdgpu_bo_kmap(pd, NULL);
+		if (ret) {
+			pr_err("amdgpu: failed to kmap PD, ret=%d\n", ret);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int sync_vm_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+			 struct dma_fence *f)
+{
+	int ret = amdgpu_sync_fence(adev, sync, f, false);
+
+	/* Sync objects can't handle multiple GPUs (contexts) updating
+	 * sync->last_vm_update. Fortunately we don't need it for
+	 * KFD's purposes, so we can just drop that fence.
+	 */
+	if (sync->last_vm_update) {
+		dma_fence_put(sync->last_vm_update);
+		sync->last_vm_update = NULL;
+	}
+
+	return ret;
+}
+
+static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo *pd = vm->root.base.bo;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
+	int ret;
+
+	ret = amdgpu_vm_update_directories(adev, vm);
+	if (ret)
+		return ret;
+
+	return sync_vm_fence(adev, sync, vm->last_update);
+}
+
+/* add_bo_to_vm - Add a BO to a VM
+ *
+ * Everything that needs to bo done only once when a BO is first added
+ * to a VM. It can later be mapped and unmapped many times without
+ * repeating these steps.
+ *
+ * 1. Allocate and initialize BO VA entry data structure
+ * 2. Add BO to the VM
+ * 3. Determine ASIC-specific PTE flags
+ * 4. Alloc page tables and directories if needed
+ * 4a.  Validate new page tables and directories
+ */
+static int add_bo_to_vm(struct amdgpu_device *adev, struct kgd_mem *mem,
+		struct amdgpu_vm *vm, bool is_aql,
+		struct kfd_bo_va_list **p_bo_va_entry)
+{
+	int ret;
+	struct kfd_bo_va_list *bo_va_entry;
+	struct amdgpu_bo *pd = vm->root.base.bo;
+	struct amdgpu_bo *bo = mem->bo;
+	uint64_t va = mem->va;
+	struct list_head *list_bo_va = &mem->bo_va_list;
+	unsigned long bo_size = bo->tbo.mem.size;
+
+	if (!va) {
+		pr_err("Invalid VA when adding BO to VM\n");
+		return -EINVAL;
+	}
+
+	if (is_aql)
+		va += bo_size;
+
+	bo_va_entry = kzalloc(sizeof(*bo_va_entry), GFP_KERNEL);
+	if (!bo_va_entry)
+		return -ENOMEM;
+
+	pr_debug("\t add VA 0x%llx - 0x%llx to vm %p\n", va,
+			va + bo_size, vm);
+
+	/* Add BO to VM internal data structures*/
+	bo_va_entry->bo_va = amdgpu_vm_bo_add(adev, vm, bo);
+	if (!bo_va_entry->bo_va) {
+		ret = -EINVAL;
+		pr_err("Failed to add BO object to VM. ret == %d\n",
+				ret);
+		goto err_vmadd;
+	}
+
+	bo_va_entry->va = va;
+	bo_va_entry->pte_flags = amdgpu_gmc_get_pte_flags(adev,
+							 mem->mapping_flags);
+	bo_va_entry->kgd_dev = (void *)adev;
+	list_add(&bo_va_entry->bo_list, list_bo_va);
+
+	if (p_bo_va_entry)
+		*p_bo_va_entry = bo_va_entry;
+
+	/* Allocate new page tables if needed and validate
+	 * them. Clearing of new page tables and validate need to wait
+	 * on move fences. We don't want that to trigger the eviction
+	 * fence, so remove it temporarily.
+	 */
+	amdgpu_amdkfd_remove_eviction_fence(pd,
+					vm->process_info->eviction_fence,
+					NULL, NULL);
+
+	ret = amdgpu_vm_alloc_pts(adev, vm, va, amdgpu_bo_size(bo));
+	if (ret) {
+		pr_err("Failed to allocate pts, err=%d\n", ret);
+		goto err_alloc_pts;
+	}
+
+	ret = vm_validate_pt_pd_bos(vm);
+	if (ret) {
+		pr_err("validate_pt_pd_bos() failed\n");
+		goto err_alloc_pts;
+	}
+
+	/* Add the eviction fence back */
+	amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
+
+	return 0;
+
+err_alloc_pts:
+	amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
+	amdgpu_vm_bo_rmv(adev, bo_va_entry->bo_va);
+	list_del(&bo_va_entry->bo_list);
+err_vmadd:
+	kfree(bo_va_entry);
+	return ret;
+}
+
+static void remove_bo_from_vm(struct amdgpu_device *adev,
+		struct kfd_bo_va_list *entry, unsigned long size)
+{
+	pr_debug("\t remove VA 0x%llx - 0x%llx in entry %p\n",
+			entry->va,
+			entry->va + size, entry);
+	amdgpu_vm_bo_rmv(adev, entry->bo_va);
+	list_del(&entry->bo_list);
+	kfree(entry);
+}
+
+static void add_kgd_mem_to_kfd_bo_list(struct kgd_mem *mem,
+				struct amdkfd_process_info *process_info)
+{
+	struct ttm_validate_buffer *entry = &mem->validate_list;
+	struct amdgpu_bo *bo = mem->bo;
+
+	INIT_LIST_HEAD(&entry->head);
+	entry->shared = true;
+	entry->bo = &bo->tbo;
+	mutex_lock(&process_info->lock);
+	list_add_tail(&entry->head, &process_info->kfd_bo_list);
+	mutex_unlock(&process_info->lock);
+}
+
+/* Reserving a BO and its page table BOs must happen atomically to
+ * avoid deadlocks. Some operations update multiple VMs at once. Track
+ * all the reservation info in a context structure. Optionally a sync
+ * object can track VM updates.
+ */
+struct bo_vm_reservation_context {
+	struct amdgpu_bo_list_entry kfd_bo; /* BO list entry for the KFD BO */
+	unsigned int n_vms;		    /* Number of VMs reserved	    */
+	struct amdgpu_bo_list_entry *vm_pd; /* Array of VM BO list entries  */
+	struct ww_acquire_ctx ticket;	    /* Reservation ticket	    */
+	struct list_head list, duplicates;  /* BO lists			    */
+	struct amdgpu_sync *sync;	    /* Pointer to sync object	    */
+	bool reserved;			    /* Whether BOs are reserved	    */
+};
+
+enum bo_vm_match {
+	BO_VM_NOT_MAPPED = 0,	/* Match VMs where a BO is not mapped */
+	BO_VM_MAPPED,		/* Match VMs where a BO is mapped     */
+	BO_VM_ALL,		/* Match all VMs a BO was added to    */
+};
+
+/**
+ * reserve_bo_and_vm - reserve a BO and a VM unconditionally.
+ * @mem: KFD BO structure.
+ * @vm: the VM to reserve.
+ * @ctx: the struct that will be used in unreserve_bo_and_vms().
+ */
+static int reserve_bo_and_vm(struct kgd_mem *mem,
+			      struct amdgpu_vm *vm,
+			      struct bo_vm_reservation_context *ctx)
+{
+	struct amdgpu_bo *bo = mem->bo;
+	int ret;
+
+	WARN_ON(!vm);
+
+	ctx->reserved = false;
+	ctx->n_vms = 1;
+	ctx->sync = &mem->sync;
+
+	INIT_LIST_HEAD(&ctx->list);
+	INIT_LIST_HEAD(&ctx->duplicates);
+
+	ctx->vm_pd = kcalloc(ctx->n_vms, sizeof(*ctx->vm_pd), GFP_KERNEL);
+	if (!ctx->vm_pd)
+		return -ENOMEM;
+
+	ctx->kfd_bo.robj = bo;
+	ctx->kfd_bo.priority = 0;
+	ctx->kfd_bo.tv.bo = &bo->tbo;
+	ctx->kfd_bo.tv.shared = true;
+	ctx->kfd_bo.user_pages = NULL;
+	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
+
+	amdgpu_vm_get_pd_bo(vm, &ctx->list, &ctx->vm_pd[0]);
+
+	ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list,
+				     false, &ctx->duplicates);
+	if (!ret)
+		ctx->reserved = true;
+	else {
+		pr_err("Failed to reserve buffers in ttm\n");
+		kfree(ctx->vm_pd);
+		ctx->vm_pd = NULL;
+	}
+
+	return ret;
+}
+
+/**
+ * reserve_bo_and_cond_vms - reserve a BO and some VMs conditionally
+ * @mem: KFD BO structure.
+ * @vm: the VM to reserve. If NULL, then all VMs associated with the BO
+ * is used. Otherwise, a single VM associated with the BO.
+ * @map_type: the mapping status that will be used to filter the VMs.
+ * @ctx: the struct that will be used in unreserve_bo_and_vms().
+ *
+ * Returns 0 for success, negative for failure.
+ */
+static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
+				struct amdgpu_vm *vm, enum bo_vm_match map_type,
+				struct bo_vm_reservation_context *ctx)
+{
+	struct amdgpu_bo *bo = mem->bo;
+	struct kfd_bo_va_list *entry;
+	unsigned int i;
+	int ret;
+
+	ctx->reserved = false;
+	ctx->n_vms = 0;
+	ctx->vm_pd = NULL;
+	ctx->sync = &mem->sync;
+
+	INIT_LIST_HEAD(&ctx->list);
+	INIT_LIST_HEAD(&ctx->duplicates);
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if ((vm && vm != entry->bo_va->base.vm) ||
+			(entry->is_mapped != map_type
+			&& map_type != BO_VM_ALL))
+			continue;
+
+		ctx->n_vms++;
+	}
+
+	if (ctx->n_vms != 0) {
+		ctx->vm_pd = kcalloc(ctx->n_vms, sizeof(*ctx->vm_pd),
+				     GFP_KERNEL);
+		if (!ctx->vm_pd)
+			return -ENOMEM;
+	}
+
+	ctx->kfd_bo.robj = bo;
+	ctx->kfd_bo.priority = 0;
+	ctx->kfd_bo.tv.bo = &bo->tbo;
+	ctx->kfd_bo.tv.shared = true;
+	ctx->kfd_bo.user_pages = NULL;
+	list_add(&ctx->kfd_bo.tv.head, &ctx->list);
+
+	i = 0;
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if ((vm && vm != entry->bo_va->base.vm) ||
+			(entry->is_mapped != map_type
+			&& map_type != BO_VM_ALL))
+			continue;
+
+		amdgpu_vm_get_pd_bo(entry->bo_va->base.vm, &ctx->list,
+				&ctx->vm_pd[i]);
+		i++;
+	}
+
+	ret = ttm_eu_reserve_buffers(&ctx->ticket, &ctx->list,
+				     false, &ctx->duplicates);
+	if (!ret)
+		ctx->reserved = true;
+	else
+		pr_err("Failed to reserve buffers in ttm.\n");
+
+	if (ret) {
+		kfree(ctx->vm_pd);
+		ctx->vm_pd = NULL;
+	}
+
+	return ret;
+}
+
+/**
+ * unreserve_bo_and_vms - Unreserve BO and VMs from a reservation context
+ * @ctx: Reservation context to unreserve
+ * @wait: Optionally wait for a sync object representing pending VM updates
+ * @intr: Whether the wait is interruptible
+ *
+ * Also frees any resources allocated in
+ * reserve_bo_and_(cond_)vm(s). Returns the status from
+ * amdgpu_sync_wait.
+ */
+static int unreserve_bo_and_vms(struct bo_vm_reservation_context *ctx,
+				 bool wait, bool intr)
+{
+	int ret = 0;
+
+	if (wait)
+		ret = amdgpu_sync_wait(ctx->sync, intr);
+
+	if (ctx->reserved)
+		ttm_eu_backoff_reservation(&ctx->ticket, &ctx->list);
+	kfree(ctx->vm_pd);
+
+	ctx->sync = NULL;
+
+	ctx->reserved = false;
+	ctx->vm_pd = NULL;
+
+	return ret;
+}
+
+static int unmap_bo_from_gpuvm(struct amdgpu_device *adev,
+				struct kfd_bo_va_list *entry,
+				struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo_va *bo_va = entry->bo_va;
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo *pd = vm->root.base.bo;
+
+	/* Remove eviction fence from PD (and thereby from PTs too as
+	 * they share the resv. object). Otherwise during PT update
+	 * job (see amdgpu_vm_bo_update_mapping), eviction fence would
+	 * get added to job->sync object and job execution would
+	 * trigger the eviction fence.
+	 */
+	amdgpu_amdkfd_remove_eviction_fence(pd,
+					    vm->process_info->eviction_fence,
+					    NULL, NULL);
+	amdgpu_vm_bo_unmap(adev, bo_va, entry->va);
+
+	amdgpu_vm_clear_freed(adev, vm, &bo_va->last_pt_update);
+
+	/* Add the eviction fence back */
+	amdgpu_bo_fence(pd, &vm->process_info->eviction_fence->base, true);
+
+	sync_vm_fence(adev, sync, bo_va->last_pt_update);
+
+	return 0;
+}
+
+static int update_gpuvm_pte(struct amdgpu_device *adev,
+		struct kfd_bo_va_list *entry,
+		struct amdgpu_sync *sync)
+{
+	int ret;
+	struct amdgpu_vm *vm;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+
+	bo_va = entry->bo_va;
+	vm = bo_va->base.vm;
+	bo = bo_va->base.bo;
+
+	/* Update the page tables  */
+	ret = amdgpu_vm_bo_update(adev, bo_va, false);
+	if (ret) {
+		pr_err("amdgpu_vm_bo_update failed\n");
+		return ret;
+	}
+
+	return sync_vm_fence(adev, sync, bo_va->last_pt_update);
+}
+
+static int map_bo_to_gpuvm(struct amdgpu_device *adev,
+		struct kfd_bo_va_list *entry, struct amdgpu_sync *sync)
+{
+	int ret;
+
+	/* Set virtual address for the allocation */
+	ret = amdgpu_vm_bo_map(adev, entry->bo_va, entry->va, 0,
+			       amdgpu_bo_size(entry->bo_va->base.bo),
+			       entry->pte_flags);
+	if (ret) {
+		pr_err("Failed to map VA 0x%llx in vm. ret %d\n",
+				entry->va, ret);
+		return ret;
+	}
+
+	ret = update_gpuvm_pte(adev, entry, sync);
+	if (ret) {
+		pr_err("update_gpuvm_pte() failed\n");
+		goto update_gpuvm_pte_failed;
+	}
+
+	return 0;
+
+update_gpuvm_pte_failed:
+	unmap_bo_from_gpuvm(adev, entry, sync);
+	return ret;
+}
+
+static int process_validate_vms(struct amdkfd_process_info *process_info)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		ret = vm_validate_pt_pd_bos(peer_vm);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int process_update_pds(struct amdkfd_process_info *process_info,
+			      struct amdgpu_sync *sync)
+{
+	struct amdgpu_vm *peer_vm;
+	int ret;
+
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		ret = vm_update_pds(peer_vm, sync);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int init_kfd_vm(struct amdgpu_vm *vm, void **process_info,
+		       struct dma_fence **ef)
+{
+	struct amdkfd_process_info *info = NULL;
+	int ret;
+
+	if (!*process_info) {
+		info = kzalloc(sizeof(*info), GFP_KERNEL);
+		if (!info)
+			return -ENOMEM;
+
+		mutex_init(&info->lock);
+		INIT_LIST_HEAD(&info->vm_list_head);
+		INIT_LIST_HEAD(&info->kfd_bo_list);
+
+		info->eviction_fence =
+			amdgpu_amdkfd_fence_create(dma_fence_context_alloc(1),
+						   current->mm);
+		if (!info->eviction_fence) {
+			pr_err("Failed to create eviction fence\n");
+			ret = -ENOMEM;
+			goto create_evict_fence_fail;
+		}
+
+		*process_info = info;
+		*ef = dma_fence_get(&info->eviction_fence->base);
+	}
+
+	vm->process_info = *process_info;
+
+	/* Validate page directory and attach eviction fence */
+	ret = amdgpu_bo_reserve(vm->root.base.bo, true);
+	if (ret)
+		goto reserve_pd_fail;
+	ret = vm_validate_pt_pd_bos(vm);
+	if (ret) {
+		pr_err("validate_pt_pd_bos() failed\n");
+		goto validate_pd_fail;
+	}
+	ret = ttm_bo_wait(&vm->root.base.bo->tbo, false, false);
+	if (ret)
+		goto wait_pd_fail;
+	amdgpu_bo_fence(vm->root.base.bo,
+			&vm->process_info->eviction_fence->base, true);
+	amdgpu_bo_unreserve(vm->root.base.bo);
+
+	/* Update process info */
+	mutex_lock(&vm->process_info->lock);
+	list_add_tail(&vm->vm_list_node,
+			&(vm->process_info->vm_list_head));
+	vm->process_info->n_vms++;
+	mutex_unlock(&vm->process_info->lock);
+
+	return 0;
+
+wait_pd_fail:
+validate_pd_fail:
+	amdgpu_bo_unreserve(vm->root.base.bo);
+reserve_pd_fail:
+	vm->process_info = NULL;
+	if (info) {
+		/* Two fence references: one in info and one in *ef */
+		dma_fence_put(&info->eviction_fence->base);
+		dma_fence_put(*ef);
+		*ef = NULL;
+		*process_info = NULL;
+create_evict_fence_fail:
+		mutex_destroy(&info->lock);
+		kfree(info);
+	}
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_create_process_vm(struct kgd_dev *kgd, void **vm,
+					  void **process_info,
+					  struct dma_fence **ef)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *new_vm;
+	int ret;
+
+	new_vm = kzalloc(sizeof(*new_vm), GFP_KERNEL);
+	if (!new_vm)
+		return -ENOMEM;
+
+	/* Initialize AMDGPU part of the VM */
+	ret = amdgpu_vm_init(adev, new_vm, AMDGPU_VM_CONTEXT_COMPUTE, 0);
+	if (ret) {
+		pr_err("Failed init vm ret %d\n", ret);
+		goto amdgpu_vm_init_fail;
+	}
+
+	/* Initialize KFD part of the VM and process info */
+	ret = init_kfd_vm(new_vm, process_info, ef);
+	if (ret)
+		goto init_kfd_vm_fail;
+
+	*vm = (void *) new_vm;
+
+	return 0;
+
+init_kfd_vm_fail:
+	amdgpu_vm_fini(adev, new_vm);
+amdgpu_vm_init_fail:
+	kfree(new_vm);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd,
+					   struct file *filp,
+					   void **vm, void **process_info,
+					   struct dma_fence **ef)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct drm_file *drm_priv = filp->private_data;
+	struct amdgpu_fpriv *drv_priv = drm_priv->driver_priv;
+	struct amdgpu_vm *avm = &drv_priv->vm;
+	int ret;
+
+	/* Already a compute VM? */
+	if (avm->process_info)
+		return -EINVAL;
+
+	/* Convert VM into a compute VM */
+	ret = amdgpu_vm_make_compute(adev, avm);
+	if (ret)
+		return ret;
+
+	/* Initialize KFD part of the VM and process info */
+	ret = init_kfd_vm(avm, process_info, ef);
+	if (ret)
+		return ret;
+
+	*vm = (void *)avm;
+
+	return 0;
+}
+
+void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm)
+{
+	struct amdkfd_process_info *process_info = vm->process_info;
+	struct amdgpu_bo *pd = vm->root.base.bo;
+
+	if (!process_info)
+		return;
+
+	/* Release eviction fence from PD */
+	amdgpu_bo_reserve(pd, false);
+	amdgpu_bo_fence(pd, NULL, false);
+	amdgpu_bo_unreserve(pd);
+
+	/* Update process info */
+	mutex_lock(&process_info->lock);
+	process_info->n_vms--;
+	list_del(&vm->vm_list_node);
+	mutex_unlock(&process_info->lock);
+
+	/* Release per-process resources when last compute VM is destroyed */
+	if (!process_info->n_vms) {
+		WARN_ON(!list_empty(&process_info->kfd_bo_list));
+
+		dma_fence_put(&process_info->eviction_fence->base);
+		mutex_destroy(&process_info->lock);
+		kfree(process_info);
+	}
+}
+
+void amdgpu_amdkfd_gpuvm_destroy_process_vm(struct kgd_dev *kgd, void *vm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+
+	if (WARN_ON(!kgd || !vm))
+		return;
+
+	pr_debug("Destroying process vm %p\n", vm);
+
+	/* Release the VM context */
+	amdgpu_vm_fini(adev, avm);
+	kfree(vm);
+}
+
+uint32_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *vm)
+{
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+
+	return avm->pd_phys_addr >> AMDGPU_GPU_PAGE_SHIFT;
+}
+
+int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
+		struct kgd_dev *kgd, uint64_t va, uint64_t size,
+		void *vm, struct kgd_mem **mem,
+		uint64_t *offset, uint32_t flags)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+	struct amdgpu_bo *bo;
+	int byte_align;
+	u32 alloc_domain;
+	u64 alloc_flags;
+	uint32_t mapping_flags;
+	int ret;
+
+	/*
+	 * Check on which domain to allocate BO
+	 */
+	if (flags & ALLOC_MEM_FLAGS_VRAM) {
+		alloc_domain = AMDGPU_GEM_DOMAIN_VRAM;
+		alloc_flags = AMDGPU_GEM_CREATE_VRAM_CLEARED;
+		alloc_flags |= (flags & ALLOC_MEM_FLAGS_PUBLIC) ?
+			AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED :
+			AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+	} else if (flags & ALLOC_MEM_FLAGS_GTT) {
+		alloc_domain = AMDGPU_GEM_DOMAIN_GTT;
+		alloc_flags = 0;
+	} else {
+		return -EINVAL;
+	}
+
+	*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+	if (!*mem)
+		return -ENOMEM;
+	INIT_LIST_HEAD(&(*mem)->bo_va_list);
+	mutex_init(&(*mem)->lock);
+	(*mem)->aql_queue = !!(flags & ALLOC_MEM_FLAGS_AQL_QUEUE_MEM);
+
+	/* Workaround for AQL queue wraparound bug. Map the same
+	 * memory twice. That means we only actually allocate half
+	 * the memory.
+	 */
+	if ((*mem)->aql_queue)
+		size = size >> 1;
+
+	/* Workaround for TLB bug on older VI chips */
+	byte_align = (adev->family == AMDGPU_FAMILY_VI &&
+			adev->asic_type != CHIP_FIJI &&
+			adev->asic_type != CHIP_POLARIS10 &&
+			adev->asic_type != CHIP_POLARIS11) ?
+			VI_BO_SIZE_ALIGN : 1;
+
+	mapping_flags = AMDGPU_VM_PAGE_READABLE;
+	if (flags & ALLOC_MEM_FLAGS_WRITABLE)
+		mapping_flags |= AMDGPU_VM_PAGE_WRITEABLE;
+	if (flags & ALLOC_MEM_FLAGS_EXECUTABLE)
+		mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE;
+	if (flags & ALLOC_MEM_FLAGS_COHERENT)
+		mapping_flags |= AMDGPU_VM_MTYPE_UC;
+	else
+		mapping_flags |= AMDGPU_VM_MTYPE_NC;
+	(*mem)->mapping_flags = mapping_flags;
+
+	amdgpu_sync_create(&(*mem)->sync);
+
+	ret = amdgpu_amdkfd_reserve_system_mem_limit(adev, size, alloc_domain);
+	if (ret) {
+		pr_debug("Insufficient system memory\n");
+		goto err_reserve_system_mem;
+	}
+
+	pr_debug("\tcreate BO VA 0x%llx size 0x%llx domain %s\n",
+			va, size, domain_string(alloc_domain));
+
+	ret = amdgpu_bo_create(adev, size, byte_align,
+				alloc_domain, alloc_flags, ttm_bo_type_device, NULL, &bo);
+	if (ret) {
+		pr_debug("Failed to create BO on domain %s. ret %d\n",
+				domain_string(alloc_domain), ret);
+		goto err_bo_create;
+	}
+	bo->kfd_bo = *mem;
+	(*mem)->bo = bo;
+
+	(*mem)->va = va;
+	(*mem)->domain = alloc_domain;
+	(*mem)->mapped_to_gpu_memory = 0;
+	(*mem)->process_info = avm->process_info;
+	add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info);
+
+	if (offset)
+		*offset = amdgpu_bo_mmap_offset(bo);
+
+	return 0;
+
+err_bo_create:
+	unreserve_system_mem_limit(adev, size, alloc_domain);
+err_reserve_system_mem:
+	mutex_destroy(&(*mem)->lock);
+	kfree(*mem);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem)
+{
+	struct amdkfd_process_info *process_info = mem->process_info;
+	unsigned long bo_size = mem->bo->tbo.mem.size;
+	struct kfd_bo_va_list *entry, *tmp;
+	struct bo_vm_reservation_context ctx;
+	struct ttm_validate_buffer *bo_list_entry;
+	int ret;
+
+	mutex_lock(&mem->lock);
+
+	if (mem->mapped_to_gpu_memory > 0) {
+		pr_debug("BO VA 0x%llx size 0x%lx is still mapped.\n",
+				mem->va, bo_size);
+		mutex_unlock(&mem->lock);
+		return -EBUSY;
+	}
+
+	mutex_unlock(&mem->lock);
+	/* lock is not needed after this, since mem is unused and will
+	 * be freed anyway
+	 */
+
+	/* Make sure restore workers don't access the BO any more */
+	bo_list_entry = &mem->validate_list;
+	mutex_lock(&process_info->lock);
+	list_del(&bo_list_entry->head);
+	mutex_unlock(&process_info->lock);
+
+	ret = reserve_bo_and_cond_vms(mem, NULL, BO_VM_ALL, &ctx);
+	if (unlikely(ret))
+		return ret;
+
+	/* The eviction fence should be removed by the last unmap.
+	 * TODO: Log an error condition if the bo still has the eviction fence
+	 * attached
+	 */
+	amdgpu_amdkfd_remove_eviction_fence(mem->bo,
+					process_info->eviction_fence,
+					NULL, NULL);
+	pr_debug("Release VA 0x%llx - 0x%llx\n", mem->va,
+		mem->va + bo_size * (1 + mem->aql_queue));
+
+	/* Remove from VM internal data structures */
+	list_for_each_entry_safe(entry, tmp, &mem->bo_va_list, bo_list)
+		remove_bo_from_vm((struct amdgpu_device *)entry->kgd_dev,
+				entry, bo_size);
+
+	ret = unreserve_bo_and_vms(&ctx, false, false);
+
+	/* Free the sync object */
+	amdgpu_sync_free(&mem->sync);
+
+	/* Free the BO*/
+	amdgpu_bo_unref(&mem->bo);
+	mutex_destroy(&mem->lock);
+	kfree(mem);
+
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+	int ret;
+	struct amdgpu_bo *bo;
+	uint32_t domain;
+	struct kfd_bo_va_list *entry;
+	struct bo_vm_reservation_context ctx;
+	struct kfd_bo_va_list *bo_va_entry = NULL;
+	struct kfd_bo_va_list *bo_va_entry_aql = NULL;
+	unsigned long bo_size;
+
+	/* Make sure restore is not running concurrently.
+	 */
+	mutex_lock(&mem->process_info->lock);
+
+	mutex_lock(&mem->lock);
+
+	bo = mem->bo;
+
+	if (!bo) {
+		pr_err("Invalid BO when mapping memory to GPU\n");
+		ret = -EINVAL;
+		goto out;
+	}
+
+	domain = mem->domain;
+	bo_size = bo->tbo.mem.size;
+
+	pr_debug("Map VA 0x%llx - 0x%llx to vm %p domain %s\n",
+			mem->va,
+			mem->va + bo_size * (1 + mem->aql_queue),
+			vm, domain_string(domain));
+
+	ret = reserve_bo_and_vm(mem, vm, &ctx);
+	if (unlikely(ret))
+		goto out;
+
+	if (check_if_add_bo_to_vm(avm, mem)) {
+		ret = add_bo_to_vm(adev, mem, avm, false,
+				&bo_va_entry);
+		if (ret)
+			goto add_bo_to_vm_failed;
+		if (mem->aql_queue) {
+			ret = add_bo_to_vm(adev, mem, avm,
+					true, &bo_va_entry_aql);
+			if (ret)
+				goto add_bo_to_vm_failed_aql;
+		}
+	} else {
+		ret = vm_validate_pt_pd_bos(avm);
+		if (unlikely(ret))
+			goto add_bo_to_vm_failed;
+	}
+
+	if (mem->mapped_to_gpu_memory == 0) {
+		/* Validate BO only once. The eviction fence gets added to BO
+		 * the first time it is mapped. Validate will wait for all
+		 * background evictions to complete.
+		 */
+		ret = amdgpu_amdkfd_bo_validate(bo, domain, true);
+		if (ret) {
+			pr_debug("Validate failed\n");
+			goto map_bo_to_gpuvm_failed;
+		}
+	}
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if (entry->bo_va->base.vm == vm && !entry->is_mapped) {
+			pr_debug("\t map VA 0x%llx - 0x%llx in entry %p\n",
+					entry->va, entry->va + bo_size,
+					entry);
+
+			ret = map_bo_to_gpuvm(adev, entry, ctx.sync);
+			if (ret) {
+				pr_err("Failed to map radeon bo to gpuvm\n");
+				goto map_bo_to_gpuvm_failed;
+			}
+
+			ret = vm_update_pds(vm, ctx.sync);
+			if (ret) {
+				pr_err("Failed to update page directories\n");
+				goto map_bo_to_gpuvm_failed;
+			}
+
+			entry->is_mapped = true;
+			mem->mapped_to_gpu_memory++;
+			pr_debug("\t INC mapping count %d\n",
+					mem->mapped_to_gpu_memory);
+		}
+	}
+
+	if (!amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) && !bo->pin_count)
+		amdgpu_bo_fence(bo,
+				&avm->process_info->eviction_fence->base,
+				true);
+	ret = unreserve_bo_and_vms(&ctx, false, false);
+
+	goto out;
+
+map_bo_to_gpuvm_failed:
+	if (bo_va_entry_aql)
+		remove_bo_from_vm(adev, bo_va_entry_aql, bo_size);
+add_bo_to_vm_failed_aql:
+	if (bo_va_entry)
+		remove_bo_from_vm(adev, bo_va_entry, bo_size);
+add_bo_to_vm_failed:
+	unreserve_bo_and_vms(&ctx, false, false);
+out:
+	mutex_unlock(&mem->process_info->lock);
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
+		struct kgd_dev *kgd, struct kgd_mem *mem, void *vm)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct amdkfd_process_info *process_info =
+		((struct amdgpu_vm *)vm)->process_info;
+	unsigned long bo_size = mem->bo->tbo.mem.size;
+	struct kfd_bo_va_list *entry;
+	struct bo_vm_reservation_context ctx;
+	int ret;
+
+	mutex_lock(&mem->lock);
+
+	ret = reserve_bo_and_cond_vms(mem, vm, BO_VM_MAPPED, &ctx);
+	if (unlikely(ret))
+		goto out;
+	/* If no VMs were reserved, it means the BO wasn't actually mapped */
+	if (ctx.n_vms == 0) {
+		ret = -EINVAL;
+		goto unreserve_out;
+	}
+
+	ret = vm_validate_pt_pd_bos((struct amdgpu_vm *)vm);
+	if (unlikely(ret))
+		goto unreserve_out;
+
+	pr_debug("Unmap VA 0x%llx - 0x%llx from vm %p\n",
+		mem->va,
+		mem->va + bo_size * (1 + mem->aql_queue),
+		vm);
+
+	list_for_each_entry(entry, &mem->bo_va_list, bo_list) {
+		if (entry->bo_va->base.vm == vm && entry->is_mapped) {
+			pr_debug("\t unmap VA 0x%llx - 0x%llx from entry %p\n",
+					entry->va,
+					entry->va + bo_size,
+					entry);
+
+			ret = unmap_bo_from_gpuvm(adev, entry, ctx.sync);
+			if (ret == 0) {
+				entry->is_mapped = false;
+			} else {
+				pr_err("failed to unmap VA 0x%llx\n",
+						mem->va);
+				goto unreserve_out;
+			}
+
+			mem->mapped_to_gpu_memory--;
+			pr_debug("\t DEC mapping count %d\n",
+					mem->mapped_to_gpu_memory);
+		}
+	}
+
+	/* If BO is unmapped from all VMs, unfence it. It can be evicted if
+	 * required.
+	 */
+	if (mem->mapped_to_gpu_memory == 0 &&
+	    !amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm) && !mem->bo->pin_count)
+		amdgpu_amdkfd_remove_eviction_fence(mem->bo,
+						process_info->eviction_fence,
+						    NULL, NULL);
+
+unreserve_out:
+	unreserve_bo_and_vms(&ctx, false, false);
+out:
+	mutex_unlock(&mem->lock);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_sync_memory(
+		struct kgd_dev *kgd, struct kgd_mem *mem, bool intr)
+{
+	struct amdgpu_sync sync;
+	int ret;
+
+	amdgpu_sync_create(&sync);
+
+	mutex_lock(&mem->lock);
+	amdgpu_sync_clone(&mem->sync, &sync);
+	mutex_unlock(&mem->lock);
+
+	ret = amdgpu_sync_wait(&sync, intr);
+	amdgpu_sync_free(&sync);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_dev *kgd,
+		struct kgd_mem *mem, void **kptr, uint64_t *size)
+{
+	int ret;
+	struct amdgpu_bo *bo = mem->bo;
+
+	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) {
+		pr_err("userptr can't be mapped to kernel\n");
+		return -EINVAL;
+	}
+
+	/* delete kgd_mem from kfd_bo_list to avoid re-validating
+	 * this BO in BO's restoring after eviction.
+	 */
+	mutex_lock(&mem->process_info->lock);
+
+	ret = amdgpu_bo_reserve(bo, true);
+	if (ret) {
+		pr_err("Failed to reserve bo. ret %d\n", ret);
+		goto bo_reserve_failed;
+	}
+
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT, NULL);
+	if (ret) {
+		pr_err("Failed to pin bo. ret %d\n", ret);
+		goto pin_failed;
+	}
+
+	ret = amdgpu_bo_kmap(bo, kptr);
+	if (ret) {
+		pr_err("Failed to map bo to kernel. ret %d\n", ret);
+		goto kmap_failed;
+	}
+
+	amdgpu_amdkfd_remove_eviction_fence(
+		bo, mem->process_info->eviction_fence, NULL, NULL);
+	list_del_init(&mem->validate_list.head);
+
+	if (size)
+		*size = amdgpu_bo_size(bo);
+
+	amdgpu_bo_unreserve(bo);
+
+	mutex_unlock(&mem->process_info->lock);
+	return 0;
+
+kmap_failed:
+	amdgpu_bo_unpin(bo);
+pin_failed:
+	amdgpu_bo_unreserve(bo);
+bo_reserve_failed:
+	mutex_unlock(&mem->process_info->lock);
+
+	return ret;
+}
+
+/** amdgpu_amdkfd_gpuvm_restore_process_bos - Restore all BOs for the given
+ *   KFD process identified by process_info
+ *
+ * @process_info: amdkfd_process_info of the KFD process
+ *
+ * After memory eviction, restore thread calls this function. The function
+ * should be called when the Process is still valid. BO restore involves -
+ *
+ * 1.  Release old eviction fence and create new one
+ * 2.  Get two copies of PD BO list from all the VMs. Keep one copy as pd_list.
+ * 3   Use the second PD list and kfd_bo_list to create a list (ctx.list) of
+ *     BOs that need to be reserved.
+ * 4.  Reserve all the BOs
+ * 5.  Validate of PD and PT BOs.
+ * 6.  Validate all KFD BOs using kfd_bo_list and Map them and add new fence
+ * 7.  Add fence to all PD and PT BOs.
+ * 8.  Unreserve all BOs
+ */
+int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
+{
+	struct amdgpu_bo_list_entry *pd_bo_list;
+	struct amdkfd_process_info *process_info = info;
+	struct amdgpu_vm *peer_vm;
+	struct kgd_mem *mem;
+	struct bo_vm_reservation_context ctx;
+	struct amdgpu_amdkfd_fence *new_fence;
+	int ret = 0, i;
+	struct list_head duplicate_save;
+	struct amdgpu_sync sync_obj;
+
+	INIT_LIST_HEAD(&duplicate_save);
+	INIT_LIST_HEAD(&ctx.list);
+	INIT_LIST_HEAD(&ctx.duplicates);
+
+	pd_bo_list = kcalloc(process_info->n_vms,
+			     sizeof(struct amdgpu_bo_list_entry),
+			     GFP_KERNEL);
+	if (!pd_bo_list)
+		return -ENOMEM;
+
+	i = 0;
+	mutex_lock(&process_info->lock);
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			vm_list_node)
+		amdgpu_vm_get_pd_bo(peer_vm, &ctx.list, &pd_bo_list[i++]);
+
+	/* Reserve all BOs and page tables/directory. Add all BOs from
+	 * kfd_bo_list to ctx.list
+	 */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+			    validate_list.head) {
+
+		list_add_tail(&mem->resv_list.head, &ctx.list);
+		mem->resv_list.bo = mem->validate_list.bo;
+		mem->resv_list.shared = mem->validate_list.shared;
+	}
+
+	ret = ttm_eu_reserve_buffers(&ctx.ticket, &ctx.list,
+				     false, &duplicate_save);
+	if (ret) {
+		pr_debug("Memory eviction: TTM Reserve Failed. Try again\n");
+		goto ttm_reserve_fail;
+	}
+
+	amdgpu_sync_create(&sync_obj);
+
+	/* Validate PDs and PTs */
+	ret = process_validate_vms(process_info);
+	if (ret)
+		goto validate_map_fail;
+
+	/* Wait for PD/PTs validate to finish */
+	/* FIXME: I think this isn't needed */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		struct amdgpu_bo *bo = peer_vm->root.base.bo;
+
+		ttm_bo_wait(&bo->tbo, false, false);
+	}
+
+	/* Validate BOs and map them to GPUVM (update VM page tables). */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+			    validate_list.head) {
+
+		struct amdgpu_bo *bo = mem->bo;
+		uint32_t domain = mem->domain;
+		struct kfd_bo_va_list *bo_va_entry;
+
+		ret = amdgpu_amdkfd_bo_validate(bo, domain, false);
+		if (ret) {
+			pr_debug("Memory eviction: Validate BOs failed. Try again\n");
+			goto validate_map_fail;
+		}
+
+		list_for_each_entry(bo_va_entry, &mem->bo_va_list,
+				    bo_list) {
+			ret = update_gpuvm_pte((struct amdgpu_device *)
+					      bo_va_entry->kgd_dev,
+					      bo_va_entry,
+					      &sync_obj);
+			if (ret) {
+				pr_debug("Memory eviction: update PTE failed. Try again\n");
+				goto validate_map_fail;
+			}
+		}
+	}
+
+	/* Update page directories */
+	ret = process_update_pds(process_info, &sync_obj);
+	if (ret) {
+		pr_debug("Memory eviction: update PDs failed. Try again\n");
+		goto validate_map_fail;
+	}
+
+	amdgpu_sync_wait(&sync_obj, false);
+
+	/* Release old eviction fence and create new one, because fence only
+	 * goes from unsignaled to signaled, fence cannot be reused.
+	 * Use context and mm from the old fence.
+	 */
+	new_fence = amdgpu_amdkfd_fence_create(
+				process_info->eviction_fence->base.context,
+				process_info->eviction_fence->mm);
+	if (!new_fence) {
+		pr_err("Failed to create eviction fence\n");
+		ret = -ENOMEM;
+		goto validate_map_fail;
+	}
+	dma_fence_put(&process_info->eviction_fence->base);
+	process_info->eviction_fence = new_fence;
+	*ef = dma_fence_get(&new_fence->base);
+
+	/* Wait for validate to finish and attach new eviction fence */
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+		validate_list.head)
+		ttm_bo_wait(&mem->bo->tbo, false, false);
+	list_for_each_entry(mem, &process_info->kfd_bo_list,
+		validate_list.head)
+		amdgpu_bo_fence(mem->bo,
+			&process_info->eviction_fence->base, true);
+
+	/* Attach eviction fence to PD / PT BOs */
+	list_for_each_entry(peer_vm, &process_info->vm_list_head,
+			    vm_list_node) {
+		struct amdgpu_bo *bo = peer_vm->root.base.bo;
+
+		amdgpu_bo_fence(bo, &process_info->eviction_fence->base, true);
+	}
+
+validate_map_fail:
+	ttm_eu_backoff_reservation(&ctx.ticket, &ctx.list);
+	amdgpu_sync_free(&sync_obj);
+ttm_reserve_fail:
+	mutex_unlock(&process_info->lock);
+	kfree(pd_bo_list);
+	return ret;
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c

@@ -114,6 +114,9 @@ union igp_info {
 	struct atom_integrated_system_info_v1_11 v11;
 };
 
+union umc_info {
+	struct atom_umc_info_v3_1 v31;
+};
 /*
  * Return vram width from integrated system info table, if available,
  * or 0 if not.
@@ -143,6 +146,94 @@ int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int convert_atom_mem_type_to_vram_type (struct amdgpu_device *adev,
+					       int atom_mem_type)
+{
+	int vram_type;
+
+	if (adev->flags & AMD_IS_APU) {
+		switch (atom_mem_type) {
+		case Ddr2MemType:
+		case LpDdr2MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR2;
+			break;
+		case Ddr3MemType:
+		case LpDdr3MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR3;
+			break;
+		case Ddr4MemType:
+		case LpDdr4MemType:
+			vram_type = AMDGPU_VRAM_TYPE_DDR4;
+			break;
+		default:
+			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+			break;
+		}
+	} else {
+		switch (atom_mem_type) {
+		case ATOM_DGPU_VRAM_TYPE_GDDR5:
+			vram_type = AMDGPU_VRAM_TYPE_GDDR5;
+			break;
+		case ATOM_DGPU_VRAM_TYPE_HBM:
+			vram_type = AMDGPU_VRAM_TYPE_HBM;
+			break;
+		default:
+			vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+			break;
+		}
+	}
+
+	return vram_type;
+}
+/*
+ * Return vram type from either integrated system info table
+ * or umc info table, if available, or 0 (TYPE_UNKNOWN) if not
+ */
+int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index;
+	u16 data_offset, size;
+	union igp_info *igp_info;
+	union umc_info *umc_info;
+	u8 frev, crev;
+	u8 mem_type;
+
+	if (adev->flags & AMD_IS_APU)
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						    integratedsysteminfo);
+	else
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						    umc_info);
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context,
+					  index, &size,
+					  &frev, &crev, &data_offset)) {
+		if (adev->flags & AMD_IS_APU) {
+			igp_info = (union igp_info *)
+				(mode_info->atom_context->bios + data_offset);
+			switch (crev) {
+			case 11:
+				mem_type = igp_info->v11.memorytype;
+				return convert_atom_mem_type_to_vram_type(adev, mem_type);
+			default:
+				return 0;
+			}
+		} else {
+			umc_info = (union umc_info *)
+				(mode_info->atom_context->bios + data_offset);
+			switch (crev) {
+			case 1:
+				mem_type = umc_info->v31.vram_type;
+				return convert_atom_mem_type_to_vram_type(adev, mem_type);
+			default:
+				return 0;
+			}
+		}
+	}
+
+	return 0;
+}
+
 union firmware_info {
 	struct atom_firmware_info_v3_1 v31;
 };
@@ -151,10 +242,6 @@ union smu_info {
 	struct atom_smu_info_v3_1 v31;
 };
 
-union umc_info {
-	struct atom_umc_info_v3_1 v31;
-};
-
 int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
 {
 	struct amdgpu_mode_info *mode_info = &adev->mode_info;

drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h

@@ -28,6 +28,7 @@ bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev)
 void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev);
+int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev);
 int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev);
 
 #endif