#asahi · 2021-09-27 15:21
marcan function-tx = 102:GPIO(0xa0, 0x102)#asahi · 2021-09-27 15:21
marcanwhich is actually in the adt!#asahi · 2021-09-27 15:21
marcanuart0 tx is at 0x23c100280#asahi · 2021-09-27 15:21
marcanmost jellybean IO pins are, yes#asahi · 2021-09-27 15:19
marcanand a scope!#asahi · 2021-09-27 15:18
marcan(I used the UART TX pin as a guinea pig, since I could poke that externally)#asahi · 2021-09-27 15:17
marcanpovik: I spent some quality time with a multimeter on a call with j_ey the other day to figure out the gpio bits :)#asahi · 2021-09-27 15:17
marcanFireFox317: yeah#asahi · 2021-09-27 14:57
marcanbut either way, I intend to spend tomorrow marathoning small kernel drivers (finally), so if I get to it, why not#asahi · 2021-09-27 14:57
marcanpovik: if the clock mux never changes, we might not need a clocksel driver#asahi · 2021-09-27 12:53
marcan(and yet it's still faster than other ARM chips *with* acceleration...)#asahi · 2021-09-27 12:51
marcanpovik: I'll put a clocksel driver on my list, having the audio driver poke that register directly is very ugly#asahi · 2021-09-27 12:49
marcanit doesn't mean memory doesn't work#asahi · 2021-09-27 12:48
marcanas in "things we might have to deal with in the memory controller"#asahi · 2021-09-27 12:48
marcanand yes, that "unified memory controller" thing doesn't make any sense unless you mean stuff like the perf switching driver I have to write#asahi · 2021-09-27 12:48
marcanyes, 2D acceleration hasn't really been a thing in a while#asahi · 2021-09-27 12:47
marcan"2D" acceleration is just 3D with no depth#asahi · 2021-09-27 12:47
marcanthese GPUs are 3D only#asahi · 2021-09-27 12:47
marcannsklaus_: accelerated 2D is not a thing#asahi · 2021-09-27 12:45
marcan:)#asahi · 2021-09-27 12:45
marcan6 consecutive clocks, one for each mca#asahi · 2021-09-27 12:45
marcanwell that explains it then#asahi · 2021-09-27 12:44
marcanI guess this means I get to write a clocksel driver after all...#asahi · 2021-09-27 12:43
marcanpovik: you sure this was mca3 and not mca1?#asahi · 2021-09-27 12:43
marcan0x18 size though, so that's exactly 6 clocks, which matches up with 6 MCA clusters#asahi · 2021-09-27 12:41
marcannot sure why mca3 would be offset 1 in that though, probably referenced somewhere#asahi · 2021-09-27 12:40
marcanmca-switch has a regblock pointing riiight into the clocks at the right place#asahi · 2021-09-27 12:40
marcanthere you go#asahi · 2021-09-27 12:40
marcan size = 0x0000000000000018, Container: #asahi · 2021-09-27 12:40
marcan addr = 0x000000003B0400D4#asahi · 2021-09-27 12:39
marcanand the answer might well be "it's hardcoded somewhere"#asahi · 2021-09-27 12:39
marcanso the question is how does it know about that clock reg#asahi · 2021-09-27 12:33
marcanobviously some things can change#asahi · 2021-09-27 12:33
marcanbecause (hopefully) it lists the clock config at boot#asahi · 2021-09-27 12:30
marcanand an NCO is exactly what I expect to be driving an audio block#asahi · 2021-09-27 12:30
marcanit would be extremely weird if it didn't#asahi · 2021-09-27 12:30
marcanyes#asahi · 2021-09-27 12:28
marcanit's entirely possible 113 is a coincidence, and the clock offset is hardcoded somewhere#asahi · 2021-09-27 12:27
marcanmclk-config also has the NCO index in the low bits#asahi · 2021-09-27 12:25
marcanthis is on the mini, and which mca?#asahi · 2021-09-27 12:25
marcanyes, it's likely the mclk input, and the two settings are switching between different (hardcoded, likely) dividers of NCO output#asahi · 2021-09-27 12:23
marcanyeah, that's changing a clock mux#asahi · 2021-09-27 12:20
marcanso the NCOs are 72~75 in the clocks#asahi · 2021-09-27 12:17
marcanso this is audio#asahi · 2021-09-27 12:17
marcanI was wondering#asahi · 2021-09-27 12:17
marcanahh, now we know what the NCOs are for#asahi · 2021-09-27 12:14
marcan(those are referencing mca-switch; the first number is the phandle)#asahi · 2021-09-27 12:14
marcanthe function-i2s_route stuff sounds useful#asahi · 2021-09-27 12:13
marcanbut most of the hardware stuff is there#asahi · 2021-09-27 12:12
marcanthose are unpersonalized, so they lack some info