Avoid DMA channel/PIO SM conflicts from scanvideo/audio #801
Conversation
audio setup claims the channel passed in so we claim an unused channel/sm, then immediately unclaim it
|
This might get less nasty, depending on what happens with the pico-extras PR I just opened... |
|
IMO Pico Extras looks... neglected? I'm tempted to just vendor the i2s driver and fix it downstream until we get something that's not in i2s purgatory. Not that we should rush to reinvent the wheel, but I'm pretty sure I've had use cases where all the buffer pool handling is just unnecessary bloat. This might even be one of those cases- IIRC the STM32 32blit port just uses a single buffer with a half complete callback. |
|
Yeah, could definitely write a simplified i2s driver (and save some RAM). Same probably applies for scanvideo, we're not using half the features and the thing we do need (mode switching, or at least the scale) isn't implemented... Guess that's another thing to look at at some point. The STM32 port... doesn't even have a buffer... (#568) |
ಠ_ಠ I had forgotten this little factoid 😬 |
This is a bit of a hack due to pico-extras not using
*_claim_unused_*(scanvideo using hardcoded numbers). Avoids more hardcoding/hacks in the SD card and DVI code.