#
✕ nick:nickchan
#asahi-dev · 2026-02-27 06:28
nickchani don't know how large my target is but i have 0x40000 iodev_write(IODEV_DOCKCHANNEL_UART, "\n"); in a loop
#asahi-dev · 2026-02-27 06:24
nickchantested some unguarded iodev_write's while not reading on the other side
#asahi-dev · 2026-02-27 06:23
nickchani don't think blocking is an issue
#asahi-dev · 2026-02-27 06:07
nickchansurely enough if i initialize the path to all 0 before calling adt_path_offset_trace then it does print the missing message
#asahi-dev · 2026-02-27 06:01
nickchanso it just used some garbage in path and grabbed the reg of who-knows-what
#asahi-dev · 2026-02-27 06:01
nickchani think it's because there's something on the stack already
#asahi-dev · 2026-02-27 06:00
nickchanokay so in testing m1n1 on a11 doesn't actually print the missing dockchannel uart message on a11
#asahi-dev · 2026-02-27 04:17
nickchangithub allows edit others comment if you have issue management access to a repo
#asahi-dev · 2026-02-27 04:16
nickchanhttps://github.com/AsahiLinux/asahi-installer/issues/83#issue-1199914456 this comment needs to be updated
#asahi-dev · 2026-02-26 16:41
nickchanI mean iodev->can_write
#asahi-dev · 2026-02-26 16:41
nickchanright and that's what iodev->can_free is for
#asahi-dev · 2026-02-26 16:39
nickchanworked well enough-ish for uartproxy
#asahi-dev · 2026-02-26 16:38
nickchanwell "!!! Failed to get dockchannel UART reg property!\n"
#asahi-dev · 2026-02-26 16:38
nickchanat least the adt_path_offset_trace() retval needs to checked to silently return
#asahi-dev · 2026-02-26 16:38
nickchani will leave properly review later but as is this is going to put an annoying message on devices without dockchannel-uart
#asahi-dev · 2026-02-26 16:36
nickchanand of course much slower than 1.5 Mbps uart
#asahi-dev · 2026-02-26 16:36
nickchanthough not by that much
#asahi-dev · 2026-02-26 16:35
nickchanthis is slower than default baud uart
#asahi-dev · 2026-02-26 16:34
nickchanfor a 0x00050000 bytes ADT
#asahi-dev · 2026-02-26 16:34
nickchantake 27 seconds to start uartproxy
#asahi-dev · 2026-02-26 16:30
nickchanthe throughput is low even compard to hardware uart
#asahi-dev · 2026-02-26 15:35
nickchansven: return read32(uart_base + DATA_RX8) >> 8; fixed it
#asahi-dev · 2026-02-26 15:33
nickchanwhich is... number of bytes remaining!
#asahi-dev · 2026-02-26 15:32
nickchanso the bottom 8 bits
#asahi-dev · 2026-02-26 15:32
nickchanwhich would get casted to u8 in the return value
#asahi-dev · 2026-02-26 15:32
nickchanin the dockchannel_uart_getbyte() getbyte function it read32 the DATA_RX8 register
#asahi-dev · 2026-02-26 15:32
nickchanso
#asahi-dev · 2026-02-26 15:20
nickchanby putting a printf right after the poll loop in dockchannel_uart_getbyte() i can see that dockchannel rx is working
#asahi-dev · 2026-02-25 00:59
nickchanmore specifically 'm1n1.proxy.UartTimeout: Expected 1 bytes, got 0 bytes'
#asahi-dev · 2026-02-25 00:54
nickchanpython doesn't receive bytes at all i think
#asahi-dev · 2026-02-25 00:53
nickchansven: seems working but only for output
#asahi-dev · 2026-02-24 09:18
nickchanlikely just a build tag messup
#asahi-dev · 2026-02-24 06:22
nickchanbridgeOS for example is unaffected
#asahi-dev · 2026-02-24 06:21
nickchanon some devices
#asahi-dev · 2026-02-24 00:44
nickchansven: could you pus h the m1n1 dockchannel patches?
#asahi-dev · 2026-02-22 13:59
nickchanah please do push once you are able to then
#asahi-dev · 2026-02-22 13:32
nickchanah right linux has a driver i should take a look at that instead of just looking at sven's messages
#asahi-dev · 2026-02-22 13:30
nickchanand i have the can_write function just returns true
#asahi-dev · 2026-02-22 13:29
nickchanbut for tx i must be dropping lots of bytes when trying to tx because i don't know anything about the tx fifo level
#asahi-dev · 2026-02-22 13:28
nickchanmaking the can_read function cache the rx register and make the getbyte function check the cached value first seems to fixed rx
#asahi-dev · 2026-02-22 13:28
nickchanright, but that implementation isn't available
#asahi-dev · 2026-02-22 13:16
nickchanso it would not be straightforward to implement iodev_can_read
#asahi-dev · 2026-02-22 13:16
nickchantrying to implelement dockchannel iodev, but querying the fifo level will result in draining a byte
#asahi-dev · 2026-02-18 12:03
nickchanby factory build I mean the earliest batch came with 13.5
#asahi-dev · 2026-02-18 12:03
nickchanit was never released over the internet
#asahi-dev · 2026-02-18 12:03
nickchannit: the m3 macbook pro did ran 13.5 as a factory build
#asahi-dev · 2026-02-18 12:02
nickchangood progress report
#asahi-dev · 2026-02-16 05:20
nickchanin macOS 15
#asahi-dev · 2026-02-16 05:19
nickchanalso: M2 Pro/Max/Ultra: No
#asahi-dev · 2026-02-16 05:18
nickchanand then in macOS 26 it's SPTM for all M2 variants and up

plus ›