drivers: display: hub12: add horizontal chaining support #98445
Conversation
heronet
commented
Oct 28, 2025
heronet
commented
- Add support for chaining multiple 32x16 panels horizontally by setting width to multiples of 32 (64, 96, 128, etc).
- Fix a minor race condition issue.
zephyrbot
requested review from
JarmouniA,
VynDragon,
danieldegrasse,
decsny,
jfischer-no,
mbolivar and
rruuaanng
drivers/display/display_hub12.c
Outdated
| } | ||
|
|
||
| k_sem_take(&data->lock, K_FOREVER); | ||
| k_mutex_lock(&data->lock, K_FOREVER); |
There was a problem hiding this comment.
Mutex seemed like a better fit since we're protecting shared data access. Also helps if multiple threads end up calling display_write at different priorities
There was a problem hiding this comment.
It appears that the submitter of this document is you. I'm curious why your first choice wasn't a mutex, and now you're back to modify it again (just asking).
There was a problem hiding this comment.
Mostly a matter of preference actually. The initial implementation with a semaphore worked fine, and this patch primarily adds display chaining. While updating, I realized that a semaphore isn’t strictly necessary here, and using a mutex could also help with priority inversion. So I replaced the semaphore with a mutex.
There was a problem hiding this comment.
Expect for priority inversion mitigation, a semaphore is lighter than & recommended over a mutex in Zephyr. A mutex makes the function unusable from an ISR.
Also calling display_write from multiple threads would be a bad practice in my opinion.
https://academy.nordicsemi.com/courses/nrf-connect-sdk-fundamentals/lessons/lesson-8-thread-synchronization/topic/mutexes/
https://academy.nordicsemi.com/courses/nrf-connect-sdk-fundamentals/lessons/lesson-8-thread-synchronization/topic/semaphores/
There was a problem hiding this comment.
Yeah, it’s a trade-off. Do you think it would be better if I reverted to semaphores? I’m fine with that.
Edit: I went for semaphore as it is more performant and recommended than mutex.
- Add support for chaining multiple 32x16 panels horizontally by setting width to multiples of 32 (64, 96, 128, etc). Signed-off-by: Siratul Islam <email@sirat.me>
- Add support for chaining multiple 32x16 panels horizontally by setting width to multiples of 32 (64, 96, 128, etc). - Also fixed a race condition issue. Signed-off-by: Siratul Islam <email@sirat.me>
heronet
force-pushed
the
drivers/display/update-hub12-matrix-x
branch
from
69c4ee3 to
6fec537
Compare
|
5 participants
