#asahi · 2021-12-02 11:06
sventhat's what vacations are for! just need to work for two more weeks and then i'll have time to do that ;)#asahi · 2021-12-02 11:03
sven*fixed#asahi · 2021-12-02 11:03
svenyeah, that'll be fixe#asahi · 2021-12-01 17:08
svenI think jannau has a nice collection for arch#asahi · 2021-12-01 17:07
sven*as long as you’re comfortable building downstream kernels with patches you can already somewhat use Linux :)#asahi · 2021-12-01 17:07
svenas long youre#asahi · 2021-12-01 17:05
sventhat’ll very likely be the only way dual booting will be possible #asahi · 2021-11-27 10:42
sven:)#asahi · 2021-11-26 16:38
svenjust making sure :D#asahi · 2021-11-26 16:37
svenor did I just confuse rx and tx?#asahi · 2021-11-26 16:37
svenhrm, I tested it from m1n1 by just writing to it without starting a transaction #asahi · 2021-11-26 16:26
svenheh, yeah. getting access to the spi signals is probably going to be a bit annoying#asahi · 2021-11-26 16:24
svenI’m just curious #asahi · 2021-11-26 16:24
svenprobably not necessary though if it still works without that write#asahi · 2021-11-26 16:23
svenyeah#asahi · 2021-11-26 16:22
svenmight be interesting to look at the signals for different values of that register. It sounds like it’ll introduce some delay sometime during the SPI transmission#asahi · 2021-11-26 09:22
sventhanks for managing that tree btw.!#asahi · 2021-11-26 09:22
svensounds like we need to hurry up with the upstreaming :D#asahi · 2021-11-25 14:52
svenhrm, okay. probably custom then#asahi · 2021-11-25 14:52
sventhat one's at least called "M3"#asahi · 2021-11-25 14:52
svenhrm, might be more similar to the thunderbolt co-proc mailbox then#asahi · 2021-11-25 14:52
svenoh, hrm. true#asahi · 2021-11-25 14:26
sveni'd expect it to write a pointer to the first 0x8000 bytes of the MMIO region (i.e. the region just before the mailbox interface)#asahi · 2021-11-25 14:22
svenit'll just map the firmware using the DART then#asahi · 2021-11-25 14:22
sven*writes#asahi · 2021-11-25 14:22
sventhe wires are to RAM, that's why you don't see them#asahi · 2021-11-25 14:19
svenif it's the same co-proc like the others it'll just put the firmware somewhere in RAM and write a pointer to some configuration register#asahi · 2021-11-25 13:38
svenj_ey: oh, and unless you enjoy hearing maz rant again: _relaxed everywhere :D#asahi · 2021-11-25 13:00
svenBIT(4) in there might be something like tx_fifo_full#asahi · 2021-11-25 12:57
sventhat should explain the FIFO_SIZE#asahi · 2021-11-25 12:57
svenj_ey: so REG_AVAIL just contains the fifo levels and the max value it takes is 0x10 here#asahi · 2021-11-25 12:45
svenyeah#asahi · 2021-11-25 12:44
svenREG_AVAIL_TXFIFO as well because it seems to do available = FIFO_SIZE - REG_AVAIL_TXFIFO#asahi · 2021-11-25 12:42
svenand yeah, i agree with kettenis that CLKCFG and CLKIDLE sound weird#asahi · 2021-11-25 12:36
svenin practise it just doesn't matter because the clock is just a dumb constant frequency clock anyway#asahi · 2021-11-25 12:35
sveni think for clk_prepare_enable the same comment i got for my watchdogv1 applies. technically you should only disable the clock after devm has unregistered the spi controller so you should use devm_add_action_or_reset or however it's called#asahi · 2021-11-25 12:27
svenhmmm… I might’ve forgotten about that. sorry :(#asahi · 2021-11-25 12:25
svenimho just start with a simple driver that doesn’t do DMA and only add that later if it makes sense#asahi · 2021-11-25 12:24
svenagreed#asahi · 2021-11-25 12:23
svenbut I still suspect that SIO just uses a busy loop then#asahi · 2021-11-25 12:22
svenyes#asahi · 2021-11-25 12:21
svenit should be fairly simple since the HW probably doesn’t even do DMA #asahi · 2021-11-25 12:21
svenj_ey: as long as there’s no obscure magic or similar weird things it might be good enough. Did you write that driver from scratch?#asahi · 2021-11-25 12:18
svenwould be nice to get a spi driver on the ML at some point#asahi · 2021-11-23 17:20
sventhere's still enough other stuff to be done to make upstream useable on these machines#asahi · 2021-11-23 17:20
svenyeah, no one has started to look at the kernel side yet#asahi · 2021-11-23 13:07
svenwhich really smells like "just read data from $interface and put it into $buffer"#asahi · 2021-11-23 13:07
svenand the protocol seems to be just sending a message with a IOVA, a size and some flags afaict#asahi · 2021-11-23 13:06
svenyeah, exactly#asahi · 2021-11-23 12:54
svenbut i kinda suspect it really just uses the normal SPI regs to allow off-loading the busy loop