David Lechner

> What do you think about committing FW images for some common platforms? Sounds like a job for a separate repo.

Did the OS prompt for any kind of authentication? It is not possible to diagnose further without additional logs.

Fixing this is likely going to break things that depend on this buggy behavoir. For example, in the `ad_pulsar` driver. https://github.com/analogdevicesinc/linux/blob/b8d0e51373d107a62a364ec7c7abce9cad6ba7cd/drivers/iio/adc/ad_pulsar.c#L359-L382 For a single conversion (reading the `_raw` sysfs attribute),...

It also looks like there is potentially another bug here. `j` is written but never used. I'm guessing it is supposed to be `writel(buf[i], sdo_addr + j); `

Also, if there are multiple xfers, e.g. of len=2, this would make the data in `sdo_addr` not contiguous between xfers. Is this going to cause a problem? I'm guessing so...

> Ideally, we would have a new specific controller op for `spi_engine_offload_load_msg()` and the API calling that op, should run `__spi_validate()` on the message. Agreed. I came to the same...

I think it is fine in the spinlock since the timer should have the `TIMER_IRQSAFE` flag. But if you want to be consistent with the upstream driver, I just put...

> Mac OS 10.11 Hmm... this could be the issue. I'm not able to test Bleak on older versions of macOS. Looks like similar reports here: https://github.com/ronaldoussoren/pyobjc/issues/220

See https://www.ev3dev.org/docs/tutorials/using-docker-to-cross-compile/ for the recommended way to cross-compile for ev3dev.

Alternately, you can statically link your C++ program and it should work.