#asahi-dev · 2022-01-17 20:13
svenI’m just glad that DCP isn’t my problem ;)#asahi-dev · 2022-01-17 20:07
sveni guess it's just the usb curse again#asahi-dev · 2022-01-17 19:53
svenhuh, no, I don’t think I ever saw that#asahi-dev · 2022-01-17 16:50
svenjannau: it should just reinitialize that screen then, shouldn't it?#asahi-dev · 2022-01-17 16:49
svenit'll still crash DCP with that fix but at least the proxy continues to work :D#asahi-dev · 2022-01-17 16:48
svendisconnect your screen#asahi-dev · 2022-01-17 16:40
svenyup. DCP crashes with "TTY> rtkit(dcp): Message 1: unsupported subpacket version" but I'm not gonna fix that#asahi-dev · 2022-01-17 16:39
svenhttps://github.com/AsahiLinux/m1n1/pull/146 should at least prevent m1n1 from getting stuck in an infinite loop#asahi-dev · 2022-01-17 16:36
svendcp crashes and then afk_recv infinite loops#asahi-dev · 2022-01-17 16:36
svenyes. I think I found the issue#asahi-dev · 2022-01-17 16:29
svenhuh, looks like the DCP stuff infinite loops or crashes early on when used with <12 (yeah, yeah, I’ll finally upgrade after fixing that)#asahi-dev · 2022-01-15 12:37
sventhere’s #dri-devel on this server fwiw#asahi-dev · 2022-01-14 11:18
svenso now I have what essentially is a stripped down version of the pcie driver for the M1#asahi-dev · 2022-01-14 11:18
sventhey have a split at a different layer. arnd first split the pcie parts and i added the quirks but the nvme maintainers didn't like that approach#asahi-dev · 2022-01-14 11:15
svenespecially that you didn't need to mess with the PCIe driver. still a bit sad i essentially had to copy/paste that one for linux#asahi-dev · 2022-01-14 11:15
svennice :)#asahi-dev · 2022-01-13 08:29
sveninteresting that the watchdog reset also works to shutdown then#asahi-dev · 2022-01-11 18:47
svenWeird#asahi-dev · 2022-01-11 18:46
svenhuh#asahi-dev · 2022-01-11 18:41
svensounds like the verbose tracing slows something down too much and breaks it then#asahi-dev · 2022-01-10 17:25
svenmarcan: once you figured out the iBoot protocol https://github.com/AsahiLinux/m1n1/pull/143 has mailbox and rtkit code (and a nvme stub, but feel free to ignore that for now. i just need that to play with nvme from python a bit more and that was a good test for the rtkit crap)#asahi-dev · 2022-01-08 15:07
svenah, this was a recent change that caused that. that explains why i never hit that so far#asahi-dev · 2022-01-08 15:06
svenenjoys problems that solve themselves#asahi-dev · 2022-01-08 13:57
svenbut i never looked into the details#asahi-dev · 2022-01-08 13:57
sveni thought those props were just forwarded by their drivers#asahi-dev · 2022-01-08 13:56
svenyeah#asahi-dev · 2022-01-08 13:52
svenah, i think the dmashim property of the SIO node contains the MMIO addresses#asahi-dev · 2022-01-08 13:48
svenyeah, i found those but no addrs near them. but there's a chance i just missed it#asahi-dev · 2022-01-08 13:47
svenhave you seen it in there? i did a very brief look but didn't find any hardcoded address #asahi-dev · 2022-01-08 13:47
svenin the SIO firmware?#asahi-dev · 2022-01-08 13:47
svenhm, if SIO is actually just a dumb software DMA it should probably get the SPI MMIO address in one of the setup messages#asahi-dev · 2022-01-08 13:43
svenright, and i think if it has dma-parent it'll go through sio which makes things a bit more complex#asahi-dev · 2022-01-08 13:43
svenChaosPrincess: i think it was called dma-parent#asahi-dev · 2022-01-08 13:41
sveni think as long as the spi node doesn't have a reference to sio (dma-parent or something like that) it the trace thing should work#asahi-dev · 2022-01-08 13:41
svencan't type today#asahi-dev · 2022-01-08 13:41
sven*sigh*#asahi-dev · 2022-01-08 13:41
sveni think as long as the spi node doesn't have a reference to sio #asahi-dev · 2022-01-08 13:38
svenand i think jannau had some code to trace that#asahi-dev · 2022-01-08 13:38
svens/the HV/macOS?#asahi-dev · 2022-01-08 13:38
svenerm#asahi-dev · 2022-01-08 13:38
svenkeep in mind the the HV uses SIO#asahi-dev · 2022-01-08 13:38
svenisn't that just for direct experimentation? or does that also work under the hv?#asahi-dev · 2022-01-08 13:37
sveni think jannau did that#asahi-dev · 2022-01-08 13:32
sveneither way, once you have a working SMC prototype and i've done one more test for this flush slowdown I can probably send the first nvme RFC#asahi-dev · 2022-01-08 13:19
sven*bits/ or nvme/dev#asahi-dev · 2022-01-08 13:19
svenmarcan: should I just push it to the bits/ branch nvme/dev? I’m mostly happy with that code anyway, I just have a one small fix queued #asahi-dev · 2022-01-08 12:58
svensure#asahi-dev · 2022-01-06 13:43
sveni suspect linear sq will break horribly over pcie, but let’s see :D#asahi-dev · 2022-01-06 13:43
svenI guess :D