adding BLE to a system without softdevice, using the 17.1 sdk (can't change) freetos no ble events

[rexxdad]

the existing code is using freetos, 
and main.c finally ends at 

vTaskStartScheduler();

 

which never returns as expected... but my ble scan and advertise do not appear to work, no events are recevied... and nrf_connect app doesn't see my device advertising

I had a previous topic open on adding the ble code, which fixed a couple build setup issues. 

debugging with OZONE

have stops on the event handlers.. 

scanning I am not filtering UUIDs in the scan,, give me everything

 

advertise start 

 

err_code = sd_ble_gap_adv_start(m_adv_handle, APP_BLE_CONN_CFG_TAG);

 

scanner start 

bool ble_stack_manager_scan_start(ble_mgr_adv_report_handler_t scan_handler)
{
ret_code_t err_code;

m_scan_handler = scan_handler;

err_code = nrf_ble_scan_start(&m_scan);

everything returns rc=0

I don't think I have to  create freetos tasks explicitly and run them 

we did have to move the existing timer 0 code to 1 so SD timer would work. 

[rexxdad]

here is the scan setup code 

#define SCAN_INTERVAL 0x00A0 // Scan interval (100 ms)

#define SCAN_WINDOW 0x0050 // Scan window (50 ms)

#define SCAN_TIMEOUT 0 // Scan timeout (0 = no timeout)

// Initialize central mode
// QLOGD("Initializing BLE in central mode");

// Initialize scanner
nrf_ble_scan_init_t scan_init_params;
memset(&scan_init_params, 0, sizeof(scan_init_params));

// Set scan parameters
scan_init_params.p_scan_param = &(ble_gap_scan_params_t){
.active = 1,
.interval = SCAN_INTERVAL,
.window = SCAN_WINDOW,
.timeout = SCAN_TIMEOUT,
.scan_phys = BLE_GAP_PHY_1MBPS,
.filter_policy = BLE_GAP_SCAN_FP_ACCEPT_ALL,
};


err_code = nrf_ble_scan_init(&m_scan, &scan_init_params, NULL);
if (err_code != NRF_SUCCESS)
{
QLOGE("Failed to initialize scanner: %ld", err_code);
return false;
}

// Initialize database discovery
err_code = ble_db_discovery_init(NULL);
if (err_code != NRF_SUCCESS)
{
QLOGE("Failed to initialize database discovery: %ld", err_code);
return false;
}

// Initialize central mode
// QLOGD("Initializing BLE in central mode");

// Initialize scanner
nrf_ble_scan_init_t scan_init_params;
memset(&scan_init_params, 0, sizeof(scan_init_params));

// Set scan parameters
scan_init_params.p_scan_param = &(ble_gap_scan_params_t){
.active = 1,
.interval = SCAN_INTERVAL,
.window = SCAN_WINDOW,
.timeout = SCAN_TIMEOUT,
.scan_phys = BLE_GAP_PHY_1MBPS,
.filter_policy = BLE_GAP_SCAN_FP_ACCEPT_ALL,
};


err_code = nrf_ble_scan_init(&m_scan, &scan_init_params, NULL);
if (err_code != NRF_SUCCESS)
{
QLOGE("Failed to initialize scanner: %ld", err_code);
return false;
}

// Initialize database discovery
err_code = ble_db_discovery_init(NULL);
if (err_code != NRF_SUCCESS)
{
QLOGE("Failed to initialize database discovery: %ld", err_code);
return false;
}

[rexxdad]

add rtt for debugging
in Controller mode, this will scan for Controllee mode endpoints, AND advertise to pair with phone 

neither appear to actually work, altho rc=0 

about 20 seconds in get hardfault,  stack return address out of range, pc A60, which shouldn't be.. like branch to 0 

00> Calling ble_stack_manager_init...
00> ble_stack_manager_init returned: TRUE
00> BLE init SUCCESS - starting advertising
00> advertising_start called
00> === BLE Init Complete ===

freetos scheduler started here

00> log_processing_task: Started
00> [00:00:03.275,078] <info> app: ble_stack_manager_init: 1-nrf_sdh_enable_request
00>
00> [00:00:03.275,274] <info> app: ble_stack_manager_init: 2-nrf_sdh_ble_default_cfg_set
00>
00> [00:00:03.275,325] <info> app: ble_stack_manager_init: 3-nrf_sdh_ble_enable
00>
00> [00:00:03.275,977] <info> app: ble_stack_manager_init: 4-nrf_sdh_freertos_init
00>
00> [00:00:03.276,419] <info> app: ble_stack_manager_init: 5-gap_params_init
00>
00> [00:00:03.276,484] <info> app: ble_stack_manager_init: 6-gap_conn_params_init
00>
00> [00:00:03.276,505] <info> app: ble_stack_manager_init: 7-gatt_init
00>
00> [00:00:03.276,523] <info> app: Registered UCI UUID with type: 2
00>
00> [00:00:03.276,767] <info> app: ble_stack_manager_init: 8-advertising_init
00>
00> [00:00:03.276,799] <info> app: Advertising will include UCI service UUID: 0x0001 (type 2)
00>
00> [00:00:03.276,923] <info> app: ble_stack_manager_init: 9-scan_init
00>
00> [00:00:03.276,927] <info> app: Initializing BLE scanning module
00>
00> [00:00:03.276,933] <info> app: ble_stack_manager_init: 10-COMPLETE
00>
00> [00:00:03.276,937] <info> app: BLE stack initialized successfully
00>
00> [00:00:03.276,972] <info> app: About to call ble_advertising_start...
00>
00> [00:00:03.283,869] <info> app: ble_advertising_start returned: 0

this was an app that used USB only. nothing connected to USB uart port on Murata 2AB 
debugger connect via pin header, and USB port socket

[rexxdad]

I’ve made progress. Lots of setup problems, but finally working

one outstanding issue, set number of peripherals 16 with larger buffer size, get ram_size 12400

set in ld file, linker shows 12400

sdh_ble_enable fails rc=4

but it reports 12400 is needed

is there a 64k limit?  Size under 64k works

but I might have 30 or more of our devices to scan and talk to

[Sigurd]

Hi!

rexxdad said:

I’ve made progress. Lots of setup problems, but finally working

Great to hear!

rexxdad said:

one outstanding issue, set number of peripherals 16 with larger buffer size, get ram_size 12400

set in ld file, linker shows 12400

sdh_ble_enable fails rc=4

but it reports 12400 is needed

is there a 64k limit?  Size under 64k works

but I might have 30 or more of our devices to scan and talk to

Yes, there is a limit, the dynamic part of the SoftDevice RAM region can be max 64 kB 

https://docs.nordicsemi.com/bundle/s140_v7.2.0_api/page/group_b_l_e_c_o_m_m_o_n_f_u_n_c_t_i_o_n_s.html#ga8a92420df90945526d4c89d09486508d

[rexxdad]

Ok, almost there

using the S140 softdevice works, debug works in Ozone
now another device, using the 52833

I have adjusted the LD file to reserve space for SD per the release notes.

in Ozone project script I load the elf (which does NOT have SD included, but does have the debugger symbols) 

then load the SD file(incl the MBR) at position 0 (as the elf positons the app isr vectors after the SD size)

then the setupTarget function (in the ozone project)  gets the sp value from data at 0

and the pc value from position 4

but I get immediate hardfault. 

the process should be the same for 52833 vs 52840, right?  Boot Record is the same

log from the Ozone startup 

Downloading Program.
Exec.Download ("$(ProjectDir)/Downloads/DW3_QM33_SDK_1.0.2.new/SDK/Firmware/BuildOutput/DW3_QM33_SDK_UCI/FreeRTOS/DWM3001CDK/DWM3001CDK-DW3_QM33_SDK_UCI-FreeRTOS.elf");
J-Link: Flash download: Bank 0 @ 0x00000000: 10 ranges affected (266240 bytes)
J-Link: Flash download: Total: 8.561s (Prepare: 0.114s, Compare: 0.222s, Erase: 5.663s, Program & Verify: 2.471s, Restore: 0.089s)
J-Link: Flash download: Program & Verify speed: 104 KB/s
Loading softdevice
Target.LoadMemory ("/home/sam/Downloads/DW3_QM33_SDK_1.0.2.new/SDK/Firmware/SDK_BSP/Nordic/NORDIC_SDK_17_1_0/components/softdevice/s113/hex/s113_nrf52_7.2.0_softdevice.hex", 0x00000000);
J-Link: Flash download: Bank 0 @ 0x00000000: Skipped. Contents already match
Target.ReadU32 (0x00000000); // returns 0x4, data is 0x20000400
Target.SetReg ("SP", 0x20000400);
Target.ReadU32 (0x00000004); // returns 0x4, data is 0xA81
Target.SetReg ("PC", 0xA81);

the LD file for app before SD is 

MEMORY
{
RAM (rwx) : ORIGIN = 0x20000000, LENGTH = 128K
FLASH (rx) : ORIGIN = 0x00000000, LENGTH = 512K - 4K - 4K /* 1M minus calib_sha/calib sections. */
CALIB_SHA (rw) : ORIGIN = 0x0007E000, LENGTH = 4K
CALIB (rw) : ORIGIN = 0x0007F000, LENGTH = 4K
}

SECTIONS
{
.application (0x00000000) :
{
. = ALIGN(4);
KEEP(*(.isr_vector)) /* Startup code */

and after 

MEMORY
{
RAM (rwx) : ORIGIN = 0x20002000, LENGTH = 128K - 8k
FLASH (rx) : ORIGIN = 0x0001E000, LENGTH = 512K - 112K - 4K - 4K /* 512 minus calib_sha/calib sections. */
CALIB_SHA (rw) : ORIGIN = 0x0007E000, LENGTH = 4K
CALIB (rw) : ORIGIN = 0x0007F000, LENGTH = 4K
}

SECTIONS
{
/* start with interrupt vectors */
isr_vectors (ORIGIN(FLASH)) :

I know the ram value isn't correct, I need to see the output of the ble_enable. 

the map shows the app start address at 3D990
and the app vector table loaded at 1E000 shows (flash origin)
00 00 02 20
91 D9 03 00  (03D991)

--------
hardfault info 
@ 0000 0000

top of stack out of range
pc=1AE44

That would be in the SD code space.. top/ending at 1E000 (-8k)= 1C000 (documented size for SD S113)
ozone setup link 
devzone.nordicsemi.com/.../segger-ozone-causes-a-soft-device-error-when-trying-to-debug-or-reset-during-debugging

[rexxdad]

I changed flash to 
FLASH (rx) : ORIGIN = 0x0001C000,

and now my code is entered.

but 
nrf_sdh_enable_request();  returns 0x1001 (already enabled?)

skip that then 
nrf_sdh_ble_default_cfg_set
returns rc=0, ram_start= 0x2002000, matching LD

then 
nrf_sdh_ble_enable(&ram_start);

returns rc=2

confused