bugfix: STM32cubeMX uses uninitialized memory for LL_USART_Init

Posted on July 11, 2018 at 19:49

Sorry for not providing this information in the first place:

STM32CubeF0 Firmware Package V1.9.0 / 25-August-2017

STMcubeMX V1.0 Version 1.0

adding the missing HardwareFlowControl entries to

'mcu/config/llConfig/USART-STM32F0xx_LLConfigs.xml'

fixed the problem (see attachment patch.txt). But this is not a complete fix! I only patched a few obviously missing entries in the supplied version and only for this mcu.

issue was in Src/usart.c:

...

<quote>

void MX_USART1_UART_Init(void)

{

// <<<<<<<<<<< The following struct is left uninitalized >>>>>>>>>>>>>>>

LL_USART_InitTypeDef USART_InitStruct;

LL_GPIO_InitTypeDef GPIO_InitStruct;

/* Peripheral clock enable */

LL_APB1_GRP2_EnableClock(LL_APB1_GRP2_PERIPH_USART1);

/**USART1 GPIO Configuration

PA2 ------> USART1_TX

*/

GPIO_InitStruct.Pin = LL_GPIO_PIN_2;

GPIO_InitStruct.Mode = LL_GPIO_MODE_ALTERNATE;

GPIO_InitStruct.Speed = LL_GPIO_SPEED_FREQ_HIGH;

GPIO_InitStruct.OutputType = LL_GPIO_OUTPUT_OPENDRAIN;

GPIO_InitStruct.Pull = LL_GPIO_PULL_UP;

GPIO_InitStruct.Alternate = LL_GPIO_AF_1;

LL_GPIO_Init(GPIOA, &GPIO_InitStruct);

/* USART1 interrupt Init */

NVIC_SetPriority(USART1_IRQn, 1);

NVIC_EnableIRQ(USART1_IRQn);

USART_InitStruct.BaudRate = 38400;

USART_InitStruct.DataWidth = LL_USART_DATAWIDTH_8B;

USART_InitStruct.StopBits = LL_USART_STOPBITS_1;

USART_InitStruct.Parity = LL_USART_PARITY_NONE;

USART_InitStruct.TransferDirection = LL_USART_DIRECTION_TX_RX;

>>>>>> USART_InitStruct.HardwareFlowControl = LL_USART_HWCONTROL_NONE; <<<<<< // THIS LINE WAS MISSING

USART_InitStruct.OverSampling = LL_USART_OVERSAMPLING_16;

LL_USART_Init(USART1, &USART_InitStruct);

LL_USART_ConfigHalfDuplexMode(USART1);

LL_USART_Enable(USART1);

}

</quote>

originally the marked line was missing leaving the 'USART_InitStruct.HardwareFlowControl'

entry uninitialized. this eventually led 'USARTx->CR3' to be modified through a

WRITE_REG((REG), (((READ_REG(REG)) & (~(CLEARMASK))) | (SETMASK)))

where CLEARMASK was (USART_CR3_RTSE | USART_CR3_CTSE)

but SETMASK contained random stack-values and in my case the DMA bits had been asserted, swamping my I2C tx register because USART1 & I2C1 share the same DMA channel....

bugs of this kind are very hard to debug, so please evaluate all other XYZ_InitStruct for complete initialization for all combinations of MCU's.

thanks for taking care of this issue

regard

ingo

ps. an email reply bounced....

________________

Attachments :

MSB2optima.ioc.zip : https://st--c.eu10.content.force.com/sfc/dist/version/download/?oid=00Db0000000YtG6&ids=0680X000006HxQh&d=%2Fa%2F0X0000000axv%2FsokmivXZN4SePbpml4KfMpMbfXoJqwwjOuv2lXJhuEg&asPdf=false

patch.txt.zip : https://st--c.eu10.content.force.com/sfc/dist/version/download/?oid=00Db0000000YtG6&ids=0680X000006HxQc&d=%2Fa%2F0X0000000axu%2F4x16dl2nA.ZhmAqHynuxQFuwMYI3.CcqPDZUBQry7_c&asPdf=false