#
✕ nick:sven
#asahi-dev · 2023-09-14 12:18
svenso it sounds like jlcpcb assembly is very hit and miss at best :/
#asahi-dev · 2023-09-13 16:30
svenhv can be run from anywhere
#asahi-dev · 2023-09-13 16:29
sven*passes
#asahi-dev · 2023-09-13 16:29
svenbut the hard reset over pd might be very useful for CI
#asahi-dev · 2023-09-13 16:29
svenit just passed through most hw and doesn’t try to virtualization anything
#asahi-dev · 2023-09-13 16:28
svenI do all my development inside that hv fwiw
#asahi-dev · 2023-09-13 16:25
sven(as a regular cdc acm device)
#asahi-dev · 2023-09-13 16:25
svenor you can run Linux inside m1n1‘s thin hv which will steal a usb port to expose a serial console
#asahi-dev · 2023-09-13 16:24
svenor just another m1/m2 machines :p
#asahi-dev · 2023-09-10 18:57
sven(maybe not for the apple case where you know about all devices but to do it correctly for the general case of arbitrary pcie busses)
#asahi-dev · 2023-09-10 18:56
svenyeah, I’d try to avoid writing that code myself. That uboot and Linux code is partly hard to follow because the whole setup is a bit tricky
#asahi-dev · 2023-09-10 18:55
svenyou’ll also have to configure all the bridges above the device
#asahi-dev · 2023-09-10 15:33
svenand, uh, maybe I misremember but I though k mentioned this a while ago when we discussed dapf. maybe I should’ve put more emphasis on just how much of a hack the adt can be
#asahi-dev · 2023-09-10 15:31
svenI‘d love to see why they need this specific split :D
#asahi-dev · 2023-09-10 15:31
svenall of device mode and half of host most goes through the first and the other half of host mode through the second one
#asahi-dev · 2023-09-10 15:30
svenand the splits makes no sense
#asahi-dev · 2023-09-10 15:30
svenAnd yeah, dwc3 also has two darts that are merged into a single one inside adt
#asahi-dev · 2023-09-10 15:30
svenThey sometimes use regs we have no clue about
#asahi-dev · 2023-09-10 15:29
svenMost darts have tunables that can just be ignored
#asahi-dev · 2023-09-10 14:36
svenyeah, that
#asahi-dev · 2023-09-10 14:36
sventhe dart driver only supports two right now, I think you can just change a define to make that three
#asahi-dev · 2023-09-09 19:38
svendoes DCP maybe already have the logic to add more than one domain?
#asahi-dev · 2023-09-09 19:37
sven+have
#asahi-dev · 2023-09-09 19:37
sveni think the reason from the subsystem maintainers was that once you more than one the sequencing is usually complicated and should be left to the driver
#asahi-dev · 2023-09-09 19:36
sven*genpd
#asahi-dev · 2023-09-09 19:36
sveniirc genod by default only supports a single power domain
#asahi-dev · 2023-09-08 17:24
svenhow did they mess them up? unaligned ics?
#asahi-dev · 2023-09-08 16:52
sven:D
#asahi-dev · 2023-09-08 16:48
sveni wouldn't want to solder some of those chips without a stencil :D
#asahi-dev · 2023-09-08 13:17
svenhm, true. weird.
#asahi-dev · 2023-09-08 13:12
svenor maybe that was just a bug that was fixed at some point
#asahi-dev · 2023-09-08 13:12
sveni thought with an iommu enabled the iommu would fully decide virt/phys mapping instead of dma-ranges which is used without an iommu
#asahi-dev · 2023-09-08 13:11
svenhrm, maybe I remember it wrong then
#asahi-dev · 2023-09-08 11:31
svenso i guess "new from the factory" actually means "stripped from some broken apple mainboard"
#asahi-dev · 2023-09-08 11:31
svenlol, my cd3217/8 "new from the factory" chips arrived from china and they are obviously reballed :D
#asahi-dev · 2023-09-08 11:26
svenso dma_set_mask_and_coherent is for device limits (e.g. this device has only 36 address bits for dma) while the iommu domain aperture is for iommu limits (e.g. only 32bit of the device address lines are actually connected to the iommu)
#asahi-dev · 2023-09-08 11:24
svenyeah, from what i can tell dma-iommu.c respects the dma mask set by dma_set_mask_and_coherent
#asahi-dev · 2023-09-08 11:22
svenif that 36bit thing indeed is an iommu limitation and not just a limitation of the dma device i guess
#asahi-dev · 2023-09-08 11:21
sventhat makes more sense
#asahi-dev · 2023-09-08 11:19
svenbut then again, dma-ranges is ignored when an iommu is present
#asahi-dev · 2023-09-08 11:19
sveniirc dma_set_mask_and_coherent is meant to express "this device can handle n bits of dma address" independently of whatever the iommu does
#asahi-dev · 2023-09-08 11:17
svenhuh, dma set coherent should work independently of the iommu subsystem I thought
#asahi-dev · 2023-09-08 10:00
svendma_set_mask_and_coherent is the thing he's talking about fwiw
#asahi-dev · 2023-09-07 18:53
svenlike, does that dma go through some additional path or whatever because the device can’t handle that many bits in the dma address??
#asahi-dev · 2023-09-07 18:52
svendma to higher iovas is *slower*? how the hell did they mess that up
#asahi-dev · 2023-08-30 15:33
svenyeah, just tested it with quicktime on macos, it also records at 30 fps by default
#asahi-dev · 2023-08-30 15:24
svenyeah, PR sounds good! and nice work :)
#asahi-dev · 2023-08-30 09:20
svenI don't think these usb2-only-type-c connectors are even made outside of china ;)
#asahi-dev · 2023-08-30 07:39
sventhat's v1 though I think
#asahi-dev · 2023-08-30 07:39
svenfwiw, since I replaced the usb connectors my board has been very stable now

plus ›