kk-thrane commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-2137655606
Yes, as @slorquet wrote. Sorry for not making that clear.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
davids5 commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-2137086065
@slorquet Yes https://github.com/apache/nuttx/pull/11990
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL
slorquet commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-2136754566
Why was this closed without commenting? Were any changes applied to upstream?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
kk-thrane closed pull request #11871: STM32H7 serial TX DMA fix
URL: https://github.com/apache/nuttx/pull/11871
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe,
davids5 commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1988110803
@kk-thrane
I am glad you went through the `nxsem_trywait` and see the contrast now.
If you consider the foregrounds looping trying to transmit. I have concerns
about the
kk-thrane commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1987826207
To my surprise, nuttx implements nxsem_trywait() using a critical section...
so yes, using a critical section instead of the semaphore provides better
performance.
Hasn't nuttx got
davids5 commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1987236662
If the only change required is fixing the typo by adding the "== 0" why not
do that? I think the CS is a few uS and a nit compared to the semaphore code
time and complexity.
--
This
kk-thrane commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1986985589
The current software in master is not working, because the test in
up_dma_txavailable() does not correctly determine whether TX DMA is active or
not. This could be fixed, but isn't it
kk-thrane commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1985624357
The half transfer interrupt is not enabled (for TX), only the completion
interrupt is enabled.
I had forgot to remove that. I have pushed version where the half transfer
davids5 commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1985595771
yes.
I am not clear on the flow, here is my concern
[Half done
kk-thrane commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1985543808
The semaphore is used to indicate when DMA is idle or not. The call to
nxsem_trywait() in up_dma_txavailable() perform this check.
Did you notice the reverts? The revert commits
davids5 commented on PR #11871:
URL: https://github.com/apache/nuttx/pull/11871#issuecomment-1985488360
How is this checking for the DMA idle?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to
kk-thrane opened a new pull request, #11871:
URL: https://github.com/apache/nuttx/pull/11871
## Summary
Fixes TX DMA for STM32H7
## Testing
Works for me on an STM32H723
--
This is an automated message from the Apache Git Service.
To respond to the message, please log
13 matches
Mail list logo