Xuartlite standalone drivers misses the errors

[raczben]

The driver reads the Xuartlite's status register more times (like here), but it does not process the error values (overrun, framing, parity) in it just in one place (at the end of the XUartLite_ReceiveBuffer function). And since the errors are "clear-on-read" type bits, there is no chance to get the errors using the drivers.

I have started a thread in this topic on Xilinx forum as well.

[RavitejaNr]

This patch is not needed.

This patch is adding a new wrapper function to read the status register. The new function “XUartLite_GetStatusReg2” is reading the status register and also calling “XUartLite_UpdateStats” function which is updating the “Rx errors in stats” (Nothing related to Tx) structure in the Instance (refer below). So, there is no need to call "XUartLite_UpdateStats" in the Tx path. Please find the patch and comments in line.

#define XUartLite_UpdateStats(InstancePtr, StatusRegister)
{
if ((StatusRegister) & XUL_SR_OVERRUN_ERROR)
{
(InstancePtr)->Stats.ReceiveOverrunErrors++;
}
if ((StatusRegister) & XUL_SR_PARITY_ERROR)
{
(InstancePtr)->Stats.ReceiveParityErrors++;
}
if ((StatusRegister) & XUL_SR_FRAMING_ERROR)
{
(InstancePtr)->Stats.ReceiveFramingErrors++;
}
}

fix: xuartlite standalone drivers misses the errors

   StatusRegister = XUartLite_GetStatusReg(InstancePtr->RegBaseAddress);
   StatusRegister = XUartLite_GetStatusReg2(InstancePtr);

*There is no need to update the Rx errors in stats everytime status register is read

    XUartLite_WriteReg(InstancePtr->RegBaseAddress,
                          XUL_CONTROL_REG_OFFSET, 0);

@@ -480,13 +503,13 @@ unsigned int XUartLite_SendBuffer(XUartLite InstancePtr)
/
* Read the status register to determine if the transmitter is full
*/
StatusRegister = XUartLite_GetStatusReg(InstancePtr->RegBaseAddress);
StatusRegister = XUartLite_GetStatusReg2(InstancePtr);

@@ -511,7 +534,7 @@ unsigned int XUartLite_SendBuffer(XUartLite *InstancePtr)
SentCount++;

           StatusRegister =
                  XUartLite_GetStatusReg(InstancePtr->RegBaseAddress);
                 XUartLite_GetStatusReg2(InstancePtr);

*The patch updates Rx error stats even in Tx path (Like Sendbuffer function) which is unnecessary.

   StatusRegisterVal = XUartLite_GetStatusReg(InstancePtr->RegBaseAddress);
   StatusRegisterVal = XUartLite_GetStatusReg2(InstancePtr);
   XUartLite_WriteReg(InstancePtr->RegBaseAddress,
                          XUL_CONTROL_REG_OFFSET, 0);

          StatusRegister =
                  XUartLite_GetStatusReg(InstancePtr->RegBaseAddress);
                 XUartLite_GetStatusReg2(InstancePtr);

           /*
           * If there is data ready to be removed, then put the next byte
                 XUartLite_UpdateStats(InstancePtr, StatusRegister);

*We can update the Rx errors stats only after receiving any data (which is already taken care in the existing driver above).

[raczben]

Hi @RavitejaNr

First of all, can you edit your post? Please use the code markdown feature, to improve readability.

This patch is needed if you want to detect all errors. This patch is needed in both TX and RX paths. Yes, it is needed in TX. The UART itself is full-duplex, asynchronous communication. This means that bytes can arrive (on the RX side), while we send a message (on the TX side).* Therefore any RX errors can occur at any point in time, during transmitting data as well.

*Yes, this (full-duplex behavior) can be disabled from the higher level of the communication (ie. in higher OSI layers), but the UART itself cannot assume this limitation, and the UART itself must support its full-duplex behavior.

[RavitejaNr]

Hello @raczben
Edited the previous post.
Yes, you are correct. The driver should not assume half duplex. There are indeed some errors missing in the existing driver.

[RavitejaNr]

Can we have a more descriptive name for the new function(XUartLite_GetStatusReg2) like
"XUartLite_Update_GetStatus" or something which gives a hint of what we are doing inside?

[raczben]

Yes, feel free to rename it, I just shared mine quick fix.
I will be happy, this issue will be solved in any way.

[raczben]

Hi,

The issue still exists in the latest release.

[vsrik]

Hi - We somehow could not get into the latest release. We will get it into the next release , we will get an Answer Record Created.