[B96B_F446VE] New platform + development of dma_asynch_serial + hwControlFlow #1470

adustm · 2015-12-11T13:03:58Z

B96B F446VE is a new STM32 platform.
Asynch serial has been implemented using mbed interface (using DEVICE_SERIAL_ASYNCH).
A new define was added : DEVICE_SERIAL_ASYNCH_DMA 1 to force the use of DMA transfer for the async serial. define it to 0 if you want to use interrupts only instead of DMA.
Serial asynch can be extended to every STM32F4 family.

Serial HW Control Flow is supported for this platform (using DEVICE_SERIAL_FC define)

…_F4_asynch_serial

…to dev_F4_asynch_serial

…_F4_asynch_serial

Need to change the SerialBase.cpp api, in order to provide the callback address to the DMA interrupt. Simple test works with USBTX and USBRX pins. To be further tested

…to dev_F4_asynch_serial

…_F4_asynch_serial

Different handler for dma rx interrupt, compatible with mbed api

…os tests

adustm · 2015-12-17T10:17:38Z

		DTCT_1	EXAMPLE_1	MBED_10	MBED_11	MBED_12	MBED_16	MBED_2	MBED_23	MBED_24	MBED_25	MBED_26	MBED_34	MBED_37	MBED_38	MBED_A1	MBED_A21	MBED_A9	MBED_BUSOUT	RTOS_1	RTOS_2	RTOS_3	RTOS_4	RTOS_5	RTOS_6	RTOS_7	RTOS_8
ARM	B96B_F446VE	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK
GCC_ARM	B96B_F446VE	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK
IAR	B96B_F446VE	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK
uARM	B96B_F446VE	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	FAIL	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK	OK

adustm · 2015-12-17T10:17:53Z

MBED_16 is failing due to a modification done recently in mbed micromaster (to be investigated and fixed separately)
MBED_37 is failing on the complete F4 family, but it is not considered as blocking.

adustm · 2015-12-17T10:24:04Z

Please find hereafter the main I used to validate the serial

#include "mbed.h"

#define TEST_HWFLOWCONTROL 1
#define TRANSMITTER_BOARD
/------------ Constant definitions --------------/
#define TOGGLE_RATE (0.5f)
/* Exported macro ------------------------------------------------------------/
#define COUNTOF(BUFFER) (sizeof(BUFFER) / sizeof((BUFFER)))
/-------- Check if platform compatible ----------/
//Serial test_connection(USBTX, USBRX); // tested ok. UART4, via ST-LINK and VirtualComPort
//Serial test_connection(PE_8, PE_7); // not tested, solder JP50 and JP51, unsolder JP48 JP49
//Serial test_connection(PC_12, PD_2); // tested ok. UART5: connector Arduino, pins D4 & D14
//Serial test_connection(PD_8, PD_9); // tested ok. UART3: use connector J10, pins 3 & 4
//Serial test_connection(PA_2, PA_3); // tested ok. UART2: use connector Arduino_A1 and J3 pin3
//Serial test_connection(PD_5, PD_6); // tested ok. UART2: use connector Arduino_D1 and Arduino_D0
//Serial test_connection(PA_9, PA_10); // tested ok. UART1: PA_9 is on JP71, PA_10 is on JP72 (JP72 to be unsoldered)

/------------------ Variables -------------------/
//LowPowerTicker blinker;
event_callback_t serialTXEventCb;
event_callback_t serialRXEventCb;
DigitalOut led2(LED2);
DigitalOut led3(LED3);
DigitalOut led4(LED4);
DigitalIn button(PD_13,PullUp);
//uint8_t aTxBuffer[2] = {0xAA, 0xAA};
uint8_t aTxBuffer[] = " *_UART_TwoBoards communication based on DMA* *_UART_TwoBoards communication based on DMA* _UART_TwoBoards communication based on DMA_* ";
/* Size of Transmission buffer /
#define TXBUFFERSIZE (COUNTOF(aTxBuffer) - 1)
/ Size of Reception buffer /
#define RXBUFFERSIZE TXBUFFERSIZE
uint8_t rx_buf[RXBUFFERSIZE];
volatile uint32_t tx_complete = 0;
volatile uint32_t rx_complete = 0;
volatile uint32_t rx_char_received = 0;
/------------------ Callbacks -------------------/

/**

This is a callback! Do not call frequency-dependent operations here.
*/
void serialTxCb(int events) {
/ handle TX callback */
if (events & SERIAL_EVENT_TX_COMPLETE) {
tx_complete = 1;
led4 = 1;
}
}
void serialRxCb(int events) {
unsigned char i;

if(events & SERIAL_EVENT_RX_CHARACTER_MATCH) {
rx_char_received = 1;
rx_complete = 1;
//Received 'S', check for buffer length
for(i = 0; i < RXBUFFERSIZE; i++) {
//Found the length!
if(rx_buf[i] == 'S') break;
}
led3 = 1;
} else if (events & SERIAL_EVENT_RX_COMPLETE) {
i = RXBUFFERSIZE - 1;
rx_complete = 1;
led3 = 1;
} else {
rx_buf[0] = 'E';
rx_buf[1] = 'R';
rx_buf[2] = 'R';
rx_buf[3] = '!';
rx_buf[4] = 0;
i = 3;
led2 = 1;
}
}

/**

@brief Compares two buffers.
@param pBuffer1, pBuffer2: buffers to be compared.
@param BufferLength: buffer's length
@RetVal 0 : pBuffer1 identical to pBuffer2
```
    >0 : pBuffer1 differs from pBuffer2
```
/
static uint16_t Buffercmp(uint8_t pBuffer1, uint8_t* pBuffer2, uint16_t BufferLength)
{
while (BufferLength--)
{
if ((*pBuffer1) != *pBuffer2)
{
return BufferLength;
}
pBuffer1++;
pBuffer2++;
}

return 0;
}

/-------------------- Main ----------------------/
int main() {
Serial test_connection(PA_9, PA_10);
int set_event = 0;
led2 = 0; led3 = 0; led4= 0;

/* Setup serial connection */
test_connection.baud(9600);

#if TEST_HWFLOWCONTROL
/* Use set_flow_control /
test_connection.set_flow_control((mbed::SerialBase::Flow)3, PA_12, PA_11);
#endif
/ Very Simple Main (tm) */
serialTXEventCb.attach(serialTxCb);
serialRXEventCb.attach(serialRxCb);

#ifdef TRANSMITTER_BOARD
while (button.read()!=RESET){
}
test_connection.write(aTxBuffer, TXBUFFERSIZE, serialTXEventCb, SERIAL_EVENT_TX_COMPLETE);

while (tx_complete==0){
  // wait there for the write completion. To be removed in case the application wants to work on something else
}
tx_complete = 0;

set_event = SERIAL_EVENT_RX_OVERFLOW | SERIAL_EVENT_RX_PARITY_ERROR | SERIAL_EVENT_RX_FRAMING_ERROR | SERIAL_EVENT_RX_OVERRUN_ERROR | SERIAL_EVENT_RX_COMPLETE;
test_connection.read(rx_buf, RXBUFFERSIZE, serialRxCb, set_event, SERIAL_RESERVED_CHAR_MATCH);

#endif
while (rx_complete==0){
// wait there for the write completion. To be removed in case the application wants to work on something else
}
rx_complete = 0;

/*## Compare the sent and received buffers ##############################*/
if(Buffercmp((uint8_t*)aTxBuffer,(uint8_t*)rx_buf,RXBUFFERSIZE))
{
  while (1){}
}

while(1) {
  led3=0; 
  wait(0.2);
  led3=1;
  wait(0.2);
}

}

0xc0170 · 2016-01-04T13:24:07Z

libraries/mbed/targets/hal/TARGET_STM/TARGET_STM32F4/serial_api.c

-    obj->databits = UART_WORDLENGTH_8B;
-    obj->stopbits = UART_STOPBITS_1;
-    obj->parity   = UART_PARITY_NONE;
+    _SERIAL_OBJ(baudrate) = 9600;


I would not start a macro with underscore followed by an uppercase letter (it's reserved in any scope , see standard C/C++).

This macro is needed now as not all F4 are up to date to enable this feature ? Because this should not be needed if serial object is same for all F4 , it should be the same?

Hi, I'll rename the macro and check your other comments below.
You're right, this is needed as long as this feature is not deployed on every F4.
Cheers

0xc0170 · 2016-01-04T13:32:25Z

There're unit test for asynch, please have a look at : https://github.com/mbedmicro/mbed/tree/master/libraries/tests/utest . I would like to see result from serial one.

If you can align the coding style to the one which is already in the file (should follow https://developer.mbed.org/teams/SDK-Development/wiki/mbed-sdk-coding-style)

0xc0170 · 2016-01-04T13:43:51Z

libraries/mbed/targets/hal/TARGET_STM/TARGET_STM32F4/serial_api.c

+uint8_t serial_tx_active(serial_t *obj)
+{
+    MBED_ASSERT(obj);
+    MBED_ASSERT(HAL_UART_STATE_BUSY_TX==0x12);


what is this assert for? Same is for BUSY_RX == 0x22 ?

0xc0170 · 2016-01-04T13:49:02Z

This would be easier to integrate if we could split this PR to smaller pieces:

adding new target to the current code base
adding flow control
add serial asynch support

Shall we do that? Sorry for taking longer to review this, but it is quite large changeset which came in before holidays :-)

adustm · 2016-01-04T16:38:31Z

Hello, is it possible to handle several PRs at several gates of development in the same branch ? How is your suggestion of several PRs usually handled ?

0xc0170 · 2016-01-05T17:09:06Z

@adustm You can do cherry pick to get only those commits adding the platform.. I'll give it a shot. It was just a suggestion for this integration to be smoother/faster.

Can you run those unit tests for this platform for serial? I shared above the links : #1470 (comment)

0xc0170 · 2016-01-06T14:59:37Z

The target was added, please send PR with adding other 2 features. I'll close this one.

adustm added 20 commits November 17, 2015 17:41

Add asyn serial ability

b5910de

Merge remote-tracking branch 'refs/remotes/mbedmicro/master' into dev…

760934d

…_F4_asynch_serial

commit latest modifications in serial async (TX Interrupt ok )

8645c17

[NUCLEO_F446RE] async serial. allow rx part

52845b8

[B96B_F446VE] Add new target

7d6c91c

Merge remote-tracking branch 'refs/remotes/origin/dev_B96B_F446VE' in…

073d91a

…to dev_F4_asynch_serial

[B96B_F446VE] add DMA to serial_asyn (not finalized)

058a95c

Merge remote-tracking branch 'refs/remotes/mbedmicro/master' into dev…

d58d597

…_F4_asynch_serial

[B96B_F446VE] Add DMA part to SERIAL_ASYNCH (OK with USBTX USBRX pins)

969b93d

Need to change the SerialBase.cpp api, in order to provide the callback address to the DMA interrupt. Simple test works with USBTX and USBRX pins. To be further tested

[B96B_F446VE] add new platform to builders scripts

95fb27a

Merge remote-tracking branch 'refs/remotes/origin/dev_B96B_F446VE' in…

1842001

…to dev_F4_asynch_serial

Merge remote-tracking branch 'refs/remotes/mbedmicro/master' into dev…

473d350

…_F4_asynch_serial

[B96B_F446VE] Add other UARTs async support

3c420c0

[B96B_F446VE] Add HWControlFlow option

3763d53

Merge branch 'master' into dev_F4_asynch_serial

809bbbe

[B96B_F449VE] better init for hw control flow

9655279

[B96B_F446VE] Revert mbed api modification to be compliant with mbed

9a3163e

Different handler for dma rx interrupt, compatible with mbed api

[B96B_F446VE] change led names to be aligned with board's naming

1f88744

[B96B_F446VE] Add define for F446VE in rtos library files to allow rt…

741e7c7

…os tests

[B96B_F446VE] add exporter files

e4fb268

adustm mentioned this pull request Dec 17, 2015

B96B_F446VE + NUCLEO_F446RE CAN API #1477

Closed

0xc0170 reviewed Jan 4, 2016
View reviewed changes

0xc0170 added the needs: work label Jan 4, 2016

0xc0170 added the needs: review label Jan 4, 2016

0xc0170 mentioned this pull request Jan 6, 2016

Add target B96B_F446VE files #1489

Merged

0xc0170 closed this Jan 6, 2016

adustm deleted the dev_F4_asynch_serial branch February 29, 2016 15:50

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[B96B_F446VE] New platform + development of dma_asynch_serial + hwControlFlow #1470

[B96B_F446VE] New platform + development of dma_asynch_serial + hwControlFlow #1470

Uh oh!

adustm commented Dec 11, 2015

Uh oh!

adustm commented Dec 17, 2015

Uh oh!

adustm commented Dec 17, 2015

Uh oh!

adustm commented Dec 17, 2015

Uh oh!

0xc0170 Jan 4, 2016

Uh oh!

adustm Jan 4, 2016

Uh oh!

0xc0170 commented Jan 4, 2016

Uh oh!

0xc0170 Jan 4, 2016

Uh oh!

0xc0170 commented Jan 4, 2016

Uh oh!

adustm commented Jan 4, 2016

Uh oh!

0xc0170 commented Jan 5, 2016

Uh oh!

0xc0170 commented Jan 6, 2016

Uh oh!

Uh oh!

[B96B_F446VE] New platform + development of dma_asynch_serial + hwControlFlow #1470

[B96B_F446VE] New platform + development of dma_asynch_serial + hwControlFlow #1470

Uh oh!

Conversation

adustm commented Dec 11, 2015

Uh oh!

adustm commented Dec 17, 2015

Uh oh!

adustm commented Dec 17, 2015

Uh oh!

adustm commented Dec 17, 2015

Uh oh!

0xc0170 Jan 4, 2016

Choose a reason for hiding this comment

Uh oh!

adustm Jan 4, 2016

Choose a reason for hiding this comment

Uh oh!

0xc0170 commented Jan 4, 2016

Uh oh!

0xc0170 Jan 4, 2016

Choose a reason for hiding this comment

Uh oh!

0xc0170 commented Jan 4, 2016

Uh oh!

adustm commented Jan 4, 2016

Uh oh!

0xc0170 commented Jan 5, 2016

Uh oh!

0xc0170 commented Jan 6, 2016

Uh oh!

Uh oh!