drivers: mss_ethernet_mac: various compilation fixes

[fcuzzocrea]

No description provided.

[nitindeshpande]

Thanks for the PR.
As mentioned in issue#5, the compilation errors due to folder structure change have been fixed. There has been a delay in publishing the change. We are hoping to get this out by next week.

[petermcs-01]

On the critical nesting point, I'd note that the portENTER/EXIT functions are not supposed to be called from ISR. I can't find the reference that caused me to put the updates to the nesting level in at the moment and I'm not 100% sure they are needed...
It may relate to the use of xSemaphoreGiveFromISR() in the LwIP mac device driver.

[fcuzzocrea]

@petermcs-01 I think you are right. I just successfully tried an external loop-back configuration in phy-less mode between mac0 and mac1, and the portENTER/EXIT functions we making the sofware go kaput when the bridge between mac0 and mac1 was in place. Removed the functions, and it worked.

Do you have any recommendation on how to implement a proper fix for this, or we just nuke it from the driver ?

Thanks!

[petermcs-01]

All the testing I've done to date over the last 4 years has been with the original adjustment of the nesting count and I'd be inclined to leave that in the driver for now until someone has tested the code properly with it removed....

[fcuzzocrea]

Alright! Which kinds of tests needs to be performed? I can help with that.

So far, the only configuration I tried is null phy loopback connection between mac0 and mac1. I wasn't able to perform the same kind of test on the ICICLE kit with an ethernet crossed cable because for some reason, rx code always fails.

Also, how can I export the nesting count variable from FreeRTOS to be available to the driver?

Thanks

[fcuzzocrea]

I saw that the #include were fixed with the 2022.08 update, but what about the direct update of the uxCriticalNesting variable in the driver?