#asahi-dev · 2023-11-24 18:16
ArcaneNibbledidn't manage to poke hard enough to figure that out#asahi-dev · 2023-11-24 18:15
ArcaneNibblenot sure#asahi-dev · 2023-11-24 18:13
ArcaneNibbleit has 4 h.265, 4 h.264, and 1 vp9#asahi-dev · 2023-11-24 18:13
ArcaneNibblelilyD has multiple#asahi-dev · 2023-11-24 18:13
ArcaneNibbleoh wait, does viola only have three VP fifos?#asahi-dev · 2023-11-24 18:12
ArcaneNibbleat least, from static analysis#asahi-dev · 2023-11-24 18:12
ArcaneNibblelike, after VP9#asahi-dev · 2023-11-24 18:11
ArcaneNibblethere's instead a mysterious thing *after* it#asahi-dev · 2023-11-24 18:11
ArcaneNibblebtw, there's no gap between 265/264 on lilyD#asahi-dev · 2023-11-24 18:06
ArcaneNibblethe lilyD firmware doesn't look like that#asahi-dev · 2023-11-24 18:06
ArcaneNibblebut yeah, we can see the status register polling you mentioned#asahi-dev · 2023-11-24 18:01
ArcaneNibbleso... apple split out some cm3 IRQs?#asahi-dev · 2023-11-24 18:00
ArcaneNibblethe others use a revision number in a completely different range#asahi-dev · 2023-11-24 18:00
ArcaneNibbleso it might be A14?#asahi-dev · 2023-11-24 17:59
ArcaneNibbleclover might be older than viola? it uses smaller numbers in the unreadably messy start() function#asahi-dev · 2023-11-24 17:53
ArcaneNibbleno idea#asahi-dev · 2023-11-24 17:50
ArcaneNibblefrom a quick eyeball of the firmwares we have on hand here, clover/viola are different from dahlia/radish/lilyC/lilyD#asahi-dev · 2023-11-24 17:45
ArcaneNibbleoh! viola firmware IRQs are totally different!#asahi-dev · 2023-11-24 17:44
ArcaneNibblewhich version are you looking at?#asahi-dev · 2023-11-24 17:44
ArcaneNibblehmm, maybe lilyD is totally different#asahi-dev · 2023-11-24 17:42
ArcaneNibblefirst one pokes a status register then calls a shared handler that logs 'CtxD', second one logs 'H?Er' (error?), and the other two do more complicated things#asahi-dev · 2023-11-24 17:41
ArcaneNibbleso groups of 4, stride of 5#asahi-dev · 2023-11-24 17:41
ArcaneNibbleoh sorry, the 5th one is always an infinite loop#asahi-dev · 2023-11-24 17:38
ArcaneNibblewhich we had been guessing was interrupts for each VP being done or having an error#asahi-dev · 2023-11-24 17:36
ArcaneNibblein groups of 5#asahi-dev · 2023-11-24 17:36
ArcaneNibblein lilyD, there's handlers in the cm3 firmware for IRQ78 and up#asahi-dev · 2023-11-24 17:33
ArcaneNibble(we're looking at a very different version of the firmware so the addresses won't match up)#asahi-dev · 2023-11-24 17:32
ArcaneNibbleeiln: oh, you mean writes to registers at 40100124 and after?#asahi-dev · 2023-11-24 17:25
ArcaneNibblebut i was thinking that for an RE'd implementation we'd either want "m3 does almost nothing, just proxies interrupts" or "m3 does a lot of offloading in a more competent way than apple"#asahi-dev · 2023-11-24 17:24
ArcaneNibbledon't quite follow which bits of code you're looking at#asahi-dev · 2023-11-24 17:19
ArcaneNibbleis the M3's memory lost on power gate?#asahi-dev · 2023-11-24 17:10
ArcaneNibbleit *also* polls status registers, yes#asahi-dev · 2023-11-24 17:10
ArcaneNibblethe firmware definitely has IRQ handlers in the vector table#asahi-dev · 2023-11-24 17:03
ArcaneNibblewhich is one thing that the M3 notably does#asahi-dev · 2023-11-24 17:03
ArcaneNibblehm, pretty sure the VP/PP interrupts aren't routed to the main CPU?#asahi-dev · 2023-11-24 16:47
ArcaneNibbleexcellent work!#asahi-dev · 2023-11-24 16:47
ArcaneNibbleooh, just found https://eiln.net/avd-notes2.html#asahi-dev · 2023-11-24 16:24
ArcaneNibblei think we'd want to completely rip out apple's M3 firmware for it though, it is terrible#asahi-dev · 2023-11-24 16:23
ArcaneNibblethe biggest blocker for AVD on my part (other than burnout) is not knowing all that much about video codecs#asahi-dev · 2023-11-24 16:16
ArcaneNibbleoh yeah, i also still have the hardware JPEG block RE'd without a linux driver, but that's of questionable utility (mjpeg accelerator?)#asahi-dev · 2023-11-24 16:14
ArcaneNibble(but it's certainly possible to treat that as a black box)#asahi-dev · 2023-11-24 16:14
ArcaneNibbleoh right, prores is also missing somebody to RE the bitstream for "prores raw"#asahi-dev · 2023-11-24 16:13
ArcaneNibbleyes#asahi-dev · 2023-11-24 16:13
ArcaneNibbleinherently#asahi-dev · 2023-11-24 16:13
ArcaneNibbleyes#asahi-dev · 2023-11-24 16:12
ArcaneNibblehardware is conceptually very simple: you just give it a frame (by putting a descriptor in a ring buffer) and it encodes/decodes it and tells you when it's done#asahi-dev · 2023-11-24 16:11
ArcaneNibbleif someone wants to tackle prores on the linux side, feel free to message me (and/or publicly here)#asahi-dev · 2023-11-24 16:10
ArcaneNibblebut it's *so* complicated and i don't really know how to trigger most operations from userspace#asahi-dev · 2023-11-24 16:10
ArcaneNibblei have basic work on scaler/csc#asahi-dev · 2023-11-24 16:10
ArcaneNibbleunless someone wants to do a char/misc proof-of-concept