#asahi-re · 2021-09-21 03:31
marcanif I don't do HV time accounting, it panics because doing certain things takes "too long"#asahi-re · 2021-09-21 03:31
marcandammit, damned if I do, damned if I don't#asahi-re · 2021-09-21 03:26
marcanif I clear the reset for ATC1_COMMON it works fine#asahi-re · 2021-09-21 03:25
marcanthen it times out waiting for the pstate change for ATC1_USB (which I guess is only a panic with debugging enabled)#asahi-re · 2021-09-21 03:25
marcanit's powering up *and resetting* the ATC1_USB_AON/ATC1_COMMON/ATC1_USB trio - but ATC1_USB depends on ATC1_COMMON, and while ATC1_COMMON is in reset, ATC1_USB can't power up (bit 0x800 - parent missing)#asahi-re · 2021-09-21 03:25
marcan[0xfffffe0013064a64] PMGR: W.4 0x23d2800a0 (ATC1_USB.pstate + 0x00) = 0x900 -> 0x8000090f#asahi-re · 2021-09-21 03:25
marcan[0xfffffe0013064a64] PMGR: W.4 0x23b700448 (ATC1_COMMON.pstate + 0x00) = 0x2300 -> 0x8000230f#asahi-re · 2021-09-21 03:25
marcan[0xfffffe0013064a64] PMGR: W.4 0x23d280090 (ATC1_USB_AON.pstate + 0x00) = 0x800000ff -> 0x800000ff#asahi-re · 2021-09-21 03:24
marcanpanic(cpu 0 caller 0xfffffe001440cbf4): "ApplePMGR: virtual void ApplePMGR::waitReg32(ApplePMGR::RegMap, UInt32, UInt32, UInt32, UInt32, UInt32):4491 " "waitReg32 1 0xa0 & 0xf0 is 0x8000090f instead of 0xf0 after 192291 ticks die 0\n" @ApplePMGR.cpp:4491#asahi-re · 2021-09-21 03:24
marcanI'm pretty sure Apple isn't testing xnu with debugging turned on, because this thing is panicing on what, by all appearances, are bugs#asahi-re · 2021-09-21 03:17
chadmedits also just ridiculously good for yields, and with a good interconnect and healthy amount of on-die cache the IO latency problems can be significantly diminished#asahi-re · 2021-09-21 03:16
chadmedMCM SoCs make sense. jim keller was at apple for a while, and i remember reading that apple wants to revive the Xserve line#asahi-re · 2021-09-20 17:16
roxfanDEBUG_USB/DEBUG_AUTH hmm#asahi-re · 2021-09-20 17:04
marcanNUMA M2? :p#asahi-re · 2021-09-20 17:04
marcanit does kind of sound like they intend to glue two SoCs together into one or something like that, judging by this multidie business#asahi-re · 2021-09-20 17:01
alyssagiggle#asahi-re · 2021-09-20 17:00
marcanand it turns out it'd do it for me :p#asahi-re · 2021-09-20 17:00
marcansee I'd been writing dump_pmgr.py all this time :p#asahi-re · 2021-09-20 17:00
marcanhttps://mrcn.st/p/M3nxXZZf#asahi-re · 2021-09-20 17:00
alyssa(I guess DISP0_CPU is the DCP)#asahi-re · 2021-09-20 17:00
alyssathose do be acronyms ....#asahi-re · 2021-09-20 17:00
alyssaApplePMGR: DISP0_BRC | ID = 0x2a | PERF = PMGR + 0x00034670#asahi-re · 2021-09-20 17:00
alyssaApplePMGR: DISP0_BLC | ID = 0x23 | PERF = PMGR + 0x00034660#asahi-re · 2021-09-20 17:00
alyssaApplePMGR: DISP0_CPU | ID = 0x22 | PERF = PMGR + 0x00034620#asahi-re · 2021-09-20 17:00
alyssaApplePMGR: DISP0_GP1 | ID = 0x08 | PERF = PMGR + 0x00034640#asahi-re · 2021-09-20 17:00
alyssaApplePMGR: DISP0_PPP | ID = 0x09 | PERF = PMGR + 0x00034650#asahi-re · 2021-09-20 17:00
alyssaApplePMGR: DISP0_GP0 | ID = 0x07 | PERF = PMGR + 0x00034630#asahi-re · 2021-09-20 16:59
marcanthen ghidra to figure out exactly how to make it work#asahi-re · 2021-09-20 16:59
marcanI was stringsing pmgr and stumbled upon that string...#asahi-re · 2021-09-20 16:59
marcanmultidie apple silicon coming up#asahi-re · 2021-09-20 16:59
alyssahow did you find that ooi?#asahi-re · 2021-09-20 16:59
marcan[Die 0] is ominous#asahi-re · 2021-09-20 16:59
alyssaholy shit nice#asahi-re · 2021-09-20 16:58
marcanit panics after that due to a timeout, but I'm sure I can fix that#asahi-re · 2021-09-20 16:58
marcanhttps://mrcn.st/p/44xr8mcE all the devs#asahi-re · 2021-09-20 16:57
alyssaregister names? bitfields? enums?#asahi-re · 2021-09-20 16:57
alyssa😁#asahi-re · 2021-09-20 16:57
marcanso if you turn on XNU debugging and pass pmgr-debug=0x8001... it dumps *everything*#asahi-re · 2021-09-20 16:56
marcanomfg#asahi-re · 2021-09-15 12:55
marcandone.#asahi-re · 2021-09-15 10:13
alyssaglhf#asahi-re · 2021-09-15 10:13
marcan:p#asahi-re · 2021-09-15 10:13
marcanalyssa: I'm adding SMP support to the hypervisor so...#asahi-re · 2021-09-15 10:12
alyssamarcan: that last line set off all sorts of alarm bells in my head. nice job, apple#asahi-re · 2021-09-15 10:12
alyssa+ elif otype == 11:#asahi-re · 2021-09-15 10:12
alyssa+ d = bool(size)#asahi-re · 2021-09-15 10:12
alyssa+ d = stream.read(size)#asahi-re · 2021-09-15 10:12
alyssa+ elif otype == 10:#asahi-re · 2021-09-15 10:10
alyssa("What about showering and eating breakfast?" "Ugh..")#asahi-re · 2021-09-15 10:09
alyssai have a bit less than 3 hrs till air time, that's enough time to port to C code right?