drm/i915: Introduce Intel PXP
Subsystems (6)
v6 cover drm/i915: Introduce Intel PXP
PXP (Protected Xe Path) is an i915 component, available on GEN12+, that helps to establish the hardware protected session and manage the status of the alive software session, as well as its life cycle.
Very minimal changes from v5:
- Update mei_pxp match code to look the same as mei_hdcp (no functional/logical changes) and drop a couple of debug prints.
- Rebase on top of object create_ext changes (removal of vanilla_object).
Tested with: https://patchwork.freedesktop.org/series/87570/
Anshuman Gupta (2):
drm/i915/pxp: Add plane decryption support
drm/i915/pxp: black pixels on pxp disabled
Daniele Ceraolo Spurio (7):
drm/i915/pxp: Define PXP component interface drm/i915/pxp: define PXP device flag and kconfig drm/i915/pxp: allocate a vcs context for pxp usage drm/i915/pxp: set KCR reg init drm/i915/pxp: interfaces for using protected objects drm/i915/pxp: start the arb session on demand drm/i915/pxp: enable PXP for integrated Gen12
Huang, Sean Z (5):
drm/i915/pxp: Implement funcs to create the TEE channel drm/i915/pxp: Create the arbitrary session after boot drm/i915/pxp: Implement arb session teardown drm/i915/pxp: Implement PXP irq handler drm/i915/pxp: Enable PXP power management
Vitaly Lubart (1):
mei: pxp: export pavp client to me client bus
drivers/gpu/drm/i915/Kconfig | 11 + drivers/gpu/drm/i915/Makefile | 9 + .../gpu/drm/i915/display/intel_atomic_plane.c | 25 ++ drivers/gpu/drm/i915/display/intel_display.c | 4 + .../drm/i915/display/intel_display_types.h | 6 + .../drm/i915/display/skl_universal_plane.c | 49 +++- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 +++-- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 42 ++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 + drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 + .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 + drivers/gpu/drm/i915/gt/intel_engine.h | 2 + drivers/gpu/drm/i915/gt/intel_gpu_commands.h | 22 +- drivers/gpu/drm/i915/gt/intel_gt.c | 5 + drivers/gpu/drm/i915/gt/intel_gt_irq.c | 7 + drivers/gpu/drm/i915/gt/intel_gt_pm.c | 15 +- drivers/gpu/drm/i915/gt/intel_gt_types.h | 3 + drivers/gpu/drm/i915/i915_drv.c | 2 + drivers/gpu/drm/i915/i915_drv.h | 4 + drivers/gpu/drm/i915/i915_pci.c | 2 + drivers/gpu/drm/i915/i915_reg.h | 48 ++++ drivers/gpu/drm/i915/intel_device_info.h | 1 + drivers/gpu/drm/i915/pxp/intel_pxp.c | 268 ++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 67 +++++ drivers/gpu/drm/i915/pxp/intel_pxp_cmd.c | 141 +++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_cmd.h | 15 + drivers/gpu/drm/i915/pxp/intel_pxp_irq.c | 100 +++++++ drivers/gpu/drm/i915/pxp/intel_pxp_irq.h | 32 +++ drivers/gpu/drm/i915/pxp/intel_pxp_pm.c | 40 +++ drivers/gpu/drm/i915/pxp/intel_pxp_pm.h | 23 ++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 172 +++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.h | 15 + drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 170 +++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_tee.h | 17 ++ .../drm/i915/pxp/intel_pxp_tee_interface.h | 37 +++ drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 55 ++++ drivers/misc/mei/Kconfig | 2 + drivers/misc/mei/Makefile | 1 + drivers/misc/mei/pxp/Kconfig | 13 + drivers/misc/mei/pxp/Makefile | 7 + drivers/misc/mei/pxp/mei_pxp.c | 229 +++++++++++++++ drivers/misc/mei/pxp/mei_pxp.h | 18 ++ include/drm/i915_component.h | 1 + include/drm/i915_pxp_tee_interface.h | 45 +++ include/uapi/drm/i915_drm.h | 58 +++- 48 files changed, 1940 insertions(+), 33 deletions(-) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_cmd.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_cmd.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_irq.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_irq.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_pm.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_pm.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_session.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_session.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_tee.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_tee.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_types.h create mode 100644 drivers/misc/mei/pxp/Kconfig create mode 100644 drivers/misc/mei/pxp/Makefile create mode 100644 drivers/misc/mei/pxp/mei_pxp.c create mode 100644 drivers/misc/mei/pxp/mei_pxp.h create mode 100644 include/drm/i915_pxp_tee_interface.h
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 1/15 drm/i915/pxp: Define PXP component interface
This will be used for communication between the i915 driver and the mei one. Defining it in a stand-alone patch to avoid circualr dependedencies between the patches modifying the 2 drivers.
Split out from an original patch from Huang, Sean Z
v2: rename the component struct (Rodrigo)
--- include/drm/i915_component.h | 1 + include/drm/i915_pxp_tee_interface.h | 45 ++++++++++++++++++++++++++++ 2 files changed, 46 insertions(+) create mode 100644 include/drm/i915_pxp_tee_interface.h
include/drm/i915_component.h +1 -0
@@ -29,6 +29,7 @@enum i915_component_type {I915_COMPONENT_AUDIO =1 ,I915_COMPONENT_HDCP , +I915_COMPONENT_PXP };/* MAX_PORT is the number of port
include/drm/i915_pxp_tee_interface.h new file +45 -0
45 lines not on this page · raw message
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
maintainerRe: [Intel-gfx] [PATCH v6 01/15] drm/i915/pxp: Define PXP component interface
Quoted 35 lines · Daniele Ceraolo Spurio
This will be used for communication between the i915 driver and the mei one. Defining it in a stand-alone patch to avoid circualr dependedencies between the patches modifying the 2 drivers.
Split out from an original patch from Huang, Sean Z
v2: rename the component struct (Rodrigo)
Signed-off-by Daniele Ceraolo Spurio <[email protected]>Cc Rodrigo Vivi <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]>--- include/drm/i915_component.h | 1 + include/drm/i915_pxp_tee_interface.h | 45 ++++++++++++++++++++++++++++ 2 files changed, 46 insertions(+) create mode 100644 include/drm/i915_pxp_tee_interface.hinclude/drm/i915_component.h +1
-0@@ -29,6 +29,7 @@enum i915_component_type {I915_COMPONENT_AUDIO =1 ,I915_COMPONENT_HDCP , +I915_COMPONENT_PXP };/* MAX_PORT is the number of port include/drm/i915_pxp_tee_interface.h new file +0
-0
include/drm/i915_pxp_tee_interface.h line 6
Quoted 6 lines
include/drm/i915_pxp_tee_interface.h new file +6
-0@@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2020 Intel Corporation +* +* Authors: +* Vitaly Lubart <[email protected]>
IMO we should avoid adding new authors lists in code comments. Git log provides accurate and up-to-date information, and I don't want patches where people add their names to authors lists.
BR,
Jani.
Quoted 39 lines
include/drm/i915_pxp_tee_interface.h new file +39
-0+*/ + +#ifndef _I915_PXP_TEE_INTERFACE_H_ +#define _I915_PXP_TEE_INTERFACE_H_ + +#include <linux/mutex.h> +#include <linux/device.h> + +/** +* struct i915_pxp_component_ops - ops for PXP services. +* @owner: Module providing the ops +* @send: sends data to PXP +* @receive: receives data from PXP +*/ +struct i915_pxp_component_ops { +/** +* @owner: owner of the module provding the ops +*/ +struct module *owner; + +int (*send)(struct device *dev,const void *message,size_t size); +int (*recv)(struct device *dev,void *buffer,size_t size); +}; + +/** +* struct i915_pxp_component - Used for communication between i915 and TEE +* drivers for the PXP services +* @tee_dev: device that provide the PXP service from TEE Bus. +* @pxp_ops: Ops implemented by TEE driver, used by i915 driver. +*/ +struct i915_pxp_component { +struct device *tee_dev; +const struct i915_pxp_component_ops *ops; + +/* To protect the above members. */ +struct mutex mutex; +}; + +#endif /* _I915_TEE_PXP_INTERFACE_H_ */
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
PATCH v6 2/15 mei: pxp: export pavp client to me client bus
In reply to Daniele Ceraolo SpurioFrom: Vitaly Lubart <[email protected]>
Export PAVP client to work with i915 driver,
for binding it uses kernel component framework.
v2:drop debug prints, refactor match code to match mei_hdcp (Tomas)
--- drivers/misc/mei/Kconfig | 2 + drivers/misc/mei/Makefile | 1 + drivers/misc/mei/pxp/Kconfig | 13 ++ drivers/misc/mei/pxp/Makefile | 7 + drivers/misc/mei/pxp/mei_pxp.c | 229 +++++++++++++++++++++++++++++++++ drivers/misc/mei/pxp/mei_pxp.h | 18 +++ 6 files changed, 270 insertions(+) create mode 100644 drivers/misc/mei/pxp/Kconfig create mode 100644 drivers/misc/mei/pxp/Makefile create mode 100644 drivers/misc/mei/pxp/mei_pxp.c create mode 100644 drivers/misc/mei/pxp/mei_pxp.h
drivers/misc/mei/Kconfig +2 -0
@@ -47,3 +47,5 @@ config INTEL_MEI_TXE Intel Bay Trail source "drivers/misc/mei/hdcp/Kconfig" +source "drivers/misc/mei/pxp/Kconfig" +
drivers/misc/mei/Makefile +1 -0
@@ -26,3 +26,4 @@ mei-$(CONFIG_EVENT_TRACING) += mei-trace.o CFLAGS_mei-trace.o = -I$(src) obj-$(CONFIG_INTEL_MEI_HDCP) += hdcp/ +obj-$(CONFIG_INTEL_MEI_PXP) += pxp/
drivers/misc/mei/pxp/Kconfig new file +13 -0
@@ -0,0 +1,13 @@ + +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020, Intel Corporation. All rights reserved. +# +config INTEL_MEI_PXP + tristate "Intel PXP services of ME Interface" + select INTEL_MEI_ME + depends on DRM_I915 + help + MEI Support for PXP Services on Intel platforms. + + Enables the ME FW services required for PXP support through + I915 display driver of Intel.
drivers/misc/mei/pxp/Makefile new file +7 -0
@@ -0,0 +1,7 @@ +# SPDX-License-Identifier: GPL-2.0 +# +# Copyright (c) 2020, Intel Corporation. All rights reserved. +# +# Makefile - PXP client driver for Intel MEI Bus Driver. + +obj-$(CONFIG_INTEL_MEI_PXP) += mei_pxp.o
drivers/misc/mei/pxp/mei_pxp.c new file +229 -0
229 lines not on this page · raw message
drivers/misc/mei/pxp/mei_pxp.h new file +18 -0
@@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* +* Copyright © 2020 Intel Corporation +* +* Authors: +* Vitaly Lubart <[email protected]> +*/ + +#ifndef __MEI_PXP_H__ +#define __MEI_PXP_H__ + +/* me_pxp_status: Enumeration of all PXP Status Codes */ +enum me_pxp_status { +ME_PXP_STATUS_SUCCESS =0x0000 , + +}; + +#endif /* __MEI_PXP_H__ */
Signature · 1 line
2.32.0
Rodrigo Vivi
maintainerRe: [PATCH v6 02/15] mei: pxp: export pavp client to me client bus
Quoted 11 lines · Daniele Ceraolo Spurio
From: Vitaly Lubart <[email protected]>
Export PAVP client to work with i915 driver,
for binding it uses kernel component framework.v2:drop debug prints, refactor match code to match
mei_hdcp(Tomas)Signed-off-by Vitaly Lubart <[email protected]>Signed-off-by Tomas Winkler <[email protected]>Signed-off-by Daniele Ceraolo Spurio <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v1
shouldn't we use the 2020-2021 in the other headers as well?
anyway, rv-b remains
Quoted 326 lines
--- drivers/misc/mei/Kconfig | 2 + drivers/misc/mei/Makefile | 1 + drivers/misc/mei/pxp/Kconfig | 13 ++ drivers/misc/mei/pxp/Makefile | 7 + drivers/misc/mei/pxp/mei_pxp.c | 229 +++++++++++++++++++++++++++++++++ drivers/misc/mei/pxp/mei_pxp.h | 18 +++ 6 files changed, 270 insertions(+) create mode 100644 drivers/misc/mei/pxp/Kconfig create mode 100644 drivers/misc/mei/pxp/Makefile create mode 100644 drivers/misc/mei/pxp/mei_pxp.c create mode 100644 drivers/misc/mei/pxp/mei_pxp.hdrivers/misc/mei/Kconfig +2
-0@@ -47,3 +47,5 @@ config INTEL_MEI_TXE Intel Bay Trail source "drivers/misc/mei/hdcp/Kconfig" +source "drivers/misc/mei/pxp/Kconfig" +drivers/misc/mei/Makefile +1
-0@@ -26,3 +26,4 @@ mei-$(CONFIG_EVENT_TRACING) += mei-trace.o CFLAGS_mei-trace.o = -I$(src) obj-$(CONFIG_INTEL_MEI_HDCP) += hdcp/ +obj-$(CONFIG_INTEL_MEI_PXP) += pxp/drivers/misc/mei/pxp/Kconfig new file +13
-0@@ -0,0 +1,13 @@ + +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020, Intel Corporation. All rights reserved. +# +config INTEL_MEI_PXP + tristate "Intel PXP services of ME Interface" + select INTEL_MEI_ME + depends on DRM_I915 + help + MEI Support for PXP Services on Intel platforms. + + Enables the ME FW services required for PXP support through + I915 display driver of Intel.drivers/misc/mei/pxp/Makefile new file +7
-0@@ -0,0 +1,7 @@ +# SPDX-License-Identifier: GPL-2.0 +# +# Copyright (c) 2020, Intel Corporation. All rights reserved. +# +# Makefile - PXP client driver for Intel MEI Bus Driver. + +obj-$(CONFIG_INTEL_MEI_PXP) += mei_pxp.odrivers/misc/mei/pxp/mei_pxp.c new file +229
-0@@ -0,0 +1,229 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* Copyright © 2020 - 2021 Intel Corporation +*/ + +/** +* DOC: MEI_PXP Client Driver +* +* The mei_pxp driver acts as a translation layer between PXP +* protocol implementer (I915) and ME FW by translating PXP +* negotiation messages to ME FW command payloads and vice versa. +*/ + +#include <linux/module.h> +#include <linux/slab.h> +#include <linux/uuid.h> +#include <linux/mei_cl_bus.h> +#include <linux/component.h> +#include <drm/drm_connector.h> +#include <drm/i915_component.h> +#include <drm/i915_pxp_tee_interface.h> + +#include "mei_pxp.h" + +/** +* mei_pxp_send_message() - Sends a PXP message to ME FW. +* @dev: device corresponding to the mei_cl_device +* @message: a message buffer to send +* @size: size of the message +* Return: 0 on Success, <0 on Failure +*/ +static int +mei_pxp_send_message (struct device *dev,const void *message,size_t size) +{ +struct mei_cl_device *cldev; +ssize_t byte; + +if (!dev || !message) +return -EINVAL ; + + cldev =to_mei_cl_device (dev); + +/* temporary drop const qualifier till the API is fixed */ + byte =mei_cldev_send (cldev, (u8 *)message, size); +if (byte <0 ) { +dev_dbg (dev,"mei_cldev_send failed. %zd\n" , byte); +return byte; + } + +return 0 ; +} + +/** +* mei_pxp_receive_message() - Receives a PXP message from ME FW. +* @dev: device corresponding to the mei_cl_device +* @buffer: a message buffer to contain the received message +* @size: size of the buffer +* Return: bytes sent on Success, <0 on Failure +*/ +static int +mei_pxp_receive_message (struct device *dev,void *buffer,size_t size) +{ +struct mei_cl_device *cldev; +ssize_t byte; + +if (!dev || !buffer) +return -EINVAL ; + + cldev =to_mei_cl_device (dev); + + byte =mei_cldev_recv (cldev, buffer, size); +if (byte <0 ) { +dev_dbg (dev,"mei_cldev_recv failed. %zd\n" , byte); +return byte; + } + +return byte; +} + +static const struct i915_pxp_component_ops mei_pxp_ops = { + .owner =THIS_MODULE , + .send = mei_pxp_send_message, + .recv = mei_pxp_receive_message, +}; + +static int mei_component_master_bind (struct device *dev) +{ +struct mei_cl_device *cldev =to_mei_cl_device (dev); +struct i915_pxp_component *comp_master =mei_cldev_get_drvdata (cldev); +int ret; + + comp_master->ops = &mei_pxp_ops; + comp_master->tee_dev = dev; + ret =component_bind_all (dev, comp_master); +if (ret <0 ) +return ret; + +return 0 ; +} + +static void mei_component_master_unbind (struct device *dev) +{ +struct mei_cl_device *cldev =to_mei_cl_device (dev); +struct i915_pxp_component *comp_master =mei_cldev_get_drvdata (cldev); + +component_unbind_all (dev, comp_master); +} + +static const struct component_master_ops mei_component_master_ops = { + .bind = mei_component_master_bind, + .unbind = mei_component_master_unbind, +}; + +/** +* mei_pxp_component_match - compare function for matching mei pxp. +* +* The function checks if the driver is i915, the subcomponent is PXP +* and the grand parent of pxp and the parent of i915 are the same +* PCH device. +* +* @dev: master device +* @subcomponent: subcomponent to match (I915_COMPONENT_PXP) +* @data: compare data (mei pxp device) +* +* Return: +* * 1 - if components match +* * 0 - otherwise +*/ +static int mei_pxp_component_match (struct device *dev,int subcomponent, +void *data) +{ +struct device *base = data; + +if (strcmp (dev->driver->name,"i915" ) || + subcomponent !=I915_COMPONENT_PXP ) +return 0 ; + + base = base->parent; +if (!base) +return 0 ; + + base = base->parent; + dev = dev->parent; + +return (base && dev && dev == base); +} + +static int mei_pxp_probe (struct mei_cl_device *cldev, +const struct mei_cl_device_id *id) +{ +struct i915_pxp_component *comp_master; +struct component_match *master_match; +int ret; + + ret =mei_cldev_enable (cldev); +if (ret <0 ) { +dev_err (&cldev->dev,"mei_cldev_enable Failed. %d\n" , ret); + goto enable_err_exit; + } + + comp_master =kzalloc (sizeof (*comp_master),GFP_KERNEL ); +if (!comp_master) { + ret = -ENOMEM ; + goto err_exit; + } + + master_match =NULL ; +component_match_add_typed (&cldev->dev, &master_match, + mei_pxp_component_match, &cldev->dev); +if (IS_ERR_OR_NULL (master_match)) { + ret = -ENOMEM ; + goto err_exit; + } + +mei_cldev_set_drvdata (cldev, comp_master); + ret =component_master_add_with_match (&cldev->dev, + &mei_component_master_ops, + master_match); +if (ret <0 ) { +dev_err (&cldev->dev,"Master comp add failed %d\n" , ret); + goto err_exit; + } + +return 0 ; + +err_exit: +mei_cldev_set_drvdata (cldev,NULL ); +kfree (comp_master); +mei_cldev_disable (cldev); +enable_err_exit: +return ret; +} + +static void mei_pxp_remove (struct mei_cl_device *cldev) +{ +struct i915_pxp_component *comp_master =mei_cldev_get_drvdata (cldev); +int ret; + +component_master_del (&cldev->dev, &mei_component_master_ops); +kfree (comp_master); +mei_cldev_set_drvdata (cldev,NULL ); + + ret =mei_cldev_disable (cldev); +if (ret) +dev_warn (&cldev->dev,"mei_cldev_disable() failed\n" ); +} + +/* fbf6fcf1-96cf-4e2e-a6a6-1bab8cbe36b1 : PAVP GUID*/ +#define MEI_GUID_PXP GUID_INIT(0xfbf6fcf1, 0x96cf, 0x4e2e, 0xA6, \ + 0xa6, 0x1b, 0xab, 0x8c, 0xbe, 0x36, 0xb1) + +static struct mei_cl_device_id mei_pxp_tbl[] = { + { .uuid =MEI_GUID_PXP , .version =MEI_CL_VERSION_ANY }, + { } +}; +MODULE_DEVICE_TABLE (mei, mei_pxp_tbl); + +static struct mei_cl_driver mei_pxp_driver = { + .id_table = mei_pxp_tbl, + .name =KBUILD_MODNAME , + .probe = mei_pxp_probe, + .remove = mei_pxp_remove, +}; + +module_mei_cl_driver (mei_pxp_driver); + +MODULE_AUTHOR ("Intel Corporation" ); +MODULE_LICENSE ("GPL" ); +MODULE_DESCRIPTION ("MEI PXP" );drivers/misc/mei/pxp/mei_pxp.h new file +18
-0@@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* +* Copyright © 2020 Intel Corporation +* +* Authors: +* Vitaly Lubart <[email protected]> +*/ + +#ifndef __MEI_PXP_H__ +#define __MEI_PXP_H__ + +/* me_pxp_status: Enumeration of all PXP Status Codes */ +enum me_pxp_status { +ME_PXP_STATUS_SUCCESS =0x0000 , + +}; + +#endif /* __MEI_PXP_H__ */ Signature · 1 line
2.32.0
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 3/15 drm/i915/pxp: define PXP device flag and kconfig
In reply to Daniele Ceraolo SpurioAhead of the PXP implementation, define the relevant define flag and kconfig option.
v2: flip kconfig default to N. Some machines have IFWIs that do not support PXP, so we need it to be an opt-in until we add support to query the caps from the mei device.
--- drivers/gpu/drm/i915/Kconfig | 11 +++++++++++ drivers/gpu/drm/i915/i915_drv.h | 4 ++++ drivers/gpu/drm/i915/intel_device_info.h | 1 + 3 files changed, 16 insertions(+)
drivers/gpu/drm/i915/Kconfig +11 -0
@@ -131,6 +131,17 @@ config DRM_I915_GVT_KVMGT Choose this option if you want to enable KVMGT support for Intel GVT-g. +config DRM_I915_PXP + bool "Enable Intel PXP support for Intel Gen12+ platform" + depends on DRM_I915 + depends on INTEL_MEI && INTEL_MEI_PXP + default n + help + PXP (Protected Xe Path) is an i915 component, available on GEN12+ + GPUs, that helps to establish the hardware protected session and + manage the status of the alive software session, as well as its life + cycle. + menu "drm/i915 Debugging" depends on DRM_I915 depends on EXPERT
drivers/gpu/drm/i915/i915_drv.h +4 -0
@@ -1724,6 +1724,10 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915,#define HAS_VRR (i915) (GRAPHICS_VER(i915) >= 12) +#define HAS_PXP (dev_priv) (IS_ENABLED(CONFIG_DRM_I915_PXP) && \ + INTEL_INFO(dev_priv)->has_pxp) && \ + VDBOX_MASK(&dev_priv->gt) +/* Only valid when HAS_DISPLAY() is true */ #define INTEL_DISPLAY_ENABLED (dev_priv) \ (drm_WARN_ON(&(dev_priv)->drm, !HAS_DISPLAY(dev_priv)), !(dev_priv)->params.disable_display)
drivers/gpu/drm/i915/intel_device_info.h +1 -0
@@ -134,6 +134,7 @@ enum intel_ppgtt_type {func (has_logical_ring_contexts); \func (has_logical_ring_elsq); \func (has_pooled_eu); \ +func (has_pxp); \func (has_rc6); \func (has_rc6p); \func (has_rps); \
Signature · 1 line
2.32.0
PATCH v6 4/15 drm/i915/pxp: allocate a vcs context for pxp usage
In reply to Daniele Ceraolo SpurioThe context is required to send the session termination commands to the VCS, which will be implemented in a follow-up patch. We can also use the presence of the context as a check of pxp initialization completion.
v2: use perma-pinned context (Chris)
v3: rename pinned_context functions (Chris)
v4: split export of pinned_context functions to a separate patch (Rodrigo)
--- drivers/gpu/drm/i915/Makefile | 4 ++ drivers/gpu/drm/i915/gt/intel_engine.h | 2 + drivers/gpu/drm/i915/gt/intel_gt.c | 5 ++ drivers/gpu/drm/i915/gt/intel_gt_types.h | 3 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 62 ++++++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 35 ++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 15 ++++++ 7 files changed, 126 insertions(+) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_types.h
drivers/gpu/drm/i915/Makefile +4 -0
@@ -275,6 +275,10 @@ i915-y += \ i915-y += i915_perf.o +# Protected execution platform (PXP) support +i915-$(CONFIG_DRM_I915_PXP) += \ + pxp/intel_pxp.o + # Post-mortem debug and GPU hang state capture i915-$(CONFIG_DRM_I915_CAPTURE_ERROR) += i915_gpu_error.o i915-$(CONFIG_DRM_I915_SELFTEST) += \
drivers/gpu/drm/i915/gt/intel_engine.h +2 -0
@@ -175,6 +175,8 @@ intel_write_status_page(struct intel_engine_cs *engine, int reg, u32 value)#define I915_GEM_HWS_SEQNO 0x40#define I915_GEM_HWS_SEQNO_ADDR (I915_GEM_HWS_SEQNO * sizeof(u32))#define I915_GEM_HWS_MIGRATE (0x42 * sizeof(u32)) +#define I915_GEM_HWS_PXP 0x60 +#define I915_GEM_HWS_PXP_ADDR (I915_GEM_HWS_PXP * sizeof(u32))#define I915_GEM_HWS_SCRATCH 0x80#define I915_HWS_CSB_BUF0_INDEX 0x10
drivers/gpu/drm/i915/gt/intel_gt.c +5 -0
@@ -21,6 +21,7 @@#include "intel_uncore.h" #include "intel_pm.h" #include "shmem_utils.h" +#include "pxp/intel_pxp.h" void intel_gt_init_early (struct intel_gt *gt,struct drm_i915_private *i915) {
@@ -662,6 +663,8 @@ int intel_gt_init(struct intel_gt *gt)intel_migrate_init (>->migrate, gt); +intel_pxp_init (>->pxp); + goto out_fw; err_gt:__intel_gt_disable (gt);
@@ -697,6 +700,8 @@ void intel_gt_driver_unregister(struct intel_gt *gt)intel_rps_driver_unregister (>->rps); +intel_pxp_fini (>->pxp); +/* * Upon unregistering the device to prevent any new users, cancel * all in-flight requests so that we can quickly unbind the active
drivers/gpu/drm/i915/gt/intel_gt_types.h +3 -0
@@ -26,6 +26,7 @@#include "intel_rps_types.h" #include "intel_migrate_types.h" #include "intel_wakeref.h" +#include "pxp/intel_pxp_types.h" struct drm_i915_private ;struct i915_ggtt ;
@@ -185,6 +186,8 @@ struct intel_gt {/* Slice/subslice/EU info */ struct sseu_dev_info sseu; } info; + +struct intel_pxp pxp; };enum intel_gt_scratch_field {
drivers/gpu/drm/i915/pxp/intel_pxp.c new file +62 -0
62 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp.h new file +35 -0
@@ -0,0 +1,35 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_H__ +#define __INTEL_PXP_H__ + +#include "gt/intel_gt_types.h" +#include "intel_pxp_types.h" + +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) +{ + return container_of(pxp, struct intel_gt, pxp); +} + +static inline bool intel_pxp_is_enabled (const struct intel_pxp *pxp) +{ +return pxp->ce; +} + +#ifdef CONFIG_DRM_I915_PXP +void intel_pxp_init (struct intel_pxp *pxp); +void intel_pxp_fini (struct intel_pxp *pxp); +#else +static inline void intel_pxp_init (struct intel_pxp *pxp) +{ +} + +static inline void intel_pxp_fini (struct intel_pxp *pxp) +{ +} +#endif + +#endif /* __INTEL_PXP_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_types.h new file +15 -0
@@ -0,0 +1,15 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_TYPES_H__ +#define __INTEL_PXP_TYPES_H__ + +struct intel_context ; + +struct intel_pxp { +struct intel_context *ce; +}; + +#endif /* __INTEL_PXP_TYPES_H__ */
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 5/15 drm/i915/pxp: Implement funcs to create the TEE channel
In reply to Daniele Ceraolo SpurioFrom: "Huang, Sean Z" <[email protected]>
Implement the funcs to create the TEE channel, so kernel can send the TEE commands directly to TEE for creating the arbitrary (default) session.
v2: fix locking, don't pollute dev_priv (Chris)
v3: wait for mei PXP component to be bound.
v4: drop the wait, as the component might be bound after i915 load completes. We'll instead check when sending a tee message.
--- drivers/gpu/drm/i915/Makefile | 3 +- drivers/gpu/drm/i915/pxp/intel_pxp.c | 13 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 77 ++++++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_tee.h | 14 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 3 + 5 files changed, 109 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_tee.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_tee.h
drivers/gpu/drm/i915/Makefile +2 -1
@@ -277,7 +277,8 @@ i915-y += i915_perf.o # Protected execution platform (PXP) support i915-$(CONFIG_DRM_I915_PXP) += \ - pxp/intel_pxp.o + pxp/intel_pxp.o \ + pxp/intel_pxp_tee.o # Post-mortem debug and GPU hang state capture i915-$(CONFIG_DRM_I915_CAPTURE_ERROR) += i915_gpu_error.o
drivers/gpu/drm/i915/pxp/intel_pxp.c +13 -0
@@ -3,6 +3,7 @@* Copyright(c) 2020 Intel Corporation. */ #include "intel_pxp.h" +#include "intel_pxp_tee.h" #include "gt/intel_context.h" #include "i915_drv.h"
@@ -50,7 +51,16 @@ void intel_pxp_init(struct intel_pxp *pxp)if (ret)return ; + ret =intel_pxp_tee_component_init (pxp); +if (ret) + goto out_context; +drm_info (>->i915->drm,"Protected Xe Path (PXP) protected content support initialized\n" ); + +return ; + +out_context: +destroy_vcs_context (pxp); }void intel_pxp_fini (struct intel_pxp *pxp)
@@ -58,5 +68,8 @@ void intel_pxp_fini(struct intel_pxp *pxp)if (!intel_pxp_is_enabled (pxp))return ; +intel_pxp_tee_component_fini (pxp); +destroy_vcs_context (pxp); + }
drivers/gpu/drm/i915/pxp/intel_pxp_tee.c new file +77 -0
77 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp_tee.h new file +14 -0
@@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_TEE_H__ +#define __INTEL_PXP_TEE_H__ + +#include "intel_pxp.h" + +int intel_pxp_tee_component_init (struct intel_pxp *pxp); +void intel_pxp_tee_component_fini (struct intel_pxp *pxp); + +#endif /* __INTEL_PXP_TEE_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +3 -0
@@ -7,8 +7,11 @@#define __INTEL_PXP_TYPES_H__struct intel_context ; +struct i915_pxp_component ;struct intel_pxp { +struct i915_pxp_component *pxp_component; +struct intel_context *ce; };
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 6/15 drm/i915/pxp: set KCR reg init
In reply to Daniele Ceraolo SpurioThe setting is required by hardware to allow us doing further protection operation such as sending commands to GPU or TEE. The register needs to be re-programmed on resume, so for simplicitly we bundle the programming with the component binding, which is automatically called on resume.
Further HW set-up operations will be added in the same location in follow-up patches, so get ready for them by using a couple of init/fini_hw wrappers instead of calling the KCR funcs directly.
v3: move programming to component binding function, rework commit msg
--- drivers/gpu/drm/i915/pxp/intel_pxp.c | 27 ++++++++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 3 +++ drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 5 +++++ 3 files changed, 35 insertions(+)
drivers/gpu/drm/i915/pxp/intel_pxp.c +27 -0
@@ -7,6 +7,24 @@#include "gt/intel_context.h" #include "i915_drv.h" +/* KCR register definitions */ +#define KCR_INIT _MMIO(0x320f0) + +/* Setting KCR Init bit is required after system boot */ +#define KCR_INIT_ALLOW_DISPLAY_ME_WRITES REG_BIT(14) + +static void kcr_pxp_enable (struct intel_gt *gt) +{ +intel_uncore_write (gt->uncore,KCR_INIT , +_MASKED_BIT_ENABLE (KCR_INIT_ALLOW_DISPLAY_ME_WRITES )); +} + +static void kcr_pxp_disable (struct intel_gt *gt) +{ +intel_uncore_write (gt->uncore,KCR_INIT , +_MASKED_BIT_DISABLE (KCR_INIT_ALLOW_DISPLAY_ME_WRITES )); +} +static int create_vcs_context (struct intel_pxp *pxp) {static struct lock_class_key pxp_lock;
@@ -71,5 +89,14 @@ void intel_pxp_fini(struct intel_pxp *pxp)intel_pxp_tee_component_fini (pxp);destroy_vcs_context (pxp); +} + +void intel_pxp_init_hw (struct intel_pxp *pxp) +{ +kcr_pxp_enable (pxp_to_gt (pxp)); +} +void intel_pxp_fini_hw (struct intel_pxp *pxp) +{ +kcr_pxp_disable (pxp_to_gt (pxp)); }
drivers/gpu/drm/i915/pxp/intel_pxp.h +3 -0
@@ -22,6 +22,9 @@ static inline bool intel_pxp_is_enabled(const struct intel_pxp *pxp)#ifdef CONFIG_DRM_I915_PXP void intel_pxp_init (struct intel_pxp *pxp);void intel_pxp_fini (struct intel_pxp *pxp); + +void intel_pxp_init_hw (struct intel_pxp *pxp); +void intel_pxp_fini_hw (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {
drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +5 -0
@@ -33,6 +33,9 @@ static int i915_pxp_tee_component_bind(struct device *i915_kdev, pxp->pxp_component = data; pxp->pxp_component->tee_dev = tee_kdev; +/* the component is required to fully start the PXP HW */ +intel_pxp_init_hw (pxp); +return 0 ; }
@@ -41,6 +44,8 @@ static void i915_pxp_tee_component_unbind(struct device *i915_kdev, {struct intel_pxp *pxp =i915_dev_to_pxp (i915_kdev); +intel_pxp_fini_hw (pxp); + pxp->pxp_component =NULL ; }
Signature · 1 line
2.32.0
PATCH v6 7/15 drm/i915/pxp: Create the arbitrary session after boot
In reply to Daniele Ceraolo SpurioFrom: "Huang, Sean Z" <[email protected]>
Create the arbitrary session, with the fixed session id 0xf, after system boot, for the case that application allocates the protected buffer without establishing any protection session. Because the hardware requires at least one alive session for protected buffer creation. This arbitrary session will need to be re-created after teardown or power event because hardware encryption key won't be valid after such cases.
The session ID is exposed as part of the uapi so it can be used as part of userspace commands.
v2: use gt->uncore->rpm (Chris)
v3: s/arb_is_in_play/arb_is_valid (Chris), move set-up to the new
init_hw function
v4: move interface defs to separate header, set arb_is valid to false
on fini (Rodrigo)
v5: handle async component binding
--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/pxp/intel_pxp.c | 7 ++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 5 ++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 74 ++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 87 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_tee.h | 3 + .../drm/i915/pxp/intel_pxp_tee_interface.h | 37 ++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 12 +++ include/uapi/drm/i915_drm.h | 3 + 10 files changed, 244 insertions(+) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_session.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_session.h create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h
drivers/gpu/drm/i915/Makefile +1 -0
@@ -278,6 +278,7 @@ i915-y += i915_perf.o # Protected execution platform (PXP) support i915-$(CONFIG_DRM_I915_PXP) += \ pxp/intel_pxp.o \ + pxp/intel_pxp_session.o \ pxp/intel_pxp_tee.o # Post-mortem debug and GPU hang state capture
drivers/gpu/drm/i915/pxp/intel_pxp.c +7 -0
@@ -3,6 +3,7 @@* Copyright(c) 2020 Intel Corporation. */ #include "intel_pxp.h" +#include "intel_pxp_session.h" #include "intel_pxp_tee.h" #include "gt/intel_context.h" #include "i915_drv.h"
@@ -65,6 +66,8 @@ void intel_pxp_init(struct intel_pxp *pxp)if (!HAS_PXP (gt->i915))return ; +mutex_init (&pxp->tee_mutex); + ret =create_vcs_context (pxp);if (ret)return ;
@@ -86,6 +89,8 @@ void intel_pxp_fini(struct intel_pxp *pxp)if (!intel_pxp_is_enabled (pxp))return ; + pxp->arb_is_valid = false; +intel_pxp_tee_component_fini (pxp);destroy_vcs_context (pxp);
@@ -94,6 +99,8 @@ void intel_pxp_fini(struct intel_pxp *pxp)void intel_pxp_init_hw (struct intel_pxp *pxp) {kcr_pxp_enable (pxp_to_gt (pxp)); + +intel_pxp_create_arb_session (pxp); }void intel_pxp_fini_hw (struct intel_pxp *pxp)
drivers/gpu/drm/i915/pxp/intel_pxp.h +5 -0
@@ -19,6 +19,11 @@ static inline bool intel_pxp_is_enabled(const struct intel_pxp *pxp)return pxp->ce; } +static inline bool intel_pxp_is_active (const struct intel_pxp *pxp) +{ +return pxp->arb_is_valid; +} +#ifdef CONFIG_DRM_I915_PXP void intel_pxp_init (struct intel_pxp *pxp);void intel_pxp_fini (struct intel_pxp *pxp);
drivers/gpu/drm/i915/pxp/intel_pxp_session.c new file +74 -0
74 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp_session.h new file +15 -0
@@ -0,0 +1,15 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_SESSION_H__ +#define __INTEL_PXP_SESSION_H__ + +#include <linux/types.h> + +struct intel_pxp ; + +int intel_pxp_create_arb_session (struct intel_pxp *pxp); + +#endif /* __INTEL_PXP_SESSION_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +87 -0
124 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp_tee.h +3 -0
@@ -11,4 +11,7 @@int intel_pxp_tee_component_init (struct intel_pxp *pxp);void intel_pxp_tee_component_fini (struct intel_pxp *pxp); +int intel_pxp_tee_cmd_create_arb_session (struct intel_pxp *pxp, +int arb_session_id); +#endif /* __INTEL_PXP_TEE_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h new file +37 -0
@@ -0,0 +1,37 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_TEE_INTERFACE_H__ +#define __INTEL_PXP_TEE_INTERFACE_H__ + +#include <linux/types.h> + +#define PXP_TEE_APIVER 0x40002 +#define PXP_TEE_ARB_CMDID 0x1e +#define PXP_TEE_ARB_PROTECTION_MODE 0x2 + +/* PXP TEE message header */ +struct pxp_tee_cmd_header { +u32 api_version; +u32 command_id; +u32 status; +/* Length of the message (excluding the header) */ +u32 buffer_len; +} __packed; + +/* PXP TEE message input to create a arbitrary session */ +struct pxp_tee_create_arb_in { +struct pxp_tee_cmd_header header; +u32 protection_mode; +u32 session_id; +} __packed; + +/* PXP TEE message output to create a arbitrary session */ +struct pxp_tee_create_arb_out { +struct pxp_tee_cmd_header header; +} __packed; + + +#endif /* __INTEL_PXP_TEE_INTERFACE_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +12 -0
@@ -6,6 +6,9 @@#ifndef __INTEL_PXP_TYPES_H__#define __INTEL_PXP_TYPES_H__ +#include <linux/mutex.h> +#include <linux/types.h> +struct intel_context ;struct i915_pxp_component ;
@@ -13,6 +16,15 @@ struct intel_pxp {struct i915_pxp_component *pxp_component;struct intel_context *ce; + +/* +* After a teardown, the arb session can still be in play on the HW +* even if the keys are gone, so we can't rely on the HW state of the +* session to know if it's valid and need to track the status in SW. +*/ +bool arb_is_valid; + + struct mutex tee_mutex; /* protects the tee channel binding */ };#endif /* __INTEL_PXP_TYPES_H__ */
include/uapi/drm/i915_drm.h +3 -0
@@ -3018,6 +3018,9 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/* ID of the protected content session managed by i915 when PXP is active */ +#define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xf +#if defined(__cplusplus) }#endif
Signature · 1 line
2.32.0
PATCH v6 8/15 drm/i915/pxp: Implement arb session teardown
In reply to Daniele Ceraolo SpurioFrom: "Huang, Sean Z" <[email protected]>
Teardown is triggered when the display topology changes and no long meets the secure playback requirement, and hardware trashes all the encryption keys for display. Additionally, we want to emit a teardown operation to make sure we're clean on boot and resume
v2: emit in the ring, use high prio request (Chris)
v3: better defines, stalling flush, cleaned up and renamed submission
funcs (Chris)
--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/gt/intel_gpu_commands.h | 22 ++- drivers/gpu/drm/i915/pxp/intel_pxp.c | 7 +- drivers/gpu/drm/i915/pxp/intel_pxp_cmd.c | 141 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_cmd.h | 15 ++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 29 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.h | 1 + 7 files changed, 212 insertions(+), 4 deletions(-) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_cmd.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_cmd.h
drivers/gpu/drm/i915/Makefile +1 -0
@@ -278,6 +278,7 @@ i915-y += i915_perf.o # Protected execution platform (PXP) support i915-$(CONFIG_DRM_I915_PXP) += \ pxp/intel_pxp.o \ + pxp/intel_pxp_cmd.o \ pxp/intel_pxp_session.o \ pxp/intel_pxp_tee.o
drivers/gpu/drm/i915/gt/intel_gpu_commands.h +19 -3
@@ -28,10 +28,13 @@#define INSTR_26_TO_24_MASK 0x7000000#define INSTR_26_TO_24_SHIFT 24 +#define __INSTR (client) ((client) << INSTR_CLIENT_SHIFT) +/* * Memory interface instructions used by the kernel */ -#define MI_INSTR (opcode, flags) (((opcode) << 23) | (flags)) +#define MI_INSTR (opcode, flags) \ + (__INSTR(INSTR_MI_CLIENT) | (opcode) << 23 | (flags))/* Many MI commands use bit 22 of the header dword for GGTT vs PPGTT */ #define MI_GLOBAL_GTT (1<<22)
@@ -57,6 +60,7 @@#define MI_SUSPEND_FLUSH MI_INSTR(0x0b, 0)#define MI_SUSPEND_FLUSH_EN (1<<0)#define MI_SET_APPID MI_INSTR(0x0e, 0) +#define MI_SET_APPID_SESSION_ID (x) ((x) << 0)#define MI_OVERLAY_FLIP MI_INSTR(0x11, 0)#define MI_OVERLAY_CONTINUE (0x0<<21)#define MI_OVERLAY_ON (0x1<<21)
@@ -146,6 +150,7 @@#define MI_STORE_REGISTER_MEM_GEN8 MI_INSTR(0x24, 2)#define MI_SRM_LRM_GLOBAL_GTT (1<<22)#define MI_FLUSH_DW MI_INSTR(0x26, 1)/* for GEN6 */ +#define MI_FLUSH_DW_PROTECTED_MEM_EN (1<<22)#define MI_FLUSH_DW_STORE_INDEX (1<<21)#define MI_INVALIDATE_TLB (1<<18)#define MI_FLUSH_DW_OP_STOREDW (1<<14)
@@ -272,6 +277,19 @@#define MI_MATH_REG_ZF 0x32#define MI_MATH_REG_CF 0x33 +/* +* Media instructions used by the kernel +*/ +#define MEDIA_INSTR (pipe, op, sub_op, flags) \ + (__INSTR(INSTR_RC_CLIENT) | (pipe) << INSTR_SUBCLIENT_SHIFT | \ + (op) << INSTR_26_TO_24_SHIFT | (sub_op) << 16 | (flags)) + +#define MFX_WAIT MEDIA_INSTR(1, 0, 0, 0) +#define MFX_WAIT_DW0_MFX_SYNC_CONTROL_FLAG REG_BIT(8) +#define MFX_WAIT_DW0_PXP_SYNC_CONTROL_FLAG REG_BIT(9) + +#define CRYPTO_KEY_EXCHANGE MEDIA_INSTR(2, 6, 9, 0) +/* * Commands used only by the command parser */
@@ -328,8 +346,6 @@#define GFX_OP_3DSTATE_BINDING_TABLE_EDIT_PS \ ((0x3<<29)|(0x3<<27)|(0x0<<24)|(0x47<<16)) -#define MFX_WAIT ((0x3<<29)|(0x1<<27)|(0x0<<16)) -#define COLOR_BLT ((0x2<<29)|(0x40<<22))#define SRC_COPY_BLT ((0x2<<29)|(0x43<<22))
drivers/gpu/drm/i915/pxp/intel_pxp.c +6 -1
@@ -98,9 +98,14 @@ void intel_pxp_fini(struct intel_pxp *pxp)void intel_pxp_init_hw (struct intel_pxp *pxp) { +int ret; +kcr_pxp_enable (pxp_to_gt (pxp)); -intel_pxp_create_arb_session (pxp); +/* always emit a full termination to clean the state */ + ret =intel_pxp_terminate_arb_session_and_global (pxp); +if (!ret) +intel_pxp_create_arb_session (pxp); }void intel_pxp_fini_hw (struct intel_pxp *pxp)
drivers/gpu/drm/i915/pxp/intel_pxp_cmd.c new file +141 -0
141 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp_cmd.h new file +15 -0
@@ -0,0 +1,15 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_CMD_H__ +#define __INTEL_PXP_CMD_H__ + +#include <linux/types.h> + +struct intel_pxp ; + +int intel_pxp_terminate_session (struct intel_pxp *pxp,u32 idx); + +#endif /* __INTEL_PXP_CMD_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_session.c +29 -0
@@ -7,6 +7,7 @@#include "i915_drv.h" #include "intel_pxp.h" +#include "intel_pxp_cmd.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" #include "intel_pxp_types.h"
@@ -15,6 +16,9 @@#define GEN12_KCR_SIP _MMIO(0x32260)/* KCR hwdrm session in play 0-31 */ +/* PXP global terminate register for session termination */ +#define PXP_GLOBAL_TERMINATE _MMIO(0x320f8) +static bool intel_pxp_session_is_in_play (struct intel_pxp *pxp,u32 id) {struct intel_gt *gt =pxp_to_gt (pxp);
@@ -72,3 +76,28 @@ int intel_pxp_create_arb_session(struct intel_pxp *pxp)return 0 ; } + +int intel_pxp_terminate_arb_session_and_global (struct intel_pxp *pxp) +{ +int ret; +struct intel_gt *gt =pxp_to_gt (pxp); + + pxp->arb_is_valid = false; + +/* terminate the hw sessions */ + ret =intel_pxp_terminate_session (pxp,ARB_SESSION ); +if (ret) { +drm_err (>->i915->drm,"Failed to submit session termination\n" ); +return ret; + } + + ret =pxp_wait_for_session_state (pxp,ARB_SESSION , false); +if (ret) { +drm_err (>->i915->drm,"Session state did not clear\n" ); +return ret; + } + +intel_uncore_write (gt->uncore,PXP_GLOBAL_TERMINATE ,1 ); + +return ret; +}
drivers/gpu/drm/i915/pxp/intel_pxp_session.h +1 -0
@@ -11,5 +11,6 @@struct intel_pxp ;int intel_pxp_create_arb_session (struct intel_pxp *pxp); +int intel_pxp_terminate_arb_session_and_global (struct intel_pxp *pxp);#endif /* __INTEL_PXP_SESSION_H__ */
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 9/15 drm/i915/pxp: Implement PXP irq handler
In reply to Daniele Ceraolo SpurioFrom: "Huang, Sean Z" <[email protected]>
The HW will generate a teardown interrupt when session termination is required, which requires i915 to submit a terminating batch. Once the HW is done with the termination it will generate another interrupt, at which point it is safe to re-create the session.
Since the termination and re-creation flow is something we want to trigger from the driver as well, use a common work function that can be called both from the irq handler and from the driver set-up flows, which has the addded benefit of allowing us to skip any extra locks because the work itself serializes the operations.
v2: use struct completion instead of bool (Chris)
v3: drop locks, clean up functions and improve comments (Chris),
move to common work function.
v4: improve comments, simplify wait logic (Rodrigo)
v5: unconditionally set interrupts, rename state_attacked var (Rodrigo)
--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/gt/intel_gt_irq.c | 7 ++ drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/pxp/intel_pxp.c | 66 +++++++++++-- drivers/gpu/drm/i915/pxp/intel_pxp.h | 8 ++ drivers/gpu/drm/i915/pxp/intel_pxp_irq.c | 99 ++++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_irq.h | 32 +++++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 54 ++++++++++- drivers/gpu/drm/i915/pxp/intel_pxp_session.h | 5 +- drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 8 +- drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 18 ++++ 11 files changed, 283 insertions(+), 16 deletions(-) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_irq.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_irq.h
drivers/gpu/drm/i915/Makefile +1 -0
@@ -279,6 +279,7 @@ i915-y += i915_perf.o i915-$(CONFIG_DRM_I915_PXP) += \ pxp/intel_pxp.o \ pxp/intel_pxp_cmd.o \ + pxp/intel_pxp_irq.o \ pxp/intel_pxp_session.o \ pxp/intel_pxp_tee.o
drivers/gpu/drm/i915/gt/intel_gt_irq.c +7 -0
@@ -13,6 +13,7 @@#include "intel_lrc_reg.h" #include "intel_uncore.h" #include "intel_rps.h" +#include "pxp/intel_pxp_irq.h" static void guc_irq_handler (struct intel_guc *guc,u16 iir) {
@@ -64,6 +65,9 @@ gen11_other_irq_handler(struct intel_gt *gt, const u8 instance,if (instance ==OTHER_GTPM_INSTANCE )return gen11_rps_irq_handler (>->rps, iir); +if (instance ==OTHER_KCR_INSTANCE ) +return intel_pxp_irq_handler (>->pxp, iir); +WARN_ONCE (1 ,"unhandled other interrupt instance=0x%x, iir=0x%x\n" , instance, iir); }
@@ -196,6 +200,9 @@ void gen11_gt_irq_reset(struct intel_gt *gt)intel_uncore_write (uncore,GEN11_GPM_WGBOXPERF_INTR_MASK , ~0 );intel_uncore_write (uncore,GEN11_GUC_SG_INTR_ENABLE ,0 );intel_uncore_write (uncore,GEN11_GUC_SG_INTR_MASK , ~0 ); + +intel_uncore_write (uncore,GEN11_CRYPTO_RSVD_INTR_ENABLE ,0 ); +intel_uncore_write (uncore,GEN11_CRYPTO_RSVD_INTR_MASK , ~0 ); }void gen11_gt_irq_postinstall (struct intel_gt *gt)
drivers/gpu/drm/i915/i915_reg.h +1 -0
@@ -8086,6 +8086,7 @@ enum {/* irq instances for OTHER_CLASS */ #define OTHER_GUC_INSTANCE 0#define OTHER_GTPM_INSTANCE 1 +#define OTHER_KCR_INSTANCE 4#define GEN11_INTR_IDENTITY_REG (x) _MMIO(0x190060 + ((x) * 4))
drivers/gpu/drm/i915/pxp/intel_pxp.c +60 -6
92 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp.h +8 -0
@@ -30,6 +30,9 @@ void intel_pxp_fini(struct intel_pxp *pxp);void intel_pxp_init_hw (struct intel_pxp *pxp);void intel_pxp_fini_hw (struct intel_pxp *pxp); + +void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp); +int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {
@@ -38,6 +41,11 @@ static inline void intel_pxp_init(struct intel_pxp *pxp)static inline void intel_pxp_fini (struct intel_pxp *pxp) { } + +static inline int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp) +{ +return 0 ; +}#endif #endif /* __INTEL_PXP_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_irq.c new file +99 -0
99 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp_irq.h new file +32 -0
@@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_IRQ_H__ +#define __INTEL_PXP_IRQ_H__ + +#include <linux/types.h> + +struct intel_pxp ; + +#define GEN12_DISPLAY_PXP_STATE_TERMINATED_INTERRUPT BIT(1) +#define GEN12_DISPLAY_APP_TERMINATED_PER_FW_REQ_INTERRUPT BIT(2) +#define GEN12_DISPLAY_STATE_RESET_COMPLETE_INTERRUPT BIT(3) + +#define GEN12_PXP_INTERRUPTS \ + (GEN12_DISPLAY_PXP_STATE_TERMINATED_INTERRUPT | \ + GEN12_DISPLAY_APP_TERMINATED_PER_FW_REQ_INTERRUPT | \ + GEN12_DISPLAY_STATE_RESET_COMPLETE_INTERRUPT) + +#ifdef CONFIG_DRM_I915_PXP +void intel_pxp_irq_enable (struct intel_pxp *pxp); +void intel_pxp_irq_disable (struct intel_pxp *pxp); +void intel_pxp_irq_handler (struct intel_pxp *pxp,u16 iir); +#else +static inline void intel_pxp_irq_handler (struct intel_pxp *pxp,u16 iir) +{ +} +#endif + +#endif /* __INTEL_PXP_IRQ_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_session.c +51 -3
73 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp_session.h +2 -3
@@ -8,9 +8,8 @@#include <linux/types.h> -struct intel_pxp ; +struct work_struct ; -int intel_pxp_create_arb_session (struct intel_pxp *pxp); -int intel_pxp_terminate_arb_session_and_global (struct intel_pxp *pxp); +void intel_pxp_session_work (struct work_struct *work);#endif /* __INTEL_PXP_SESSION_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +4 -4
@@ -80,6 +80,7 @@ static int i915_pxp_tee_component_bind(struct device *i915_kdev, {struct drm_i915_private *i915 =kdev_to_i915 (i915_kdev);struct intel_pxp *pxp =i915_dev_to_pxp (i915_kdev); +int ret;mutex_lock (&pxp->tee_mutex); pxp->pxp_component = data;
@@ -88,15 +89,14 @@ static int i915_pxp_tee_component_bind(struct device *i915_kdev,/* the component is required to fully start the PXP HW */ intel_pxp_init_hw (pxp); - -if (!pxp->arb_is_valid) { + ret =intel_pxp_wait_for_arb_start (pxp); +if (ret) {drm_err (&i915->drm,"Failed to create arb session during bind\n" );intel_pxp_fini_hw (pxp); pxp->pxp_component =NULL ; -return -EIO ; } -return 0 ; +return ret; } static void i915_pxp_tee_component_unbind(struct device *i915_kdev,
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +18 -0
@@ -6,8 +6,10 @@#ifndef __INTEL_PXP_TYPES_H__#define __INTEL_PXP_TYPES_H__ +#include <linux/completion.h> #include <linux/mutex.h> #include <linux/types.h> +#include <linux/workqueue.h> struct intel_context ;struct i915_pxp_component ;
@@ -25,6 +27,22 @@ struct intel_pxp {bool arb_is_valid;struct mutex tee_mutex;/* protects the tee channel binding */ + +/* +* If the HW perceives an attack on the integrity of the encryption it +* will invalidate the keys and expect SW to re-initialize the session. +* We keep track of this state to make sure we only re-start the arb +* session when required. +*/ +bool hw_state_invalidated; + +bool irq_enabled; +struct completion termination; + +struct work_struct session_work; +u32 session_events;/* protected with gt->irq_lock */ +#define PXP_TERMINATION_REQUEST BIT(0) +#define PXP_TERMINATION_COMPLETE BIT(1) };#endif /* __INTEL_PXP_TYPES_H__ */
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH v6 09/15] drm/i915/pxp: Implement PXP irq handler
Quoted 507 lines · Daniele Ceraolo Spurio
From: "Huang, Sean Z" <[email protected]>
The HW will generate a teardown interrupt when session termination is required, which requires i915 to submit a terminating batch. Once the HW is done with the termination it will generate another interrupt, at which point it is safe to re-create the session.
Since the termination and re-creation flow is something we want to trigger from the driver as well, use a common work function that can be called both from the irq handler and from the driver set-up flows, which has the addded benefit of allowing us to skip any extra locks because the work itself serializes the operations.
v2: use struct completion instead of bool (Chris)
v3: drop locks, clean up functions and improve comments (Chris),
move to common work function.
v4: improve comments, simplify wait logic (Rodrigo)
v5: unconditionally set interrupts, renamestate_attackedvar (Rodrigo)Signed-off-by Huang, Sean Z <[email protected]>Signed-off-by Daniele Ceraolo Spurio <[email protected]>Cc Chris Wilson <[email protected]>Cc Rodrigo Vivi <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/gt/intel_gt_irq.c | 7 ++ drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/pxp/intel_pxp.c | 66 +++++++++++-- drivers/gpu/drm/i915/pxp/intel_pxp.h | 8 ++ drivers/gpu/drm/i915/pxp/intel_pxp_irq.c | 99 ++++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_irq.h | 32 +++++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 54 ++++++++++- drivers/gpu/drm/i915/pxp/intel_pxp_session.h | 5 +- drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 8 +- drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 18 ++++ 11 files changed, 283 insertions(+), 16 deletions(-) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_irq.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_irq.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -279,6 +279,7 @@ i915-y += i915_perf.o i915-$(CONFIG_DRM_I915_PXP) += \ pxp/intel_pxp.o \ pxp/intel_pxp_cmd.o \ + pxp/intel_pxp_irq.o \ pxp/intel_pxp_session.o \ pxp/intel_pxp_tee.odrivers/gpu/drm/i915/gt/intel_gt_irq.c +7
-0@@ -13,6 +13,7 @@#include "intel_lrc_reg.h" #include "intel_uncore.h" #include "intel_rps.h" +#include "pxp/intel_pxp_irq.h" static void guc_irq_handler (struct intel_guc *guc,u16 iir) {@@ -64,6 +65,9 @@ gen11_other_irq_handler(struct intel_gt *gt, const u8 instance,if (instance ==OTHER_GTPM_INSTANCE )return gen11_rps_irq_handler (>->rps, iir); +if (instance ==OTHER_KCR_INSTANCE ) +return intel_pxp_irq_handler (>->pxp, iir); +WARN_ONCE (1 ,"unhandled other interrupt instance=0x%x, iir=0x%x\n" , instance, iir); }@@ -196,6 +200,9 @@ void gen11_gt_irq_reset(struct intel_gt *gt)intel_uncore_write (uncore,GEN11_GPM_WGBOXPERF_INTR_MASK , ~0 );intel_uncore_write (uncore,GEN11_GUC_SG_INTR_ENABLE ,0 );intel_uncore_write (uncore,GEN11_GUC_SG_INTR_MASK , ~0 ); + +intel_uncore_write (uncore,GEN11_CRYPTO_RSVD_INTR_ENABLE ,0 ); +intel_uncore_write (uncore,GEN11_CRYPTO_RSVD_INTR_MASK , ~0 ); }void gen11_gt_irq_postinstall (struct intel_gt *gt)drivers/gpu/drm/i915/i915_reg.h +1
-0@@ -8086,6 +8086,7 @@ enum {/* irq instances for OTHER_CLASS */ #define OTHER_GUC_INSTANCE 0#define OTHER_GTPM_INSTANCE 1 +#define OTHER_KCR_INSTANCE 4#define GEN11_INTR_IDENTITY_REG (x) _MMIO(0x190060 + ((x) * 4))drivers/gpu/drm/i915/pxp/intel_pxp.c +60
-6@@ -2,7 +2,9 @@/* * Copyright(c) 2020 Intel Corporation. */ +#include <linux/workqueue.h> #include "intel_pxp.h" +#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" #include "gt/intel_context.h" @@ -68,6 +70,16 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +/* +* we'll use the completion to check if there is a termination pending, +* so we start it as completed and we reinit it when a termination +* is triggered. +*/ +init_completion (&pxp->termination); +complete_all (&pxp->termination); + +INIT_WORK (&pxp->session_work, intel_pxp_session_work); + ret =create_vcs_context (pxp);if (ret)return ;@@ -96,19 +108,61 @@ void intel_pxp_fini(struct intel_pxp *pxp)destroy_vcs_context (pxp); } -void intel_pxp_init_hw (struct intel_pxp *pxp) +void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp) { -int ret; + pxp->arb_is_valid = false; +reinit_completion (&pxp->termination); +} + +static void intel_pxp_queue_termination (struct intel_pxp *pxp) +{ +struct intel_gt *gt =pxp_to_gt (pxp); + +/* +* We want to get the same effect as if we received a termination +* interrupt, so just pretend that we did. +*/ +spin_lock_irq (>->irq_lock); +intel_pxp_mark_termination_in_progress (pxp); + pxp->session_events |=PXP_TERMINATION_REQUEST ; +queue_work (system_unbound_wq, &pxp->session_work); +spin_unlock_irq (>->irq_lock); +} +/* +* the arb session is restarted from the irq work when we receive the +* termination completion interrupt +*/ +int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp) +{ +if (!intel_pxp_is_enabled (pxp)) +return 0 ; + +if (!wait_for_completion_timeout (&pxp->termination, +msecs_to_jiffies (100 ))) +return -ETIMEDOUT ; + +if (!pxp->arb_is_valid) +return -EIO ; + +return 0 ; +} + +void intel_pxp_init_hw (struct intel_pxp *pxp) +{kcr_pxp_enable (pxp_to_gt (pxp)); +intel_pxp_irq_enable (pxp); -/* always emit a full termination to clean the state */ - ret =intel_pxp_terminate_arb_session_and_global (pxp); -if (!ret) -intel_pxp_create_arb_session (pxp); +/* +* the session could've been attacked while we weren't loaded, so +* handle it as if it was and re-create it. +*/ +intel_pxp_queue_termination (pxp); }void intel_pxp_fini_hw (struct intel_pxp *pxp) {kcr_pxp_disable (pxp_to_gt (pxp)); + +intel_pxp_irq_disable (pxp); }drivers/gpu/drm/i915/pxp/intel_pxp.h +8
-0@@ -30,6 +30,9 @@ void intel_pxp_fini(struct intel_pxp *pxp);void intel_pxp_init_hw (struct intel_pxp *pxp);void intel_pxp_fini_hw (struct intel_pxp *pxp); + +void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp); +int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -38,6 +41,11 @@ static inline void intel_pxp_init(struct intel_pxp *pxp)static inline void intel_pxp_fini (struct intel_pxp *pxp) { } + +static inline int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp) +{ +return 0 ; +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_irq.c new file +99
-0@@ -0,0 +1,99 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright(c) 2020 Intel Corporation. +*/ +#include <linux/workqueue.h> +#include "intel_pxp.h" +#include "intel_pxp_irq.h" +#include "intel_pxp_session.h" +#include "gt/intel_gt_irq.h" +#include "i915_irq.h" +#include "i915_reg.h" + +/** +* intel_pxp_irq_handler - Handles PXP interrupts. +* @pxp: pointer to pxp struct +* @iir: interrupt vector +*/ +void intel_pxp_irq_handler (struct intel_pxp *pxp,u16 iir) +{ +struct intel_gt *gt =pxp_to_gt (pxp); + +if (GEM_WARN_ON (!intel_pxp_is_enabled (pxp))) +return ; + +lockdep_assert_held (>->irq_lock); + +if (unlikely (!iir)) +return ; + +if (iir & (GEN12_DISPLAY_PXP_STATE_TERMINATED_INTERRUPT | +GEN12_DISPLAY_APP_TERMINATED_PER_FW_REQ_INTERRUPT )) { +/* immediately mark PXP as inactive on termination */ +intel_pxp_mark_termination_in_progress (pxp); + pxp->session_events |=PXP_TERMINATION_REQUEST ; + } + +if (iir &GEN12_DISPLAY_STATE_RESET_COMPLETE_INTERRUPT ) + pxp->session_events |=PXP_TERMINATION_COMPLETE ; + +if (pxp->session_events) +queue_work (system_unbound_wq, &pxp->session_work); +} + +static inline void __pxp_set_interrupts (struct intel_gt *gt,u32 interrupts) +{ +struct intel_uncore *uncore = gt->uncore; +const u32 mask = interrupts <<16 ; + +intel_uncore_write (uncore,GEN11_CRYPTO_RSVD_INTR_ENABLE , mask); +intel_uncore_write (uncore,GEN11_CRYPTO_RSVD_INTR_MASK , ~mask); +} + +static inline void pxp_irq_reset (struct intel_gt *gt) +{ +spin_lock_irq (>->irq_lock); +gen11_gt_reset_one_iir (gt,0 ,GEN11_KCR ); +spin_unlock_irq (>->irq_lock); +} + +void intel_pxp_irq_enable (struct intel_pxp *pxp) +{ +struct intel_gt *gt =pxp_to_gt (pxp); + +spin_lock_irq (>->irq_lock); + +if (!pxp->irq_enabled) +WARN_ON_ONCE (gen11_gt_reset_one_iir (gt,0 ,GEN11_KCR )); + +__pxp_set_interrupts (gt,GEN12_PXP_INTERRUPTS ); + pxp->irq_enabled = true; + +spin_unlock_irq (>->irq_lock); +} + +void intel_pxp_irq_disable (struct intel_pxp *pxp) +{ +struct intel_gt *gt =pxp_to_gt (pxp); + +/* +* We always need to submit a global termination when we re-enable the +* interrupts, so there is no need to make sure that the session state +* makes sense at the end of this function. Just make sure this is not +* called in a path were the driver consider the session as valid and +* doesn't call a termination on restart. +*/ +GEM_WARN_ON (intel_pxp_is_active (pxp)); + +spin_lock_irq (>->irq_lock); + + pxp->irq_enabled = false; +__pxp_set_interrupts (gt,0 ); + +spin_unlock_irq (>->irq_lock); +intel_synchronize_irq (gt->i915); + +pxp_irq_reset (gt); + +flush_work (&pxp->session_work); +}drivers/gpu/drm/i915/pxp/intel_pxp_irq.h new file +32
-0@@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_IRQ_H__ +#define __INTEL_PXP_IRQ_H__ + +#include <linux/types.h> + +struct intel_pxp ; + +#define GEN12_DISPLAY_PXP_STATE_TERMINATED_INTERRUPT BIT(1) +#define GEN12_DISPLAY_APP_TERMINATED_PER_FW_REQ_INTERRUPT BIT(2) +#define GEN12_DISPLAY_STATE_RESET_COMPLETE_INTERRUPT BIT(3) + +#define GEN12_PXP_INTERRUPTS \ + (GEN12_DISPLAY_PXP_STATE_TERMINATED_INTERRUPT | \ + GEN12_DISPLAY_APP_TERMINATED_PER_FW_REQ_INTERRUPT | \ + GEN12_DISPLAY_STATE_RESET_COMPLETE_INTERRUPT) + +#ifdef CONFIG_DRM_I915_PXP +void intel_pxp_irq_enable (struct intel_pxp *pxp); +void intel_pxp_irq_disable (struct intel_pxp *pxp); +void intel_pxp_irq_handler (struct intel_pxp *pxp,u16 iir); +#else +static inline void intel_pxp_irq_handler (struct intel_pxp *pxp,u16 iir) +{ +} +#endif + +#endif /* __INTEL_PXP_IRQ_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +51
-3@@ -48,7 +48,7 @@ static int pxp_wait_for_session_state(struct intel_pxp *pxp, u32 id, bool in_plareturn ret; } -int intel_pxp_create_arb_session (struct intel_pxp *pxp) +static int pxp_create_arb_session (struct intel_pxp *pxp) {struct intel_gt *gt =pxp_to_gt (pxp);int ret;@@ -77,12 +77,13 @@ int intel_pxp_create_arb_session(struct intel_pxp *pxp)return 0 ; } -int intel_pxp_terminate_arb_session_and_global (struct intel_pxp *pxp) +static int pxp_terminate_arb_session_and_global (struct intel_pxp *pxp) {int ret;struct intel_gt *gt =pxp_to_gt (pxp); - pxp->arb_is_valid = false; +/* must mark termination in progress calling this function */ +GEM_WARN_ON (pxp->arb_is_valid);/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );@@ -101,3 +102,50 @@ int intel_pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)return ret; } + +static void pxp_terminate (struct intel_pxp *pxp) +{ +int ret; + + pxp->hw_state_invalidated = true; + +/* +* if we fail to submit the termination there is no point in waiting for +* it to complete. PXP will be marked as non-active until the next +* termination is issued. +*/ + ret =pxp_terminate_arb_session_and_global (pxp); +if (ret) +complete_all (&pxp->termination); +} + +static void pxp_terminate_complete (struct intel_pxp *pxp) +{ +/* Re-create the arb session after teardown handle complete */ +if (fetch_and_zero (&pxp->hw_state_invalidated)) +pxp_create_arb_session (pxp); + +complete_all (&pxp->termination); +} + +void intel_pxp_session_work (struct work_struct *work) +{ +struct intel_pxp *pxp =container_of (work,typeof (*pxp), session_work); +struct intel_gt *gt =pxp_to_gt (pxp); +u32 events =0 ; + +spin_lock_irq (>->irq_lock); + events =fetch_and_zero (&pxp->session_events); +spin_unlock_irq (>->irq_lock); + +if (!events) +return ; + +if (events &PXP_TERMINATION_REQUEST ) { + events &= ~PXP_TERMINATION_COMPLETE ; +pxp_terminate (pxp); + } + +if (events &PXP_TERMINATION_COMPLETE ) +pxp_terminate_complete (pxp); +}drivers/gpu/drm/i915/pxp/intel_pxp_session.h +2
-3@@ -8,9 +8,8 @@#include <linux/types.h> -struct intel_pxp ; +struct work_struct ; -int intel_pxp_create_arb_session (struct intel_pxp *pxp); -int intel_pxp_terminate_arb_session_and_global (struct intel_pxp *pxp); +void intel_pxp_session_work (struct work_struct *work);#endif /* __INTEL_PXP_SESSION_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +4
-4@@ -80,6 +80,7 @@ static int i915_pxp_tee_component_bind(struct device *i915_kdev, {struct drm_i915_private *i915 =kdev_to_i915 (i915_kdev);struct intel_pxp *pxp =i915_dev_to_pxp (i915_kdev); +int ret;mutex_lock (&pxp->tee_mutex); pxp->pxp_component = data;@@ -88,15 +89,14 @@ static int i915_pxp_tee_component_bind(struct device *i915_kdev,/* the component is required to fully start the PXP HW */ intel_pxp_init_hw (pxp); - -if (!pxp->arb_is_valid) { + ret =intel_pxp_wait_for_arb_start (pxp); +if (ret) {drm_err (&i915->drm,"Failed to create arb session during bind\n" );intel_pxp_fini_hw (pxp); pxp->pxp_component =NULL ; -return -EIO ; } -return 0 ; +return ret; } static void i915_pxp_tee_component_unbind(struct device *i915_kdev,drivers/gpu/drm/i915/pxp/intel_pxp_types.h +2
-0@@ -6,8 +6,10 @@#ifndef __INTEL_PXP_TYPES_H__#define __INTEL_PXP_TYPES_H__ +#include <linux/completion.h> #include <linux/mutex.h> #include <linux/types.h> +#include <linux/workqueue.h> struct intel_context ;struct i915_pxp_component ;
drivers/gpu/drm/i915/pxp/intel_pxp_types.h struct intel_pxp { · line 36
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +7
-0@@ -25,6 +27,22 @@ struct intel_pxp {bool arb_is_valid;struct mutex tee_mutex;/* protects the tee channel binding */ + +/* +* If the HW perceives an attack on the integrity of the encryption it +* will invalidate the keys and expect SW to re-initialize the session. +* We keep track of this state to make sure we only re-start the arb +* session when required. +*/
Please convert these into kerneldoc, pull them somewhere into driver docs, and then document the locking rules for all the members here. At least where relevant. Even if the locking rules are "only &session_work mutates this completion" or whatever the rules are.
Yeah I know this is very late, but I've read too much i915-gem/gt/core locking code, and it's way too much incomprehensible overall. And probably worse, not documented at all. So we need to step up here. -Daniel
Quoted 19 lines
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +9
-0+bool hw_state_invalidated; + +bool irq_enabled; +struct completion termination; + +struct work_struct session_work; +u32 session_events;/* protected with gt->irq_lock */ +#define PXP_TERMINATION_REQUEST BIT(0) +#define PXP_TERMINATION_COMPLETE BIT(1) };#endif /* __INTEL_PXP_TYPES_H__ */ Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
PATCH v6 10/15 drm/i915/pxp: interfaces for using protected objects
In reply to Daniele Ceraolo SpurioThis api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the RESET_STATS ioctl to report the context invalidation to userspace.
This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on proto_ctx, update kerneldoc
v6: rebase on obj create_ext changes
--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)
drivers/gpu/drm/i915/gem/i915_gem_context.c +62 -6
@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h"
@@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {
@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;
@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :
@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :
@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)
@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;
@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :
@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :
@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :
@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }
drivers/gpu/drm/i915/gem/i915_gem_context.h +18 -0
@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); } +static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags); +} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);
drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2 -0
@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans
@@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;
drivers/gpu/drm/i915/gem/i915_gem_create.c +57 -18
119 lines not on this page · raw message
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +39 -1
@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h"
@@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;
@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);
@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);
@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ; + goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)
drivers/gpu/drm/i915/gem/i915_gem_object.c +6 -0
@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h"
@@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;
@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));
drivers/gpu/drm/i915/gem/i915_gem_object.h +12 -0
@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9 -0
@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags
@@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;
drivers/gpu/drm/i915/pxp/intel_pxp.c +89 -0
104 lines not on this page · raw message
drivers/gpu/drm/i915/pxp/intel_pxp.h +15 -0
@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);
@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {
@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3 -0
@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5 -0
@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h>
@@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */
include/uapi/drm/i915_drm.h +54 -1
86 lines not on this page · raw message
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Rodrigo Vivi
maintainerRe: [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 15 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.
Quoted 7 lines
This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchanges
The "rebase" word led me to think it was a small change caused only by rebasing conflicts, but then I spotted something on i915_gem_create.c that I didn't remember seeing before.
Since it took me a while to understand what was going on, let me try to summarize to see if I got it right and to check with others (Jason in special)
if we are aligned with the recent directions:
With the removal of the vanilla_object from the create_ext, the addition of the flags was needed.
Then, instead of adding a new user_flags it adds the I915_BO_PROTECTED to the existent flags (what is cleaner) but then it does
/* Add any flag set by create_ext options */
flags |= ext_flags;
on the new user_create_ext.
What shouldn't be a problem because the ext_flags is really only the I915_BO_PROTECTED set by our new extension and immutable.
But I'm just trying to see if we are not opening any holes to accept some undesired flags.
Apparently not.
Am I getting things right? Anything else we should check in here?
Thanks,
Rodrigo.
Quoted 801 lines
Signed-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +18
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); } +static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags); +} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +39
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ; + goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +89
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp) +{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 1 line
2.32.0
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 20 lines · Rodrigo Vivi
Quoted 19 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
Quoted 6 lines
Quoted 3 lines
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesThe "rebase" word led me to think it was a small change caused only by rebasing conflicts, but then I spotted something on
i915_gem_create.cthat I didn't remember seeing before.
Apologies for not being clear.
Quoted 8 lines
Since it took me a while to understand what was going on, let me try to summarize to see if I got it right and to check with others (Jason in special)
if we are aligned with the recent directions:With the removal of the
vanilla_objectfrom thecreate_ext, the addition of the flags was needed.
yes.
Quoted 9 lines
Then, instead of adding a new
user_flagsit adds theI915_BO_PROTECTEDto the existent flags (what is cleaner) but then it does/* Add any flag set by
create_extoptions */
flags |=ext_flags;on the new
user_create_ext.
Quoted 7 lines
What shouldn't be a problem because the
ext_flagsis really only theI915_BO_PROTECTEDset by our new extension and immutable.But I'm just trying to see if we are not opening any holes to accept some undesired flags.
Apparently not.
ext_flags are not set directly by the user, but by the create_ext extension functions, based on the extensions the user has provided. This should allow the kernel to perform any required checks in the extension functions before setting the flag (like it happens in this case for the PXP option).
Quoted 2 lines
Am I getting things right? Anything else we should check in here?
yup, got it right.
Daniele
Quoted 804 lines
Thanks,
Rodrigo.Quoted 800 lines
Signed-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +18
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); } +static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags); +} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +39
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ; + goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +89
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp) +{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 1 line
2.32.0
Re: [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 66 lines · Daniele Ceraolo Spurio
Quoted 20 lines · Rodrigo Vivi
Quoted 19 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
Quoted 6 lines
Quoted 3 lines
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesThe "rebase" word led me to think it was a small change caused only by rebasing conflicts, but then I spotted something on
i915_gem_create.cthat I didn't remember seeing before.Apologies for not being clear.
Quoted 8 lines
Since it took me a while to understand what was going on, let me try to summarize to see if I got it right and to check with others (Jason in special)
if we are aligned with the recent directions:With the removal of the
vanilla_objectfrom thecreate_ext, the addition of the flags was needed.yes.
Quoted 9 lines
Then, instead of adding a new
user_flagsit adds theI915_BO_PROTECTEDto the existent flags (what is cleaner) but then it does/* Add any flag set by
create_extoptions */
flags |=ext_flags;on the new
user_create_ext.Quoted 7 lines
What shouldn't be a problem because the
ext_flagsis really only theI915_BO_PROTECTEDset by our new extension and immutable.But I'm just trying to see if we are not opening any holes to accept some undesired flags.
Apparently not.
ext_flagsare not set directly by the user, but by thecreate_extextension functions, based on the extensions the user has provided. This should allow the kernel to perform any required checks in the extension functions before setting the flag (like it happens in this case for the PXP option).Quoted 1 line
Quoted 3 lines
Quoted 1 line
Am I getting things right? Anything else we should check in here?
yup, got it right.
Great, thanks for confirming
so I'm okay with:
Quoted 752 lines
Daniele
Quoted 748 lines
Thanks,
Rodrigo.Quoted 744 lines
Signed-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +18
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); } +static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags); +} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +39
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ; + goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +89
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp) +{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 1 line
2.32.0
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
In reply to Daniele Ceraolo SpurioQuoted 206 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 120
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);
Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
This ctx->flags is atomic, unorderd, and that means you need barriers and everything.
If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 6 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);
For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 217 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +28
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +7
-0@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;
Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 135 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;
drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 203
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)
There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
- kref_get_unless_zero
- test_bit (that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops like intel_context_set_banned or i915_gem_context_set_invalid also need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.
Quoted 233 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +54
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 551 lines · Daniel Vetter
Quoted 206 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 120
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 6 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 217 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +28
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +7
-0@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 77 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +1
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm;
Quoted 7 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h struct drm_i915_gem_object { · line 547
Quoted 7 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +7
-0+/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link;
This
- seems to be accessed locklessly, which needs giant amounts of comments and justifications to prove a) it's needed and b) it's correct against all races
- we already have a per-object lock in the form of dma_resv_lock. Why can't we use that to protect pxp object state?
Looking needs to be done with extreme care, or we're just burrying ourselves into something that cannot be understood and maintained anymore. -Daniel
Quoted 75 lines
Quoted 51 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +1
-0+/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 203
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 213
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link);
We have a per
Quoted 228 lines
Quoted 223 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +44
-0+ } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfxSignature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 549 lines · Daniel Vetter
Quoted 548 lines · Daniel Vetter
Quoted 206 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 120
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 6 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 217 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +28
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +7
-0@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 77 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +1
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm;
Quoted 13 lines
Quoted 7 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h struct drm_i915_gem_object { · line 547
Quoted 7 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +7
-0+/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link;This
- seems to be accessed locklessly, which needs giant amounts of comments and justifications to prove a) it's needed and b) it's correct against all races
- we already have a per-object lock in the form ofdma_resv_lock. Why can't we use that to protect pxp object state?
This is always modified under pxp->lock. The only lockless access we do is a list_empty() check to see if the protection is valid, and that's because the check is inherently racy anyway (state can change at any point before or after we've checked) so adding a lock wouldn't really help. Note that the list can only be set at object creation time and it is only cleared either during the first invalidation or during object destruction.
Daniele
Quoted 316 lines
Looking needs to be done with extreme care, or we're just burrying ourselves into something that cannot be understood and maintained anymore. -Daniel
Quoted 74 lines
Quoted 51 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +1
-0+/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 203
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 213
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link);We have a per
Quoted 227 lines
Quoted 223 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +44
-0+ } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfxSignature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 528 lines · Daniele Ceraolo Spurio
Quoted 516 lines · Daniel Vetter
Quoted 515 lines · Daniel Vetter
Quoted 196 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 119
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 6 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 199 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +28
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2786
Quoted 8 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +7
-0@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 74 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +6
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. Quoted 8 lines
Quoted 2 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h struct drm_i915_gem_object { · line 547
Quoted 2 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +2
-0+*/ +struct list_head pxp_link;This
- seems to be accessed locklessly, which needs giant amounts of comments and justifications to prove a) it's needed and b) it's correct against all races
- we already have a per-object lock in the form ofdma_resv_lock. Why can't we use that to protect pxp object state?
Quoted 6 lines
This is always modified under
pxp->lock. The only lockless access we do is alist_empty()check to see if the protection is valid, and that's because the check is inherently racy anyway (state can change at any point before or after we've checked) so adding a lock wouldn't really help. Note that the list can only be set at object creation time and it is only cleared either during the first invalidation or during object destruction.
Yeah, but every time we have races you need to think through whether really nothing bad can ever happen. I pondered this a bit and came up with the following idea:
- every time we get a pxp reset, we increment a generation counter. That should never race, because we should be able to sufficiently protect pxp state with some mutext.
- when we allocate a protect context or a protected buffer, we remember the current generation counter. This can race, but also, it doesn't matter because userspace is supposed allocate buffers after the context was allocated. This pxp_generation is invariant over the lifetime of an object/context, at least for now. We can sample this pxp_generation holding the same lock as the pxp reset, which avoids another pile of potentially confusion races.
- for execbuf context side we entirely rely on the existing context banning.
- for the buffer objects we compare the object's pxp_generation against the contexts. If they don't match we fail.
This means no locking to protected this, no additional races to ponder. Also pretty easy way reset a buffer for the non-robust version (if that ever happens), we just update the pxp_generation.
-Daniel
Quoted 302 lines
Daniele
Quoted 298 lines
Looking needs to be done with extreme care, or we're just burrying ourselves into something that cannot be understood and maintained anymore. -Daniel
Quoted 69 lines
Quoted 46 lines
drivers/gpu/drm/i915/gem/i915_gem_object_types.h +1
-0+/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 202
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 213
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link);We have a per
Quoted 214 lines
Quoted 210 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +44
-0+ } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfxSignature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
In reply to Daniel VetterQuoted 220 lines · Daniel Vetter
Quoted 206 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 120
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
Quoted 6 lines
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
The BANNED flag is at the intel_context level, while I needed something at the gem_context level to report out. Can make it a bool if you prefer.
The invalidation process is inherently racy (the teardown can always occur after we've checked the flag) so no amount of locking can help there. The flag itself is only set once and never cleared, so no risk of a set and clear racing.
Quoted 9 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 7 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
It is immutable. Should we have a ctx->immutable_flags, or better a bool?
Quoted 225 lines
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 220 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +28
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt);
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 7 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +7
-0+if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); +if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
This is to bail out early and report the error to userspace. If the invalidation flag gets set after this check then we'll still catch it later as you said, but IMO there is value in bailing out early when possible, especially since it is a low-effort check. The invalidation flag can't be cleared, so no risk of rejecting something that shouldn't be.
Daniele
Quoted 392 lines
Quoted 136 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 203
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.Quoted 233 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +54
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 220 lines · Daniele Ceraolo Spurio
Quoted 210 lines · Daniel Vetter
Quoted 196 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 119
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
Quoted 6 lines
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
Quoted 6 lines
The BANNED flag is at the
intel_contextlevel, while I needed something at thegem_contextlevel to report out. Can make it a bool if you prefer. The invalidation process is inherently racy (the teardown can always occur after we've checked the flag) so no amount of locking can help there. The flag itself is only set once and never cleared, so no risk of a set and clear racing.
Yes the invalidate is racy, but we still need to make guarantees internally in the kernel that nothing escapes and falls through cracks.
Wrt reporting out, this should be exactly the same flow as a gpu reset that results in a banned gem context. We really shouldn't duplicate infrastructure here.
Quoted 10 lines
Quoted 8 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 6 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
It is immutable. Should we have a
ctx->immutable_flags, or better a bool?
Yeah just make it a bool, and add a kerneldoc comment that it's immutable and never changes.
Quoted 217 lines
Quoted 208 lines
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 202 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +30
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp);Quoted 8 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 5 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +5
-0+if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 5 lines
This is to bail out early and report the error to userspace. If the invalidation flag gets set after this check then we'll still catch it later as you said, but IMO there is value in bailing out early when possible, especially since it is a low-effort check. The invalidation flag can't be cleared, so no risk of rejecting something that shouldn't be.
Nah, that just papers over races and means you now have to validate more code for correctness - the later catching still must happen, but it's now much harder to trigger.
Also, the context banning should do this for us already. Maybe a good idea to audit and document that, to make sure it's working correctly and we're using it correctly for pxp, but this wheel should exist and we don't need a new one.
Except if the uapi promise is that we send back a different errno, but that again should be integrated into the context banning stuff. -Daniel
Quoted 370 lines
Daniele
Quoted 366 lines
Quoted 123 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 202
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.Quoted 220 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +54
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 219 lines · Daniel Vetter
Quoted 218 lines · Daniele Ceraolo Spurio
Quoted 216 lines · Daniel Vetter
Quoted 196 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 119
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
Quoted 8 lines
Quoted 6 lines
The BANNED flag is at the
intel_contextlevel, while I needed something at thegem_contextlevel to report out. Can make it a bool if you prefer. The invalidation process is inherently racy (the teardown can always occur after we've checked the flag) so no amount of locking can help there. The flag itself is only set once and never cleared, so no risk of a set and clear racing.Yes the invalidate is racy, but we still need to make guarantees internally in the kernel that nothing escapes and falls through cracks.
We can't make that guarantee, because we get the invalidation interrupt after the HW has already processed all the invalidations. A protected object might already be in use on an engine (in which case it is too late to pull it out) or have been submitted for a flip when we get the interrupt; we could also end up using the invalidated objects before we manage to process the interrupt. The whole invalidation tracking is best effort by HW design, AFAIU it was deemed acceptable to potentially get 1 corrupted frame as long as we do our best to keep the window short.
Quoted 4 lines
Wrt reporting out, this should be exactly the same flow as a gpu reset that results in a banned gem context. We really shouldn't duplicate infrastructure here.
AFAICS there is no banning of gem_context (unless it is named something different). There is banning of intel_context and full on banning of the file. There was an explicit request from mesa to add something new at the reset_stats level because what we already had didn't work. If there is a different solution that works for them I'm happy to switch (+ Lionel for comments).
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 8 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
Quoted 3 lines
Quoted 1 line
It is immutable. Should we have a
ctx->immutable_flags, or better a bool?Yeah just make it a bool, and add a kerneldoc comment that it's immutable and never changes.
Ok. The fact that this is immutable is already in the doc for the create_ext, but I'll add it next to the variable as well.
Quoted 220 lines
Quoted 215 lines
Quoted 207 lines
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 202 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +30
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp);Quoted 8 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 5 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +5
-0+if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 5 lines
This is to bail out early and report the error to userspace. If the invalidation flag gets set after this check then we'll still catch it later as you said, but IMO there is value in bailing out early when possible, especially since it is a low-effort check. The invalidation flag can't be cleared, so no risk of rejecting something that shouldn't be.
Quoted 8 lines
Nah, that just papers over races and means you now have to validate more code for correctness - the later catching still must happen, but it's now much harder to trigger.
Also, the context banning should do this for us already. Maybe a good idea to audit and document that, to make sure it's working correctly and we're using it correctly for pxp, but this wheel should exist and we don't need a new one.
Ok, I can rely on the intel_context being banned directly (it's checked earlier) but as mentioned above we do need but-in from mesa to drop the new invalidation flag at the gem_context level.
Daniele
Quoted 372 lines
Except if the uapi promise is that we send back a different errno, but that again should be integrated into the context banning stuff. -Daniel
Quoted 367 lines
Daniele
Quoted 365 lines
Quoted 122 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 202
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.Quoted 220 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +54
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH v6 10/15] drm/i915/pxp: interfaces for using protected objects
Quoted 231 lines · Daniele Ceraolo Spurio
Quoted 219 lines · Daniel Vetter
Quoted 218 lines · Daniele Ceraolo Spurio
Quoted 216 lines · Daniel Vetter
Quoted 196 lines · Daniele Ceraolo Spurio
This api allow user mode to create protected buffers and to mark contexts as making use of such objects. Only when using contexts marked in such a way is the execution guaranteed to work as expected.
Contexts can only be marked as using protected content at creation time (i.e. the parameter is immutable) and they must be both bannable and not recoverable.
All protected objects and contexts that have backing storage will be considered invalid when the PXP session is destroyed and all new submissions using them will be rejected. All intel contexts within the invalidated gem contexts will be marked banned. A new flag has been added to the
RESET_STATSioctl to report the context invalidation to userspace.This patch was previously sent as 2 separate patches, which have been squashed following a request to have all the uapi in a single patch. I've retained the s-o-b from both.
v5: squash patches, rebase on
proto_ctx, update kerneldocv6: rebase on obj
create_extchangesSigned-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Bommu Krishnaiah <[email protected]>Cc Rodrigo Vivi <[email protected]>Cc Chris Wilson <[email protected]>Cc Lionel Landwerlin <[email protected]>Cc Jason Ekstrand <[email protected]>Cc Daniel Vetter <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]> #v5--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 68 ++++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_context.h | 18 ++++ .../gpu/drm/i915/gem/i915_gem_context_types.h | 2 + drivers/gpu/drm/i915/gem/i915_gem_create.c | 75 ++++++++++++---- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 40 ++++++++- drivers/gpu/drm/i915/gem/i915_gem_object.c | 6 ++ drivers/gpu/drm/i915/gem/i915_gem_object.h | 12 +++ .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++ drivers/gpu/drm/i915/pxp/intel_pxp.c | 89 +++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp.h | 15 ++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 3 + drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 5 ++ include/uapi/drm/i915_drm.h | 55 +++++++++++- 13 files changed, 371 insertions(+), 26 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +62
-6@@ -77,6 +77,8 @@#include "gt/intel_gpu_commands.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_gem_context.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -241,6 +243,25 @@ static int proto_context_set_persistence(struct drm_i915_private *i915,return 0 ; } +static int proto_context_set_protected (struct drm_i915_private *i915, +struct i915_gem_proto_context *pc, +bool protected) +{ +int ret =0 ; + +if (!intel_pxp_is_enabled (&i915->gt.pxp)) + ret = -ENODEV ; +else if (!protected) + pc->user_flags &= ~BIT (UCONTEXT_PROTECTED ); +else if ((pc->user_flags &BIT (UCONTEXT_RECOVERABLE )) || + !(pc->user_flags &BIT (UCONTEXT_BANNABLE ))) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_PROTECTED ); + +return ret; +} +static struct i915_gem_proto_context *proto_context_create (struct drm_i915_private *i915,unsigned int flags) {@@ -686,6 +707,8 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) pc->user_flags |= BIT(UCONTEXT_BANNABLE); + else if (pc->user_flags & BIT(UCONTEXT_PROTECTED)) + ret = -EPERM ; else pc->user_flags &= ~BIT(UCONTEXT_BANNABLE);break ;@@ -693,10 +716,12 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) - pc->user_flags |=BIT (UCONTEXT_RECOVERABLE ); -else +else if (!args->value) pc->user_flags &= ~BIT (UCONTEXT_RECOVERABLE ); +else if (pc->user_flags &BIT (UCONTEXT_PROTECTED )) + ret = -EPERM ; +else + pc->user_flags |=BIT (UCONTEXT_RECOVERABLE );break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -724,6 +749,11 @@ static int set_proto_ctx_param(struct drm_i915_file_private *fpriv, args->value);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =proto_context_set_protected (fpriv->dev_priv, pc, + args->value); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -1798,6 +1828,18 @@ static int set_priority(struct i915_gem_context *ctx,return 0 ; } +static int get_protected (struct i915_gem_context *ctx, +struct drm_i915_gem_context_param *args) +{ +if (!intel_pxp_is_enabled (&ctx->i915->gt.pxp)) +return -ENODEV ; + + args->size =0 ; + args->value =i915_gem_context_uses_protected_content (ctx); + +return 0 ; +} +static int ctx_setparam (struct drm_i915_file_private *fpriv,struct i915_gem_context *ctx,struct drm_i915_gem_context_param *args)@@ -1821,6 +1863,8 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret = -EPERM ; else if (args->value) i915_gem_context_set_bannable(ctx); + else if (i915_gem_context_uses_protected_content(ctx)) + ret = -EPERM ;/* can't clear this for protected contexts */ elsei915_gem_context_clear_bannable (ctx );break ;@@ -1828,10 +1872,12 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv,case I915_CONTEXT_PARAM_RECOVERABLE :if (args->size) ret = -EINVAL ; -else if (args->value) -i915_gem_context_set_recoverable (ctx); -else +else if (!args->value)i915_gem_context_clear_recoverable (ctx); +else if (i915_gem_context_uses_protected_content (ctx)) + ret = -EPERM ;/* can't set this for protected contexts */ +else +i915_gem_context_set_recoverable (ctx);break ;case I915_CONTEXT_PARAM_PRIORITY :@@ -1846,6 +1892,7 @@ static int ctx_setparam(struct drm_i915_file_private *fpriv, ret =set_persistence (ctx, args);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT :case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_RINGSIZE :@@ -2174,6 +2221,10 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data, args->value =i915_gem_context_is_persistent (ctx);break ; +case I915_CONTEXT_PARAM_PROTECTED_CONTENT : + ret =get_protected (ctx, args); +break ; +case I915_CONTEXT_PARAM_NO_ZEROMAP :case I915_CONTEXT_PARAM_BAN_PERIOD :case I915_CONTEXT_PARAM_ENGINES :@@ -2250,6 +2301,11 @@ int i915_gem_context_reset_stats_ioctl(struct drm_device *dev, args->batch_active =atomic_read (&ctx->guilty_count); args->batch_pending =atomic_read (&ctx->active_count); +/* re-use args->flags for output flags */ + args->flags =0 ; +if (i915_gem_context_invalidated (ctx)) + args->flags |=I915_CONTEXT_INVALIDATED ; +i915_gem_context_put (ctx);return 0 ; }drivers/gpu/drm/i915/gem/i915_gem_context.h +0
-0@@ -108,6 +108,24 @@ i915_gem_context_clear_user_engines(struct i915_gem_context *ctx)clear_bit (CONTEXT_USER_ENGINES , &ctx->flags); }drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 119
Quoted 10 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +10
-0+static inline bool +i915_gem_context_invalidated (const struct i915_gem_context *ctx) +{ +return test_bit (CONTEXT_INVALID , &ctx->flags); +} + +static inline void +i915_gem_context_set_invalid (struct i915_gem_context *ctx) +{ +set_bit (CONTEXT_INVALID , &ctx->flags);Do we _really_ need a new bit in this already very complex state machinery, and can't we reuse the BANNED flag instead?
This
ctx->flagsis atomic, unorderd, and that means you need barriers and everything.If you don't actually need the atomic state bit machinery because you're using simple locking, then pls use your own boolean, and document by which lock it's protected.
Quoted 8 lines
Quoted 6 lines
The BANNED flag is at the
intel_contextlevel, while I needed something at thegem_contextlevel to report out. Can make it a bool if you prefer. The invalidation process is inherently racy (the teardown can always occur after we've checked the flag) so no amount of locking can help there. The flag itself is only set once and never cleared, so no risk of a set and clear racing.Yes the invalidate is racy, but we still need to make guarantees internally in the kernel that nothing escapes and falls through cracks.
Quoted 8 lines
We can't make that guarantee, because we get the invalidation interrupt after the HW has already processed all the invalidations. A protected object might already be in use on an engine (in which case it is too late to pull it out) or have been submitted for a flip when we get the interrupt; we could also end up using the invalidated objects before we manage to process the interrupt. The whole invalidation tracking is best effort by HW design, AFAIU it was deemed acceptable to potentially get 1 corrupted frame as long as we do our best to keep the window short.
That's not what I meant with races, but the races in our book-keeping. If we ban a context someone needs to clean up all the pending requests. If one slips through and then non one cleans that up, we have a potential problem. Since the context banning stuff has this solved already (hopefully at least, would be good to document how it works exactly) it shouldn't be a problem.
Quoted 10 lines
Quoted 3 lines
Wrt reporting out, this should be exactly the same flow as a gpu reset that results in a banned gem context. We really shouldn't duplicate infrastructure here.
AFAICS there is no banning of
gem_context(unless it is named something different). There is banning ofintel_contextand full on banning of the file. There was an explicit request from mesa to add something new at thereset_statslevel because what we already had didn't work. If there is a different solution that works for them I'm happy to switch (+ Lionel for comments).
Hm the only case where I see a difference between banning gem context and banning each intel_context is when you use legacy engine map, where we lazily instantiate stuff. Mesa shouldn't be using that anymore, so banning all the intel_context should be equivalent. Maybe we should require the new engine map stuff.
But I guess it wasn't so what exactly escaped here?
There's two concerns I have:
- When I discussed our entire context uapi with Jason a while ago we agreed that the gem context should be as much as possible a pure container, without semantics. The real thing should be the intel_contexts underneath.
- Locking gets really nasty when we have to go backwards from intel_context to gem context (it currently is, I have some plans to fix that up). So if you add uapi guarantees where we have to go backwards, things get potentially very interesting. Not impossible, but definitely more fun than I maybe want.
Cheers, Daniel
Quoted 626 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h i915_gem_context_clear_user_engines(struct i915_gem_context *ctx) · line 126
Quoted 9 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +6
-0+} + +static inline bool +i915_gem_context_uses_protected_content (const struct i915_gem_context *ctx) +{ +return test_bit (UCONTEXT_PROTECTED , &ctx->user_flags);For immutable state (I really hope this is immutable) pls don't reuse the atomic bitfield of mutable state, but create a flag of your own.
Quoted 3 lines
Quoted 1 line
It is immutable. Should we have a
ctx->immutable_flags, or better a bool?Yeah just make it a bool, and add a kerneldoc comment that it's immutable and never changes.
Ok. The fact that this is immutable is already in the doc for the
create_ext, but I'll add it next to the variable as well.Quoted 220 lines
Quoted 215 lines
Quoted 207 lines
Also please document all the rules around how this is set/changed in the kerneldoc header comments for the data structure. Finally if you never set it except at creation.
Quoted 202 lines
drivers/gpu/drm/i915/gem/i915_gem_context.h +2
-0+} +/* i915_gem_context.c */ void i915_gem_init__contexts (struct drm_i915_private *i915);drivers/gpu/drm/i915/gem/i915_gem_context_types.h +2
-0@@ -301,6 +301,7 @@ struct i915_gem_context {#define UCONTEXT_BANNABLE 2#define UCONTEXT_RECOVERABLE 3#define UCONTEXT_PERSISTENCE 4 +#define UCONTEXT_PROTECTED 5/** * @flags: small set of booleans @@ -308,6 +309,7 @@ struct i915_gem_context {unsigned long flags;#define CONTEXT_CLOSED 0#define CONTEXT_USER_ENGINES 1 +#define CONTEXT_INVALID 2/** @mutex: guards everything that isn't engines or handles_vma */ struct mutex mutex;drivers/gpu/drm/i915/gem/i915_gem_create.c +57
-18@@ -6,6 +6,7 @@#include "gem/i915_gem_ioctls.h" #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_region.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_trace.h" @@ -82,21 +83,11 @@ static int i915_gem_publish(struct drm_i915_gem_object *obj,return 0 ; } -/** -* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT -* @i915: i915 private -* @size: size of the buffer, in bytes -* @placements: possible placement regions, in priority order -* @n_placements: number of possible placement regions -* -* This function is exposed primarily for selftests and does very little -* error checking. It is assumed that the set of placement regions has -* already been verified to be valid. -*/ -struct drm_i915_gem_object * -__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, -struct intel_memory_region **placements, -unsigned int n_placements) +static struct drm_i915_gem_object * +__i915_gem_object_create_user_ext (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements, +unsigned int ext_flags) {struct intel_memory_region *mr = placements[0 ];struct drm_i915_gem_object *obj;@@ -135,6 +126,12 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,GEM_BUG_ON (size != obj->base.size); +/* Add any flag set by create_ext options */ + flags |= ext_flags; + +if (i915_gem_object_is_protected (obj)) +intel_pxp_object_add (obj); +trace_i915_gem_object_create (obj);return obj;@@ -145,6 +142,26 @@ __i915_gem_object_create_user(struct drm_i915_private *i915, u64 size,return ERR_PTR (ret); } +/** +* Creates a new object using the same path as DRM_I915_GEM_CREATE_EXT +* @i915: i915 private +* @size: size of the buffer, in bytes +* @placements: possible placement regions, in priority order +* @n_placements: number of possible placement regions +* +* This function is exposed primarily for selftests and does very little +* error checking. It is assumed that the set of placement regions has +* already been verified to be valid. +*/ +struct drm_i915_gem_object * +__i915_gem_object_create_user (struct drm_i915_private *i915,u64 size, +struct intel_memory_region **placements, +unsigned int n_placements) +{ +return __i915_gem_object_create_user_ext (i915, size, placements, + n_placements,0 ); +} + int i915_gem_dumb_create(struct drm_file *file, struct drm_device *dev,@@ -224,6 +241,7 @@ struct create_ext {struct drm_i915_private *i915;struct intel_memory_region *placements[INTEL_REGION_UNKNOWN ];unsigned int n_placements; +unsigned long flags; }; static void repr_placements(char *buf, size_t size,@@ -356,8 +374,28 @@ static int ext_set_placements(struct i915_user_extension __user *base,return set_placements (&ext, data); } +static int ext_set_protected (struct i915_user_extension __user *base,void *data) +{ +struct drm_i915_gem_create_ext_protected_content ext; +struct create_ext *ext_data = data; + +if (copy_from_user (&ext, base,sizeof (ext))) +return -EFAULT ; + +if (ext.flags) +return -EINVAL ; + +if (!intel_pxp_is_enabled (&ext_data->i915->gt.pxp)) +return -ENODEV ; + + ext_data->flags |=I915_BO_PROTECTED ; + +return 0 ; +} +static const i915_user_extension_fn create_extensions[] = { [I915_GEM_CREATE_EXT_MEMORY_REGIONS ] = ext_set_placements, + [I915_GEM_CREATE_EXT_PROTECTED_CONTENT ] = ext_set_protected, };/** @@ -392,9 +430,10 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data, ext_data.n_placements =1 ; } - obj =__i915_gem_object_create_user (i915, args->size, - ext_data.placements, - ext_data.n_placements); + obj =__i915_gem_object_create_user_ext (i915, args->size, + ext_data.placements, + ext_data.n_placements, + ext_data.flags);if (IS_ERR (obj))return PTR_ERR (obj);drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +30
-1@@ -21,6 +21,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_ring.h" +#include "pxp/intel_pxp.h" +#include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -751,6 +753,11 @@ static int eb_select_context(struct i915_execbuffer *eb)if (unlikely (IS_ERR (ctx)))return PTR_ERR (ctx); +if (i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +return -EACCES ; + } + eb->gem_context = ctx;if (rcu_access_pointer (ctx->vm)) eb->invalid_flags |=EXEC_OBJECT_NEEDS_GTT ;@@ -819,7 +826,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) do { struct drm_i915_gem_object *obj; struct i915_vma *vma; - int err; + int err = 0; rcu_read_lock(); vma = radix_tree_lookup(&eb->gem_context->handles_vma, handle);@@ -833,6 +840,26 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle)if (unlikely (!obj))return ERR_PTR (-ENOENT ); +/* +* If the user has opted-in for protected-object tracking, make +* sure the object encryption can be used. +* We only need to do this when the object is first used with +* this context, because the context itself will be banned when +* the protected objects become invalid. +*/ +if (i915_gem_context_uses_protected_content (eb->gem_context) && +i915_gem_object_is_protected (obj)) { +if (!intel_pxp_is_active (&vm->gt->pxp)) + err = -ENODEV ; +else if (!i915_gem_object_has_valid_protection (obj)) + err = -ENOEXEC ; + +if (err) { +i915_gem_object_put (obj); +return ERR_PTR (err); + } + } + vma =i915_vma_instance (obj, vm,NULL );if (IS_ERR (vma)) {i915_gem_object_put (obj);@@ -2752,6 +2779,17 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt); +if (i915_gem_context_uses_protected_content (eb->gem_context)) { + err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp);Quoted 8 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c eb_select_engine(struct i915_execbuffer *eb) · line 2788
Quoted 5 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +5
-0+if (err) + goto err; + +if (i915_gem_context_invalidated (eb->gem_context)) { + err = -EACCES ;Shouldn't the normal banned context handling takee care of anything that slips through? Rolling your own racy invalidation checks doesn't look like a good idea.
Quoted 5 lines
This is to bail out early and report the error to userspace. If the invalidation flag gets set after this check then we'll still catch it later as you said, but IMO there is value in bailing out early when possible, especially since it is a low-effort check. The invalidation flag can't be cleared, so no risk of rejecting something that shouldn't be.
Quoted 8 lines
Nah, that just papers over races and means you now have to validate more code for correctness - the later catching still must happen, but it's now much harder to trigger.
Also, the context banning should do this for us already. Maybe a good idea to audit and document that, to make sure it's working correctly and we're using it correctly for pxp, but this wheel should exist and we don't need a new one.
Ok, I can rely on the
intel_contextbeing banned directly (it's checked earlier) but as mentioned above we do need but-in from mesa to drop the new invalidation flag at thegem_contextlevel.Daniele
Quoted 372 lines
Except if the uapi promise is that we send back a different errno, but that again should be integrated into the context banning stuff. -Daniel
Quoted 367 lines
Daniele
Quoted 365 lines
Quoted 122 lines
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +4
-0+ goto err; + } + } +if (!test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { err =intel_context_alloc_state (ce);if (err)drivers/gpu/drm/i915/gem/i915_gem_object.c +6
-0@@ -25,6 +25,7 @@#include <linux/sched/mm.h> #include "display/intel_frontbuffer.h" +#include "pxp/intel_pxp.h" #include "i915_drv.h" #include "i915_gem_clflush.h" #include "i915_gem_context.h" @@ -73,6 +74,8 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,INIT_LIST_HEAD (&obj->lut_list);spin_lock_init (&obj->lut_lock); +INIT_LIST_HEAD (&obj->pxp_link); +spin_lock_init (&obj->mmo.lock); obj->mmo.offsets =RB_ROOT ;@@ -231,6 +234,9 @@ void __i915_gem_free_object(struct drm_i915_gem_object *obj)spin_unlock (&obj->vma.lock); } +if (i915_gem_object_has_valid_protection (obj)) +intel_pxp_object_remove (obj); +__i915_gem_object_free_mmaps (obj);GEM_BUG_ON (!list_empty (&obj->lut_list));drivers/gpu/drm/i915/gem/i915_gem_object.h +12
-0@@ -269,6 +269,18 @@ i915_gem_object_clear_tiling_quirk(struct drm_i915_gem_object *obj)clear_bit (I915_TILING_QUIRK_BIT , &obj->flags); } +static inline bool +i915_gem_object_is_protected (const struct drm_i915_gem_object *obj) +{ +return obj->flags &I915_BO_PROTECTED ; +} + +static inline bool +i915_gem_object_has_valid_protection (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !list_empty (&obj->pxp_link); +} +static inline bool i915_gem_object_type_has (const struct drm_i915_gem_object *obj,unsigned long flags)drivers/gpu/drm/i915/gem/i915_gem_object_types.h +9
-0@@ -298,6 +298,7 @@ struct drm_i915_gem_object { I915_BO_ALLOC_USER)#define I915_BO_READONLY BIT(4)#define I915_TILING_QUIRK_BIT 5/* unknown swizzling; do not release! */ +#define I915_BO_PROTECTED BIT(6)/** * @mem_flags - Mutable placement-related flags @@ -537,6 +538,14 @@ struct drm_i915_gem_object { bool created:1; } ttm; +/* +* When the PXP session is invalidated, we need to mark all protected +* objects as invalid. To easily do so we add them all to a list. The +* presence on the list is used to check if the encryption is valid or +* not. +*/ +struct list_head pxp_link; +/** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17;drivers/gpu/drm/i915/pxp/intel_pxp.c +25
-0@@ -7,6 +7,7 @@#include "intel_pxp_irq.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" +#include "gem/i915_gem_context.h" #include "gt/intel_context.h" #include "i915_drv.h" @@ -70,6 +71,9 @@ void intel_pxp_init(struct intel_pxp *pxp)mutex_init (&pxp->tee_mutex); +spin_lock_init (&pxp->lock); +INIT_LIST_HEAD (&pxp->protected_objects); +/* * we'll use the completion to check if there is a termination pending, * so we start it as completed and we reinit it when a termination @@ -166,3 +170,88 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp)intel_pxp_irq_disable (pxp); } + +int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp; + +if (!intel_pxp_is_enabled (pxp)) +return -ENODEV ; + +if (!list_empty (&obj->pxp_link)) +return -EEXIST ; + +spin_lock_irq (&pxp->lock); +list_add (&obj->pxp_link, &pxp->protected_objects); +spin_unlock_irq (&pxp->lock); + +return 0 ; +} + +void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +struct intel_pxp *pxp = &to_i915 (obj->base.dev)->gt.pxp;drivers/gpu/drm/i915/pxp/intel_pxp.c void intel_pxp_fini_hw(struct intel_pxp *pxp) · line 202
Quoted 10 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +10
-0+ +if (!intel_pxp_is_enabled (pxp)) +return ; + +spin_lock_irq (&pxp->lock); +list_del_init (&obj->pxp_link); +spin_unlock_irq (&pxp->lock); +} + +void intel_pxp_invalidate (struct intel_pxp *pxp)There is a lot of locking going on here. Please make sure it's all properly documented what the rules are in kerneldoc, even for existing stuff.
Especially anytime there's anything tricky going on, like:
-kref_get_unless_zero
-test_bit(that's an unorderd atomic, so either you don't actually need the atomic, or you need a pile of barriers and comments about them)
- hiddent atomic ops likeintel_context_set_bannedori915_gem_context_set_invalidalso need great care, least because the current rules aren't documented really. So step one here before using them is documenting the rules of how banning works.Quoted 220 lines
drivers/gpu/drm/i915/pxp/intel_pxp.c +54
-0+{ +struct drm_i915_private *i915 =pxp_to_gt (pxp)->i915; +struct drm_i915_gem_object *obj, *tmp; +struct i915_gem_context *ctx, *cn; + +/* delete objects that have been used with the invalidated session */ +spin_lock_irq (&pxp->lock); +list_for_each_entry_safe (obj, tmp, &pxp->protected_objects, pxp_link) { +if (i915_gem_object_has_pages (obj)) +list_del_init (&obj->pxp_link); + } +spin_unlock_irq (&pxp->lock); + +/* ban all contexts marked as protected */ +spin_lock_irq (&i915->gem.contexts.lock); +list_for_each_entry_safe (ctx, cn, &i915->gem.contexts.list, link) { +struct i915_gem_engines_iter it; +struct intel_context *ce; + +if (!kref_get_unless_zero (&ctx->ref)) +continue ; + +if (likely (!i915_gem_context_uses_protected_content (ctx)) || +i915_gem_context_invalidated (ctx)) { +i915_gem_context_put (ctx); +continue ; + } + +spin_unlock_irq (&i915->gem.contexts.lock); + +/* +* Note that by the time we get here the HW keys are already +* long gone, so any batch using them that's already on the +* engines is very likely a lost cause (and it has probably +* already hung the engine). Therefore, we skip attempting to +* pull the running context out of the HW and we prioritize +* bringing the session back as soon as possible. +*/ +for_each_gem_engine (ce,i915_gem_context_lock_engines (ctx), it) { +/* only invalidate if at least one ce was allocated */ +if (test_bit (CONTEXT_ALLOC_BIT , &ce->flags)) { +intel_context_set_banned (ce); +i915_gem_context_set_invalid (ctx); + } + } +i915_gem_context_unlock_engines (ctx); + +spin_lock_irq (&i915->gem.contexts.lock); +list_safe_reset_next (ctx, cn, link); +i915_gem_context_put (ctx); + } +spin_unlock_irq (&i915->gem.contexts.lock); +} +drivers/gpu/drm/i915/pxp/intel_pxp.h +15
-0@@ -9,6 +9,8 @@#include "gt/intel_gt_types.h" #include "intel_pxp_types.h" +struct drm_i915_gem_object ; +static inline struct intel_gt *pxp_to_gt (const struct intel_pxp *pxp) { return container_of(pxp, struct intel_gt, pxp);@@ -33,6 +35,11 @@ void intel_pxp_fini_hw(struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp);int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_object_add (struct drm_i915_gem_object *obj); +void intel_pxp_object_remove (struct drm_i915_gem_object *obj); + +void intel_pxp_invalidate (struct intel_pxp *pxp);#else static inline void intel_pxp_init (struct intel_pxp *pxp) {@@ -46,6 +53,14 @@ static inline int intel_pxp_wait_for_arb_start(struct intel_pxp *pxp) {return 0 ; } + +static inline int intel_pxp_object_add (struct drm_i915_gem_object *obj) +{ +return 0 ; +} +static inline void intel_pxp_object_remove (struct drm_i915_gem_object *obj) +{ +}#endif #endif /* __INTEL_PXP_H__ */ drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3
-0@@ -85,6 +85,9 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); +/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {drivers/gpu/drm/i915/pxp/intel_pxp_types.h +5
-0@@ -7,7 +7,9 @@#define __INTEL_PXP_TYPES_H__#include <linux/completion.h> +#include <linux/list.h> #include <linux/mutex.h> +#include <linux/spinlock.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -43,6 +45,9 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) + +spinlock_t lock;/* protects the objects list */ + struct list_head protected_objects; };#endif /* __INTEL_PXP_TYPES_H__ */ include/uapi/drm/i915_drm.h +54
-1@@ -1843,12 +1843,32 @@ struct drm_i915_gem_context_param {* attempted to use it, never re-use this context param number. */ #define I915_CONTEXT_PARAM_RINGSIZE 0xc + +/* +* I915_CONTEXT_PARAM_PROTECTED_CONTENT: +* +* Mark that the context makes use of protected content, which will result +* in the context being invalidated when the protected content session is. The +* invalidation is reported back to userspace via the RESET_STATS ioctl (see +* relevant doc for details). +* This flag can only be set at context creation time and, when set to true, +* must be preceded by an explicit setting of I915_CONTEXT_PARAM_RECOVERABLE +* to false. This flag can't be set to true in conjunction with setting the +* I915_CONTEXT_PARAM_BANNABLE flag to false. +* +* In addition to the normal failure cases, setting this flag during context +* creation can result in the following errors: +* +* -ENODEV: feature not available +* -EPERM: trying to mark a recoverable or not bannable context as protected +*/ +#define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd/* Must be kept compact -- no holes and well documented */ __u64 value; }; -/* +/** * Context SSEU programming * * It may be necessary for either functional or performance reason to configure @@ -2181,6 +2201,12 @@ struct drm_i915_reg_read {struct drm_i915_reset_stats {__u32 ctx_id;__u32 flags; +/* +* contexts marked as using protected content are invalidated when the +* protected content session dies. Submission of invalidated contexts +* is rejected with -EACCES. +*/ +#define I915_CONTEXT_INVALIDATED 0x1/* All resets since boot/module reload, for all contexts */ __u32 reset_count;@@ -2959,8 +2985,12 @@ struct drm_i915_gem_create_ext {* * For I915_GEM_CREATE_EXT_MEMORY_REGIONS usage see * struct drm_i915_gem_create_ext_memory_regions. +* +* For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see +* struct drm_i915_gem_create_ext_protected_content. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 +#define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1__u64 extensions; };@@ -3018,6 +3048,29 @@ struct drm_i915_gem_create_ext_memory_regions {__u64 regions; }; +/** +* struct drm_i915_gem_create_ext_protected_content - The +* I915_OBJECT_PARAM_PROTECTED_CONTENT extension. +* +* If this extension is provided, buffer contents are expected to be protected +* by PXP encryption and require decryption for scan out and processing. This +* is only possible on platforms that have PXP enabled, on all other scenarios +* using this extension will cause the ioctl to fail and return -ENODEV. The +* flags parameter is reserved for future expansion and must currently be set +* to zero. +* +* The buffer contents are considered invalid after a PXP session teardown. +* +* The encryption is guaranteed to be processed correctly only if the object +* is submitted with a context created using the +* I915_CONTEXT_PARAM_PROTECTED_CONTENT flag. This will also enable extra checks +* at submission time on the validity of the objects involved. +*/ +struct drm_i915_gem_create_ext_protected_content { +struct i915_user_extension base; +__u32 flags; +}; +/* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xfSignature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
PATCH v6 11/15 drm/i915/pxp: start the arb session on demand
In reply to Daniele Ceraolo SpurioNow that we can handle destruction and re-creation of the arb session, we can postpone the start of the session to the first submission that requires it, to avoid keeping it running with no user.
--- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 8 ++-- drivers/gpu/drm/i915/pxp/intel_pxp.c | 37 ++++++++++++------- drivers/gpu/drm/i915/pxp/intel_pxp.h | 5 ++- drivers/gpu/drm/i915/pxp/intel_pxp_irq.c | 2 +- drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 6 +-- drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 10 +---- drivers/gpu/drm/i915/pxp/intel_pxp_types.h | 2 + 7 files changed, 39 insertions(+), 31 deletions(-)
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +5 -3
@@ -2780,9 +2780,11 @@ eb_select_engine(struct i915_execbuffer *eb)intel_gt_pm_get (ce->engine->gt);if (i915_gem_context_uses_protected_content (eb->gem_context)) { - err =intel_pxp_wait_for_arb_start (&ce->engine->gt->pxp); -if (err) - goto err; +if (!intel_pxp_is_active (&ce->engine->gt->pxp)) { + err =intel_pxp_start (&ce->engine->gt->pxp); +if (err) + goto err; + }if (i915_gem_context_invalidated (eb->gem_context)) { err = -EACCES ;
drivers/gpu/drm/i915/pxp/intel_pxp.c +24 -13
@@ -82,6 +82,7 @@ void intel_pxp_init(struct intel_pxp *pxp)init_completion (&pxp->termination);complete_all (&pxp->termination); +mutex_init (&pxp->arb_mutex);INIT_WORK (&pxp->session_work, intel_pxp_session_work); ret =create_vcs_context (pxp);
@@ -118,7 +119,7 @@ void intel_pxp_mark_termination_in_progress(struct intel_pxp *pxp)reinit_completion (&pxp->termination); } -static void intel_pxp_queue_termination (struct intel_pxp *pxp) +static void pxp_queue_termination (struct intel_pxp *pxp) {struct intel_gt *gt =pxp_to_gt (pxp);
@@ -137,31 +138,41 @@ static void intel_pxp_queue_termination(struct intel_pxp *pxp)* the arb session is restarted from the irq work when we receive the * termination completion interrupt */ -int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp) +int intel_pxp_start (struct intel_pxp *pxp) { +int ret =0 ; +if (!intel_pxp_is_enabled (pxp)) -return 0 ; +return -ENODEV ; + +mutex_lock (&pxp->arb_mutex); + +if (pxp->arb_is_valid) + goto unlock; + +pxp_queue_termination (pxp);if (!wait_for_completion_timeout (&pxp->termination, -msecs_to_jiffies (100 ))) -return -ETIMEDOUT ; +msecs_to_jiffies (100 ))) { + ret = -ETIMEDOUT ; + goto unlock; + } + +/* make sure the compiler doesn't optimize the double access */ +barrier ();if (!pxp->arb_is_valid) -return -EIO ; + ret = -EIO ; -return 0 ; +unlock: +mutex_unlock (&pxp->arb_mutex); +return ret; }void intel_pxp_init_hw (struct intel_pxp *pxp) {kcr_pxp_enable (pxp_to_gt (pxp));intel_pxp_irq_enable (pxp); - -/* -* the session could've been attacked while we weren't loaded, so -* handle it as if it was and re-create it. -*/ -intel_pxp_queue_termination (pxp); }void intel_pxp_fini_hw (struct intel_pxp *pxp)
drivers/gpu/drm/i915/pxp/intel_pxp.h +3 -2
@@ -34,7 +34,8 @@ void intel_pxp_init_hw(struct intel_pxp *pxp);void intel_pxp_fini_hw (struct intel_pxp *pxp);void intel_pxp_mark_termination_in_progress (struct intel_pxp *pxp); -int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp); + +int intel_pxp_start (struct intel_pxp *pxp);int intel_pxp_object_add (struct drm_i915_gem_object *obj);void intel_pxp_object_remove (struct drm_i915_gem_object *obj);
@@ -49,7 +50,7 @@ static inline void intel_pxp_fini(struct intel_pxp *pxp) { } -static inline int intel_pxp_wait_for_arb_start (struct intel_pxp *pxp) +static inline int intel_pxp_start (struct intel_pxp *pxp) {return 0 ; }
drivers/gpu/drm/i915/pxp/intel_pxp_irq.c +1 -1
@@ -31,7 +31,7 @@ void intel_pxp_irq_handler(struct intel_pxp *pxp, u16 iir) GEN12_DISPLAY_APP_TERMINATED_PER_FW_REQ_INTERRUPT)) {/* immediately mark PXP as inactive on termination */ intel_pxp_mark_termination_in_progress (pxp ); - pxp->session_events |=PXP_TERMINATION_REQUEST ; + pxp->session_events |=PXP_TERMINATION_REQUEST |PXP_INVAL_REQUIRED ; }if (iir &GEN12_DISPLAY_STATE_RESET_COMPLETE_INTERRUPT )
drivers/gpu/drm/i915/pxp/intel_pxp_session.c +3 -3
@@ -85,9 +85,6 @@ static int pxp_terminate_arb_session_and_global(struct intel_pxp *pxp)/* must mark termination in progress calling this function */ GEM_WARN_ON (pxp->arb_is_valid); -/* invalidate protected objects */ -intel_pxp_invalidate (pxp); -/* terminate the hw sessions */ ret =intel_pxp_terminate_session (pxp,ARB_SESSION );if (ret) {
@@ -144,6 +141,9 @@ void intel_pxp_session_work(struct work_struct *work)if (!events)return ; +if (events &PXP_INVAL_REQUIRED ) +intel_pxp_invalidate (pxp); +if (events &PXP_TERMINATION_REQUEST ) { events &= ~PXP_TERMINATION_COMPLETE ;pxp_terminate (pxp);
drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +1 -9
@@ -78,9 +78,7 @@ static int intel_pxp_tee_io_message(struct intel_pxp *pxp,static int i915_pxp_tee_component_bind (struct device *i915_kdev,struct device *tee_kdev,void *data) { -struct drm_i915_private *i915 =kdev_to_i915 (i915_kdev);struct intel_pxp *pxp =i915_dev_to_pxp (i915_kdev); -int ret;mutex_lock (&pxp->tee_mutex); pxp->pxp_component = data;
@@ -89,14 +87,8 @@ static int i915_pxp_tee_component_bind(struct device *i915_kdev,/* the component is required to fully start the PXP HW */ intel_pxp_init_hw (pxp); - ret =intel_pxp_wait_for_arb_start (pxp); -if (ret) { -drm_err (&i915->drm,"Failed to create arb session during bind\n" ); -intel_pxp_fini_hw (pxp); - pxp->pxp_component =NULL ; - } -return ret; +return 0 ; } static void i915_pxp_tee_component_unbind(struct device *i915_kdev,
drivers/gpu/drm/i915/pxp/intel_pxp_types.h +2 -0
@@ -26,6 +26,7 @@ struct intel_pxp {* even if the keys are gone, so we can't rely on the HW state of the * session to know if it's valid and need to track the status in SW. */ +struct mutex arb_mutex;/* protects arb session start */ bool arb_is_valid;struct mutex tee_mutex;/* protects the tee channel binding */
@@ -45,6 +46,7 @@ struct intel_pxp {u32 session_events;/* protected with gt->irq_lock */ #define PXP_TERMINATION_REQUEST BIT(0)#define PXP_TERMINATION_COMPLETE BIT(1) +#define PXP_INVAL_REQUIRED BIT(2)spinlock_t lock;/* protects the objects list */ struct list_head protected_objects;
Signature · 1 line
2.32.0
PATCH v6 12/15 drm/i915/pxp: Enable PXP power management
In reply to Daniele Ceraolo SpurioFrom: "Huang, Sean Z" <[email protected]>
During the power event S3+ sleep/resume, hardware will lose all the encryption keys for every hardware session, even though the session state might still be marked as alive after resume. Therefore, we should consider the session as dead on suspend and invalidate all the objects. The session will be automatically restarted on the first protected submission on resume.
v2: runtime suspend also invalidates the keys
v3: fix return codes, simplify rpm ops (Chris), use the new worker func
v4: invalidate the objects on suspend, don't re-create the arb sesson on resume (delayed to first submission).
v5: move irq changes back to irq patch (Rodrigo)
--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/gt/intel_gt_pm.c | 15 +++++++- drivers/gpu/drm/i915/i915_drv.c | 2 + drivers/gpu/drm/i915/pxp/intel_pxp_irq.c | 1 + drivers/gpu/drm/i915/pxp/intel_pxp_pm.c | 40 ++++++++++++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_pm.h | 23 +++++++++++ drivers/gpu/drm/i915/pxp/intel_pxp_session.c | 38 ++++++++++++++----- drivers/gpu/drm/i915/pxp/intel_pxp_tee.c | 9 +++++ 8 files changed, 118 insertions(+), 11 deletions(-) create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_pm.c create mode 100644 drivers/gpu/drm/i915/pxp/intel_pxp_pm.h
drivers/gpu/drm/i915/Makefile +1 -0
@@ -280,6 +280,7 @@ i915-$(CONFIG_DRM_I915_PXP) += \ pxp/intel_pxp.o \ pxp/intel_pxp_cmd.o \ pxp/intel_pxp_irq.o \ + pxp/intel_pxp_pm.o \ pxp/intel_pxp_session.o \ pxp/intel_pxp_tee.o
drivers/gpu/drm/i915/gt/intel_gt_pm.c +14 -1
@@ -18,6 +18,7 @@#include "intel_rc6.h" #include "intel_rps.h" #include "intel_wakeref.h" +#include "pxp/intel_pxp_pm.h" static void user_forcewake (struct intel_gt *gt,bool suspend) {
@@ -262,6 +263,8 @@ int intel_gt_resume(struct intel_gt *gt)intel_uc_resume (>->uc); +intel_pxp_resume (>->pxp); +user_forcewake (gt, false); out_fw:
@@ -296,6 +299,7 @@ void intel_gt_suspend_prepare(struct intel_gt *gt)user_forcewake (gt, true);wait_for_suspend (gt); +intel_pxp_suspend (>->pxp);intel_uc_suspend (>->uc); }
@@ -346,6 +350,7 @@ void intel_gt_suspend_late(struct intel_gt *gt)void intel_gt_runtime_suspend (struct intel_gt *gt) { +intel_pxp_suspend (>->pxp);intel_uc_runtime_suspend (>->uc);GT_TRACE (gt,"\n" );
@@ -353,11 +358,19 @@ void intel_gt_runtime_suspend(struct intel_gt *gt)int intel_gt_runtime_resume (struct intel_gt *gt) { +int ret; +GT_TRACE (gt,"\n" );intel_gt_init_swizzling (gt);intel_ggtt_restore_fences (gt->ggtt); -return intel_uc_runtime_resume (>->uc); + ret =intel_uc_runtime_resume (>->uc); +if (ret) +return ret; + +intel_pxp_resume (>->pxp); + +return 0 ; }static ktime_t __intel_gt_get_awake_time (const struct intel_gt *gt)
drivers/gpu/drm/i915/i915_drv.c +2 -0
@@ -67,6 +67,8 @@#include "gt/intel_gt_pm.h" #include "gt/intel_rc6.h" +#include "pxp/intel_pxp_pm.h" +#include "i915_debugfs.h" #include "i915_drv.h" #include "i915_ioc32.h"
drivers/gpu/drm/i915/pxp/intel_pxp_irq.c +1 -0
@@ -9,6 +9,7 @@#include "gt/intel_gt_irq.h" #include "i915_irq.h" #include "i915_reg.h" +#include "intel_runtime_pm.h" /** * intel_pxp_irq_handler - Handles PXP interrupts.
drivers/gpu/drm/i915/pxp/intel_pxp_pm.c new file +40 -0
@@ -0,0 +1,40 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright(c) 2020 Intel Corporation. +*/ + +#include "intel_pxp.h" +#include "intel_pxp_irq.h" +#include "intel_pxp_pm.h" +#include "intel_pxp_session.h" + +void intel_pxp_suspend (struct intel_pxp *pxp) +{ +if (!intel_pxp_is_enabled (pxp)) +return ; + + pxp->arb_is_valid = false; + +/* invalidate protected objects */ +intel_pxp_invalidate (pxp); + +intel_pxp_fini_hw (pxp); + + pxp->hw_state_invalidated = false; +} + +void intel_pxp_resume (struct intel_pxp *pxp) +{ +if (!intel_pxp_is_enabled (pxp)) +return ; + +/* +* The PXP component gets automatically unbound when we go into S3 and +* re-bound after we come out, so in that scenario we can defer the +* hw init to the bind call. +*/ +if (!pxp->pxp_component) +return ; + +intel_pxp_init_hw (pxp); +}
drivers/gpu/drm/i915/pxp/intel_pxp_pm.h new file +23 -0
@@ -0,0 +1,23 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright(c) 2020, Intel Corporation. All rights reserved. +*/ + +#ifndef __INTEL_PXP_PM_H__ +#define __INTEL_PXP_PM_H__ + +#include "i915_drv.h" + +#ifdef CONFIG_DRM_I915_PXP +void intel_pxp_suspend (struct intel_pxp *pxp); +void intel_pxp_resume (struct intel_pxp *pxp); +#else +static inline void intel_pxp_suspend (struct intel_pxp *pxp) +{ +} +static inline void intel_pxp_resume (struct intel_pxp *pxp) +{ +} +#endif + +#endif /* __INTEL_PXP_PM_H__ */
drivers/gpu/drm/i915/pxp/intel_pxp_session.c +28 -10
@@ -21,29 +21,36 @@static bool intel_pxp_session_is_in_play (struct intel_pxp *pxp,u32 id) { -struct intel_gt *gt =pxp_to_gt (pxp); +struct intel_uncore *uncore =pxp_to_gt (pxp)->uncore;intel_wakeref_t wakeref;u32 sip =0 ; -with_intel_runtime_pm (gt->uncore->rpm, wakeref) - sip = intel_uncore_read(gt->uncore, GEN12_KCR_SIP); +/* if we're suspended the session is considered off */ +with_intel_runtime_pm_if_in_use (uncore->rpm, wakeref) + sip = intel_uncore_read(uncore, GEN12_KCR_SIP);return sip &BIT (id); }static int pxp_wait_for_session_state (struct intel_pxp *pxp,u32 id,bool in_play) { -struct intel_gt *gt =pxp_to_gt (pxp); +struct intel_uncore *uncore =pxp_to_gt (pxp)->uncore;intel_wakeref_t wakeref;u32 mask =BIT (id);int ret; -with_intel_runtime_pm (gt->uncore->rpm, wakeref) - ret = intel_wait_for_register(gt->uncore, -GEN12_KCR_SIP , - mask, - in_play ? mask :0 , -100 ); +/* if we're suspended the session is considered off */ + wakeref =intel_runtime_pm_get_if_in_use (uncore->rpm); +if (!wakeref) +return in_play ? -ENODEV :0 ; + + ret =intel_wait_for_register (uncore, +GEN12_KCR_SIP , + mask, + in_play ? mask :0 , +100 ); + +intel_runtime_pm_put (uncore->rpm, wakeref);return ret; }
@@ -132,6 +139,7 @@ void intel_pxp_session_work(struct work_struct *work) {struct intel_pxp *pxp =container_of (work,typeof (*pxp), session_work);struct intel_gt *gt =pxp_to_gt (pxp); +intel_wakeref_t wakeref;u32 events =0 ;spin_lock_irq (>->irq_lock);
@@ -144,6 +152,14 @@ void intel_pxp_session_work(struct work_struct *work)if (events &PXP_INVAL_REQUIRED )intel_pxp_invalidate (pxp); +/* +* If we're processing an event while suspending then don't bother, +* we're going to re-init everything on resume anyway. +*/ + wakeref =intel_runtime_pm_get_if_in_use (gt->uncore->rpm); +if (!wakeref) +return ; +if (events &PXP_TERMINATION_REQUEST ) { events &= ~PXP_TERMINATION_COMPLETE ;pxp_terminate (pxp);
@@ -151,4 +167,6 @@ void intel_pxp_session_work(struct work_struct *work)if (events &PXP_TERMINATION_COMPLETE )pxp_terminate_complete (pxp); + +intel_runtime_pm_put (gt->uncore->rpm, wakeref); }
drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +9 -0
@@ -78,16 +78,25 @@ static int intel_pxp_tee_io_message(struct intel_pxp *pxp,static int i915_pxp_tee_component_bind (struct device *i915_kdev,struct device *tee_kdev,void *data) { +struct drm_i915_private *i915 =kdev_to_i915 (i915_kdev);struct intel_pxp *pxp =i915_dev_to_pxp (i915_kdev); +intel_wakeref_t wakeref;mutex_lock (&pxp->tee_mutex); pxp->pxp_component = data; pxp->pxp_component->tee_dev = tee_kdev;mutex_unlock (&pxp->tee_mutex); +/* if we are suspended, the HW will be re-initialized on resume */ + wakeref =intel_runtime_pm_get_if_in_use (&i915->runtime_pm); +if (!wakeref) +return 0 ; +/* the component is required to fully start the PXP HW */ intel_pxp_init_hw (pxp); +intel_runtime_pm_put (&i915->runtime_pm, wakeref); +return 0 ; }
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 13/15 drm/i915/pxp: Add plane decryption support
In reply to Daniele Ceraolo SpurioFrom: Anshuman Gupta <[email protected]>
Add support to enable/disable PLANE_SURF Decryption Request bit. It requires only to enable plane decryption support when following condition met.
1. PXP session is enabled.
2. Buffer object is protected.
v2:
- Used gen fb obj user_flags instead gem_object_metadata. [Krishna]
v3:
- intel_pxp_gem_object_status() API changes.
v4: use intel_pxp_is_active (Daniele)
v5: rebase and use the new protected object status checker (Daniele)
v6: used plane state for plane_decryption to handle async flip
as suggested by Ville.
v7: check pxp session while plane decrypt state computation. [Ville]
removed pointless code. [Ville]
v8 (Daniele): update PXP check
--- .../gpu/drm/i915/display/intel_atomic_plane.c | 16 ++++++++++++++++ drivers/gpu/drm/i915/display/intel_display.c | 4 ++++ .../gpu/drm/i915/display/intel_display_types.h | 3 +++ .../gpu/drm/i915/display/skl_universal_plane.c | 15 ++++++++++++--- drivers/gpu/drm/i915/i915_reg.h | 1 + 5 files changed, 36 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/display/intel_atomic_plane.c +16 -0
@@ -41,6 +41,7 @@#include "intel_display_types.h" #include "intel_pm.h" #include "intel_sprite.h" +#include "pxp/intel_pxp.h" static void intel_plane_state_reset (struct intel_plane_state *plane_state,struct intel_plane *plane)
@@ -383,6 +384,14 @@ intel_crtc_get_plane(struct intel_crtc *crtc, enum plane_id plane_id)return NULL ; } +static int bo_has_valid_encryption (const struct drm_i915_gem_object *obj) +{ +struct drm_i915_private *i915 =to_i915 (obj->base.dev); + +return i915_gem_object_has_valid_protection (obj) && +intel_pxp_is_active (&i915->gt.pxp); +} +int intel_plane_atomic_check (struct intel_atomic_state *state,struct intel_plane *plane) {
@@ -397,6 +406,7 @@ int intel_plane_atomic_check(struct intel_atomic_state *state,intel_atomic_get_old_crtc_state (state, crtc);struct intel_crtc_state *new_crtc_state =intel_atomic_get_new_crtc_state (state, crtc); +const struct drm_framebuffer *fb;if (new_crtc_state && new_crtc_state->bigjoiner_slave) { struct intel_plane *master_plane =
@@ -413,6 +423,12 @@ int intel_plane_atomic_check(struct intel_atomic_state *state, new_master_plane_state, crtc); + fb = new_plane_state->hw.fb; +if (fb) + new_plane_state->decrypt =bo_has_valid_encryption (intel_fb_obj (fb)); +else + new_plane_state->decrypt = old_plane_state->decrypt; + new_plane_state->uapi.visible = false;if (!new_crtc_state)return 0 ;
drivers/gpu/drm/i915/display/intel_display.c +4 -0
@@ -9891,6 +9891,10 @@ static int intel_atomic_check_async(struct intel_atomic_state *state)drm_dbg_kms (&i915->drm,"Color range cannot be changed in async flip\n" );return -EINVAL ; } + +/* plane decryption is allow to change only in synchronous flips */ +if (old_plane_state->decrypt != new_plane_state->decrypt) +return -EINVAL ; }return 0 ;
drivers/gpu/drm/i915/display/intel_display_types.h +3 -0
@@ -629,6 +629,9 @@ struct intel_plane_state {struct intel_fb_view view; +/* Plane pxp decryption state */ +bool decrypt; +/* plane control register */ u32 ctl;
drivers/gpu/drm/i915/display/skl_universal_plane.c +12 -3
@@ -18,6 +18,7 @@#include "intel_sprite.h" #include "skl_scaler.h" #include "skl_universal_plane.h" +#include "pxp/intel_pxp.h" static const u32 skl_plane_formats[] = {DRM_FORMAT_C8 ,
@@ -1024,7 +1025,7 @@ skl_program_plane(struct intel_plane *plane,u8 alpha = plane_state->hw.alpha >>8 ;u32 plane_color_ctl =0 , aux_dist =0 ;unsigned long irqflags; -u32 keymsk, keymax; +u32 keymsk, keymax, plane_surf;u32 plane_ctl = plane_state->ctl; plane_ctl |=skl_plane_ctl_crtc (crtc_state);
@@ -1113,8 +1114,16 @@ skl_program_plane(struct intel_plane *plane,* the control register just before the surface register. */ intel_de_write_fw (dev_priv,PLANE_CTL (pipe, plane_id), plane_ctl); -intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), -intel_plane_ggtt_offset (plane_state) + surf_addr); + plane_surf =intel_plane_ggtt_offset (plane_state) + surf_addr; + +/* +* FIXME: pxp session invalidation can hit any time even at time of commit +* or after the commit, display content will be garbage. +*/ +if (plane_state->decrypt) + plane_surf |=PLANE_SURF_DECRYPT ; + +intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), plane_surf);spin_unlock_irqrestore (&dev_priv->uncore.lock, irqflags); }
drivers/gpu/drm/i915/i915_reg.h +1 -0
@@ -7345,6 +7345,7 @@ enum {#define _PLANE_SURF_3 (pipe) _PIPE(pipe, _PLANE_SURF_3_A, _PLANE_SURF_3_B)#define PLANE_SURF (pipe, plane) \ _MMIO_PLANE(plane, _PLANE_SURF_1(pipe), _PLANE_SURF_2(pipe)) +#define PLANE_SURF_DECRYPT REG_BIT(2)#define _PLANE_OFFSET_1_B 0x711a4#define _PLANE_OFFSET_2_B 0x712a4
Signature · 1 line
2.32.0
PATCH v6 13/15 drm/i915/pxp: Add plane decryption support
Add support to enable/disable PLANE_SURF Decryption Request bit. It requires only to enable plane decryption support when following condition met.
1. PXP session is enabled.
2. Buffer object is protected.
v2:
- Used gen fb obj user_flags instead gem_object_metadata. [Krishna]
v3:
- intel_pxp_gem_object_status() API changes.
v4: use intel_pxp_is_active (Daniele)
v5: rebase and use the new protected object status checker (Daniele)
v6: used plane state for plane_decryption to handle async flip
as suggested by Ville.
v7: check pxp session while plane decrypt state computation. [Ville]
removed pointless code. [Ville]
v8 (Daniele): update PXP check
v9: move decrypt check after icl_check_nv12_planes() when overlays
have fb set (Juston)
--- drivers/gpu/drm/i915/display/intel_display.c | 27 +++++++++++++++++++ .../drm/i915/display/intel_display_types.h | 3 +++ .../drm/i915/display/skl_universal_plane.c | 15 ++++++++--- drivers/gpu/drm/i915/i915_reg.h | 1 + 4 files changed, 43 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/display/intel_display.c +27 -0
@@ -70,6 +70,8 @@#include "gt/intel_rps.h" #include "gt/gen8_ppgtt.h" +#include "pxp/intel_pxp.h" +#include "g4x_dp.h" #include "g4x_hdmi.h" #include "i915_drv.h"
@@ -9610,13 +9612,24 @@ static int intel_bigjoiner_add_affected_planes(struct intel_atomic_state *state)return 0 ; } +static int bo_has_valid_encryption (const struct drm_i915_gem_object *obj) +{ +struct drm_i915_private *i915 =to_i915 (obj->base.dev); + +return i915_gem_object_has_valid_protection (obj) && +intel_pxp_is_active (&i915->gt.pxp); +} +static int intel_atomic_check_planes (struct intel_atomic_state *state) {struct drm_i915_private *dev_priv =to_i915 (state->base.dev);struct intel_crtc_state *old_crtc_state, *new_crtc_state;struct intel_plane_state *plane_state;struct intel_plane *plane; +struct intel_plane_state *new_plane_state; +struct intel_plane_state *old_plane_state;struct intel_crtc *crtc; +const struct drm_framebuffer *fb;int i, ret; ret =icl_add_linked_planes (state);
@@ -9664,6 +9677,16 @@ static int intel_atomic_check_planes(struct intel_atomic_state *state)return ret; } +for_each_new_intel_plane_in_state (state, plane, plane_state, i) { + new_plane_state =intel_atomic_get_new_plane_state (state, plane); + old_plane_state =intel_atomic_get_old_plane_state (state, plane); + fb = new_plane_state->hw.fb; +if (fb) + new_plane_state->decrypt =bo_has_valid_encryption (intel_fb_obj (fb)); +else + new_plane_state->decrypt = old_plane_state->decrypt; + } +return 0 ; }
@@ -9950,6 +9973,10 @@ static int intel_atomic_check_async(struct intel_atomic_state *state)drm_dbg_kms (&i915->drm,"Color range cannot be changed in async flip\n" );return -EINVAL ; } + +/* plane decryption is allow to change only in synchronous flips */ +if (old_plane_state->decrypt != new_plane_state->decrypt) +return -EINVAL ; }return 0 ;
drivers/gpu/drm/i915/display/intel_display_types.h +3 -0
@@ -629,6 +629,9 @@ struct intel_plane_state {struct intel_fb_view view; +/* Plane pxp decryption state */ +bool decrypt; +/* plane control register */ u32 ctl;
drivers/gpu/drm/i915/display/skl_universal_plane.c +12 -3
@@ -18,6 +18,7 @@#include "intel_sprite.h" #include "skl_scaler.h" #include "skl_universal_plane.h" +#include "pxp/intel_pxp.h" static const u32 skl_plane_formats[] = {DRM_FORMAT_C8 ,
@@ -1024,7 +1025,7 @@ skl_program_plane(struct intel_plane *plane,u8 alpha = plane_state->hw.alpha >>8 ;u32 plane_color_ctl =0 , aux_dist =0 ;unsigned long irqflags; -u32 keymsk, keymax; +u32 keymsk, keymax, plane_surf;u32 plane_ctl = plane_state->ctl; plane_ctl |=skl_plane_ctl_crtc (crtc_state);
@@ -1113,8 +1114,16 @@ skl_program_plane(struct intel_plane *plane,* the control register just before the surface register. */ intel_de_write_fw (dev_priv,PLANE_CTL (pipe, plane_id), plane_ctl); -intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), -intel_plane_ggtt_offset (plane_state) + surf_addr); + plane_surf =intel_plane_ggtt_offset (plane_state) + surf_addr; + +/* +* FIXME: pxp session invalidation can hit any time even at time of commit +* or after the commit, display content will be garbage. +*/ +if (plane_state->decrypt) + plane_surf |=PLANE_SURF_DECRYPT ; + +intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), plane_surf);spin_unlock_irqrestore (&dev_priv->uncore.lock, irqflags); }
drivers/gpu/drm/i915/i915_reg.h +1 -0
@@ -7347,6 +7347,7 @@ enum {#define _PLANE_SURF_3 (pipe) _PIPE(pipe, _PLANE_SURF_3_A, _PLANE_SURF_3_B)#define PLANE_SURF (pipe, plane) \ _MMIO_PLANE(plane, _PLANE_SURF_1(pipe), _PLANE_SURF_2(pipe)) +#define PLANE_SURF_DECRYPT REG_BIT(2)#define _PLANE_OFFSET_1_B 0x711a4#define _PLANE_OFFSET_2_B 0x712a4
Signature · 1 line
2.26.2
Re: [Intel-gfx] [PATCH v6 13/15] drm/i915/pxp: Add plane decryption support
Quoted 32 lines
-----Original Message-----
From: Gupta, Anshuman <[email protected]>
Sent: Thursday, August 12, 2021 1:47 PM
To: [email protected]
Cc: Shankar, Uma <[email protected]>; Gupta, Anshuman <[email protected]>; Bommu, Krishnaiah <[email protected]>; Huang Sean Z <[email protected]>; Gaurav, Kumar <[email protected]>; Ville Syrjälä <[email protected]>; Ceraolo Spurio, Daniele <[email protected]>; Li, Juston <[email protected]>; Vivi, Rodrigo <[email protected]>
Subject: [PATCH v6 13/15] drm/i915/pxp: Add plane decryption supportAdd support to enable/disable
PLANE_SURFDecryption Request bit.
It requires only to enable plane decryption support when following condition met.
1. PXP session is enabled.
2. Buffer object is protected.v2:
- Used gen fb objuser_flagsinsteadgem_object_metadata. [Krishna]v3:
-intel_pxp_gem_object_status()API changes.v4: use
intel_pxp_is_active(Daniele)v5: rebase and use the new protected object status checker (Daniele)
v6: used plane state for
plane_decryptionto handle async flip
as suggested by Ville.
Quoted 7 lines
v7: check pxp session while plane decrypt state computation. [Ville]
removed pointless code. [Ville]v8 (Daniele): update PXP check
v9: move decrypt check after
icl_check_nv12_planes()when overlays
have fb set (Juston)
Looks Good to me.
Quoted 157 lines
Cc Bommu Krishnaiah <[email protected]>Cc Huang Sean Z <[email protected]>Cc Gaurav Kumar <[email protected]>Cc Ville Syrjälä <[email protected]>Signed-off-by Anshuman Gupta <[email protected]>Signed-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Juston Li <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]>--- drivers/gpu/drm/i915/display/intel_display.c | 27 +++++++++++++++++++ .../drm/i915/display/intel_display_types.h | 3 +++ .../drm/i915/display/skl_universal_plane.c | 15 ++++++++--- drivers/gpu/drm/i915/i915_reg.h | 1 + 4 files changed, 43 insertions(+), 3 deletions(-)drivers/gpu/drm/i915/display/intel_display.c +23
-0@@ -70,6 +70,8 @@#include "gt/intel_rps.h" #include "gt/gen8_ppgtt.h" +#include "pxp/intel_pxp.h" +#include "g4x_dp.h" #include "g4x_hdmi.h" #include "i915_drv.h" @@ -9610,13 +9612,24 @@ static int intel_bigjoiner_add_affected_planes(struct intel_atomic_state *state)return 0 ; } +static int bo_has_valid_encryption (const struct drm_i915_gem_object +*obj) { +struct drm_i915_private *i915 =to_i915 (obj->base.dev); + +return i915_gem_object_has_valid_protection (obj) && +intel_pxp_is_active (&i915->gt.pxp); +} +static int intel_atomic_check_planes (struct intel_atomic_state *state) {struct drm_i915_private *dev_priv =to_i915 (state->base.dev);struct intel_crtc_state *old_crtc_state, *new_crtc_state;struct intel_plane_state *plane_state;struct intel_plane *plane; +struct intel_plane_state *new_plane_state; +struct intel_plane_state *old_plane_state;struct intel_crtc *crtc; +const struct drm_framebuffer *fb;int i, ret; ret =icl_add_linked_planes (state);@@ -9664,6 +9677,16 @@ static int intel_atomic_check_planes(struct intel_atomic_state *state)return ret; } +for_each_new_intel_plane_in_state (state, plane, plane_state, i) { + new_plane_state =intel_atomic_get_new_plane_state (state, plane); + old_plane_state =intel_atomic_get_old_plane_state (state, plane); + fb = new_plane_state->hw.fb; +if (fb) + new_plane_state->decrypt =bo_has_valid_encryption (intel_fb_obj (fb)); +else + new_plane_state->decrypt = old_plane_state->decrypt; + } +drivers/gpu/drm/i915/display/intel_display.c +4
-0return 0 ; }@@ -9950,6 +9973,10 @@ static int intel_atomic_check_async(struct intel_atomic_state *state) drm_dbg_kms(&i915->drm, "Color range cannot be changed in async flip\n"); return -EINVAL ; } + +/* plane decryption is allow to change only in synchronous flips */ +if (old_plane_state->decrypt != new_plane_state->decrypt) +return -EINVAL ; }drivers/gpu/drm/i915/display/intel_display.c +0
-0return 0 ;drivers/gpu/drm/i915/display/intel_display_types.h +3
-0@@ -629,6 +629,9 @@ struct intel_plane_state {struct intel_fb_view view; +/* Plane pxp decryption state */ +bool decrypt; +/* plane control register */ u32 ctl;drivers/gpu/drm/i915/display/skl_universal_plane.c +2
-1@@ -18,6 +18,7 @@#include "intel_sprite.h" #include "skl_scaler.h" #include "skl_universal_plane.h" +#include "pxp/intel_pxp.h" static const u32 skl_plane_formats[] = {DRM_FORMAT_C8 ,@@ -1024,7 +1025,7 @@ skl_program_plane(struct intel_plane *plane,u8 alpha = plane_state->hw.alpha >>8 ;u32 plane_color_ctl =0 , aux_dist =0 ;unsigned long irqflags; -u32 keymsk, keymax; +u32 keymsk, keymax, plane_surf;u32 plane_ctl = plane_state->ctl; plane_ctl |= skl_plane_ctl_crtc(crtc_state); @@ -1113,8 +1114,16 @@skl_program_plane(struct
intel_plane*plane,drivers/gpu/drm/i915/display/skl_universal_plane.c +10
-2* the control register just before the surface register. */ intel_de_write_fw (dev_priv,PLANE_CTL (pipe, plane_id), plane_ctl); -intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), -intel_plane_ggtt_offset (plane_state) + surf_addr); + plane_surf =intel_plane_ggtt_offset (plane_state) + surf_addr; + +/* +* FIXME: pxp session invalidation can hit any time even at time of commit +* or after the commit, display content will be garbage. +*/ +if (plane_state->decrypt) + plane_surf |=PLANE_SURF_DECRYPT ; + +intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), plane_surf); spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags); } diff --git
a/drivers/gpu/drm/i915/i915_reg.hb/drivers/gpu/drm/i915/i915_reg.hindexf76eaed8e855..57f22f54d9f7100644drivers/gpu/drm/i915/i915_reg.h +1
-0@@ -7347,6 +7347,7 @@ enum {#define _PLANE_SURF_3 (pipe) _PIPE(pipe, _PLANE_SURF_3_A, _PLANE_SURF_3_B)#define PLANE_SURF (pipe, plane) \ _MMIO_PLANE(plane, _PLANE_SURF_1(pipe), _PLANE_SURF_2(pipe)) +#define PLANE_SURF_DECRYPT REG_BIT(2)#define _PLANE_OFFSET_1_B 0x711a4#define
_PLANE_OFFSET_2_B0x712a4
--
2.26.2
PATCH v6 14/15 drm/i915/pxp: black pixels on pxp disabled
In reply to Daniele Ceraolo SpurioFrom: Anshuman Gupta <[email protected]>
When protected sufaces has flipped and pxp session is disabled, display black pixels by using plane color CTM correction.
v2:
- Display black pixels in async flip too.
v3:
- Removed the black pixels logic for async flip. [Ville]
- Used plane state to force black pixels. [Ville]
v4 (Daniele): update pxp_is_borked check.
--- .../gpu/drm/i915/display/intel_atomic_plane.c | 13 +++++- .../drm/i915/display/intel_display_types.h | 3 ++ .../drm/i915/display/skl_universal_plane.c | 36 ++++++++++++++- drivers/gpu/drm/i915/i915_reg.h | 46 +++++++++++++++++++ 4 files changed, 95 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/display/intel_atomic_plane.c +11 -2
@@ -392,6 +392,11 @@ static int bo_has_valid_encryption(const struct drm_i915_gem_object *obj)intel_pxp_is_active (&i915->gt.pxp); } +static bool pxp_is_borked (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !bo_has_valid_encryption (obj); +} +int intel_plane_atomic_check (struct intel_atomic_state *state,struct intel_plane *plane) {
@@ -424,10 +429,14 @@ int intel_plane_atomic_check(struct intel_atomic_state *state, crtc); fb = new_plane_state->hw.fb; -if (fb) +if (fb) { new_plane_state->decrypt =bo_has_valid_encryption (intel_fb_obj (fb)); -else + new_plane_state->force_black =pxp_is_borked (intel_fb_obj (fb)); + + }else { new_plane_state->decrypt = old_plane_state->decrypt; + new_plane_state->force_black = old_plane_state->force_black; + } new_plane_state->uapi.visible = false;if (!new_crtc_state)
drivers/gpu/drm/i915/display/intel_display_types.h +3 -0
@@ -632,6 +632,9 @@ struct intel_plane_state {/* Plane pxp decryption state */ bool decrypt; +/* Plane state to display black pixels when pxp is borked */ +bool force_black; +/* plane control register */ u32 ctl;
drivers/gpu/drm/i915/display/skl_universal_plane.c +35 -1
@@ -1002,6 +1002,33 @@ static u32 skl_surf_address(const struct intel_plane_state *plane_state, } } +static void intel_load_plane_csc_black (struct intel_plane *intel_plane) +{ +struct drm_i915_private *dev_priv =to_i915 (intel_plane->base.dev); +enum pipe pipe = intel_plane->pipe; +enum plane_id plane = intel_plane->id; +u16 postoff =0 ; + +drm_dbg_kms (&dev_priv->drm,"plane color CTM to black %s:%d\n" , + intel_plane->base.name, plane); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,0 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,1 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,2 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,3 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,4 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,5 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,0 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,1 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,2 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,0 ), postoff); +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,1 ), postoff); +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,2 ), postoff); +} + static void skl_program_plane(struct intel_plane *plane, const struct intel_crtc_state *crtc_state,
@@ -1115,14 +1142,21 @@ skl_program_plane(struct intel_plane *plane,*/ intel_de_write_fw (dev_priv,PLANE_CTL (pipe, plane_id), plane_ctl); plane_surf =intel_plane_ggtt_offset (plane_state) + surf_addr; + plane_color_ctl =intel_de_read_fw (dev_priv,PLANE_COLOR_CTL (pipe, plane_id));/* * FIXME: pxp session invalidation can hit any time even at time of commit * or after the commit, display content will be garbage. */ -if (plane_state->decrypt) +if (plane_state->decrypt) { plane_surf |=PLANE_SURF_DECRYPT ; + }else if (plane_state->force_black) { +intel_load_plane_csc_black (plane); + plane_color_ctl |=PLANE_COLOR_PLANE_CSC_ENABLE ; + } +intel_de_write_fw (dev_priv,PLANE_COLOR_CTL (pipe, plane_id), + plane_color_ctl);intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), plane_surf);spin_unlock_irqrestore (&dev_priv->uncore.lock, irqflags);
drivers/gpu/drm/i915/i915_reg.h +46 -0
58 lines not on this page · raw message
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 14/15 drm/i915/pxp: black pixels on pxp disabled
When protected sufaces has flipped and pxp session is disabled, display black pixels by using plane color CTM correction.
v2:
- Display black pixels in async flip too.
v3:
- Removed the black pixels logic for async flip. [Ville]
- Used plane state to force black pixels. [Ville]
v4 (Daniele): update pxp_is_borked check.
v5: rebase on top of v9 plane decryption moving the decrypt check
(Juston)
--- drivers/gpu/drm/i915/display/intel_display.c | 12 ++++- .../drm/i915/display/intel_display_types.h | 3 ++ .../drm/i915/display/skl_universal_plane.c | 36 ++++++++++++++- drivers/gpu/drm/i915/i915_reg.h | 46 +++++++++++++++++++ 4 files changed, 94 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/display/intel_display.c +10 -2
@@ -9620,6 +9620,11 @@ static int bo_has_valid_encryption(const struct drm_i915_gem_object *obj)intel_pxp_is_active (&i915->gt.pxp); } +static bool pxp_is_borked (const struct drm_i915_gem_object *obj) +{ +return i915_gem_object_is_protected (obj) && !bo_has_valid_encryption (obj); +} +static int intel_atomic_check_planes (struct intel_atomic_state *state) {struct drm_i915_private *dev_priv =to_i915 (state->base.dev);
@@ -9681,10 +9686,13 @@ static int intel_atomic_check_planes(struct intel_atomic_state *state) new_plane_state =intel_atomic_get_new_plane_state (state, plane); old_plane_state =intel_atomic_get_old_plane_state (state, plane); fb = new_plane_state->hw.fb; -if (fb) +if (fb) { new_plane_state->decrypt =bo_has_valid_encryption (intel_fb_obj (fb)); -else + new_plane_state->force_black =pxp_is_borked (intel_fb_obj (fb)); + }else { new_plane_state->decrypt = old_plane_state->decrypt; + new_plane_state->force_black = old_plane_state->force_black; + } }return 0 ;
drivers/gpu/drm/i915/display/intel_display_types.h +3 -0
@@ -632,6 +632,9 @@ struct intel_plane_state {/* Plane pxp decryption state */ bool decrypt; +/* Plane state to display black pixels when pxp is borked */ +bool force_black; +/* plane control register */ u32 ctl;
drivers/gpu/drm/i915/display/skl_universal_plane.c +35 -1
@@ -1002,6 +1002,33 @@ static u32 skl_surf_address(const struct intel_plane_state *plane_state, } } +static void intel_load_plane_csc_black (struct intel_plane *intel_plane) +{ +struct drm_i915_private *dev_priv =to_i915 (intel_plane->base.dev); +enum pipe pipe = intel_plane->pipe; +enum plane_id plane = intel_plane->id; +u16 postoff =0 ; + +drm_dbg_kms (&dev_priv->drm,"plane color CTM to black %s:%d\n" , + intel_plane->base.name, plane); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,0 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,1 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,2 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,3 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,4 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,5 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,0 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,1 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,2 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,0 ), postoff); +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,1 ), postoff); +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,2 ), postoff); +} + static void skl_program_plane(struct intel_plane *plane, const struct intel_crtc_state *crtc_state,
@@ -1115,14 +1142,21 @@ skl_program_plane(struct intel_plane *plane,*/ intel_de_write_fw (dev_priv,PLANE_CTL (pipe, plane_id), plane_ctl); plane_surf =intel_plane_ggtt_offset (plane_state) + surf_addr; + plane_color_ctl =intel_de_read_fw (dev_priv,PLANE_COLOR_CTL (pipe, plane_id));/* * FIXME: pxp session invalidation can hit any time even at time of commit * or after the commit, display content will be garbage. */ -if (plane_state->decrypt) +if (plane_state->decrypt) { plane_surf |=PLANE_SURF_DECRYPT ; + }else if (plane_state->force_black) { +intel_load_plane_csc_black (plane); + plane_color_ctl |=PLANE_COLOR_PLANE_CSC_ENABLE ; + } +intel_de_write_fw (dev_priv,PLANE_COLOR_CTL (pipe, plane_id), + plane_color_ctl);intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), plane_surf);spin_unlock_irqrestore (&dev_priv->uncore.lock, irqflags);
drivers/gpu/drm/i915/i915_reg.h +46 -0
58 lines not on this page · raw message
Signature · 1 line
2.26.2
Re: [Intel-gfx] [PATCH v6 14/15] drm/i915/pxp: black pixels on pxp disabled
Quoted 19 lines
-----Original Message-----
From: Gupta, Anshuman <[email protected]>
Sent: Thursday, August 12, 2021 1:50 PM
To: [email protected]
Cc: Shankar, Uma <[email protected]>; Gupta, Anshuman <[email protected]>; Ville Syrjälä <[email protected]>; Gaurav, Kumar <[email protected]>; Ceraolo Spurio, Daniele <[email protected]>; Li, Juston <[email protected]>; Vivi, Rodrigo <[email protected]>
Subject: [PATCH v6 14/15] drm/i915/pxp: black pixels on pxp disabledWhen protected sufaces has flipped and pxp session is disabled, display black pixels by using plane color CTM correction.
v2:
- Display black pixels in async flip too.
Quoted 8 lines
v3:
- Removed the black pixels logic for async flip. [Ville]
- Used plane state to force black pixels. [Ville]v4 (Daniele): update
pxp_is_borkedcheck.v5: rebase on top of v9 plane decryption moving the decrypt check
(Juston)
Looks Good to me.
Quoted 209 lines
Cc Ville Syrjälä <[email protected]>Cc Gaurav Kumar <[email protected]>Cc Shankar Uma <[email protected]>Signed-off-by Anshuman Gupta <[email protected]>Signed-off-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Juston Li <[email protected]>Reviewed-by Rodrigo Vivi <[email protected]>--- drivers/gpu/drm/i915/display/intel_display.c | 12 ++++- .../drm/i915/display/intel_display_types.h | 3 ++ .../drm/i915/display/skl_universal_plane.c | 36 ++++++++++++++- drivers/gpu/drm/i915/i915_reg.h | 46 +++++++++++++++++++ 4 files changed, 94 insertions(+), 3 deletions(-)drivers/gpu/drm/i915/display/intel_display.c +4
-0@@ -9620,6 +9620,11 @@ static int bo_has_valid_encryption(const struct drm_i915_gem_object *obj) intel_pxp_is_active(&i915->gt.pxp); } +static bool pxp_is_borked (const struct drm_i915_gem_object *obj) { +return i915_gem_object_is_protected (obj) && +!bo_has_valid_encryption (obj); } +static int intel_atomic_check_planes (struct intel_atomic_state *state) { struct drm_i915_private *dev_priv = to_i915(state->base.dev); @@ -9681,10 +9686,13 @@ static int
intel_atomic_check_planes(struct intel_atomic_state *state)
new_plane_state= intel_atomic_get_new_plane_state(state,
plane);drivers/gpu/drm/i915/display/intel_display.c +1
-1old_plane_state =intel_atomic_get_old_plane_state (state, plane); fb = new_plane_state->hw.fb; -if (fb) +if (fb) { new_plane_state->decrypt =bo_has_valid_encryption(intel_fb_obj(fb));
drivers/gpu/drm/i915/display/intel_display.c +1
-1-else + new_plane_state->force_black =pxp_is_borked(intel_fb_obj(fb));
drivers/gpu/drm/i915/display/intel_display.c +2
-0+ }else { new_plane_state->decrypt = old_plane_state->decrypt; + new_plane_state->force_black = old_plane_state-Quoted 1 line
force_black;drivers/gpu/drm/i915/display/intel_display.c +1
-0+ } }return 0 ;drivers/gpu/drm/i915/display/intel_display_types.h +3
-0@@ -632,6 +632,9 @@ struct intel_plane_state {/* Plane pxp decryption state */ bool decrypt; +/* Plane state to display black pixels when pxp is borked */ +bool force_black; +/* plane control register */ u32 ctl;drivers/gpu/drm/i915/display/skl_universal_plane.c +27
-0@@ -1002,6 +1002,33 @@ static u32 skl_surf_address(const struct intel_plane_state *plane_state, } } +static void intel_load_plane_csc_black (struct intel_plane *intel_plane) +{ +struct drm_i915_private *dev_priv =to_i915 (intel_plane->base.dev); +enum pipe pipe = intel_plane->pipe; +enum plane_id plane = intel_plane->id; +u16 postoff =0 ; + +drm_dbg_kms (&dev_priv->drm,"plane color CTM to black %s:%d\n" , + intel_plane->base.name, plane); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,0 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,1 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,2 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,3 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,4 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_COEFF (pipe, plane,5 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,0 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,1 ),0 ); +intel_de_write_fw (dev_priv,PLANE_CSC_PREOFF (pipe, plane,2 ),0 ); + +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,0 ), postoff); +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,1 ), postoff); +intel_de_write_fw (dev_priv,PLANE_CSC_POSTOFF (pipe, plane,2 ), +postoff); } + static void skl_program_plane(struct intel_plane *plane,const struct
intel_crtc_state*crtc_state, @@ -1115,14 +1142,21
@@ skl_program_plane(structintel_plane*plane,drivers/gpu/drm/i915/display/skl_universal_plane.c +9
-1*/ intel_de_write_fw (dev_priv,PLANE_CTL (pipe, plane_id), plane_ctl); plane_surf =intel_plane_ggtt_offset (plane_state) + surf_addr; + plane_color_ctl =intel_de_read_fw (dev_priv,PLANE_COLOR_CTL (pipe, +plane_id));/* * FIXME: pxp session invalidation can hit any time even at time of commit * or after the commit, display content will be garbage. */ -if (plane_state->decrypt) +if (plane_state->decrypt) { plane_surf |=PLANE_SURF_DECRYPT ; + }else if (plane_state->force_black) { +intel_load_plane_csc_black (plane); + plane_color_ctl |=PLANE_COLOR_PLANE_CSC_ENABLE ; + } +intel_de_write_fw (dev_priv,PLANE_COLOR_CTL (pipe, plane_id), + plane_color_ctl);intel_de_write_fw (dev_priv,PLANE_SURF (pipe, plane_id), plane_surf); spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags); diff --git
a/drivers/gpu/drm/i915/i915_reg.hb/drivers/gpu/drm/i915/i915_reg.hindex57f22f54d9f7..db714ea91997100644drivers/gpu/drm/i915/i915_reg.h +1
-0@@ -7224,6 +7224,7 @@ enum {#define _PLANE_COLOR_CTL_3_A 0x703CC/* GLK+ */ #define PLANE_COLOR_PIPE_GAMMA_ENABLE (1 << 30)/* Pre-ICL */ #define PLANE_COLOR_YUV_RANGE_CORRECTION_DISABLE (1 << 28) +#define PLANE_COLOR_PLANE_CSC_ENABLE REG_BIT(21)/* ICL+ */
#define PLANE_COLOR_INPUT_CSC_ENABLE (1 << 20) /* ICL+ */ #define PLANE_COLOR_PIPE_CSC_ENABLE (1 << 23) /* Pre-ICL */ #define PLANE_COLOR_CSC_MODE_BYPASS (0 << 17)drivers/gpu/drm/i915/i915_reg.h +25
-0@@ -11341,6 +11342,51 @@ enum skl_power_gate { _PAL_PREC_MULTI_SEG_DATA_A, \ _PAL_PREC_MULTI_SEG_DATA_B) +#define _MMIO_PLANE_GAMC (plane, i, a, b) _MMIO(_PIPE(plane, a, b) + +(i) * 4) + +/* Plane CSC Registers */ +#define _PLANE_CSC_RY_GY_1_A 0x70210 +#define _PLANE_CSC_RY_GY_2_A 0x70310 + +#define _PLANE_CSC_RY_GY_1_B 0x71210 +#define _PLANE_CSC_RY_GY_2_B 0x71310 + +#define _PLANE_CSC_RY_GY_1 (pipe) _PIPE(pipe, _PLANE_CSC_RY_GY_1_A, \ + _PLANE_CSC_RY_GY_1_B) +#define _PLANE_CSC_RY_GY_2 (pipe) _PIPE(pipe, _PLANE_INPUT_CSC_RY_GY_2_A, \ + _PLANE_INPUT_CSC_RY_GY_2_B) +#define PLANE_CSC_COEFF (pipe, plane, index) _MMIO_PLANE(plane, \ + _PLANE_CSC_RY_GY_1(pipe) + (index) * 4, \ + _PLANE_CSC_RY_GY_2(pipe) + (index) * 4) + +#define _PLANE_CSC_PREOFF_HI_1_A 0x70228 +#define _PLANE_CSC_PREOFF_HI_2_A 0x70328 + +#define _PLANE_CSC_PREOFF_HI_1_B 0x71228 +#define _PLANE_CSC_PREOFF_HI_2_B 0x71328 + +#define _PLANE_CSC_PREOFF_HI_1 (pipe) _PIPE(pipe,
_PLANE_CSC_PREOFF_HI_1_A, \drivers/gpu/drm/i915/i915_reg.h +2
-0+ _PLANE_CSC_PREOFF_HI_1_B) +#define _PLANE_CSC_PREOFF_HI_2 (pipe) _PIPE(pipe,
_PLANE_CSC_PREOFF_HI_2_A, \drivers/gpu/drm/i915/i915_reg.h +2
-0+ _PLANE_CSC_PREOFF_HI_2_B) +#define PLANE_CSC_PREOFF (pipe, plane, index) _MMIO_PLANE(plane,
_PLANE_CSC_PREOFF_HI_1(pipe)+ \drivers/gpu/drm/i915/i915_reg.h +1
-0+ (index) * 4,
_PLANE_CSC_PREOFF_HI_2(pipe)+ \drivers/gpu/drm/i915/i915_reg.h +9
-0+ (index) * 4) + +#define _PLANE_CSC_POSTOFF_HI_1_A 0x70234 +#define _PLANE_CSC_POSTOFF_HI_2_A 0x70334 + +#define _PLANE_CSC_POSTOFF_HI_1_B 0x71234 +#define _PLANE_CSC_POSTOFF_HI_2_B 0x71334 + +#define _PLANE_CSC_POSTOFF_HI_1 (pipe) _PIPE(pipe,
_PLANE_CSC_POSTOFF_HI_1_A, \drivers/gpu/drm/i915/i915_reg.h +2
-0+ _PLANE_CSC_POSTOFF_HI_1_B) +#define _PLANE_CSC_POSTOFF_HI_2 (pipe) _PIPE(pipe,
_PLANE_CSC_POSTOFF_HI_2_A, \drivers/gpu/drm/i915/i915_reg.h +2
-0+ _PLANE_CSC_POSTOFF_HI_2_B) +#define PLANE_CSC_POSTOFF (pipe, plane, index) _MMIO_PLANE(plane,
_PLANE_CSC_POSTOFF_HI_1(pipe)+ \drivers/gpu/drm/i915/i915_reg.h +1
-0+ (index) * 4,
_PLANE_CSC_POSTOFF_HI_2(pipe)+ \drivers/gpu/drm/i915/i915_reg.h +2
-1+ (index) * 4) +/* pipe CSC & degamma/gamma LUTs on CHV */ #define _CGM_PIPE_A_CSC_COEFF01 (VLV_DISPLAY_BASE + 0x67900)#define _CGM_PIPE_A_CSC_COEFF23 (VLV_DISPLAY_BASE + 0x67904) --2.26.2
PATCH v6 15/15 drm/i915/pxp: enable PXP for integrated Gen12
In reply to Daniele Ceraolo SpurioNote that discrete cards can support PXP as well, but we haven't tested on those yet so keeping it disabled for now.
--- drivers/gpu/drm/i915/i915_pci.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/gpu/drm/i915/i915_pci.c +2 -0
@@ -877,6 +877,7 @@ static const struct intel_device_info jsl_info = { }, \ TGL_CURSOR_OFFSETS, \ .has_global_mocs = 1, \ + .has_pxp = 1, \ .display.has_dsb = 1 static const struct intel_device_info tgl_info = {
@@ -903,6 +904,7 @@ static const struct intel_device_info rkl_info = {#define DGFX_FEATURES \ .memory_regions = REGION_SMEM | REGION_LMEM | REGION_STOLEN_LMEM, \ .has_llc = 0, \ + .has_pxp = 0, \ .has_snoop = 1, \ .is_dgfx = 1
Signature · 6 lines
2.32.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI[Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: Introduce Intel PXP (rev2)
In reply to Daniele Ceraolo Spurio== Series Details ==
== Series Details ==
Series: drm/i915: Introduce Intel PXP (rev2)
URL : https://patchwork.freedesktop.org/series/90503/
State : warning
== Summary ==
$ dim checkpatch origin/drm-tip 3eda4d7baf4a drm/i915/pxp: Define PXP component interface -:31: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #31: new file mode 100644
total: 0 errors, 1 warnings, 0 checks, 52 lines checked1ee35a3e836c mei: pxp: export pavp client to me client bus
-:36: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #36:
new file mode 100644
total: 0 errors, 1 warnings, 0 checks, 276 lines checkedc61305748e40 drm/i915/pxp: define PXP device flag and kconfig
-:46: ERROR:COMPLEX_MACRO: Macros with complex values should be enclosed in parentheses #46: FILE: drivers/gpu/drm/i915/i915_drv.h:1727:
+#define HAS_PXP(dev_priv) (IS_ENABLED(CONFIG_DRM_I915_PXP) && \
+ INTEL_INFO(dev_priv)->has_pxp) && \
+ VDBOX_MASK(&dev_priv->gt)
-:46: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'dev_priv' - possible side-effects? #46: FILE: drivers/gpu/drm/i915/i915_drv.h:1727:
+#define HAS_PXP(dev_priv) (IS_ENABLED(CONFIG_DRM_I915_PXP) && \
+ INTEL_INFO(dev_priv)->has_pxp) && \
+ VDBOX_MASK(&dev_priv->gt)
total: 1 errors, 0 warnings, 1 checks, 34 lines checkedaa416e099e6f drm/i915/pxp: allocate a vcs context for pxp usage
-:98: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #98:
new file mode 100644
-:122: ERROR:TRAILING_STATEMENTS: trailing statements should be on next line #122: FILE: drivers/gpu/drm/i915/pxp/intel_pxp.c:20:
+ for (engine = gt->engine_class[VIDEO_DECODE_CLASS][0]; !engine; engine++);
total: 1 errors, 1 warnings, 0 checks, 168 lines checked7ffdfca48a9e drm/i915/pxp: Implement funcs to create the TEE channel
-:75: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #75:
new file mode 100644
total: 0 errors, 1 warnings, 0 checks, 142 lines checked6f6679322f8d drm/i915/pxp: set KCR reg init57f92b5ff2d8 drm/i915/pxp: Create the arbitrary session after boot
-:98: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #98:
new file mode 100644
-:383: CHECK:LINE_SPACING: Please don't use multiple blank lines #383: FILE: drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h:36:
+
+
total: 0 errors, 1 warnings, 1 checks, 339 lines checked 26740ebc8086 drm/i915/pxp: Implement arb session teardown -:63: CHECK:SPACING: spaces preferred around that '<<' (ctx:VxV) #63: FILE: drivers/gpu/drm/i915/gt/intel_gpu_commands.h:153:
+#define MI_FLUSH_DW_PROTECTED_MEM_EN (1<<22)
^
-:117: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #117:
new file mode 100644
total: 0 errors, 1 warnings, 1 checks, 283 lines checked363a76eda1be drm/i915/pxp: Implement PXP irq handler
-:211: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #211:
new file mode 100644
total: 0 errors, 1 warnings, 0 checks, 423 lines checked89558164d429 drm/i915/pxp: interfaces for using protected objects
-:673: CHECK:LINE_SPACING: Please use a blank line after function/struct/union/enum declarations #673: FILE: drivers/gpu/drm/i915/pxp/intel_pxp.h:61:
+}
+static inline void intel_pxp_object_remove(struct drm_i915_gem_object *obj)
total: 0 errors, 0 warnings, 1 checks, 676 lines checkedfe8ab2429443 drm/i915/pxp: start the arb session on demand8ad86e692de6 drm/i915/pxp: Enable PXP power management
-:120: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #120:
new file mode 100644
-:188: CHECK:LINE_SPACING: Please use a blank line after function/struct/union/enum declarations #188: FILE: drivers/gpu/drm/i915/pxp/intel_pxp_pm.h:18:
+}
+static inline void intel_pxp_resume(struct intel_pxp *pxp)
total: 0 errors, 1 warnings, 1 checks, 232 lines checked0f8e7697ee99 drm/i915/pxp: Add plane decryption support9f4496a68386 drm/i915/pxp: black pixels on pxp disabled
-:165: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'pipe' - possible side-effects? #165: FILE: drivers/gpu/drm/i915/i915_reg.h:11389:
+#define PLANE_CSC_COEFF(pipe, plane, index) _MMIO_PLANE(plane, \
+ _PLANE_CSC_RY_GY_1(pipe) + (index) * 4, \
+ _PLANE_CSC_RY_GY_2(pipe) + (index) * 4)
-:165: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'index' - possible side-effects? #165: FILE: drivers/gpu/drm/i915/i915_reg.h:11389:
+#define PLANE_CSC_COEFF(pipe, plane, index) _MMIO_PLANE(plane, \
+ _PLANE_CSC_RY_GY_1(pipe) + (index) * 4, \
+ _PLANE_CSC_RY_GY_2(pipe) + (index) * 4)
-:166: WARNING:LONG_LINE: line length of 102 exceeds 100 columns #166: FILE: drivers/gpu/drm/i915/i915_reg.h:11390:
+ _PLANE_CSC_RY_GY_1(pipe) + (index) * 4, \
-:179: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'pipe' - possible side-effects? #179: FILE: drivers/gpu/drm/i915/i915_reg.h:11403:
+#define PLANE_CSC_PREOFF(pipe, plane, index) _MMIO_PLANE(plane, _PLANE_CSC_PREOFF_HI_1(pipe) + \
+ (index) * 4, _PLANE_CSC_PREOFF_HI_2(pipe) + \
+ (index) * 4)
-:179: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'index' - possible side-effects? #179: FILE: drivers/gpu/drm/i915/i915_reg.h:11403:
+#define PLANE_CSC_PREOFF(pipe, plane, index) _MMIO_PLANE(plane, _PLANE_CSC_PREOFF_HI_1(pipe) + \
+ (index) * 4, _PLANE_CSC_PREOFF_HI_2(pipe) + \
+ (index) * 4)
-:180: WARNING:LONG_LINE: line length of 105 exceeds 100 columns #180: FILE: drivers/gpu/drm/i915/i915_reg.h:11404:
+ (index) * 4, _PLANE_CSC_PREOFF_HI_2(pipe) + \
-:193: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'pipe' - possible side-effects? #193: FILE: drivers/gpu/drm/i915/i915_reg.h:11417:
+#define PLANE_CSC_POSTOFF(pipe, plane, index) _MMIO_PLANE(plane, _PLANE_CSC_POSTOFF_HI_1(pipe) + \
+ (index) * 4, _PLANE_CSC_POSTOFF_HI_2(pipe) + \
+ (index) * 4)
-:193: CHECK:MACRO_ARG_REUSE: Macro argument reuse 'index' - possible side-effects? #193: FILE: drivers/gpu/drm/i915/i915_reg.h:11417:
+#define PLANE_CSC_POSTOFF(pipe, plane, index) _MMIO_PLANE(plane, _PLANE_CSC_POSTOFF_HI_1(pipe) + \
+ (index) * 4, _PLANE_CSC_POSTOFF_HI_2(pipe) + \
+ (index) * 4)
-:194: WARNING:LONG_LINE: line length of 106 exceeds 100 columns #194: FILE: drivers/gpu/drm/i915/i915_reg.h:11418:
+ (index) * 4, _PLANE_CSC_POSTOFF_HI_2(pipe) + \
total: 0 errors, 3 warnings, 6 checks, 149 lines checked 8893b90612e6 drm/i915/pxp: enable PXP for integrated Gen12
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI[Intel-gfx] ✗ Fi.CI.SPARSE: warning for drm/i915: Introduce Intel PXP (rev2)
In reply to Daniele Ceraolo Spurio== Series Details ==
== Series Details ==
Series: drm/i915: Introduce Intel PXP (rev2)
URL : https://patchwork.freedesktop.org/series/90503/
State : warning
== Summary ==
$ dim sparse --fast origin/drm-tip Sparse version: v0.6.2 Fast mode used, each commit won't be checked separately. - +drivers/gpu/drm/i915/display/intel_display.c:1900:21: expected struct i915_vma *[assigned] vma +drivers/gpu/drm/i915/display/intel_display.c:1900:21: got void [noderef] __iomem *[assigned] iomem +drivers/gpu/drm/i915/display/intel_display.c:1900:21: warning: incorrect type in assignment (different address spaces) +drivers/gpu/drm/i915/gem/i915_gem_context.c:1440:34: expected struct i915_address_space *vm +drivers/gpu/drm/i915/gem/i915_gem_context.c:1440:34: got struct i915_address_space [noderef] __rcu *vm +drivers/gpu/drm/i915/gem/i915_gem_context.c:1440:34: warning: incorrect type in argument 1 (different address spaces) +drivers/gpu/drm/i915/gem/selftests/mock_context.c:43:25: expected struct i915_address_space [noderef] __rcu *vm +drivers/gpu/drm/i915/gem/selftests/mock_context.c:43:25: got struct i915_address_space * +drivers/gpu/drm/i915/gem/selftests/mock_context.c:43:25: warning: incorrect type in assignment (different address spaces) +drivers/gpu/drm/i915/gem/selftests/mock_context.c:60:34: expected struct i915_address_space *vm +drivers/gpu/drm/i915/gem/selftests/mock_context.c:60:34: got struct i915_address_space [noderef] __rcu *vm +drivers/gpu/drm/i915/gem/selftests/mock_context.c:60:34: warning: incorrect type in argument 1 (different address spaces) +drivers/gpu/drm/i915/gt/intel_engine_stats.h:27:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:27:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:27:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:32:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:32:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:49:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:49:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:49:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:56:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_engine_stats.h:56:9: warning: trying to copy expression type 31 +drivers/gpu/drm/i915/gt/intel_reset.c:1392:5: warning: context imbalance in 'intel_gt_reset_trylock' - different lock contexts for basic block +drivers/gpu/drm/i915/gt/intel_ring_submission.c:1268:24: warning: Using plain integer as NULL pointer +drivers/gpu/drm/i915/i915_perf.c:1443:15: warning: memset with byte count of 16777216 +drivers/gpu/drm/i915/i915_perf.c:1497:15: warning: memset with byte count of 16777216 +./include/asm-generic/bitops/find.h:112:45: warning: shift count is negative (-262080) +./include/asm-generic/bitops/find.h:32:31: warning: shift count is negative (-262080) +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read64' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_write16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_write32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_write8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read64' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_write16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_write32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_write8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read64' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_write16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_write32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_write8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read64' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_write16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_write32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_write8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen8_write16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen8_write32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen8_write8' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'xehp_fwtable_write16' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'xehp_fwtable_write32' - different lock contexts for basic block +./include/linux/spinlock.h:409:9: warning: context imbalance in 'xehp_fwtable_write8' - different lock contexts for basic block +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +./include/linux/stddef.h:17:9: this was the original definition +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined +/usr/lib/gcc/x86_64-linux-gnu/8/include/stddef.h:417:9: warning: preprocessor token offsetof redefined
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI[Intel-gfx] ✗ Fi.CI.DOCS: warning for drm/i915: Introduce Intel PXP (rev2)
In reply to Daniele Ceraolo Spurio== Series Details ==
== Series Details ==
Series: drm/i915: Introduce Intel PXP (rev2)
URL : https://patchwork.freedesktop.org/series/90503/
State : warning
== Summary ==
$ make htmldocs 2>&1 > /dev/null | grep i915 ./include/uapi/drm/i915_drm.h:1872: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst ./include/uapi/drm/i915_drm.h:1872: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst ./include/uapi/drm/i915_drm.h:1872: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst ./include/uapi/drm/i915_drm.h:1872: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst ./include/uapi/drm/i915_drm.h:3072: warning: Function parameter or member 'base' not described in 'drm_i915_gem_create_ext_protected_content' ./include/uapi/drm/i915_drm.h:3072: warning: Function parameter or member 'flags' not described in 'drm_i915_gem_create_ext_protected_content'
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI[Intel-gfx] ✓ Fi.CI.BAT: success for drm/i915: Introduce Intel PXP (rev2)
In reply to Daniele Ceraolo Spurio== Series Details ==
== Series Details ==
Series: drm/i915: Introduce Intel PXP (rev2)
URL : https://patchwork.freedesktop.org/series/90503/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_10416 -> Patchwork_20735
====================================================
Summary
-------
**SUCCESS**
No regressions found.
External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/index.html
Known issues
------------
Here are the changes found in Patchwork_20735 that come from known issues:
### IGT changes ###
#### Issues hit ####
* igt@amdgpu/amd_basic@semaphore:
- fi-bdw-5557u: NOTRUN -> [SKIP][1] ([fdo#109271]) +29 similar issues
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/fi-bdw-5557u/igt@amdgpu/[email protected]
* igt@core_hotunplug@unbind-rebind:
- fi-bdw-5557u: NOTRUN -> [WARN][2] ([i915#3718])
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/fi-bdw-5557u/igt@[email protected]
* igt@i915_pm_rpm@basic-rte:
- fi-bdw-5557u: NOTRUN -> [FAIL][3] ([i915#579])
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/fi-bdw-5557u/igt@[email protected]
* igt@i915_selftest@live@late_gt_pm:
- fi-bsw-nick: [PASS][4] -> [DMESG-FAIL][5] ([i915#2927])
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10416/fi-bsw-nick/igt@i915_selftest@live@late_gt_pm.html
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/fi-bsw-nick/igt@i915_selftest@live@late_gt_pm.html
* igt@kms_chamelium@dp-crc-fast:
- fi-bdw-5557u: NOTRUN -> [SKIP][6] ([fdo#109271] / [fdo#111827]) +8 similar issues
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/fi-bdw-5557u/igt@[email protected]
* igt@runner@aborted:
- fi-bsw-nick: NOTRUN -> [FAIL][7] ([fdo#109271] / [i915#1436])
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/fi-bsw-nick/igt@[email protected]
[fdo#109271]: https://bugs.freedesktop.org/show_bug.cgi?id=109271
[fdo#111827]: https://bugs.freedesktop.org/show_bug.cgi?id=111827
[i915#1436]: https://gitlab.freedesktop.org/drm/intel/issues/1436
[i915#2927]: https://gitlab.freedesktop.org/drm/intel/issues/2927
[i915#3718]: https://gitlab.freedesktop.org/drm/intel/issues/3718
[i915#579]: https://gitlab.freedesktop.org/drm/intel/issues/579
Participating hosts (40 -> 35)
------------------------------
Missing (5): fi-ilk-m540 fi-hsw-4200u fi-bsw-cyan fi-ctg-p8600 fi-bdw-samus
Build changes
-------------
* Linux: CI_DRM_10416 -> Patchwork_20735
CI-20190529: 20190529
CI_DRM_10416: 68506f9070522fa084cb3a8e5e0de44ee63b61ca @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_6155: 4b51398dcd7559012b85776e7353d516ff1e6ce6 @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
Patchwork_20735: 8893b90612e65adef2c17852100f8eabe83e7724 @ git://anongit.freedesktop.org/gfx-ci/linux
== Linux commits ==
8893b90612e6 drm/i915/pxp: enable PXP for integrated Gen12 9f4496a68386 drm/i915/pxp: black pixels on pxp disabled 0f8e7697ee99 drm/i915/pxp: Add plane decryption support 8ad86e692de6 drm/i915/pxp: Enable PXP power management fe8ab2429443 drm/i915/pxp: start the arb session on demand 89558164d429 drm/i915/pxp: interfaces for using protected objects 363a76eda1be drm/i915/pxp: Implement PXP irq handler 26740ebc8086 drm/i915/pxp: Implement arb session teardown 57f92b5ff2d8 drm/i915/pxp: Create the arbitrary session after boot 6f6679322f8d drm/i915/pxp: set KCR reg init7ffdfca48a9e drm/i915/pxp: Implement funcs to create the TEE channel aa416e099e6f drm/i915/pxp: allocate a vcs context for pxp usage c61305748e40 drm/i915/pxp: define PXP device flag and kconfig 1ee35a3e836c mei: pxp: export pavp client to me client bus 3eda4d7baf4a drm/i915/pxp: Define PXP component interface
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20735/index.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx