[FL-1952] BLE bonding fix (#805)

* furi-hal-bt: add mutex guarding core2 state
* ble-glue: configure ble keys storage in SRAM2
* bt: add load and save ble keys in internal storage
* bt: improve work furi_hal_bt API
* bt: rework app_entry -> ble_glue
* bt: apply changes for f6 target
* desktop: remove furi check
* ble-glue: comment NVM in SRAM2 configuration
* FuriHal: fix flash controller state corruption, fix incorrect semaphore release, implement C1-C2 flash controller access according to spec. Gui: change logging level.
* Libs: better lfs integration with lfs_config.
* Ble: switch C2 NVM to RAM.
* FuriHalCrypto: ensure that core2 is alive before sending shci commands
* Ble: fix incorrect nvm buffer size

Co-authored-by: あく <alleteam@gmail.com>
This commit is contained in:
gornekich
2021-11-04 20:26:41 +03:00
committed by GitHub
parent bb9c464a13
commit 3225f40870
39 changed files with 1184 additions and 568 deletions

View File

@@ -29,6 +29,37 @@
* Semaphores
* THIS SHALL NO BE CHANGED AS THESE SEMAPHORES ARE USED AS WELL ON THE CM0+
*****************************************************************************/
/**
* Index of the semaphore used the prevent conflicts after standby sleep.
* Each CPUs takes this semaphore at standby wakeup until conclicting elements are restored.
*/
#define CFG_HW_PWR_STANDBY_SEMID 10
/**
* The CPU2 may be configured to store the Thread persistent data either in internal NVM storage on CPU2 or in
* SRAM2 buffer provided by the user application. This can be configured with the system command SHCI_C2_Config()
* When the CPU2 is requested to store persistent data in SRAM2, it can write data in this buffer at any time when needed.
* In order to read consistent data with the CPU1 from the SRAM2 buffer, the flow should be:
* + CPU1 takes CFG_HW_THREAD_NVM_SRAM_SEMID semaphore
* + CPU1 reads all persistent data from SRAM2 (most of the time, the goal is to write these data into an NVM managed by CPU1)
* + CPU1 releases CFG_HW_THREAD_NVM_SRAM_SEMID semaphore
* CFG_HW_THREAD_NVM_SRAM_SEMID semaphore makes sure CPU2 does not update the persistent data in SRAM2 at the same time CPU1 is reading them.
* There is no timing constraint on how long this semaphore can be kept.
*/
#define CFG_HW_THREAD_NVM_SRAM_SEMID 9
/**
* The CPU2 may be configured to store the BLE persistent data either in internal NVM storage on CPU2 or in
* SRAM2 buffer provided by the user application. This can be configured with the system command SHCI_C2_Config()
* When the CPU2 is requested to store persistent data in SRAM2, it can write data in this buffer at any time when needed.
* In order to read consistent data with the CPU1 from the SRAM2 buffer, the flow should be:
* + CPU1 takes CFG_HW_BLE_NVM_SRAM_SEMID semaphore
* + CPU1 reads all persistent data from SRAM2 (most of the time, the goal is to write these data into an NVM managed by CPU1)
* + CPU1 releases CFG_HW_BLE_NVM_SRAM_SEMID semaphore
* CFG_HW_BLE_NVM_SRAM_SEMID semaphore makes sure CPU2 does not update the persistent data in SRAM2 at the same time CPU1 is reading them.
* There is no timing constraint on how long this semaphore can be kept.
*/
#define CFG_HW_BLE_NVM_SRAM_SEMID 8
/**
* Index of the semaphore used by CPU2 to prevent the CPU1 to either write or erase data in flash
* The CPU1 shall not either write or erase in flash when this semaphore is taken by the CPU2