13:07 <narmstrong> bpi-m5 is failing to boot on linux-next, bisecting with the board in my lab!!!

13:08 <narmstrong> weird it's only SM1, and very early

19:34 <f_> narmstrong: may I pm?

23:46 <xdarklight> montjoie: narmstrong: I did some more debugging with CH348 and it's actually the *TX* path that makes the 2048 byte case fail. I found a way to make things much simpler, unfortunately it's not stable yet: https://github.com/xdarklight/ch348/blob/main/ch348.c#L333-L355 (I'm passing max bulk_out_size - TX header size to usb_serial_generic_write, then I'm waiting for the TX done status. we can do this regardless of the requested count as long as we

23:46 <xdarklight> return the number of bytes that we have transferred. TTY core will take care of calling .write() again with the remaining bytes that we did not transmit). sometimes it works perfectly, other times it causes the 1024 and 2048 byte cases to fail as it doesn't seem to transmit all data. but that's a story for another time, good night