Hi, isn't it possible that your hal_frame_read be called again prior rf_rx_completion_handler is called, and because of this calling spi_message_init on a spi_message that is in spi queue yet? 2016-07-06 10:25 GMT-03:00 Moti Cohen <motic.mail@gmail.com>:
Hi all,
I'm trying to write a kernel device driver for a Freescale imx6 processor based board.
This driver is using spi to communicate with a radio module (using linux spi.c).
I managed to wr/rd from the module, however once I'm trying to read more than 64 bytes I'm getting coruppted data. here are some details:
The SPI reading process goes as follows:
1. Interrupt function is invoked from SPI device via dedicated GPIO line.
2. The interrupt function performs asynchronous SPI read in the following order:
2.1 radio_trx_end_callback is called, as it is the registered call back function.
2.2 radio_trx_end_callback calls hal_frame_read function which prepares spi_message and spi_transfer structures and fill them with relevant data: spi_device , data tx, transfer len , complete handler , …
3. The read itself is done by calling to function spi_async
4. The completion handler function is called and prints the read message, but trying to read more than 64 bytes in one transaction causes the read data to be corrupted.
Any idea why am I seeing this behavior ?
Is there (in Linux) any definition for the size to be read via the SPI ?
The relevant variables and functions can be seen below:
Thanks in advance for helping..
__________________________
uint8_t rx_buf[200];
uint8_t reg[200] = {HAL_TRX_CMD_FR, 0xff, 0xff};
struct spi_transfer transfer;
struct spi_message spi_msg;
// interrupt function
//-----------------------------------------------------------------------------------------------------------------------
static irq_handler_t radio_trx_end_callback(unsigned int irq, void *dev_id, struct pt_regs *regs)
{
hal_frame_read(irq, dev_id);
return (irq_handler_t) IRQ_HANDLED; // Announce that the IRQ has been handled correctly
}
//read function
//-----------------------------------------------------------------------------------------------------------------------
void hal_frame_read(unsigned int irq, void *dev_id)
{
int status;
spi_device->irq = irq;
spi_message_init(&spi_msg);
spi_msg.spi = spi_device;
spi_msg.complete = rf_rx_completion_handler;
spi_msg.context = NULL;
memset((uint8_t*)&transfer, 0, sizeof(transfer));
memset(rx_buf, 0xFF, sizeof(rx_buf));
memset(reg, 0xFF, sizeof(reg));
reg[0] = HAL_TRX_CMD_FR;
transfer.tx_buf = reg;
transfer.len =64;
transfer.rx_buf = rx_buf;
spi_message_add_tail(&transfer, &spi_msg);
memset(rx_buf, 0x0, sizeof(rx_buf));
status = spi_async(spi_device, &spi_msg);
return;
}
//read completion handler
//-----------------------------------------------------------------------------------------------------------------------
static void rf_rx_completion_handler(void *arg)
{
int ii;
//print the received message
printk("\npure message: rx_buf [1]-2=%02X\n ", rx_buf [1]-2);
for (ii=0; ii< rx_buf [1]-2; ii++)
printk("%02X ", rx_buf [2+ii]);
printk("\nframe len=%X actual len=%X status=%i\n",spi_msg.frame_length, spi_msg.actual_length, spi_msg.status);
send_up_rf_event_to_workque(TRX_END_EVENT);
}
_______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
-- "Do or do not. There is no try" Yoda Master