#
✕ nick:sven
#asahi · 2021-06-03 16:14
svenhm, that may change once we tackle thunderbolt though
#asahi · 2021-06-03 16:12
svennot yet. so the way i understand it we can actually program a sid mapping using some pcie config regs?
#asahi · 2021-06-03 16:09
svennice!
#asahi · 2021-06-03 16:07
svenyeah, mine too
#asahi · 2021-06-03 16:05
svenyup, this unit has been around for at least 7 years now with the only difference being the hierarchy and the available clocks
#asahi · 2021-06-03 16:01
svenkettenis: sounds like a discussion for the mailing list though :-)
#asahi · 2021-06-03 15:59
svenpipcet[m]: why do you think the self-reference is spurious?
#asahi · 2021-06-03 15:58
sveni could also just get rid of the... yeah that
#asahi · 2021-06-03 15:57
svenhm.. probably
#asahi · 2021-06-03 15:52
svenno
#asahi · 2021-06-03 15:51
sventhe linux maintainers ideally want a single node for the clock controller. that self-reference is one way to represent the same thing in the device tree
#asahi · 2021-06-03 15:50
svenremember how the clock nodes like corellium did them sometimes had references to each other?
#asahi · 2021-06-03 15:44
svenwould that also work with openbsd and u-boot?
#asahi · 2021-06-03 15:43
svennot sure if the maintainers will like that
#asahi · 2021-06-03 15:43
svenotherwise it would get stuck in a EDEFER loop
#asahi · 2021-06-03 15:43
svenit requires a small hack. i just had to register the clock provider with "empty" clocks before assigning parents.
#asahi · 2021-06-03 15:36
svenkettenis: so this actually doesn't look too bad for a clock binding imho https://github.com/AsahiLinux/linux/commit/db975d0b41911068768f0bb23998e2b7d4d59807#diff-845c07f30f0f94856b22d7b50c0ff3d27a961abeffefcdfab01f2e8cd64569ccR138
#asahi · 2021-06-03 15:23
svenso you want to run xnu in a vm essentially to debug it?
#asahi · 2021-06-03 11:57
svenmaybe some parts of it allow to select a subset
#asahi · 2021-06-03 11:57
sventhere's this 32bit command register that has the flush bit and i doubt that they need 32 commands
#asahi · 2021-06-03 11:56
sveni wonder if there is a way to only flush a certain iova range though
#asahi · 2021-06-03 11:56
svenyeah, maybe
#asahi · 2021-06-03 10:25
svenit doesn't hurt to have that there but the code sequence is kinda dumb then
#asahi · 2021-06-03 10:25
svenyup
#asahi · 2021-06-03 10:25
sveni wanted to see which streamid pcie uses and a simple way to do that it to setup empty pagetables and enable translation
#asahi · 2021-06-03 10:24
svenyes
#asahi · 2021-06-03 10:24
svenah. i also forgot a apple_dart_hw_enable_translation i left in there for debugging
#asahi · 2021-06-03 10:22
sventrue
#asahi · 2021-06-03 10:22
sveni think iboot doesn't leave any dart with a stream remap around
#asahi · 2021-06-03 10:21
svennow that i think about that again... it's probably not required
#asahi · 2021-06-03 10:20
svenyeah, it's because the hw only allows 32bit writes
#asahi · 2021-06-03 10:19
svenit's just 0,1,2,3,4,5,...
#asahi · 2021-06-03 10:11
svenoh, there'll definitely be a v4
#asahi · 2021-06-03 10:07
svenj_ey: let's just go with that story then ;)
#asahi · 2021-06-03 10:07
sveni also messed up their name. they should've been called fake_bypass
#asahi · 2021-06-03 10:06
svenactually, let's make that yes! i obviously skipped that because it's part of the hack! :P
#asahi · 2021-06-03 10:06
svenerm.. nope
#asahi · 2021-06-03 09:57
svenwhich sounds like a huge hack
#asahi · 2021-06-03 09:57
svenunless i setup clocks 0,...,i-3, i-2, i-1 to be possible parents of clock
#asahi · 2021-06-03 09:56
svendoesn't sound like that's the correct mechanism then :/
#asahi · 2021-06-03 09:55
svenso i think linux actually requires the clock that's specified in assigned-clock-parents to already be a possible parent
#asahi · 2021-06-03 09:48
svenhm. so assigned-clock-parents is when i e.g. have a clock that can have one out of N parents and i want to select a specific default, isn't it?
#asahi · 2021-06-03 09:21
svenkettenis: so.. clocks. it looks like we'll have to find a way to do with a single clock node. we discussed assigned-clocks at some point and concluded that wouldn't work, didn't we?
#asahi · 2021-06-03 07:41
svenabsolutely. i'm almost ready for the next version of the dart driver
#asahi · 2021-06-03 07:37
svenbut pcie works just fine now
#asahi · 2021-06-03 07:37
svenmostly kettenis, i just complained ;)
#asahi · 2021-06-02 21:32
svenah :)
#asahi · 2021-06-02 21:30
svenshouldn't they only fire due to translation faults?
#asahi · 2021-06-02 21:25
sveni think the interrupts for the dart nodes make sense either way
#asahi · 2021-06-02 21:21
svenif it's acceptable it should at least give us enough confidence that we can work with the dt bindings like this

plus ›