#
✕ nick:bloom
#asahi · 2021-06-06 13:09
bloomMore intriguing is
#asahi · 2021-06-06 13:09
bloom^ That's in the sgx node.
#asahi · 2021-06-06 13:08
bloom gfx-shared-region-size = 262144
#asahi · 2021-06-06 13:08
bloom gfx-shared-region-base = 42949378048
#asahi · 2021-06-06 13:07
bloomsven: note gfx-asc has iommu-parent = 217 = mapper-gfx-asc
#asahi · 2021-06-06 13:01
bloomi guess
#asahi · 2021-06-06 12:58
bloom mapper-gfx-asc {
#asahi · 2021-06-06 12:58
bloom compatible = [iommu-mapper,gfx]
#asahi · 2021-06-06 12:57
bloombut it's /not/ a DART node, unlike every other iommu in the ADT
#asahi · 2021-06-06 12:57
bloomit's right there in the ADT as iommu-mapper/gfx or something
#asahi · 2021-06-06 12:54
bloomthere's at least two MMUs in there.
#asahi · 2021-06-06 12:53
bloomFrom userspace's PoV, stuff in GPU memory has very different CPU vs GPU virtual addresses.
#asahi · 2021-06-06 12:53
bloomsven: I'm not sure I buy this.
#asahi · 2021-06-06 12:51
bloomI wonder how that works with FairPlay stuff using the GPU
#asahi · 2021-06-06 12:50
bloomWhat are those? 🙃
#asahi · 2021-06-06 12:50
bloomI don't read patents, sorry.
#asahi · 2021-06-06 12:49
bloomsven: what's this about sprr?
#asahi · 2021-06-06 12:41
bloomas well as, what's inside the shmem? is it just the buffer prepared by userspace passed through unmodified? would be convenient, but seems unlikely.
#asahi · 2021-06-06 12:36
bloomPossibly the IOP mailbox itself.
#asahi · 2021-06-06 12:35
bloomSo then the question is where does that shmem come from?
#asahi · 2021-06-06 12:35
bloomI did a little digging on IOP, and that's normal -- it's meant to be low bandwidth, with shared memory used for high bandwidth transaction.
#asahi · 2021-06-06 12:34
bloomsgx is the GPU itself but even less MMIO poking here
#asahi · 2021-06-06 12:34
bloomNot too much interesting here most of the time, rather than just a "there is new work for you GPU, wake the f up"
#asahi · 2021-06-06 12:33
bloomAs hinted by the compatible, gfx-asc is just exposing IOP mailbox
#asahi · 2021-06-06 12:33
bloomI started mmiotracing and looking at ADT etc
#asahi · 2021-06-06 12:33
bloomit does have its own iommu-mapper in the ADT, but.. not DART
#asahi · 2021-06-06 12:33
bloomsven: Yeah, the lack of DART for the GPU is odd
#asahi · 2021-06-06 00:27
bloomdual booting..
#asahi · 2021-06-06 00:27
bloomI need dev tools on here grumble
#asahi · 2021-06-05 23:49
bloomexpensive lesson to learn given how slow hv is to boot
#asahi · 2021-06-05 23:49
bloomnote to self: writing bad things to fb format bits can wedge the whole system, even in macos
#asahi · 2021-06-05 23:37
bloomI can poke the display controller regs with macOS booted from hv
#asahi · 2021-06-05 23:36
bloomok, this is awesome
#asahi · 2021-06-05 23:31
bloomincluding controlling the curosr plane apparently? :p
#asahi · 2021-06-05 23:31
bloomyeah ok.
#asahi · 2021-06-05 23:29
bloomyeah sorry im confused
#asahi · 2021-06-05 23:28
bloomOk, I have hypervisor macos. Cool.
#asahi · 2021-06-05 23:28
bloomthanks wiki
#asahi · 2021-06-05 23:28
bloomPMP: "Power Management Processor?". Handles power functionality
#asahi · 2021-06-05 23:28
bloomoh der no
#asahi · 2021-06-05 23:28
bloomwonder if pmp is NVMe
#asahi · 2021-06-05 23:27
bloomyeah that would help lol
#asahi · 2021-06-05 23:25
bloombetween how slow boot is and that this is all being fed over a serial port err
#asahi · 2021-06-05 23:25
bloomhow do I actually.. do anything with this?
#asahi · 2021-06-05 23:25
bloomok it finally boot, nice
#asahi · 2021-06-05 23:23
bloomsven: what is pmp[17] and why is it so noisy during boot?
#asahi · 2021-06-05 23:23
bloomoh there it is
#asahi · 2021-06-05 23:22
bloomnow i see a progress bar come up and lots of MMIO spammed at me but the display has gone black so I don't know if it's making progress
#asahi · 2021-06-05 23:22
bloomsven: etc
#asahi · 2021-06-05 23:22
bloomneeded to rebuild + chainload m1n1

plus ›