TDA4VM: The HwiP_disable() function does not work

Hi Xiang,

What SDK are you using? TI SYS/BIOS is no longer supported with the latest TDA4VM SDKs

regards

Suman

Hi Suman,

Thanks for your reply.

The SDK we use is  pdk_jacinto_07_03_00_29,and the version is indeed signtly older.

regards

Xiang

Hi Xiang,

Sys/BIOS has been descoped for a while now, will need some time to look into this.

Can you please share a sample code which I can quickly try to reproduce this issue on my end?

Regards,
Parth

Hi Parth,

Thanks for your reply. 

We modified the file in the SDK, the path and name is /pdk_jacinto_0703_00_29/packages/ti/drv/i2c/src/v1/I2C_v1.c, and the modified sample code is as follows:

uintptr_t gatekey;

/* generate start */
gatekey = HwiP_disable();
I2CMasterStart(hwAttrs->baseAddr);

while ((object->writeCountIdx != 0U) && (timeout != 0))
{
/* wait for transmit ready or error */
while(((I2CMasterIntRawStatusEx(hwAttrs->baseAddr, I2C_INT_TRANSMIT_READY) == 0U) && \
(I2CMasterIntRawStatusEx(hwAttrs->baseAddr, I2C_INT_ARBITRATION_LOST | \
I2C_INT_NO_ACK | \
I2C_INT_ACCESS_ERROR | \
I2C_INT_STOP_CONDITION ) == 0U)) && \
(timeout != 0))
{
I2C_v1_udelay(I2C_DELAY_USEC);
if (I2C_checkTimeout(&uSecTimeout))
{
timeout--;
}
}

errStat = I2CMasterIntRawStatusEx(hwAttrs->baseAddr, I2C_INT_ARBITRATION_LOST | \
I2C_INT_NO_ACK | \
I2C_INT_ACCESS_ERROR);

/* if we get an error, do a stop and return failure */
if (errStat != 0U)
/* if we get an error, do a stop and return failure */
{
fatalError = 1U;
break;
}
/* write byte and increase data pointer to next byte */
I2CMasterDataPut(hwAttrs->baseAddr, *(object->writeBufIdx));
(object->writeBufIdx)++;

/* clear transmit ready interrupt */
I2CMasterIntClearEx(hwAttrs->baseAddr, I2C_INT_TRANSMIT_READY);

/* update number of bytes written */
object->writeCountIdx--;
}
HwiP_restore(gatekey);

The above is where we have made changes, and the full document is attached.3302.I2C_v1.c

Regards,

Xiang Gao

Hi Xiang,

Thanks for the details, will try this once and provide the feedback

Regards,
Parth

Hi Xiang,

Looked into this, the changes seems fine. I see it being used at multiple places in SDK.
Can you please share the details on how you are validating this? 

Xiang Gao said:

In addition, it is possible to close a specific interrupt by calling the function HwiP_disableInterrupt (int32_t interruptNum).

Also, yes, this should be possible to disable specific interrupts using this API

Regards,
Parth

Hi Parth,

Our verification steps are as follows:

1. Add a test variable in the interrupt function of CAN communication, which will increase 1 when receiving a CAN message;

2.Add a delay in the critical section code, and the delay time is two cycles of CAN message;

3.Print out the test variable when entering and exiting critical section code;

We expect that all interrupts should be closed after executing the HwiP_disable() function to enters the critical section code, and the test variable will not increase, so the printed result should be the same. But in the actual test, whether the HwiP_disable() function is executed or not, we get the same result. The test variable still increase after we execute the HwiP_disable() function. So we think HwiP_disable() function does not work and the global interrupt cannot be closed.

Regards,

Xiang Gao

Hi Parth,

Please ask if you have verified the steps I provided? And whether you have obtained the same results as me?We urgently need to address this issue as soon as possible.

Hope to reply soon,thanks.

Regards,

Xiang Gao

Hi Xiang & Parth,

I have am facing a very similar issue with HwiP_disable() and HwiP_restore() functions. I have registered a callback function for a hardware interrupt using the function HwiP_create(). After disabling the interrupt I tried to generate a hardware interrupt using HwiP_post(). But I am still able to go to the callback function. Have you been able to find the reason or a potential solution for it? 

Regards

Rajaram

Hi Xiang, Rajaram,

I believe this issue was resolved in another discussion.

In a short summary, please refer to the osal test in the SDK for reference. We can see the interrupts getting disabled there after HwiP_disable.

Regards,
Parth