Andrew Gordon

> > (btw, I just noticed the SDK doesn't even build on FreeBSD because it tries to pull in system C++ headers that get very confused by the embedded environment...

In addition to the full-blown customisation suggested above, may I also suggest a simple mechanism for the board file to specify a different crystal frequency? At the moment it's all...

> Can you not already set that by defining `XOSC_MHZ`? It appears to accept a range 1-50. Unfortunately not: 1) XOSC_MHZ is hard-defined so you can't override it (rather than...

Yes, this is a really evil bug in the existing implementation of stdio_set_chars_available_callback() for USB. Yet another report of it on the forums yesterday, triggered by the pico-examples [pico-examples/pico_w/wifi/iperf/picow_iperf.c](https://github.com/raspberrypi/pico-examples/blob/eca13acf57916a0bd5961028314006983894fc84/pico_w/wifi/iperf/picow_iperf.c#L50) which...

True it's really a stdio_usb bug, but here it could trivially be fixed to make the example work with the existing stdio_usb. I've commented over on https://github.com/raspberrypi/pico-examples/issues/1603 with a suggested...

I've proposed a fix over on the other issue: https://github.com/raspberrypi/pico-sdk/issues/1603#issuecomment-1963941166

It's all using SDK headers, and doesn't behave like a volatile problem. I disassembled the compiler output and it looked plausible. It works fine with the DMB instruction in there,...

OK, I tried it with polling FSTAT instead of INTS0 and it worked without the DMB, but this isn't really meaningful since the loop test now has to get much...

I think the difference is you don't have the pio->txf[SM_NO] = b; right at the foot of your loops (and you haven't taken advantage of the optimisation that if you...

OK, here you go. ``` diff --git a/pio/irq_test/irq_test.c b/pio/irq_test/irq_test.c index 8545b78..2143e5a 100644 --- a/pio/irq_test/irq_test.c +++ b/pio/irq_test/irq_test.c @@ -24,6 +24,7 @@ enum type { IRQ_A, IRQ_B, IRQ_C, + IRQ_D, LAST_TYPE, };...