#
✕ nick:amarioguy
#asahi-dev · 2023-03-14 12:23
amarioguymarcan: ah thanks for checking
#asahi-dev · 2023-03-14 12:09
amarioguyjust double checked
#asahi-dev · 2023-03-14 12:09
amarioguywait my tests were with bit 27 = 0
#asahi-dev · 2023-03-14 12:06
amarioguygotta redo some test then
#asahi-dev · 2023-03-14 12:06
amarioguyahhhh crap
#asahi-dev · 2023-03-14 12:00
amarioguy(test results are serious - that is the behavior i see, and the same is true of the Uncore PMCs too)
#asahi-dev · 2023-03-14 11:59
amarioguy(sorta joking but you know would be nice)
#asahi-dev · 2023-03-14 11:59
amarioguymarcan: quick feature request, can we please get an "bit set/bit clear" msr command in m1n1, would make bit twiddling much easier :P
#asahi-dev · 2023-03-14 11:58
amarioguyESR_EL1 = 0x6239fc11
#asahi-dev · 2023-03-14 11:58
amarioguymaz: good news, when HCR_EL2.TIDCP = 1, all the impdef PMU regs i tested reads/writes from EL1 with, trap to EL2
#asahi-dev · 2023-03-14 11:50
amarioguytesting that now
#asahi-dev · 2023-03-14 11:50
amarioguythe impdef pmus?
#asahi-dev · 2023-03-14 11:49
amarioguythink you mentioned wanting to know the effects of this on that?
#asahi-dev · 2023-03-14 11:49
amarioguymaz: that was mainly to clarify that this bit does not affect read/write of the impdef stuff since that's controlled by separate bits
#asahi-dev · 2023-03-14 11:36
amarioguymaz: not trapped as in it's undefined
#asahi-dev · 2023-03-14 11:33
amarioguymaz: got it, there should be smth that allows that given apple's official virt solution needs that to work
#asahi-dev · 2023-03-14 11:22
amarioguyshould be sufficient, poke me if you need anything else tested, testing config was t6002 Mac Studio in m1n1 console
#asahi-dev · 2023-03-14 11:21
amarioguymaz: https://gist.github.com/amarioguy/f7c2fcc90ca673e511130204e381ac4a
#asahi-dev · 2023-03-13 19:50
amarioguyanything else you need me to test?
#asahi-dev · 2023-03-13 19:50
amarioguymaz: ok got results for MDCR, trapped regs and effects on Apple PMU writes
#asahi-dev · 2023-03-13 18:12
amarioguymaz: heads up running those pmuv3 tests now
#asahi-dev · 2023-03-06 18:18
amarioguywould the asahi arm sysreg docs page work for that or would you need it somewhere a bit more visible?
#asahi-dev · 2023-03-06 18:18
amarioguymaz: yes will do
#asahi-dev · 2023-03-06 18:11
amarioguynot at a mac atm :/
#asahi-dev · 2023-03-06 18:11
amarioguymaz: ahh gotcha, i'll verify that stuff tonight then
#asahi-dev · 2023-03-06 18:03
amarioguyHACR_EL2, bit 56
#asahi-dev · 2023-03-06 17:56
amarioguybut generally the same deal - we return stuff from cycle counter (PMC0), and i'm not sure if windows needs more than the cycle counter
#asahi-dev · 2023-03-06 17:56
amarioguywhat imbushuo did was similar but i think he returned some derived data, i'll have to ask him in a sec to see what specifically he returned
#asahi-dev · 2023-03-06 17:55
amarioguymaz: in my case specifically (bare metal or as close to it as possible) i just return the data from Apple's proprietary PMCs
#asahi-dev · 2023-03-06 17:39
amarioguyyep
#asahi-dev · 2023-03-06 17:38
amarioguy(gonna be a small PR too, to add it to m1n1's apple regs json)
#asahi-dev · 2023-03-06 17:38
amarioguyi'm updating asahi docs to document that hacr bit
#asahi-dev · 2023-03-06 17:38
amarioguyj`ey_, maz: i was involved in making that work to some degree, yes emulated pmuv3
#asahi-dev · 2023-03-06 17:32
amarioguypretty cool demo of m2 kvm
#asahi-dev · 2023-03-06 17:31
amarioguyhttps://twitter.com/imbushuo/status/1629839633281589248?s=20
#asahi-dev · 2023-03-06 17:31
amarioguymaz: seen this tweet yet?
#asahi-dev · 2023-02-18 17:47
amarioguyalso some irl stuff too
#asahi-dev · 2023-02-18 17:46
amarioguyoh hey been a while since i was here, lot had happened in the interim... (including my motherboard's nvme slot going bad which is... '''fun''')
#asahi-dev · 2023-01-21 03:25
amarioguyCPMU_AIC_PMI the macro of interest
#asahi-dev · 2023-01-21 03:25
amarioguyso that tracks
#asahi-dev · 2023-01-21 03:25
amarioguyotherwise fiq
#asahi-dev · 2023-01-21 03:25
amarioguy(the aic mode is set in xnu via a macro per device)
#asahi-dev · 2023-01-21 03:24
amarioguygot it that makes sense, looked at an older xnu drop it seemed to be defined for at least H9P/A10 but on newer source drops this does not appear to apply for h13
#asahi-dev · 2023-01-21 03:21
amarioguyah fair
#asahi-dev · 2023-01-21 03:20
amarioguydo m series chips only have the option to deliver these as FIQs?
#asahi-dev · 2023-01-21 03:20
amarioguyjannau: one question on pmu stuff, why do we use the FIQ mode in linux instead of the AIC mode (ostensibly should deliver as an IRQ)? is that just a legacy holdover mode from idevices or smth?
#asahi-dev · 2023-01-20 11:18
amarioguyjannau: gotcha thanks
#asahi-dev · 2023-01-20 00:06
amarioguy(for apple pmcs to clarify lol)
#asahi-dev · 2023-01-20 00:06
amarioguythe asahi docs are slightly unclear on this
#asahi-dev · 2023-01-20 00:06
amarioguybtw is pmc0 always the cpu cycle register?

plus ›