Processor Boot Management TISCI Description¶
Introduction¶
This chapter provides information about the messaging APIs for the processor boot control. APIs are divided into the following two sets:
- Book keeping APIs - Meant to control access to allow a reasonable usage scenario of processors.
- Processor Control APIs - Meant to be the actual Processor Core controls.
Book Keeping APIs:
| TISCI Message ID | Message Name | Available in GP | Avialable in HS-FS |
|---|---|---|---|
| 0xC000 | TISCI_MSG_PROC_REQUEST. | Yes | Yes |
| 0xC001 | TISCI_MSG_PROC_RELEASE. | Yes | Yes |
| 0xC005 | TISCI_MSG_PROC_HANDOVER. | Yes | Yes |
Processor Control APIs:
| TISCI Message ID | Message Name | Available in GP | Avialable in HS-FS |
|---|---|---|---|
| 0xC100 | TISCI_MSG_PROC_SET_CONFIG | Yes | Yes |
| 0xC101 | TISCI_MSG_PROC_SET_CONTROL | Yes | Yes |
| 0xC120 | TISCI_MSG_PROC_AUTH_BOOT | No | Yes* |
| 0xC400 | TISCI_MSG_PROC_GET_STATUS | Yes | Yes |
| 0xC401 | TISCI_MSG_PROC_WAIT_STATUS | Yes | Yes |
The APIs use the overall concepts explained in the following sections.
ID definition¶
To help identify the various entities invovled, the following IDs are involved:
- HOST_ID is the concept of identifying processing entities (SoC specific)
- PROC_ID is the specific hardware processor instance involved (SoC specific)
Access control definition¶
Access control is enforced via the Book keeping APIs. The basic definition of access is as follows:
- We identify a requester (just like rest of TISCI) using Host ID (or plain Host).
- By default, we permit all processors to be controlled by any other Host (no policing).
- Board configuration provides: per PROC_ID, a limited list of “permitted host IDs”
which are permitted to control a processor. BUT with the following conditions:
- Only one host can control a processor at a time
- A host(with control) can hand over control of a processor to another host in the permitted list.
- “recovery master” host id is identified in board cfg and this host can override the ownership already established.
In addition, ONLY a secure host can request for an authenticated image access.
API access control will be as follows:
- request_processor: Only one host can get control from access list. However, “recovery master” host can override previously allocated master.
- handover_processor: Only the host with current control of the processor.
- release_processor: Only the host with current control of the processor.
- set_processor_config: Only the host with current control of the processor.
- get_processor_config: Any host within the access control list.
- set_processor_control: Only the host with current control of the processor.
- get_processor_control: Any host within the access control list.
- authenticate_and_start_image: Only a secure host with current control of the processor.
- set_processor_suspend_ready: Only the host with current control of processor can report that the processor is ready to suspend
- get_processor_wake_reason: Any host within the access control list.
Sequencing of APIs¶
The boot APIs must be used in correct sequence with the device and clock APIs to be operational. This sequence would be specific to the processor involved.
Rationale: by encoding the sequence of processor boot along with boot configuration API will complicate management of processors, since the operational requirements are extremely varied. Some processors need the boot configuration done, clocked, and specific MMU/RAT configuration be done prior to be released from reset. Many of the operations may require interaction with specific processor internals that is hard to make generic and usecase independent.
Example usage of APIs¶
Example 1: Boot a processor from one core and let it self manage:
- Boot host: (optionally) PM apis to control clock and hold processor in reset. This might be necessary for some processors to get the boot vector registers to be accessible in the first place.
- Boot host: request_processor
- Boot host: set_processor_config -> set bootvector
- Alternatively call authenticate and start image API.
- Boot host: handover_processor -> give control to ‘processor host’
- Boot host: PM apis to control clock and release processor from reset. ‘processor host’ gets active. It is probably better to release the processor from reset after handover, in case the ‘processor host’ tries to do additional operations. however this depends on usecase - use a sane judgement as to how to sequence the APIs.
- Processor host: (continues to boot and do other TISCI APIs permitted)
- Processor host: set_processor_suspend_ready -> ready to suspend
- Processor WFI Wakeup:
- Processor host: get_processor_wake_reason -> get reason for wakeup. …. until all operations are complete…
- Processor host: set_processor_suspend_ready -> state it is going down
- Processor host: release_processor -> Mark no longer required
- Processor WFI (NOTE: a processor cannot switch off it’s own clocks -> so in case of a self managed processor shutdown sequence, it will need System firmware to help do the last stages of operation - typically this will involve either shutdown OR reset sequence).
Example 2: Boot a processor from one core and recover from another core:
- Boot host: request_processor
- Boot host: set_processor_config -> set bootvector
- Boot host: handover_processor -> give control to ‘processor host’
- Boot host: PM apis to control clock and release processor boot host dies
- Recovery Host: request_processor
- Recovery Host: set_processor_config -> force power off
- Recovery Host: PM apis to control clock and force power off of processor
- Recovery Host: Additional cleanup API calls …. Restoration sequence….
Processor Boot API Description¶
This Section goes into details on the various APIs involved for processor control
Note
Reference Chapter 5: SoC Family Specific Documentation to see Host IDs and Processor IDs for your SoC.
Book Keeping APIs¶
These are the top level APIs that provide access control knobs for processor operation APIs to be valid.
TISCI_MSG_PROC_REQUEST - Request Processor¶
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_REQUEST (0xC000U)
Message to get a Processor
struct tisci_msg_proc_request_req
Request for physical processor control request
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
| processor_id | u8 | Specifies a Processor ID. See the SoC Family Specific Documentation Chapter of the TISCI User Guide for accepted values. |
Provides a means for either the “recovery master” host or another host in the permitted access list to request for a physical processor control.
struct tisci_msg_proc_request_resp
Request for physical processor control response
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
Although this message is essentially empty and contains only a header a full data structure is created for consistency in implementation. ACK response: The processor access is permitted for the host if processor is un-claimed AND host is permitted to control the processor OR if the host is the recovery master. NAK response: The processor access is not permitted.
TISCI_MSG_PROC_RELEASE - Release Processor¶
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_RELEASE (0xC001U)
Message to release a Processor
struct tisci_msg_proc_release_req
Release physical processor control request
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
| processor_id | u8 | ID of processor to release. |
Provides a means for the host with current control to relinquish a physical processor control.
struct tisci_msg_proc_release_resp
Release physical processor control response
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
Although this message is essentially empty and contains only a header a full data structure is created for consistency in implementation. ACK Response: The host had control over the processor and is confirmed to be released to “free pool”. NAK Response: The processor access is not permitted.
TISCI_MSG_PROC_HANDOVER - Handover Processor¶
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_HANDOVER (0xC005U)
Message to handover a Processor
struct tisci_msg_proc_handover_req
Request to handover control of a processor to another host if permitted.
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
| processor_id | u8 | Specifies a Processor ID. |
| host_id | u8 | Specifies the new host to hand over control to. |
Provides a means for the host with current control to relinquish a physical processor control to another host in the permitted list.
struct tisci_msg_proc_handover_resp
Response to handover of control of a processor to another host if permitted.
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
Although this message is essentially empty and contains only a header a full data structure is created for consistency in implementation. ACK Response: The host had control over the processor and is confirmed to be released to “free pool”. NAK Response: The processor access is not permitted.
Processor Control APIs¶
These are granular control APIs for control of the processor state themselves. These APIs need to be used in conjunction with standard Power Management APIs.
Note
Not all APIs are supported for all processors. See Processor Specific Flags for core specific flags (implies the applicable APIs are valid for that type of core).
TISCI_MSG_PROC_SET_CONFIG - Set Processor Configuration¶
Purpose: Provides a means for the host with current control to do the base configuration of the processor.
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_SET_CONFIG (0xC100U)
Message to Set the processor configuration
struct tisci_msg_proc_set_config_req
Processor Boot Configuration
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Message header |
| processor_id | u8 | ID of processor |
| bootvector_lo | u32 | Lower 32bit (Little Endian) of boot vector |
| bootvector_hi | u32 | Higher 32bit (Little Endian) of boot vector |
| config_flags_1_set | u32 | Optional Processor specific Config Flags to set. Setting a bit here implies required bit has to be set to 1. |
| config_flags_1_clear | u32 | Optional Processor specific Config Flags to clear. Setting a bit here implies required bit has to be cleared to 0. |
Note
Boot vector address and config are processor specific configurations. This may be done in separate invocations as required in processor specific startup sequence.
Note
Note bits 6:0 of the R5F Vector Table address are always 0. The TISCI message will return failure if the bits 6:0 of the bootvector_lo are non-zero. Refer the section on CTRLMMR_*_BOOTVECT_LO for the R5F in the TRM.
struct tisci_msg_proc_set_config_resp
Response to Processor Boot Configuration message.
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
Although this message is essentially empty and contains only a header a full data structure is created for consistency in implementation. ACK Response: The host had control over the processor and requested operation is successful NAK Response: The processor access is not permitted or the operation failed.
Attention
Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
TISCI_MSG_PROC_SET_CONTROL - Set Processor Control Flags¶
Purpose: Provides a means for the host with current control to setup limited control flags in specific cases.
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_SET_CONTROL (0xC101U)
Message to Set the processor control
struct tisci_msg_proc_set_control_req
Optional processor specific message for sequence control
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Message header |
| processor_id | u8 | ID of processor |
| control_flags_1_set | u32 | Optional Processor specific Control Flags to set. Setting a bit here implies required bit has to be set to 1. |
| control_flags_1_clear | u32 | Optional Processor specific Control Flags to clear. Setting a bit here implies required bit has to be cleared to 0. |
struct tisci_msg_proc_set_control_resp
Response to optional processor specific message for sequence control
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
Although this message is essentially empty and contains only a header a full data structure is created for consistency in implementation. ACK Response: The host had control over the processor and requested operation is successful NAK Response: The processor access is not permitted or the operation failed.
Attention
Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
TISCI_MSG_PROC_AUTH_BOOT - Authenticate Image and Configure Processor¶
Purpose: Provides a means for the host with current control to do the following:
- Authenticate and load a binary using the certificate provided information
- Use certificate information also to setup critical processor specific flags which is similar to
- Use certificate information also to configure firewalls. This feature can be used to protect the binary being authenticated. Set Processor Configuration.
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_AUTH_BOOT (0xC120U)
Message to do authenticated boot configuration of a processor
struct tisci_msg_proc_auth_boot_req
Authenticate and start image
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Message header |
| certificate_address_lo | u32 | Lower 32bit (Little Endian) of certificate |
| certificate_address_hi | u32 | Higher 32bit (Little Endian) of certificate |
Attention
- The certificate itself shall contain relevant information about the processor flags and configuration information.
- See Hosts description for the corresponding SoC to identify which hosts are secure and which are not.
- On HS-FS device type, only image integrity check is done. The root of trust key comparison check is skipped.
- Only the unowned firewalls/regions or those owned by the requesting host would be configurable through the certificate.
- The user must specify the host which would own the firewall/regions being configured. Ownership would be transfered to this host. It is mandatory to specify a host in the certificate to configure firewalls. Skipping this would result in the API failure.
Attention
Please see Security X509 Certificate Documentation for certificate format.
struct tisci_msg_proc_auth_boot_resp
Response to authenticate and start image request
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
| image_address_lo | u32 | Lower 32bit (Little Endian) of image |
| image_address_hi | u32 | Higher 32bit (Little Endian) of image |
| image_size | u32 | Size of the binary |
ACK Response: The host had control over the processor and requested operation is successful NAK Response: The processor access is not permitted or the operation failed. IMPORTANT: Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
Attention
Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
TISCI_MSG_PROC_GET_STATUS - Get Processor Status¶
Purpose: Provides a means for hosts in the permitted list to get the status of a physical processor. This is required for the hosts to sequence events in the correct order
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_GET_STATUS (0xC400U)
Message to Get the processor status
struct tisci_msg_proc_get_status_req
Processor Status request
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Message header |
| processor_id | u8 | ID of processor |
struct tisci_msg_proc_get_status_resp
Processor Status Response
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Message header |
| processor_id | u8 | ID of processor |
| bootvector_lo | u32 | Lower 32bit (Little Endian) of boot vector |
| bootvector_hi | u32 | Higher 32bit (Little Endian) of boot vector |
| config_flags_1 | u32 | Optional Processor specific Config Flags set |
| control_flags_1 | u32 | Optional Processor specific Control Flags set |
| status_flags_1 | u32 | Optional Processor specific Status Flags set |
ACK Response: The host had control over the processor and requested operation is successful NAK Response: The processor access is not permitted or the operation failed. IMPORTANT: Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
Attention
Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
TISCI_MSG_PROC_WAIT_STATUS - Wait for Processor Status¶
Purpose: Provides a means for hosts in the permitted list to wait for the status of a physical processor.
Warning
This API that has impact on firmware performance and is typically only required for the hosts to sequence events in the correct order when executing from the processor itself. In short, avoid if possible.
The worst case delay could be upto: (register operations and code overheads) + num_wait_iterations(255) * delay_per_iteration_us(255) + delay_before_iteration_loop_start_us (255) = 65.280 milli seconds + (register operations and code overheads)
Usage:
| Message Type | Normal |
| Secure Queue Only? | No |
TISCI Message ID
TISCI_MSG_PROC_WAIT_STATUS (0xC401U)
Message to Wait for processor status
struct tisci_msg_proc_status_wait_req
Processor Status Wait
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Message header |
| processor_id | u8 | ID of processor |
| num_wait_iterations | u8 | 1-255, Total number of iterations we will check before we will timeout and give up |
| num_match_iterations | u8 | 1-255, How many iterations should we have continued status to account for status bits glitching. This is to make sure that match occurs for consecutive checks. This implies that the worst case should consider that the stable time should at the worst be num_wait_iterations num_match_iterations to prevent timeout. |
| delay_per_iteration_us | u8 | 0-255, Specifies how long to wait (in micro seconds) between each status checks. This is the minimum duration, and overhead of register reads and checks are on top of this and can vary based on varied conditions. |
| delay_before_iteration_loop_start_us | u8 | 0-255, Specifies how long to wait (in micro seconds) before the very first check in the first iteration of status check loop. This is the minimum duration, and overhead of register reads and checks are |
| status_flags_1_set_all_wait | u32 | If non-zero, Specifies that all bits of the status matching this field requested MUST be ‘1’. |
| status_flags_1_set_any_wait | u32 | If non-zero, Specifies that at least one of the bits matching this field requested MUST be ‘1’. |
| status_flags_1_clr_all_wait | u32 | If non-zero, Specifies that all bits of the status matching this field requested MUST be ‘0’. |
| status_flags_1_clr_any_wait | u32 | If non-zero, Specifies that at least one of the bits matching this field requested MUST be ‘0’. |
struct tisci_msg_proc_status_wait_resp
Processor Status Wait Response
| Parameter | Type | Description |
|---|---|---|
| hdr | struct tisci_header | Generic TISCI message header. |
Although this message is essentially empty and contains only a header a full data structure is created for consistency in implementation. ACK Response: The status requested was achieved within the constraints provided in request. NAK Response: The processor access is not permitted or operation failed or timedout.
Warning
At least one of status_flags_1_set_all_wait, status_flags_1_set_any_wait, status_flags_1_clr_all_wait or status_flags_1_clr_any_wait must be requested.
Flags and sequences desired are very specific to processor and SoC involved. Please refer to appropriate documentation for accurate sequencing and status information.
Note
A trivial example for a hypothetical processor status wait could have the parameters as follows:
# No optional bits to be set
status_flags_1_set_all_wait = 0
# Either WFI OR WFE to be set as 1
status_flags_1_set_any_wait = WFI | WFE
# CLK_STOP must be to be cleared
status_flags_1_clr_all_wait = CLK_STOP
# No optional status bits to be cleared
status_flags_1_clr_any_wait = 0
# Check status every ~2 us
delay_per_iteration_us = 2
# Do not wait to start checks
delay_before_iteration_loop_start_us = 0
# Check for 10 times before timing out
num_wait_iterations = 10
# Must match at least for 2 consecutive iterations
num_match_iterations = 2
Attention
Reason for failure is NOT provided to prevent security attacks by scan. If permitted, System firmware logs shall provide relevant failure information.
Generic Processor Flags¶
- config_flags_1 Field Reserved for Generic usage
| Flag Name | Bit Offset | Description |
|---|---|---|
| GEN_IGN_BOOTVECTOR | 28 | Valid only for config_flags_1_clear Flag indicating that SYSFW should not use the bootvector_lo and bootvector_hi fields in the tisci_msg_proc_set_config_req structure. |
Processor Specific Flags¶
This section lists the flags that are specific to each processor types. These flags apply to the processor control APIs.
Note
The System Firmware does not modify the silicon defaults unless the specific message to update the processor config/control flags is received.
ARMV8¶
- config_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| DBG_EN | 0 | invasive Debug |
| DBG_NIDEN | 1 | Non-invasive Debug |
| DBG_SPIDEN | 2 | Secure invasive Debug |
| DBG_SPNIDEN | 3 | Secure Non-invasive Debug |
| ARCH32 | 8 | 32bit mode (if disabled, implies 64 bit mode) |
- control_flags_1 Fields:
| Flag Name | Bit Offset | Description |
|---|---|---|
| ACINACTM | 0 | Snoop coherency interface state |
| AINACTS | 1 | ACP interface state |
| L2FLUSHREQ | 8 | SoC level L2 Flush Request - See L2FLUSH_DONE status for completion |
Note
Please refer to SoC Technical Reference Manual and ARM Technical Reference for information on sequencing required for startup and shutdown of CPUs in a cluster. Applying these flags to any CPU in a cluster affects the corresponding cluster.
- status_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| WFE | 0 | Set if the core is in WFE state |
| WFI | 1 | Set if the core is in WFI state |
| L2FLUSH_DONE | 4 | Set if the cluster’s L2 flush done is complete |
| STANDBYWFIL2 | 5 | Set if the cluster’s L2 WFI is achieved |
ARM R5¶
- config_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| DBG_EN | 0 | invasive Debug |
| DBG_NIDEN | 1 | Non-invasive Debug |
| LOCKSTEP | 8 | On Write - Enable Lockstep (if permitted) - (1: Lockstep enabled, 0 - Lockstep disabled). On Read - Core Status - (1: Core in lockstep mode, 0 - Core in split mode) |
| TE_INIT | 9 | Exception handling state at reset (0 - ARM, 1 - Thumb) |
| NMFI_EN | 10 | Enable Core Non-Maskable Fast Interrupts |
| TCM_RSTBASE | 11 | Core A/BTCM Reset Base address Indicator (0 - BTCM located at address 0x0, 1 - ATCM located at address 0x0) |
| BTCM_EN | 12 | Enable Core BTCM RAM at reset |
| ATCM_EN | 13 | Enable Core ATCM RAM at reset |
| MEM_INIT_DIS | 14 | Disables SRAM initialization (TCM, etc) at reset. (Not available on all devices. Please read TRM before trying to set this field.) (TRM Section to be referred to is: Chapter 5 (“Device Configuration”) -> Section 5.1 (“5.1 Control Module (CTRL_MMR)”) -> CTRLMMR_SEC_CLSTRx_CFG Register) |
| SINGLE_CORE | 15 | Single / Dual CPU Mode CLSTR_CFG Value 0 = Both CPUs are enabled, 1 = CPU1 Core is disabled (Not available on all devices. Please read TRM before trying to set this field.) (TRM Section to be referred to is: Chapter 5 (“Device Configuration”) -> Section 5.1 (“5.1 Control Module (CTRL_MMR)”) -> CTRLMMR_SEC_CLSTRx_CFG Register) |
- control_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| CORE_HALT | 0 | Halt Core |
| R5_LPSC | 1 | Command for R5F LPSC Control (1 - LPSC ON, 0 - LPSC OFF) Reads will always return 0. You are expected to read the PM status via the PM message TISCI_MSG_GET_DEVICE |
| R5_RESET | 2 | R5F Reset control command (1- Assert Reset, 0 - Deassert Reset) |
The usage of R5_LPSC and R5_RESET is only for below mentioned devices, - MCU R5F on the J721E, J7200, J721S2 and J784S2 family of devices where the RM and PM functions run on the MCU R5F. - DM R5F on the AM62x, AM62Ax and AM62Px family of devices where the RM and PM functions run on the DM R5F.
The MCU R5F and DM R5F (where the RM and PM functions run) cannot turn off/on by itself during a re-boot sequence which requires the following fundamental sequence of steps:
- Wait for the R5F to hit WFI. (MCU at this point is running WFI and cannot run steps 2 and 3)
- Power off the R5F by writing to PSC registers.
- Power on the R5F by writing to PSC registers.
In order to handle this the TIFS provides a special function for handling the LPSC control only for MCU R5F core 0 and core 1 which can enable performing the sequence of powering down and powering up the MCU R5Fs once the MCU R5F hits WFI.
Similar to the LPSC control for power off and power on, the TIFS also controls the reset control for R5F LPSC reset controls as the MCU R5F cannot handle its own resets.
For all other devices in the SoC the MCU R5F handles the LPSC configuration. For the devices like AM6 the TIFS, Power Management and Resource management run on the DMSC. Hence these controls are not applicable to these devices.
- status_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| WFE | 0 | Set if the core is in WFE state |
| WFI | 1 | Set if the core is in WFI state |
| CLK_GATED | 2 | Core Clock Stopped due to WFI or WFE state |
| LOCKSTEP_PERMITTED | 8 | Is Lockstep configuration permitted - 1: yes, 0 - no |
| SINGLECORE_ONLY | 9 | Is Single Core only configuration enabled - 1: yes, 0 - no |
C7x DSP¶
- config_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| L2_PIPELINE_LATENCY_VALUE | 0-3 | Set C7x Corepac L2 Pipeline latency. valid values: 1 to 5 |
| L2_ACCESS_LATENCY_VALUE | 4-7 | Set C7x Corepac L2 Access latency. valid values: 2 to 5 |
- control_flags_1 Fields
Not supported.
- status_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| WFE | 0 | Set if the core is in WFE state |
| WFI | 1 | Set if the core is in WFI state |
C6x DSP¶
- config_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| SSCLK_MODE_DIV_CLK_MODE_VALUE | 0-2 | Controls the C66 clock rate for cluster logic and bus interfaces. (See warning note below) 0x1 - Div2 clock mode. 0x2 - Div3 clock mode. 0x3 - Div4 clock mode. |
Warning
NOTE: Values of SSCLK_MODE_DIV_CLK_MODE_VALUE are TRM value + 1 for avoiding ‘0’ as a valid value which cannot be distinguished from values we are not attempting to set
- control_flags_1 Fields
Not supported.
- status_flags_1 Fields
Not supported.
ARM M4F¶
Note
The M4 processor driver is dual purpose:
- Support for general purpose M4 CPUs found on many of the supported SoCs today.
- Support for the HSM M4 processor found in the SMS subsystem
Flags described below may only be relevant to one class of M4 processors and will be clarified as necessary
Warning
NOTE: M4F always starts from 0x0 from internal TCM RAM. There is no specific bootvector configuration possible. We need to request the device with reset enabled and load the TCM RAM prior to releasing the reset to allow the processor to execute.
- config_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| DBG_EN | 0 | invasive Debug for general purpose M4 |
| DBG_NIDEN | 1 | Non-invasive Debug for general purpose M4 |
| HSM_DBG_EN | 2 | Debug enable for HSM M4 |
- control_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| HSM M4 RESET | 0 | M4F Reset control command (1- Assert Reset, 0 - Deassert Reset) This particular control is applicable only for HSM M4 processors since the TIFS must act as a secure gatekeeper for HSM reset control. This is not appicable to general purpose M4 processors. |
- status_flags_1 Fields
| Flag Name | Bit Offset | Description |
|---|---|---|
| WFI | 1 | Set if the core is in WFI state |