[PATCH v7 03/10] arm_ffa: introduce Arm FF-A low-level driver
Abdellatif El Khlifi
abdellatif.elkhlifi at arm.com
Fri Nov 11 15:36:11 CET 2022
On Wed, Nov 09, 2022 at 12:51:26PM +0100, Jens Wiklander wrote:
> On Mon, Nov 07, 2022 at 07:20:48PM +0000, Abdellatif El Khlifi wrote:
> > Add the core driver implementing Arm Firmware Framework for Armv8-A v1.0
> >
> > The Firmware Framework for Arm A-profile processors (FF-A v1.0) [1]
> > describes interfaces (ABIs) that standardize communication
> > between the Secure World and Normal World leveraging TrustZone
> > technology.
> >
> > This driver uses 64-bit registers as per SMCCCv1.2 spec and comes
> > on top of the SMCCC layer. The driver provides the FF-A ABIs needed for
> > querying the FF-A framework from the secure world.
> >
> > The driver uses SMC32 calling convention which means using the first
> > 32-bit data of the Xn registers.
> >
> > All supported ABIs come with their 32-bit version except FFA_RXTX_MAP
> > which has 64-bit version supported.
> >
> > Both 32-bit and 64-bit direct messaging are supported which allows both
> > 32-bit and 64-bit clients to use the FF-A bus.
> >
> > In U-Boot FF-A design, FF-A is considered as a discoverable bus.
> > The Secure World is considered as one entity to communicate with
> > using the FF-A bus. FF-A communication is handled by one device and
> > one instance (the bus). This FF-A driver takes care of all the
> > interactions between Normal world and Secure World.
> >
> > The driver exports its operations to be used by upper layers.
> >
> > Exported operations:
> >
> > - partition_info_get
> > - sync_send_receive
> > - rxtx_unmap
> >
> > For more details please refer to the driver documentation [2].
> >
> > [1]: https://developer.arm.com/documentation/den0077/latest/
> > [2]: doc/arch/arm64.ffa.rst
> >
> > Signed-off-by: Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > Cc: Tom Rini <trini at konsulko.com>
> > Cc: Simon Glass <sjg at chromium.org>
> > Cc: Ilias Apalodimas <ilias.apalodimas at linaro.org>
> > Cc: Jens Wiklander <jens.wiklander at linaro.org>
> >
> > ---
> >
> > Changelog:
> > ===============
> >
> > v7:
> >
> > * add support for 32-bit direct messaging
> > * rename be_uuid_str_to_le_bin() to uuid_str_to_le_bin()
> > * improve the declaration of error handling mapping
> > * stating in doc/arch/arm64.ffa.rst that EFI runtime is not supported
> >
> > v6:
> >
> > * drop use of EFI runtime support (We decided with Linaro to add this later)
> > * drop discovery from initcalls (discovery will be on demand by FF-A users)
> > * set the alignment of the RX/TX buffers to the larger translation granule size
> > * move FF-A RX/TX buffers unmapping at ExitBootServices() to a separate commit
> > * update the documentation and move it to doc/arch/arm64.ffa.rst
> >
> > v4:
> >
> > * add doc/README.ffa.drv
> > * moving the FF-A driver work to drivers/firmware/arm-ffa
> > * use less #ifdefs in lib/efi_loader/efi_boottime.c and replace
> > #if defined by #if CONFIG_IS_ENABLED
> > * improving error handling by mapping the FF-A errors to standard errors
> > and logs
> > * replacing panics with an error log and returning an error code
> > * improving features discovery in FFA_FEATURES by introducing
> > rxtx_min_pages private data field
> > * add ffa_remove and ffa_unbind functions
> > * improve how the driver behaves when bus discovery is done more than
> > once
> >
> > v3:
> >
> > * align the interfaces of the U-Boot FF-A driver with those in the linux
> > FF-A driver
> > * remove the FF-A helper layer
> > * make the U-Boot FF-A driver independent from EFI
> > * provide an optional config that enables copying the driver data to EFI
> > runtime section at ExitBootServices service
> > * use 64-bit version of FFA_RXTX_MAP, FFA_MSG_SEND_DIRECT_{REQ, RESP}
> >
> > v2:
> >
> > * make FF-A bus discoverable using device_{bind, probe} APIs
> > * remove device tree support
> >
> > v1:
> >
> > * introduce FF-A bus driver with device tree support
> >
> > MAINTAINERS | 7 +
> > doc/arch/arm64.ffa.rst | 218 ++++
> > doc/arch/index.rst | 1 +
> > drivers/Kconfig | 2 +
> > drivers/Makefile | 1 +
> > drivers/firmware/arm-ffa/Kconfig | 30 +
> > drivers/firmware/arm-ffa/Makefile | 6 +
> > drivers/firmware/arm-ffa/arm-ffa-uclass.c | 16 +
> > drivers/firmware/arm-ffa/arm_ffa_prv.h | 205 ++++
> > drivers/firmware/arm-ffa/core.c | 1310 +++++++++++++++++++++
> > include/arm_ffa.h | 93 ++
> > include/dm/uclass-id.h | 4 +
> > 12 files changed, 1893 insertions(+)
> > create mode 100644 doc/arch/arm64.ffa.rst
> > create mode 100644 drivers/firmware/arm-ffa/Kconfig
> > create mode 100644 drivers/firmware/arm-ffa/Makefile
> > create mode 100644 drivers/firmware/arm-ffa/arm-ffa-uclass.c
> > create mode 100644 drivers/firmware/arm-ffa/arm_ffa_prv.h
> > create mode 100644 drivers/firmware/arm-ffa/core.c
> > create mode 100644 include/arm_ffa.h
> >
> > diff --git a/MAINTAINERS b/MAINTAINERS
> > index 1cf99c1393..450b5725ce 100644
> > --- a/MAINTAINERS
> > +++ b/MAINTAINERS
> > @@ -263,6 +263,13 @@ F: drivers/net/cortina_ni.h
> > F: drivers/net/phy/ca_phy.c
> > F: configs/cortina_presidio-asic-pnand_defconfig
> >
> > +ARM FF-A
> > +M: Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > +S: Maintained
> > +F: doc/arch/arm64.ffa.rst
> > +F: drivers/firmware/arm-ffa/
> > +F: include/arm_ffa.h
> > +
> > ARM FREESCALE IMX
> > M: Stefano Babic <sbabic at denx.de>
> > M: Fabio Estevam <festevam at gmail.com>
> > diff --git a/doc/arch/arm64.ffa.rst b/doc/arch/arm64.ffa.rst
> > new file mode 100644
> > index 0000000000..dfcec82e45
> > --- /dev/null
> > +++ b/doc/arch/arm64.ffa.rst
> > @@ -0,0 +1,218 @@
> > +.. SPDX-License-Identifier: GPL-2.0+
> > +
> > +Arm FF-A Driver
> > +===============
> > +
> > +Summary
> > +-------
> > +
> > +FF-A stands for Firmware Framework for Arm A-profile processors.
> > +
> > +FF-A specifies interfaces that enable a pair of software sandboxes to
> > +communicate with each other. A sandbox aka partition could
> > +be a VM in the Normal or Secure world, an application in S-EL0, or a
> > +Trusted OS in S-EL1.
> > +
> > +This FF-A driver implements the interfaces to communicate with partitions in
> > +the Secure world aka Secure partitions (SPs).
> > +
> > +The driver specifically focuses on communicating with SPs that isolate portions
> > +of EFI runtime services that must run in a protected environment which is
> > +inaccessible by the Host OS or Hypervisor. Examples of such services are
> > +set/get variables.
> > +
> > +FF-A driver uses the SMC ABIs defined by the FF-A specification to:
> > +
> > +- Discover the presence of SPs of interest
> > +- Access an SP's service through communication protocols
> > + e.g. EFI MM communication protocol
> > +
> > +At this stage of development the FF-A driver supports EFI boot time only.
> > +
> > +Runtime support will be added in future developments.
> > +
> > +FF-A and SMC specifications
> > +-------------------------------------------
> > +
> > +The current implementation of the driver relies on FF-A specification v1.0
> > +and uses SMC32 calling convention which means using the first 32-bit data of the
> > +Xn registers.
> > +
> > +At this stage we only need the FF-A v1.0 features.
> > +
> > +The driver has been tested with OP-TEE which supports SMC32 calling convention.
> > +
> > +For more details please refer to the FF-A v1.0 spec:
> > +https://documentation-service.arm.com/static/5fb7e8a6ca04df4095c1d65e?token=
> > +
> > +Hypervisors are supported if they are configured to trap SMC calls.
> > +
> > +The FF-A driver uses 64-bit registers as per SMCCCv1.2 specification.
> > +
> > +For more details please refer to the SMC Calling Convention v1.2 spec:
> > +https://documentation-service.arm.com/static/5f8edaeff86e16515cdbe4c6?token=
> > +
> > +Supported hardware
> > +--------------------------------
> > +
> > +Aarch64 plaforms
> > +
> > +Configuration
> > +----------------------
> > +
> > +CONFIG_ARM_FFA_TRANSPORT
> > + Enables the FF-A bus driver. Turn this on if you want to use FF-A
> > + communication.
> > +
> > +CONFIG_SANDBOX_FFA
> > + Enables FF-A Sandbox driver. This emulates the FF-A ABIs handling under
> > + Sandbox and provides functional tests for FF-A.
> > +
> > +FF-A ABIs under the hood
> > +---------------------------------------
> > +
> > +Invoking an FF-A ABI involves providing to the secure world/hypervisor the
> > +expected arguments from the ABI.
> > +
> > +The ABI arguments are stored in x0 to x7 registers. Then, an SMC instruction
> > +is executed.
> > +
> > +At the secure side level or hypervisor the ABI is handled at a higher exception
> > +level and the arguments are read and processed.
> > +
> > +The response is put back through x0 to x7 registers and control is given back
> > +to the U-Boot FF-A driver (non-secure world).
> > +
> > +The driver reads the response and processes it accordingly.
> > +
> > +This methodology applies to all the FF-A ABIs in the driver.
> > +
> > +FF-A bus discovery in U-Boot
> > +-------------------------------------------
> > +
> > +When CONFIG_ARM_FFA_TRANSPORT is enabled, the FF-A bus is discovered on
> > +demand by the clients (users).
> > +
> > +Clients can discover the FF-A bus using ffa_bus_discover() API which triggers the
> > +discovery process.
> > +
> > +ffa_bus_discover() creates, binds and probes the arm_ffa device using
> > +device_{bind, probe} APIs.
> > +
> > +The discovery process consists in communicating with secure world (or hypervisor)
> > +and querying specific data.
> > +
> > +The discovery process takes place during the arm_ffa device probing which is
> > +handled by ffa_probe().
> > +
> > +The FF-A bus discovery is successful and the bus is ready for use when these
> > +operations succeed:
> > +
> > +- querying the FF-A framework version
> > +- querying from secure world the U-Boot endpoint ID
> > +- querying from secure world the RX/TX mapping features
> > +- mapping the RX/TX buffers
> > +- querying from secure world all the partitions information
> > +
> > +Discovery failure results in a probing failure and the arm_ffa device is
> > +destroyed.
> > +
> > +Requirements for clients
> > +-------------------------------------
> > +
> > +When using the FF-A bus with EFI, clients must:
> > +
> > +- Query SPs in EFI boot time mode using the service UUID.
> > +- Unmap RX/TX buffers before EFI runtime mode starts.
> > +
> > +The RX/TX buffers are only available at EFI boot time. Querying partitions is
> > +done at boot time and data is cached for future use.
> > +
> > +RX/TX buffers should be unmapped by the user before EFI runtime mode
> > +starts. The driver provides a bus operation for that: rxtx_unmap()
> > +
> > +If RX/TX buffers created by U-Boot are not unmapped and by
> > +consequence becoming available at EFI runtime, secure world will get confused
> > +about RX/TX buffers ownership (U-Boot vs kernel).
> > +
> > +When invoking FF-A direct messaging, clients should specify which ABI protocol
> > +they want to use (32-bit vs 64-bit). Selecting the protocol means using
> > +the 32-bit or 64-bit version of FFA_MSG_SEND_DIRECT_{REQ, RESP}.
> > +The calling convention stays the same: SMC32.
> > +
> > +The bus driver layer
> > +------------------------------
> > +
> > +The driver comes on top of the SMCCC layer and is implemented in
> > +drivers/firmware/arm-ffa/core.c
> > +
> > +The driver provides the following features:
> > +
> > +- Support for the 32-bit version of the following ABIs:
> > +
> > +FFA_VERSION
> > +FFA_ID_GET
> > +FFA_FEATURES
> > +FFA_PARTITION_INFO_GET
> > +FFA_RXTX_UNMAP
> > +FFA_RX_RELEASE
> > +FFA_RUN
> > +FFA_ERROR
> > +FFA_SUCCESS
> > +FFA_INTERRUPT
> > +FFA_MSG_SEND_DIRECT_REQ
> > +FFA_MSG_SEND_DIRECT_RESP
> > +
> > +- Support for the 64-bit version of the following ABIs:
> > +
> > +FFA_RXTX_MAP
> > +FFA_MSG_SEND_DIRECT_REQ
> > +FFA_MSG_SEND_DIRECT_RESP
> > +
> > +- Processing the received data from the secure world/hypervisor and caching it
> > +
> > +- Hiding from upper layers the FF-A protocol and registers details. Upper
> > + layers focus on exchanged data, the driver takes care of how to transport
> > + that to the secure world/hypervisor using FF-A
> > +
> > +- The driver provides callbacks to be used by clients to access the FF-A bus:
> > +
> > +partition_info_get
> > +sync_send_receive
> > +rxtx_unmap
> > +
> > +- FF-A bus discovery makes sure FF-A framework is responsive and compatible
> > + with the driver
> > +
> > +- FF-A bus can be compiled and used without EFI
> > +
> > +Using armffa command
> > +-----------------------------------
> > +
> > +armffa is an implementation defined command showcasing how to use the FF-A driver and how to invoke
> > +its operations.
> > +
> > +This provides a guidance to the client developers on how to call the FF-A bus
> > +interfaces.
> > +
> > +Usage:
> > +
> > +armffa <sub-command> <arguments>
> > +
> > +sub-commands:
> > +
> > + getpart <partition UUID>
> > +
> > + lists the partition(s) info
> > +
> > + ping <partition ID>
> > +
> > + sends a data pattern to the specified partition
> > +
> > + devlist
> > +
> > + displays the arm_ffa device info
> > +
> > +Contributors
> > +------------
> > + * Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > diff --git a/doc/arch/index.rst b/doc/arch/index.rst
> > index 792d9182c3..8d1ab0ad4e 100644
> > --- a/doc/arch/index.rst
> > +++ b/doc/arch/index.rst
> > @@ -8,6 +8,7 @@ Architecture-specific doc
> >
> > arc
> > arm64
> > + arm64.ffa
> > m68k
> > mips
> > nios2
> > diff --git a/drivers/Kconfig b/drivers/Kconfig
> > index 8b6fead351..b06b1ae481 100644
> > --- a/drivers/Kconfig
> > +++ b/drivers/Kconfig
> > @@ -6,6 +6,8 @@ source "drivers/core/Kconfig"
> >
> > source "drivers/adc/Kconfig"
> >
> > +source "drivers/firmware/arm-ffa/Kconfig"
> > +
> > source "drivers/ata/Kconfig"
> >
> > source "drivers/axi/Kconfig"
> > diff --git a/drivers/Makefile b/drivers/Makefile
> > index a1700c819d..b86c1d2a42 100644
> > --- a/drivers/Makefile
> > +++ b/drivers/Makefile
> > @@ -111,6 +111,7 @@ obj-y += iommu/
> > obj-y += smem/
> > obj-y += thermal/
> > obj-$(CONFIG_TEE) += tee/
> > +obj-$(CONFIG_ARM_FFA_TRANSPORT) += firmware/arm-ffa/
> > obj-y += axi/
> > obj-y += ufs/
> > obj-$(CONFIG_W1) += w1/
> > diff --git a/drivers/firmware/arm-ffa/Kconfig b/drivers/firmware/arm-ffa/Kconfig
> > new file mode 100644
> > index 0000000000..e4914b9bc7
> > --- /dev/null
> > +++ b/drivers/firmware/arm-ffa/Kconfig
> > @@ -0,0 +1,30 @@
> > +# SPDX-License-Identifier: GPL-2.0
> > +
> > +config ARM_FFA_TRANSPORT
> > + bool "Enable Arm Firmware Framework for Armv8-A driver"
> > + depends on DM && ARM64
> > + select ARM_SMCCC
> > + select LIB_UUID
> > + select DEVRES
> > + help
> > + The Firmware Framework for Arm A-profile processors (FF-A)
> > + describes interfaces (ABIs) that standardize communication
> > + between the Secure World and Normal World leveraging TrustZone
> > + technology.
> > +
> > + This driver is based on FF-A specification v1.0 and uses SMC32
> > + calling convention.
> > +
> > + FF-A specification:
> > +
> > + https://developer.arm.com/documentation/den0077/a/?lang=en
> > +
> > + In U-Boot FF-A design, FF-A is considered as a discoverable bus.
> > + The Secure World is considered as one entity to communicate with
> > + using the FF-A bus.
> > + FF-A communication is handled by one device and one instance (the bus).
> > + This FF-A driver takes care of all the interactions between Normal world
> > + and Secure World.
> > +
> > + For more details about the FF-A driver, please refer to doc/arch/arm64.ffa.rst
> > +
> > diff --git a/drivers/firmware/arm-ffa/Makefile b/drivers/firmware/arm-ffa/Makefile
> > new file mode 100644
> > index 0000000000..043a8915be
> > --- /dev/null
> > +++ b/drivers/firmware/arm-ffa/Makefile
> > @@ -0,0 +1,6 @@
> > +# SPDX-License-Identifier: GPL-2.0+
> > +#
> > +# (C) Copyright 2022
> > +# Abdellatif El Khlifi, Arm Limited, abdellatif.elkhlifi at arm.com.
> > +
> > +obj-y += arm-ffa-uclass.o core.o
> > diff --git a/drivers/firmware/arm-ffa/arm-ffa-uclass.c b/drivers/firmware/arm-ffa/arm-ffa-uclass.c
> > new file mode 100644
> > index 0000000000..7d9695d289
> > --- /dev/null
> > +++ b/drivers/firmware/arm-ffa/arm-ffa-uclass.c
> > @@ -0,0 +1,16 @@
> > +// SPDX-License-Identifier: GPL-2.0+
> > +/*
> > + * (C) Copyright 2022 ARM Limited
> > + * Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > + */
> > +
> > +#include <common.h>
> > +#include <dm.h>
> > +#include <asm/global_data.h>
> > +
> > +DECLARE_GLOBAL_DATA_PTR;
> > +
> > +UCLASS_DRIVER(ffa) = {
> > + .name = "ffa",
> > + .id = UCLASS_FFA,
> > +};
> > diff --git a/drivers/firmware/arm-ffa/arm_ffa_prv.h b/drivers/firmware/arm-ffa/arm_ffa_prv.h
> > new file mode 100644
> > index 0000000000..ad7430ada9
> > --- /dev/null
> > +++ b/drivers/firmware/arm-ffa/arm_ffa_prv.h
> > @@ -0,0 +1,205 @@
> > +/* SPDX-License-Identifier: GPL-2.0+ */
> > +/*
> > + * (C) Copyright 2022 ARM Limited
> > + * Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > + */
> > +
> > +#ifndef __ARM_FFA_PRV_H
> > +#define __ARM_FFA_PRV_H
> > +
> > +#include <arm_ffa.h>
> > +#include <linux/bitfield.h>
> > +#include <linux/bitops.h>
> > +#include <linux/arm-smccc.h>
> > +
> > +/*
> > + * This header is private. It is exclusively used by the FF-A driver
> > + */
> > +
> > +/* FF-A core driver name */
> > +#define FFA_DRV_NAME "arm_ffa"
> > +
> > +/* FF-A driver version definitions */
> > +
> > +#define MAJOR_VERSION_MASK GENMASK(30, 16)
> > +#define MINOR_VERSION_MASK GENMASK(15, 0)
> > +#define GET_FFA_MAJOR_VERSION(x) \
> > + ((u16)(FIELD_GET(MAJOR_VERSION_MASK, (x))))
> > +#define GET_FFA_MINOR_VERSION(x) \
> > + ((u16)(FIELD_GET(MINOR_VERSION_MASK, (x))))
> > +#define PACK_VERSION_INFO(major, minor) \
> > + (FIELD_PREP(MAJOR_VERSION_MASK, (major)) | \
> > + FIELD_PREP(MINOR_VERSION_MASK, (minor)))
> > +
> > +#define FFA_MAJOR_VERSION (1)
> > +#define FFA_MINOR_VERSION (0)
> > +#define FFA_VERSION_1_0 \
> > + PACK_VERSION_INFO(FFA_MAJOR_VERSION, FFA_MINOR_VERSION)
> > +
> > +/* Endpoint ID mask (u-boot endpoint ID) */
> > +
> > +#define GET_SELF_ENDPOINT_ID_MASK GENMASK(15, 0)
> > +#define GET_SELF_ENDPOINT_ID(x) \
> > + ((u16)(FIELD_GET(GET_SELF_ENDPOINT_ID_MASK, (x))))
> > +
> > +#define PREP_SELF_ENDPOINT_ID_MASK GENMASK(31, 16)
> > +#define PREP_SELF_ENDPOINT_ID(x) \
> > + (FIELD_PREP(PREP_SELF_ENDPOINT_ID_MASK, (x)))
> > +
> > +/* Partition endpoint ID mask (partition with which u-boot communicates with) */
> > +
> > +#define PREP_PART_ENDPOINT_ID_MASK GENMASK(15, 0)
> > +#define PREP_PART_ENDPOINT_ID(x) \
> > + (FIELD_PREP(PREP_PART_ENDPOINT_ID_MASK, (x)))
> > +
> > +/*
> > + * Definitions of the Arm FF-A interfaces supported by the Arm FF-A driver
> > + */
> > +
> > +#define FFA_SMC(calling_convention, func_num) \
> > + ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, (calling_convention), \
> > + ARM_SMCCC_OWNER_STANDARD, (func_num))
> > +
> > +#define FFA_SMC_32(func_num) FFA_SMC(ARM_SMCCC_SMC_32, (func_num))
> > +#define FFA_SMC_64(func_num) FFA_SMC(ARM_SMCCC_SMC_64, (func_num))
> > +
> > +enum ffa_abis {
> > + FFA_ERROR = 0x60,
> > + FFA_SUCCESS = 0x61,
> > + FFA_INTERRUPT = 0x62,
> > + FFA_VERSION = 0x63,
> > + FFA_FEATURES = 0x64,
> > + FFA_RX_RELEASE = 0x65,
> > + FFA_RXTX_MAP = 0x66,
> > + FFA_RXTX_UNMAP = 0x67,
> > + FFA_PARTITION_INFO_GET = 0x68,
> > + FFA_ID_GET = 0x69,
> > + FFA_RUN = 0x6D,
> > + FFA_MSG_SEND_DIRECT_REQ = 0x6F,
> > + FFA_MSG_SEND_DIRECT_RESP = 0x70,
>
> Strange indentation above.
>
> > +
> > + /* to be updated when adding new FFA IDs */
> > + FFA_FIRST_ID = FFA_ERROR, /* lowest number ID*/
> > + FFA_LAST_ID = FFA_MSG_SEND_DIRECT_RESP, /* highest number ID*/
> > +};
> > +
> > +enum ffa_abi_errcode {
> > + NOT_SUPPORTED = 1,
> > + INVALID_PARAMETERS,
> > + NO_MEMORY,
> > + BUSY,
> > + INTERRUPTED,
> > + DENIED,
> > + RETRY,
> > + ABORTED,
> > + MAX_NUMBER_FFA_ERR
> > +};
> > +
> > +/* container structure and helper macros to map between an FF-A error and relevant error log */
> > +struct ffa_abi_errmap {
> > + char *err_str[MAX_NUMBER_FFA_ERR];
> > +};
> > +
> > +#define FFA_ERRMAP_COUNT (FFA_LAST_ID - FFA_FIRST_ID + 1)
> > +#define FFA_ID_TO_ERRMAP_ID(ffa_id) ((ffa_id) - FFA_FIRST_ID)
> > +
> > +/* The FF-A SMC function definitions */
> > +
> > +typedef struct arm_smccc_1_2_regs ffa_value_t;
> > +typedef void (*invoke_ffa_fn_t)(ffa_value_t args, ffa_value_t *res);
> > +
> > +/*
> > + * struct ffa_partition_uuid - 16 bytes UUID transmitted by FFA_PARTITION_INFO_GET
> > + * @a1-4: 32-bit words access to the UUID data
> > + *
> > + */
> > +struct ffa_partition_uuid {
> > + u32 a1; /* w1 */
> > + u32 a2; /* w2 */
> > + u32 a3; /* w3 */
> > + u32 a4; /* w4 */
> > +};
> > +
> > +/**
> > + * enum ffa_rxtx_buf_sizes - minimum sizes supported
> > + * for the RX/TX buffers
> > + */
> > +enum ffa_rxtx_buf_sizes {
> > + RXTX_4K,
> > + RXTX_64K,
> > + RXTX_16K
> > +};
> > +
> > +/**
> > + * struct ffa_rxtxpair - structure hosting the RX/TX buffers virtual addresses
> > + * @rxbuf: virtual address of the RX buffer
> > + * @txbuf: virtual address of the TX buffer
> > + * @rxtx_min_pages: RX/TX buffers minimum size in pages
> > + *
> > + * Data structure hosting the virtual addresses of the mapped RX/TX buffers
> > + * These addresses are used by the FF-A functions that use the RX/TX buffers
> > + */
> > +struct ffa_rxtxpair {
> > + u64 rxbuf; /* virtual address */
> > + u64 txbuf; /* virtual address */
> > + size_t rxtx_min_pages; /* minimum number of pages in each of the RX/TX buffers */
> > +};
> > +
> > +/**
> > + * struct ffa_partition_desc - the secure partition descriptor
> > + * @info: partition information
> > + * @sp_uuid: the secure partition UUID
> > + *
> > + * Each partition has its descriptor containing the partitions information and the UUID
> > + */
> > +struct ffa_partition_desc {
> > + struct ffa_partition_info info;
> > + struct ffa_partition_uuid sp_uuid;
> > +};
> > +
> > +/**
> > + * struct ffa_partitions - descriptors for all secure partitions
> > + * @count: The number of partitions descriptors
> > + * @descs The partitions descriptors table
> > + *
> > + * This data structure contains the partitions descriptors table
> > + */
> > +struct ffa_partitions {
> > + u32 count;
> > + struct ffa_partition_desc *descs; /* virtual address */
> > +};
> > +
> > +/**
> > + * struct ffa_prvdata - the driver private data structure
> > + *
> > + * @dev: The arm_ffa device under u-boot driver model
> > + * @ffa_ops: The driver operations structure
> > + * @fwk_version: FF-A framework version
> > + * @id: u-boot endpoint ID
> > + * @partitions: The partitions descriptors structure
> > + * @pair: The RX/TX buffers pair
> > + * @invoke_ffa_fn: The function executing the FF-A function
> > + *
> > + * The driver data structure hosting all resident data.
> > + */
> > +struct ffa_prvdata {
> > + struct udevice *dev;
> > + struct ffa_bus_ops ffa_ops;
> > + u32 fwk_version;
> > + u16 id;
> > + struct ffa_partitions partitions;
> > + struct ffa_rxtxpair pair;
> > + invoke_ffa_fn_t invoke_ffa_fn;
> > +};
> > +
> > +/**
> > + * ffa_device_get - create, bind and probe the arm_ffa device
> > + */
> > +int ffa_device_get(void);
> > +
> > +/**
> > + * ffa_bus_prvdata_get - bus driver private data getter
> > + */
> > +struct ffa_prvdata **ffa_bus_prvdata_get(void);
> > +
> > +#endif
> > diff --git a/drivers/firmware/arm-ffa/core.c b/drivers/firmware/arm-ffa/core.c
> > new file mode 100644
> > index 0000000000..b955e5187b
> > --- /dev/null
> > +++ b/drivers/firmware/arm-ffa/core.c
> > @@ -0,0 +1,1310 @@
> > +// SPDX-License-Identifier: GPL-2.0+
> > +/*
> > + * (C) Copyright 2022 ARM Limited
> > + * Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > + */
> > +
> > +#include "arm_ffa_prv.h"
> > +#include <asm/global_data.h>
> > +#include <common.h>
> > +#include <dm.h>
> > +#include <dm/device-internal.h>
> > +#include <dm/devres.h>
> > +#include <dm/root.h>
> > +#include <linux/errno.h>
> > +#include <linux/sizes.h>
> > +#include <log.h>
> > +#include <malloc.h>
> > +#include <string.h>
> > +#include <uuid.h>
> > +
> > +DECLARE_GLOBAL_DATA_PTR;
> > +
> > +/**
> > + * The device private data structure containing all the
> > + * data read from secure world
> > + */
> > +struct ffa_prvdata *ffa_priv_data;
> > +
> > +/* Error mapping declarations */
> > +
> > +int ffa_to_std_errmap[MAX_NUMBER_FFA_ERR] = {
> > + [NOT_SUPPORTED] = -EOPNOTSUPP,
> > + [INVALID_PARAMETERS] = -EINVAL,
> > + [NO_MEMORY] = -ENOMEM,
> > + [BUSY] = -EBUSY,
> > + [INTERRUPTED] = -EINTR,
> > + [DENIED] = -EACCES,
> > + [RETRY] = -EAGAIN,
> > + [ABORTED] = -ECANCELED,
> > +};
> > +
> > +struct ffa_abi_errmap err_msg_map[FFA_ERRMAP_COUNT] = {
> > + [FFA_ID_TO_ERRMAP_ID(FFA_VERSION)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: A Firmware Framework implementation does not exist",
> > + },
> > + },
> > + [FFA_ID_TO_ERRMAP_ID(FFA_ID_GET)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: This function is not implemented at this FF-A instance",
> > + },
> > + },
> > + [FFA_ID_TO_ERRMAP_ID(FFA_FEATURES)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: FFA_RXTX_MAP is not implemented at this FF-A instance",
> > + },
> > + },
> > + [FFA_ID_TO_ERRMAP_ID(FFA_PARTITION_INFO_GET)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: This function is not implemented at this FF-A instance",
> > + [INVALID_PARAMETERS] =
> > + "INVALID_PARAMETERS: Unrecognized UUID",
> > + [NO_MEMORY] =
> > + "NO_MEMORY: Results cannot fit in RX buffer of the caller",
> > + [BUSY] =
> > + "BUSY: RX buffer of the caller is not free",
> > + [DENIED] =
> > + "DENIED: Callee is not in a state to handle this request",
> > + },
> > + },
> > + [FFA_ID_TO_ERRMAP_ID(FFA_RXTX_UNMAP)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: FFA_RXTX_UNMAP is not implemented at this FF-A instance",
> > + [INVALID_PARAMETERS] =
> > + "INVALID_PARAMETERS: No buffer pair registered on behalf of the caller",
> > + },
> > + },
> > + [FFA_ID_TO_ERRMAP_ID(FFA_RX_RELEASE)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: FFA_RX_RELEASE is not implemented at this FF-A instance",
> > + [DENIED] =
> > + "DENIED: Caller did not have ownership of the RX buffer",
> > + },
> > + },
> > + [FFA_ID_TO_ERRMAP_ID(FFA_RXTX_MAP)] = {
> > + {
> > + [NOT_SUPPORTED] =
> > + "NOT_SUPPORTED: This function is not implemented at this FF-A instance",
> > + [INVALID_PARAMETERS] =
> > + "INVALID_PARAMETERS: Field(s) in input parameters incorrectly encoded",
> > + [NO_MEMORY] =
> > + "NO_MEMORY: Not enough memory",
> > + [DENIED] =
> > + "DENIED: Buffer pair already registered",
> > + },
> > + },
> > +};
>
> Indentation above looks strange and inconsistent in some places. This
> entire approach is in my opinion a bit complicated, no big deal, if no
> one else complains I suppose it's OK.
>
> > +
> > +/**
> > + * ffa_to_std_errno - convert FF-A error code to standard error code
> > + * @ffa_errno: Error code returned by the FF-A ABI
> > + *
> > + * This function maps the given FF-A error code as specified
> > + * by the spec to a u-boot standard error code.
> > + *
> > + * Return:
> > + *
> > + * The standard error code on success. . Otherwise, failure
> > + */
> > +int ffa_to_std_errno(int ffa_errno)
> > +{
> > + int err_idx = -ffa_errno;
> > +
> > + /* map the FF-A error code to the standard u-boot error code */
> > + if (err_idx > 0 && err_idx < MAX_NUMBER_FFA_ERR)
> > + return ffa_to_std_errmap[err_idx];
> > + return -EINVAL;
> > +}
> > +
> > +/**
> > + * ffa_print_error_log - print the error log corresponding to the selected FF-A ABI
> > + * @ffa_id: FF-A ABI ID
> > + * @ffa_errno: Error code returned by the FF-A ABI
> > + *
> > + * This function maps the FF-A error code to the error log relevant to the
> > + * selected FF-A ABI. Then the error log is printed.
> > + *
> > + * Return:
> > + *
> > + * 0 on success. . Otherwise, failure
> > + */
> > +int ffa_print_error_log(u32 ffa_id, int ffa_errno)
> > +{
> > + int err_idx = -ffa_errno, abi_idx = 0;
> > +
> > + /* map the FF-A error code to the corresponding error log */
> > +
> > + if (err_idx <= 0 || err_idx >= MAX_NUMBER_FFA_ERR)
> > + return -EINVAL;
> > +
> > + if (ffa_id < FFA_FIRST_ID || ffa_id > FFA_LAST_ID)
> > + return -EINVAL;
> > +
> > + abi_idx = FFA_ID_TO_ERRMAP_ID(ffa_id);
> > + if (abi_idx < 0 || abi_idx >= FFA_ERRMAP_COUNT)
> > + return -EINVAL;
> > +
> > + if (!err_msg_map[abi_idx].err_str[err_idx])
> > + return -EINVAL;
> > +
> > + ffa_err("%s", err_msg_map[abi_idx].err_str[err_idx]);
> > +
> > + return 0;
> > +}
> > +
> > +/*
> > + * Driver core functions
> > + */
> > +
> > +/**
> > + * ffa_remove_device - removes the arm_ffa device
> > + * @dev: the device to be removed
> > + *
> > + * This function makes sure the arm_ffa device is removed
> > + * No need to free the kmalloced data when the device is destroyed.
> > + * It's automatically done by devm management by
> > + * device_remove() -> device_free() -> devres_release_probe().
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +int ffa_remove_device(struct udevice *dev)
> > +{
> > + int ret;
> > +
> > + if (!dev) {
> > + ffa_err("no udevice found");
> > + return -ENODEV;
> > + }
> > +
> > + ret = device_remove(dev, DM_REMOVE_NORMAL);
> > + if (ret) {
> > + ffa_err("unable to remove. err:%d\n", ret);
> > + return ret;
> > + }
> > +
> > + ffa_info("device removed and freed");
> > +
> > + ret = device_unbind(dev);
> > + if (ret) {
> > + ffa_err("unable to unbind. err:%d\n", ret);
> > + return ret;
> > + }
> > +
> > + ffa_info("device unbound");
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_device_get - create, bind and probe the arm_ffa device
> > + *
> > + * This function makes sure the arm_ffa device is
> > + * created, bound to this driver, probed and ready to use.
> > + * Arm FF-A transport is implemented through a single u-boot
> > + * device managing the FF-A bus (arm_ffa).
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +int ffa_device_get(void)
> > +{
> > + int ret;
> > + struct udevice *dev = NULL;
> > +
> > + ret = device_bind(dm_root(), DM_DRIVER_GET(arm_ffa), FFA_DRV_NAME, NULL, ofnode_null(),
> > + &dev);
> > + if (ret)
> > + return ret;
> > +
> > + /* The FF-A bus discovery succeeds when probing is successful */
> > + ret = device_probe(dev);
> > + if (ret) {
> > + ffa_err("arm_ffa device probing failed");
> > + ffa_remove_device(dev);
> > + return ret;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_get_version - FFA_VERSION handler function
> > + *
> > + * This function implements FFA_VERSION FF-A function
> > + * to get from the secure world the FF-A framework version
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_get_version(void)
> > +{
> > + u16 major, minor;
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_VERSION), .a1 = FFA_VERSION_1_0,
> > + }, &res);
> > +
> > + ffa_errno = res.a0;
> > + if (ffa_errno < 0) {
> > + ffa_print_error_log(FFA_VERSION, ffa_errno);
> > + return ffa_to_std_errno(ffa_errno);
> > + }
> > +
> > + major = GET_FFA_MAJOR_VERSION(res.a0);
> > + minor = GET_FFA_MINOR_VERSION(res.a0);
> > +
> > + ffa_info("FF-A driver %d.%d\nFF-A framework %d.%d",
> > + FFA_MAJOR_VERSION, FFA_MINOR_VERSION, major, minor);
> > +
> > + if ((major == FFA_MAJOR_VERSION && minor >= FFA_MINOR_VERSION)) {
> > + ffa_info("Versions are compatible ");
> > +
> > + ffa_priv_data->fwk_version = res.a0;
> > +
> > + return 0;
> > + }
> > +
> > + ffa_err("versions are incompatible\nExpected: %d.%d , Found: %d.%d\n",
> > + FFA_MAJOR_VERSION, FFA_MINOR_VERSION, major, minor);
> > +
> > + return -EPROTONOSUPPORT;
> > +}
> > +
> > +/**
> > + * ffa_get_endpoint_id - FFA_ID_GET handler function
> > + *
> > + * This function implements FFA_ID_GET FF-A function
> > + * to get from the secure world u-boot endpoint ID
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_get_endpoint_id(void)
> > +{
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_ID_GET),
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS)) {
> > + ffa_priv_data->id = GET_SELF_ENDPOINT_ID((u32)res.a2);
> > + ffa_info("endpoint ID is %u", ffa_priv_data->id);
> > +
> > + return 0;
> > + }
> > +
> > + ffa_errno = res.a2;
> > +
> > + ffa_print_error_log(FFA_ID_GET, ffa_errno);
> > +
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * ffa_set_rxtx_buffers_pages_cnt - sets the minimum number of pages in each of the RX/TX buffers
> > + * @prop_field: properties field obtained from FFA_FEATURES ABI
> > + *
> > + * This function sets the minimum number of pages
> > + * in each of the RX/TX buffers in the private data structure
> > + *
> > + * Return:
> > + *
> > + * buf_4k_pages points to the returned number of pages
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_set_rxtx_buffers_pages_cnt(u32 prop_field)
> > +{
> > + if (!ffa_priv_data)
> > + return -EINVAL;
> > +
> > + switch (prop_field) {
> > + case RXTX_4K:
> > + ffa_priv_data->pair.rxtx_min_pages = 1;
> > + break;
> > + case RXTX_16K:
> > + ffa_priv_data->pair.rxtx_min_pages = 4;
> > + break;
> > + case RXTX_64K:
> > + ffa_priv_data->pair.rxtx_min_pages = 16;
> > + break;
> > + default:
> > + ffa_err("RX/TX buffer size not supported");
> > + return -EINVAL;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_get_rxtx_map_features - FFA_FEATURES handler function with FFA_RXTX_MAP argument
> > + *
> > + * This function implements FFA_FEATURES FF-A function
> > + * to retrieve the FFA_RXTX_MAP features
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_get_rxtx_map_features(void)
> > +{
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_FEATURES),
> > + .a1 = FFA_SMC_64(FFA_RXTX_MAP),
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS))
> > + return ffa_set_rxtx_buffers_pages_cnt(res.a2);
> > +
> > + ffa_errno = res.a2;
> > + ffa_print_error_log(FFA_FEATURES, ffa_errno);
> > +
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * ffa_free_rxtx_buffers - frees the RX/TX buffers
> > + *
> > + * This function frees the RX/TX buffers
> > + *
> > + */
> > +static void ffa_free_rxtx_buffers(void)
> > +{
> > + ffa_info("Freeing RX/TX buffers");
> > +
> > + if (ffa_priv_data->pair.rxbuf) {
> > + free((void *)ffa_priv_data->pair.rxbuf);
> > + ffa_priv_data->pair.rxbuf = 0;
> > + }
> > +
> > + if (ffa_priv_data->pair.txbuf) {
> > + free((void *)ffa_priv_data->pair.txbuf);
> > + ffa_priv_data->pair.txbuf = 0;
> > + }
> > +}
> > +
> > +/**
> > + * ffa_alloc_rxtx_buffers - allocates the RX/TX buffers
> > + *
> > + * This function is used by ffa_map_rxtx_buffers to allocate
> > + * the RX/TX buffers before mapping them. The allocated memory is physically
> > + * contiguous since memalign ends up calling malloc which allocates
> > + * contiguous memory in u-boot.
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_alloc_rxtx_buffers(void)
> > +{
> > + u64 bytes;
> > +
> > + ffa_info("Using %lu 4KB page(s) for RX/TX buffers size",
> > + ffa_priv_data->pair.rxtx_min_pages);
> > +
> > + bytes = ffa_priv_data->pair.rxtx_min_pages * SZ_4K;
> > +
> > + /*
> > + * The alignment of the RX and TX buffers must be equal
> > + * to the larger translation granule size
> > + */
> > +
> > + ffa_priv_data->pair.rxbuf = (u64)memalign(bytes, bytes);
> > + if (!ffa_priv_data->pair.rxbuf) {
> > + ffa_err("failure to allocate RX buffer");
> > + return -ENOBUFS;
> > + }
> > +
> > + ffa_info("RX buffer at virtual address 0x%llx", ffa_priv_data->pair.rxbuf);
> > +
> > + ffa_priv_data->pair.txbuf = (u64)memalign(bytes, bytes);
> > + if (!ffa_priv_data->pair.txbuf) {
> > + free((void *)ffa_priv_data->pair.rxbuf);
> > + ffa_priv_data->pair.rxbuf = 0;
> > + ffa_err("failure to allocate the TX buffer");
> > + return -ENOBUFS;
> > + }
> > +
> > + ffa_info("TX buffer at virtual address 0x%llx", ffa_priv_data->pair.txbuf);
> > +
> > + /*
> > + * make sure the buffers are cleared before use
> > + */
> > + memset((void *)ffa_priv_data->pair.rxbuf, 0, bytes);
> > + memset((void *)ffa_priv_data->pair.txbuf, 0, bytes);
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_map_rxtx_buffers - FFA_RXTX_MAP handler function
> > + *
> > + * This function implements FFA_RXTX_MAP FF-A function
> > + * to map the RX/TX buffers
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_map_rxtx_buffers(void)
> > +{
> > + int ret;
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + ret = ffa_alloc_rxtx_buffers();
> > + if (ret)
> > + return ret;
> > +
> > + /*
> > + * we need to pass the physical addresses of the RX/TX buffers
> > + * in u-boot physical/virtual mapping is 1:1
> > + * no need to convert from virtual to physical
> > + */
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_64(FFA_RXTX_MAP),
> > + .a1 = ffa_priv_data->pair.txbuf,
> > + .a2 = ffa_priv_data->pair.rxbuf,
> > + .a3 = ffa_priv_data->pair.rxtx_min_pages,
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS)) {
> > + ffa_info("RX/TX buffers mapped");
> > + return 0;
> > + }
> > +
> > + ffa_errno = res.a2;
> > + ffa_print_error_log(FFA_RXTX_MAP, ffa_errno);
> > +
> > + ffa_free_rxtx_buffers();
> > +
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * ffa_unmap_rxtx_buffers - FFA_RXTX_UNMAP handler function
> > + *
> > + * This function implements FFA_RXTX_UNMAP FF-A function
> > + * to unmap the RX/TX buffers
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_unmap_rxtx_buffers(void)
> > +{
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_RXTX_UNMAP),
> > + .a1 = PREP_SELF_ENDPOINT_ID(ffa_priv_data->id),
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS)) {
> > + ffa_free_rxtx_buffers();
> > + return 0;
> > + }
> > +
> > + ffa_errno = res.a2;
> > + ffa_print_error_log(FFA_RXTX_UNMAP, ffa_errno);
> > +
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * ffa_release_rx_buffer - FFA_RX_RELEASE handler function
> > + *
> > + * This function invokes FFA_RX_RELEASE FF-A function
> > + * to release the ownership of the RX buffer
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_release_rx_buffer(void)
> > +{
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_RX_RELEASE),
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS))
> > + return 0;
> > +
> > + ffa_errno = res.a2;
> > + ffa_print_error_log(FFA_RX_RELEASE, ffa_errno);
> > +
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * ffa_uuid_are_identical - checks whether two given UUIDs are identical
> > + * @uuid1: first UUID
> > + * @uuid2: second UUID
> > + *
> > + * This function is used by ffa_read_partitions_info to search
> > + * for a UUID in the partitions descriptors table
> > + *
> > + * Return:
> > + *
> > + * 1 when UUIDs match. Otherwise, 0
> > + */
> > +bool ffa_uuid_are_identical(const struct ffa_partition_uuid *uuid1,
> > + const struct ffa_partition_uuid *uuid2)
> > +{
> > + if (!uuid1 || !uuid2)
> > + return 0;
> > +
> > + return !memcmp(uuid1, uuid2, sizeof(struct ffa_partition_uuid));
> > +}
> > +
> > +/**
> > + * ffa_read_partitions_info - reads the data queried by FFA_PARTITION_INFO_GET
> > + * and saves it in the private structure
> > + * @count: The number of partitions queried
> > + * @part_uuid: Pointer to the partition(s) UUID
> > + *
> > + * This function reads the partitions information
> > + * returned by the FFA_PARTITION_INFO_GET and saves it in the private
> > + * data structure.
> > + *
> > + * Return:
> > + *
> > + * The private data structure is updated with the partition(s) information
> > + * 0 is returned on success. Otherwise, failure
> > + */
> > +static int ffa_read_partitions_info(u32 count, struct ffa_partition_uuid *part_uuid)
> > +{
> > + if (!count) {
> > + ffa_err("no partition detected");
> > + return -ENODATA;
> > + }
> > +
> > + ffa_info("Reading partitions data from the RX buffer");
> > +
> > + if (!part_uuid) {
> > + /*
> > + * querying information of all partitions
> > + */
> > + u64 buf_bytes;
> > + u64 data_bytes;
> > + u32 desc_idx;
> > + struct ffa_partition_info *parts_info;
> > +
> > + data_bytes = count * sizeof(struct ffa_partition_desc);
> > +
> > + buf_bytes = ffa_priv_data->pair.rxtx_min_pages * SZ_4K;
> > +
> > + if (data_bytes > buf_bytes) {
> > + ffa_err("partitions data size exceeds the RX buffer size:");
> > + ffa_err(" sizes in bytes: data %llu , RX buffer %llu ",
> > + data_bytes,
> > + buf_bytes);
> > +
> > + return -ENOMEM;
> > + }
> > +
> > + ffa_priv_data->partitions.descs = devm_kmalloc(ffa_priv_data->dev, data_bytes,
> > + __GFP_ZERO);
> > + if (!ffa_priv_data->partitions.descs) {
> > + ffa_err("cannot allocate partitions data buffer");
> > + return -ENOMEM;
> > + }
> > +
> > + parts_info = (struct ffa_partition_info *)ffa_priv_data->pair.rxbuf;
> > +
> > + for (desc_idx = 0 ; desc_idx < count ; desc_idx++) {
> > + ffa_priv_data->partitions.descs[desc_idx].info =
> > + parts_info[desc_idx];
> > +
> > + ffa_info("Partition ID %x : info cached",
> > + ffa_priv_data->partitions.descs[desc_idx].info.id);
> > + }
> > +
> > + ffa_priv_data->partitions.count = count;
> > +
> > + ffa_info("%d partition(s) found and cached", count);
> > +
> > + } else {
> > + u32 rx_desc_idx, cached_desc_idx;
> > + struct ffa_partition_info *parts_info;
> > + u8 desc_found;
> > +
> > + parts_info = (struct ffa_partition_info *)ffa_priv_data->pair.rxbuf;
> > +
> > + /*
> > + * search for the SP IDs read from the RX buffer
> > + * in the already cached SPs.
> > + * Update the UUID when ID found.
> > + */
> > + for (rx_desc_idx = 0; rx_desc_idx < count ; rx_desc_idx++) {
> > + desc_found = 0;
> > +
> > + /*
> > + * search the current ID in the cached partitions
> > + */
> > + for (cached_desc_idx = 0;
> > + cached_desc_idx < ffa_priv_data->partitions.count;
> > + cached_desc_idx++) {
> > + /*
> > + * save the UUID
> > + */
> > + if (ffa_priv_data->partitions.descs[cached_desc_idx].info.id ==
> > + parts_info[rx_desc_idx].id) {
> > + ffa_priv_data->partitions.descs[cached_desc_idx].sp_uuid =
> > + *part_uuid;
> > +
> > + desc_found = 1;
> > + break;
> > + }
> > + }
> > +
> > + if (!desc_found)
> > + return -ENODATA;
> > + }
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_query_partitions_info - invokes FFA_PARTITION_INFO_GET and saves partitions data
> > + *
> > + * @part_uuid: Pointer to the partition(s) UUID
> > + * @pcount: Pointer to the number of partitions variable filled when querying
> > + *
> > + * This function executes the FFA_PARTITION_INFO_GET
> > + * to query the partitions data. Then, it calls ffa_read_partitions_info
> > + * to save the data in the private data structure.
> > + *
> > + * After reading the data the RX buffer is released using ffa_release_rx_buffer
> > + *
> > + * Return:
> > + *
> > + * When part_uuid is NULL, all partitions data are retrieved from secure world
> > + * When part_uuid is non NULL, data for partitions matching the given UUID are
> > + * retrieved and the number of partitions is returned
> > + * 0 is returned on success. Otherwise, failure
> > + */
> > +static int ffa_query_partitions_info(struct ffa_partition_uuid *part_uuid,
> > + u32 *pcount)
> > +{
> > + struct ffa_partition_uuid query_uuid = {0};
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > +
> > + /*
> > + * If a UUID is specified. Information for one or more
> > + * partitions in the system is queried. Otherwise, information
> > + * for all installed partitions is queried
> > + */
> > +
> > + if (part_uuid) {
> > + if (!pcount)
> > + return -EINVAL;
> > +
> > + query_uuid = *part_uuid;
> > + } else if (pcount) {
> > + return -EINVAL;
> > + }
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_PARTITION_INFO_GET),
> > + .a1 = query_uuid.a1,
> > + .a2 = query_uuid.a2,
> > + .a3 = query_uuid.a3,
> > + .a4 = query_uuid.a4,
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS)) {
> > + int ret;
> > +
> > + /*
> > + * res.a2 contains the count of partition information descriptors
> > + * populated in the RX buffer
> > + */
> > + if (res.a2) {
> > + ret = ffa_read_partitions_info((u32)res.a2, part_uuid);
> > + if (ret) {
> > + ffa_err("failed to read partition(s) data , error (%d)", ret);
> > + ffa_release_rx_buffer();
> > + return -EINVAL;
> > + }
> > + }
> > +
> > + /*
> > + * return the SP count (when querying using a UUID)
> > + */
> > + if (pcount)
> > + *pcount = (u32)res.a2;
> > +
> > + /*
> > + * After calling FFA_PARTITION_INFO_GET the buffer ownership
> > + * is assigned to the consumer (u-boot). So, we need to give
> > + * the ownership back to the SPM or hypervisor
> > + */
> > + ret = ffa_release_rx_buffer();
> > +
> > + return ret;
> > + }
> > +
> > + ffa_errno = res.a2;
> > + ffa_print_error_log(FFA_PARTITION_INFO_GET, ffa_errno);
> > +
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * ffa_get_partitions_info - FFA_PARTITION_INFO_GET handler function
> > + *
> > + * The passed arguments:
> > + * Mode 1: When getting from the driver the number of
> > + * secure partitions:
> > + * @uuid_str: pointer to the UUID string
> > + * @parts_size: pointer to the variable that contains the number of partitions
> > + * The variable will be set by the driver
> > + * @buffer: NULL
> > + *
> > + * Mode 2: When requesting the driver to return the
> > + * partitions information:
> > + * @uuid_str: pointer to the UUID string
> > + * @parts_size: pointer to the size of the SPs information buffer in bytes
> > + * @buffer: pointer to SPs information buffer
> > + * (allocated by the client).
> > + * The buffer will be filled by the driver
> > + *
> > + * This function queries the secure partition data from
> > + * the private data structure. If not found, it invokes FFA_PARTITION_INFO_GET
> > + * FF-A function to query the partition information from secure world.
> > + *
> > + * A client of the FF-A driver should know the UUID of the service it wants to
> > + * access. It should use the UUID to request the FF-A driver to provide the
> > + * partition(s) information of the service. The FF-A driver uses
> > + * PARTITION_INFO_GET to obtain this information. This is implemented through
> > + * ffa_get_partitions_info function.
> > + * A new FFA_PARTITION_INFO_GET call is issued (first one performed through
> > + * ffa_cache_partitions_info) allowing to retrieve the partition(s) information.
> > + * They are not saved (already done). We only update the UUID in the cached area.
> > + * This assumes that partitions data does not change in the secure world.
> > + * Otherwise u-boot will have an outdated partition data. The benefit of caching
> > + * the information in the FF-A driver is to accommodate discovery after
> > + * ExitBootServices().
> > + *
> > + * When invoked through a client request, ffa_get_partitions_info should be
> > + * called twice. First call is to get from the driver the number of secure
> > + * partitions (SPs) associated to a particular UUID.
> > + * Then, the caller (client) allocates the buffer to host the SPs data and
> > + * issues a 2nd call. Then, the driver fills the SPs data in the pre-allocated
> > + * buffer.
> > + *
> > + * To achieve the mechanism described above, ffa_get_partitions_info uses the
> > + * following functions:
> > + * ffa_read_partitions_info
> > + * ffa_query_partitions_info
> > + *
> > + * Return:
> > + *
> > + * @parts_size: When pointing to the number of partitions variable, the number is
> > + * set by the driver.
> > + * When pointing to the partitions information buffer size, the buffer will be
> > + * filled by the driver.
>
> This is confusing, it's better if you use the same unit all the time.
> Since you below require the size to be an even multiple of sizeof(struct
> ffa_partition_info) you should perhaps use that unit.
>
> > + *
> > + * On success 0 is returned. Otherwise, failure
> > + */
> > +static int ffa_get_partitions_info(const char *uuid_str,
> > + u32 *parts_size, struct ffa_partition_info *buffer)
> > +{
> > + /*
> > + * fill_data:
> > + * 0: return the SP count
> > + * 1: fill SP data and return it to the caller
> > + * -1: undefined mode
> > + */
> > + int fill_data = -1;
>
> In the code below this is treated as a bool, it's never tested for -1,
> so you could as well make this a bool.
>
> > + u32 desc_idx, client_desc_idx;
> > + struct ffa_partition_uuid part_uuid = {0};
> > + u32 client_desc_max_cnt;
> > + u32 parts_found = 0;
> > +
> > + if (!ffa_priv_data->partitions.count || !ffa_priv_data->partitions.descs) {
> > + ffa_err("no partition installed");
> > + return -EINVAL;
> > + }
> > +
> > + if (!uuid_str) {
> > + ffa_err("no UUID provided");
> > + return -EINVAL;
> > + }
> > +
> > + if (!parts_size) {
> > + ffa_err("no size/count provided");
> > + return -EINVAL;
> > + }
> > +
> > + if (uuid_str_to_le_bin(uuid_str, (unsigned char *)&part_uuid)) {
> > + ffa_err("invalid UUID");
> > + return -EINVAL;
> > + }
> > +
> > + if (!buffer) {
> > + /* Mode 1: getting the number of secure partitions */
> > +
> > + fill_data = 0;
> > +
> > + ffa_info("Preparing for checking partitions count");
> > +
> > + } else if ((*parts_size >= sizeof(struct ffa_partition_info)) &&
> > + !(*parts_size % sizeof(struct ffa_partition_info))) {
> > + /* Mode 2: retrieving the partitions information */
> > +
> > + fill_data = 1;
> > +
> > + client_desc_idx = 0;
> > +
> > + /*
> > + * number of empty descriptors preallocated by the caller
> > + */
> > + client_desc_max_cnt = *parts_size / sizeof(struct ffa_partition_info);
> > +
> > + ffa_info("Preparing for filling partitions info");
> > +
> > + } else {
> > + ffa_err("invalid function arguments provided");
> > + return -EINVAL;
> > + }
> > +
> > + ffa_info("Searching partitions using the provided UUID");
> > +
> > + /*
> > + * search in the cached partitions
> > + */
> > + for (desc_idx = 0;
> > + desc_idx < ffa_priv_data->partitions.count;
> > + desc_idx++) {
> > + if (ffa_uuid_are_identical(&ffa_priv_data->partitions.descs[desc_idx].sp_uuid,
> > + &part_uuid)) {
> > + ffa_info("Partition ID %x matches the provided UUID",
> > + ffa_priv_data->partitions.descs[desc_idx].info.id);
> > +
> > + parts_found++;
> > +
> > + if (fill_data) {
> > + /*
> > + * trying to fill the partition info in the input buffer
> > + */
> > +
> > + if (client_desc_idx < client_desc_max_cnt) {
> > + buffer[client_desc_idx++] =
> > + ffa_priv_data->partitions.descs[desc_idx].info;
> > + continue;
> > + }
> > +
> > + ffa_err("failed to fill the current descriptor client buffer full");
> > + return -ENOBUFS;
> > + }
> > + }
> > + }
> > +
> > + if (!parts_found) {
>
> By asking for a UUID never found ffa_query_partitions_info() will be
> called again, even if the rx/tx buffers needed are not available any
> longer. Since you populate the cache in ffa_probe() below, perhaps it
> would be better to only rely on the cache after that point.
Thanks for pointing out this.
As stated in the documentation under "Requirements for clients" paragraph:
Clients must:
- Query SPs in EFI boot time mode using the service UUID.
- Unmap RX/TX buffers before EFI runtime mode starts.
The RX/TX buffers are only available at EFI boot time. Querying partitions is
done at boot time and data is cached for future use.
So, when the RX/TX buffers are no longer available, we are at EFI
runtime. At that level querying is not allowed and ffa_get_partitions_info()
is not available. Only direct messaging is available.
At boot time, when an SP is queryed ffa_get_partitions_info() looks at
the cached information first. If the UUID is not found, it will
communicate with secure world to try getting the information.
By the way, all the previous comments on this v7 patch will be addressed in v8.
Thanks.
>
> > + int ret;
> > +
> > + ffa_info("No partition found. Querying framework ...");
> > +
> > + ret = ffa_query_partitions_info(&part_uuid, &parts_found);
> > +
> > + if (ret == 0) {
> > + if (!fill_data) {
> > + *parts_size = parts_found;
> > +
> > + ffa_info("Number of partition(s) found matching the UUID: %d",
> > + parts_found);
> > + } else {
> > + /*
> > + * If SPs data detected, they are already in the private data
> > + * structure, retry searching SP data again to return them
> > + * to the caller
> > + */
> > + if (parts_found)
> > + ret = ffa_get_partitions_info(uuid_str, parts_size, buffer);
> > + else
> > + ret = -ENODATA;
> > + }
> > + }
> > +
> > + return ret;
> > + }
> > +
> > + /* partition(s) found */
> > + if (!fill_data)
> > + *parts_size = parts_found;
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_cache_partitions_info - Queries and saves all secure partitions data
> > + *
> > + * This function invokes FFA_PARTITION_INFO_GET FF-A
> > + * function to query from secure world all partitions information.
> > + *
> > + * The FFA_PARTITION_INFO_GET call is issued with nil UUID as an argument.
> > + * All installed partitions information are returned. We cache them in the
> > + * resident private data structure and we keep the UUID field empty
> > + * (in FF-A 1.0 UUID is not provided by the partition descriptor)
> > + *
> > + * This function is called at the device probing level.
> > + * ffa_cache_partitions_info uses ffa_query_partitions_info to get the data
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_cache_partitions_info(void)
> > +{
> > + return ffa_query_partitions_info(NULL, NULL);
> > +}
> > +
> > +/**
> > + * ffa_msg_send_direct_req - FFA_MSG_SEND_DIRECT_{REQ,RESP} handler function
> > + * @dst_part_id: destination partition ID
> > + * @msg: pointer to the message data preallocated by the client (in/out)
> > + * @is_smc64: select 64-bit or 32-bit FF-A ABI
> > + *
> > + * This function implements FFA_MSG_SEND_DIRECT_{REQ,RESP}
> > + * FF-A functions.
> > + *
> > + * FFA_MSG_SEND_DIRECT_REQ is used to send the data to the secure partition.
> > + * The response from the secure partition is handled by reading the
> > + * FFA_MSG_SEND_DIRECT_RESP arguments.
> > + *
> > + * The maximum size of the data that can be exchanged is 40 bytes which is
> > + * sizeof(struct ffa_send_direct_data) as defined by the FF-A specification 1.0
> > + * in the section relevant to FFA_MSG_SEND_DIRECT_{REQ,RESP}
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_msg_send_direct_req(u16 dst_part_id, struct ffa_send_direct_data *msg, bool is_smc64)
> > +{
> > + ffa_value_t res = {0};
> > + int ffa_errno;
> > + u64 req_mode, resp_mode;
> > +
> > + if (!ffa_priv_data || !ffa_priv_data->invoke_ffa_fn)
> > + return -EINVAL;
> > +
> > + /* No partition installed */
> > + if (!ffa_priv_data->partitions.count || !ffa_priv_data->partitions.descs)
> > + return -ENODEV;
>
> This check isn't needed. What if the partition ID is known by other
> means?
I'm happy to remove this check. I'd like to add a comment explaining
what the other means are. Could you please explain what are they ?
>
> > +
> > + if (is_smc64) {
> > + req_mode = FFA_SMC_64(FFA_MSG_SEND_DIRECT_REQ);
> > + resp_mode = FFA_SMC_64(FFA_MSG_SEND_DIRECT_RESP);
> > + } else {
> > + req_mode = FFA_SMC_32(FFA_MSG_SEND_DIRECT_REQ);
> > + resp_mode = FFA_SMC_32(FFA_MSG_SEND_DIRECT_RESP);
> > + }
> > +
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = req_mode,
> > + .a1 = PREP_SELF_ENDPOINT_ID(ffa_priv_data->id) |
> > + PREP_PART_ENDPOINT_ID(dst_part_id),
> > + .a2 = 0,
> > + .a3 = msg->data0,
> > + .a4 = msg->data1,
> > + .a5 = msg->data2,
> > + .a6 = msg->data3,
> > + .a7 = msg->data4,
> > + }, &res);
> > +
> > + while (res.a0 == FFA_SMC_32(FFA_INTERRUPT))
> > + ffa_priv_data->invoke_ffa_fn((ffa_value_t){
> > + .a0 = FFA_SMC_32(FFA_RUN),
> > + .a1 = res.a1,
> > + }, &res);
> > +
> > + if (res.a0 == FFA_SMC_32(FFA_SUCCESS)) {
> > + /* Message sent with no response */
> > + return 0;
> > + }
> > +
> > + if (res.a0 == resp_mode) {
> > + /*
> > + * Message sent with response
> > + * extract the return data
> > + */
> > + msg->data0 = res.a3;
> > + msg->data1 = res.a4;
> > + msg->data2 = res.a5;
> > + msg->data3 = res.a6;
> > + msg->data4 = res.a7;
> > +
> > + return 0;
> > + }
> > +
> > + ffa_errno = res.a2;
> > + return ffa_to_std_errno(ffa_errno);
> > +}
> > +
> > +/**
> > + * __arm_ffa_fn_smc - SMC wrapper
> > + * @args: FF-A ABI arguments to be copied to Xn registers
> > + * @res: FF-A ABI return data to be copied from Xn registers
> > + *
> > + * Calls low level SMC assembly function
> > + *
> > + * Return: void
> > + */
> > +void __arm_ffa_fn_smc(ffa_value_t args, ffa_value_t *res)
> > +{
> > + arm_smccc_1_2_smc(&args, res);
> > +}
> > +
> > +/**
> > + * ffa_set_smc_conduit - Set the SMC conduit
> > + *
> > + * This function selects the SMC conduit by setting the driver invoke function
> > + * to SMC assembly function
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_set_smc_conduit(void)
> > +{
> > + ffa_priv_data->invoke_ffa_fn = __arm_ffa_fn_smc;
> > +
> > + if (!ffa_priv_data->invoke_ffa_fn) {
> > + ffa_err("failure to set the invoke function");
> > + return -EINVAL;
> > + }
> > +
> > + ffa_info("Conduit is SMC");
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_set_bus_ops - Set the bus driver operations
> > + *
> > + * Setting the driver callbacks.
> > + *
> > + */
> > +static void ffa_set_bus_ops(void)
> > +{
> > + ffa_priv_data->ffa_ops.partition_info_get = ffa_get_partitions_info;
> > + ffa_priv_data->ffa_ops.sync_send_receive = ffa_msg_send_direct_req;
> > + ffa_priv_data->ffa_ops.rxtx_unmap = ffa_unmap_rxtx_buffers;
> > +}
> > +
> > +/**
> > + * ffa_alloc_prvdata - allocate the driver main data structure and sets the device
> > + * @dev: the arm_ffa device
> > + *
> > + * This function creates the main data structure embedding all the driver data.
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_alloc_prvdata(struct udevice *dev)
> > +{
> > + if (!dev) {
> > + ffa_err("no udevice found");
> > + return -ENODEV;
> > + }
> > +
> > + /* The device is registered with the DM. Let's create the driver main data structure*/
> > +
> > + ffa_priv_data = devm_kmalloc(dev, sizeof(struct ffa_prvdata), __GFP_ZERO);
> > + if (!ffa_priv_data) {
> > + ffa_err("can not allocate the driver main data structure");
> > + return -ENOMEM;
> > + }
> > +
> > + ffa_priv_data->dev = dev;
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_probe - The driver probe function
> > + * @dev: the arm_ffa device
> > + *
> > + * Probing is done at boot time and triggered by the uclass device discovery.
> > + * At probe level the following actions are done:
> > + * - setting the conduit
> > + * - querying the FF-A framework version
> > + * - querying from secure world the u-boot endpoint ID
> > + * - querying from secure world the supported features of FFA_RXTX_MAP
> > + * - mapping the RX/TX buffers
> > + * - querying from secure world all the partitions information
> > + *
> > + * All data queried from secure world is saved in the resident private data structure.
> > + *
> > + * The probe will fail if either FF-A framework is not detected or the
> > + * FF-A requests are not behaving correctly. This ensures that the
> > + * driver is not installed and its operations are not exported to the clients.
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +static int ffa_probe(struct udevice *dev)
> > +{
> > + int ret;
> > +
> > + ret = ffa_alloc_prvdata(dev);
> > + if (ret != 0)
> > + return ret;
> > +
> > + ffa_set_bus_ops();
> > +
> > + ret = ffa_set_smc_conduit();
> > + if (ret != 0)
> > + return ret;
> > +
> > + ret = ffa_get_version();
> > + if (ret != 0)
> > + return ret;
> > +
> > + ret = ffa_get_endpoint_id();
> > + if (ret != 0)
> > + return ret;
> > +
> > + ret = ffa_get_rxtx_map_features();
> > + if (ret != 0)
> > + return ret;
> > +
> > + ret = ffa_map_rxtx_buffers();
> > + if (ret != 0)
> > + return ret;
> > +
> > + ret = ffa_cache_partitions_info();
> > + if (ret != 0) {
> > + ffa_free_rxtx_buffers();
> > + return ret;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_remove - The driver remove function
> > + * @dev: the arm_ffa device
> > + * When the device is about to be removed , unmap the RX/TX buffers and free the memory
> > + * Return:
> > + *
> > + * 0 on success.
> > + */
> > +static int ffa_remove(struct udevice *dev)
> > +{
> > + ffa_info("removing the device");
> > +
> > + ffa_unmap_rxtx_buffers();
> > +
> > + if (ffa_priv_data->pair.rxbuf || ffa_priv_data->pair.txbuf)
> > + ffa_free_rxtx_buffers();
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_unbind - The driver unbind function
> > + * @dev: the arm_ffa device
> > + * After the device is removed and memory freed the device is unbound
> > + * Return:
> > + *
> > + * 0 on success.
> > + */
> > +static int ffa_unbind(struct udevice *dev)
> > +{
> > + ffa_info("unbinding the device , private data already released");
> > +
> > + ffa_priv_data = NULL;
> > +
> > + return 0;
> > +}
> > +
> > +/**
> > + * ffa_bus_ops_get - bus driver operations getter
> > + *
> > + * Return:
> > + * This function returns a pointer to the driver operations structure
> > + */
> > +const struct ffa_bus_ops *ffa_bus_ops_get(void)
> > +{
> > + return &ffa_priv_data->ffa_ops;
> > +}
> > +
> > +/**
> > + * ffa_bus_prvdata_get - bus driver private data getter
> > + *
> > + * Return:
> > + * This function returns a pointer to the main private data structure
> > + */
> > +struct ffa_prvdata **ffa_bus_prvdata_get(void)
>
> Why a pointer to a pointer, isn't "struct ffa_prvdata *" enough?
Because ffa_priv_data is a pointer. ffa_bus_prvdata_get() returns an
address of a pointer so the returned type should be struct ffa_prvdata
**
Otherwise, a compiler warning:
drivers/firmware/arm-ffa/core.c: In function ‘ffa_bus_prvdata_get’:
drivers/firmware/arm-ffa/core.c:1278:9: warning: return from incompatible pointer type [-Wincompatible-pointer-types]
return &ffa_priv_data;
^~~~~~~~~~~~~~
>
> > +{
> > + return &ffa_priv_data;
> > +}
> > +
> > +/**
> > + * ffa_bus_discover - discover FF-A bus and probe the arm_ffa device
> > + *
> > + * This function makes sure the FF-A bus is discoverable.
> > + * When probing succeeds FF-A discovery is done. The arm_ffa device is ready to use.
> > + *
> > + * When the bus was already discovered successfully the discovery will not run again.
> > + *
> > + * Arm FF-A transport is implemented through arm_ffa u-boot device managing the FF-A
> > + * communication.
> > + * All FF-A clients should use the arm_ffa device to use the FF-A transport.
> > + *
> > + * Return:
> > + *
> > + * 0 on success. Otherwise, failure
> > + */
> > +int ffa_bus_discover(void)
> > +{
> > + int ret = 0;
> > +
> > + if (!ffa_priv_data)
> > + ret = ffa_device_get();
> > +
> > + return ret;
> > +}
> > +
> > +/**
> > + * Declaring the arm_ffa driver under UCLASS_FFA
> > + */
> > +
> > +U_BOOT_DRIVER(arm_ffa) = {
> > + .name = FFA_DRV_NAME,
> > + .id = UCLASS_FFA,
> > + .probe = ffa_probe,
> > + .remove = ffa_remove,
> > + .unbind = ffa_unbind,
> > +};
> > diff --git a/include/arm_ffa.h b/include/arm_ffa.h
> > new file mode 100644
> > index 0000000000..b0c8a18926
> > --- /dev/null
> > +++ b/include/arm_ffa.h
> > @@ -0,0 +1,93 @@
> > +/* SPDX-License-Identifier: GPL-2.0+ */
> > +/*
> > + * (C) Copyright 2022 ARM Limited
> > + * Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > + */
> > +
> > +#ifndef __ARM_FFA_H
> > +#define __ARM_FFA_H
> > +
> > +#include <linux/printk.h>
> > +
> > +/*
> > + * This header is public. It can be used by clients to access
> > + * data structures and definitions they need
> > + */
> > +
> > +/*
> > + * Macros for displaying logs
> > + */
> > +
> > +#define ffa_info(fmt, ...) pr_info("[FFA] " fmt "\n", ##__VA_ARGS__)
> > +#define ffa_err(fmt, ...) pr_err("[FFA] " fmt "\n", ##__VA_ARGS__)
> > +
> > +/*
> > + * struct ffa_partition_info - Partition information descriptor
> > + * @id: Partition ID
> > + * @exec_ctxt: Execution context count
> > + * @properties: Partition properties
> > + *
> > + * Data structure containing information about partitions instantiated in the system
> > + * This structure is filled with the data queried by FFA_PARTITION_INFO_GET
> > + */
> > +struct __packed ffa_partition_info {
> > + u16 id;
> > + u16 exec_ctxt;
> > +/* partition supports receipt of direct requests */
> > +#define FFA_PARTITION_DIRECT_RECV BIT(0)
> > +/* partition can send direct requests. */
> > +#define FFA_PARTITION_DIRECT_SEND BIT(1)
> > +/* partition can send and receive indirect messages. */
> > +#define FFA_PARTITION_INDIRECT_MSG BIT(2)
> > + u32 properties;
> > +};
>
> Perhaps this has been discussed before. Why is this packed? Is it to allow
> unaligned access or to be sure that there is not implicitly added padding?
> The Linux kernel does seem to need it.
When not using __packed the compiler will add paddings.
ffa_partition_info structure is used for reading SP information
from the secure world.
The issue arises when the non secure world and the secure world
have different architectures (Aarch64 vs Aarch32) or different
endianess. In these cases, the data will be corrupted.
I think we need to use __packed for all the comms structures.
I'm aware ffa_partition_info in the kernel is not packed. I'm happy
to remove __packed from here to align it with Linux.
But please share your thoughts about the padding issues when
working with different architecures/endianess.
Thanks.
>
> > +
> > +/*
> > + * struct ffa_send_direct_data - Data structure hosting the data
> > + * used by FFA_MSG_SEND_DIRECT_{REQ,RESP}
> > + * @data0-4: Data read/written from/to x3-x7 registers
> > + *
> > + * Data structure containing the data to be sent by FFA_MSG_SEND_DIRECT_REQ
> > + * or read from FFA_MSG_SEND_DIRECT_RESP
> > + */
> > +
> > +/* For use with FFA_MSG_SEND_DIRECT_{REQ,RESP} which pass data via registers */
> > +struct __packed ffa_send_direct_data {
> > + unsigned long data0; /* w3/x3 */
> > + unsigned long data1; /* w4/x4 */
> > + unsigned long data2; /* w5/x5 */
> > + unsigned long data3; /* w6/x6 */
> > + unsigned long data4; /* w7/x7 */
> > +};
>
> Why is this __packed? It seems unnecessary.
Yes, in this specific case packing is not necessary. I'll remove this in
v8.
>
> Cheers,
> Jens
>
> > +
> > +/**
> > + * struct ffa_bus_ops - The driver operations structure
> > + * @partition_info_get: callback for the FFA_PARTITION_INFO_GET
> > + * @sync_send_receive: callback for the FFA_MSG_SEND_DIRECT_REQ
> > + * @rxtx_unmap: callback for the FFA_RXTX_UNMAP
> > + *
> > + * The data structure providing all the operations supported by the driver.
> > + * This structure is EFI runtime resident.
> > + */
> > +struct ffa_bus_ops {
> > + int (*partition_info_get)(const char *uuid_str,
> > + u32 *parts_size, struct ffa_partition_info *buffer);
> > + int (*sync_send_receive)(u16 dst_part_id, struct ffa_send_direct_data *msg, bool is_smc64);
> > + int (*rxtx_unmap)(void);
> > +};
> > +
> > +/**
> > + * The device driver and the Uclass driver public functions
> > + */
> > +
> > +/**
> > + * ffa_bus_ops_get - driver operations getter
> > + */
> > +const struct ffa_bus_ops *ffa_bus_ops_get(void);
> > +
> > +/**
> > + * ffa_bus_discover - discover FF-A bus and probes the arm_ffa device
> > + */
> > +int ffa_bus_discover(void);
> > +
> > +#endif
> > diff --git a/include/dm/uclass-id.h b/include/dm/uclass-id.h
> > index 4b2c323452..fb59d4f356 100644
> > --- a/include/dm/uclass-id.h
> > +++ b/include/dm/uclass-id.h
> > @@ -4,6 +4,9 @@
> > *
> > * (C) Copyright 2012
> > * Pavel Herrmann <morpheus.ibis at gmail.com>
> > + *
> > + * (C) Copyright 2022 ARM Limited
> > + * Abdellatif El Khlifi <abdellatif.elkhlifi at arm.com>
> > */
> >
> > #ifndef _DM_UCLASS_ID_H
> > @@ -55,6 +58,7 @@ enum uclass_id {
> > UCLASS_EFI_MEDIA, /* Devices provided by UEFI firmware */
> > UCLASS_ETH, /* Ethernet device */
> > UCLASS_ETH_PHY, /* Ethernet PHY device */
> > + UCLASS_FFA, /* Arm Firmware Framework for Armv8-A */
> > UCLASS_FIRMWARE, /* Firmware */
> > UCLASS_FPGA, /* FPGA device */
> > UCLASS_FUZZING_ENGINE, /* Fuzzing engine */
> > --
> > 2.17.1
> >
More information about the U-Boot
mailing list