#asahi-dev · 2024-08-04 07:48
chadmedwill have to do some tests once we rebase#asahi-dev · 2024-08-04 07:48
chadmedhmm schedutil might beat the crap interactivity allegations with 6.12 #asahi-dev · 2024-07-30 16:35
mazscreaming*#asahi-dev · 2024-07-30 16:35
mazyup. I really wish there was a creaming conflict.#asahi-dev · 2024-07-30 16:29
jannauso it was just testing the uninitialized pointer on the stack?#asahi-dev · 2024-07-30 16:29
mazjannau: yeah, DWC3 is the only one I able to use. I'll poke around once I have rebased the NV queue.#asahi-dev · 2024-07-30 16:27
jannauyes dwc3, display (currently only on macbooks with hdmi port) and isp but I guess you don't care about the latter two#asahi-dev · 2024-07-30 16:23
mazDWC3 is the one with multiple IOMMUs, right?#asahi-dev · 2024-07-30 16:23
mazI'm restarting with a full tree (I had to reduce the whole thing to 4 branches).#asahi-dev · 2024-07-30 16:22
mazNo conflicts. it just applied gracefully.#asahi-dev · 2024-07-30 16:21
jannauworth checking if our other fixes for devices with multiple iommus are still needed#asahi-dev · 2024-07-30 16:20
jannaumaz: probably more than a workaround. How did the conflicts get resolved? I don't see how the commit could make any sense in 6.11-rc1's of_iommu_xlate()#asahi-dev · 2024-07-30 15:59
mazjannau: reverting "iommu/of: Handle missing iommu_ops as probe deferral in of_iommu_xlate" works around it. it somehow clashes with the new way iommu_ops are handled.#asahi-dev · 2024-07-30 07:59
mpsand I guess it will need mesa-asahi-20240727#asahi-dev · 2024-07-30 07:56
mpscould someone create verified tag release#asahi-dev · 2024-07-30 07:28
mazjannau: ah, that'd explain it. yeah, 6.10 was almost a walk in the park, but 6.11 is a bit more elusive. if doesn't help that I caught something on my way back from the US, leaving me with only 20% of a brain.#asahi-dev · 2024-07-30 07:26
jannauasahi-wip is now on 6.10 but other than the conflicts in io-pgtable-dart .free_pages() the rebase was uneventful#asahi-dev · 2024-07-30 07:24
jannaumaz: nvme has its own "iommu"/address filter (drivers/soc/apple/sart.c) bypassing the iommu subsystem. display or sound would be other drivers using dart#asahi-dev · 2024-07-30 07:05
mazjannau: the darts definitely probe, and nvme works, so *something* is OK. I'll try and track to see whether PCIe correctly attaches to its IOMMU, as I think there were some changes in that area.#asahi-dev · 2024-07-29 17:25
jannauamarioguy: "ranges" translates / moves address space#asahi-dev · 2024-07-29 17:25
nicolas17macOS 15.1(!) beta 1 was just released (apparently Apple Intelligence is being delayed and won't be on iOS 18.0 / macOS 15.0 so there will be parallel .0 and .1 betas for a while)#asahi-dev · 2024-07-29 17:20
jannaumaz: iommu probes / works as expected? though I would not be surprised if dwc3 and pcie break indepedently#asahi-dev · 2024-07-29 16:19
amarioguy`<0x02000000 0x0 0xc0000000 0x5 0xc0000000 0x0 0x40000000>;` for anyone who's unsure of what address range i'm trying to access#asahi-dev · 2024-07-29 16:10
amarioguywhenever i'm trying to use them, seems the range is undefined (address below 0xFFFFFFFF, untranslated) in ranges#asahi-dev · 2024-07-29 16:08
amarioguythe ones defined in "ranges" property i mean#asahi-dev · 2024-07-29 16:07
amarioguyare the 32-bit PCIe BARs (the ones defined in /soc/pcie0 in the FDT) mapped in the HV?#asahi-dev · 2024-07-29 15:11
mazM1 mini still works fine with vanilla upstream, so it must be something related to the M2 pro support that is part of the asahi branch.#asahi-dev · 2024-07-29 15:09
mazjannau: starting with 6.11-rc1, and it looks like most things related to USB fail to probe correctly (both PCI and DWC3). I'm trying to figure out what went wrong...#asahi-dev · 2024-07-28 15:57
jannauapparently broken in 13.5 but working with 12.3#asahi-dev · 2024-07-28 12:35
jannausigh, somehow USB-C hotplug got broken under the hypervisor. I don't think it's m1n1. I've already gone back to v1.3.3#asahi-dev · 2024-07-24 03:50
hgupta1235Hi Everyone, I want to contribute to the project but I don't have kernel development expoerience but I do know C++ and Python How can I get started?#asahi-dev · 2024-07-22 20:34
tpw_rulesgood to learn more#asahi-dev · 2024-07-22 20:34
jannauI know, zzywysm was pointing to that change#asahi-dev · 2024-07-22 20:33
tpw_ruleshttps://github.com/tpwrules/nixos-apple-silicon/commit/01954941192855fbfbc1693bc02e66e8f3c37309#asahi-dev · 2024-07-22 20:31
tpw_rulesah, yeah i did not think to use the hypervisor. and came up with the same solution#asahi-dev · 2024-07-22 20:31
jannauyes, booting with the hypervisor is clear about that. see https://pagure.io/fedora-asahi/fedora-asahi-remix-scripts/pull-request/7 for some details#asahi-dev · 2024-07-22 20:27
tpw_rules(i think then the watchdog just reset it). i'm just wondering if there was a kernel panic or something i somehow missed.#asahi-dev · 2024-07-22 20:25
tpw_ruleshow do you know you got an SError? a few months ago i put in a lot of effort to find that suspending it did result in a system hang#asahi-dev · 2024-07-22 20:13
jannegin the driver's probe() reading from the BAR results in SError when the device was runtime suspended#asahi-dev · 2024-07-22 20:12
jannegany idea why the generic PCI runtime suspend breaks the gl9755 sd card reader?#asahi-dev · 2024-07-21 11:37
chadmedella-0: nice!#asahi-dev · 2024-07-21 11:29
ella-0chadmed: Finaly got round to getting the disabling UI stuff upstream. https://github.com/lsp-plugins/lsp-plugins/issues/412#asahi-dev · 2024-07-20 02:02
i509vcbbut that won't happen any time soon#asahi-dev · 2024-07-20 02:02
i509vcbmaybe if I can get an old M1 mini for like $50 I can try#asahi-dev · 2024-07-20 02:02
chadmedso if you had bad luck and a slightly heat warped bottom case you could short 50 V straight into the display controller#asahi-dev · 2024-07-20 02:01
chadmednot even that, just stupid crap like how on the intel macs the display backlight power pin was next to the edp data pins on the top housing connector#asahi-dev · 2024-07-20 02:01
i509vcbyeah I've heard the stories about people with screwed up usb cables plugging those into a mac and having it damage something#asahi-dev · 2024-07-20 01:59
chadmedgiven apples approach to the electrical design of their products id say yes it would#asahi-dev · 2024-07-20 01:59
i509vcbI was working on a USB-C PD sniffing device for some else and just checked up on the page to see if it could run dual purpose as an alternative to the central scrutinizer#asahi-dev · 2024-07-20 01:58
i509vcbunfortunately no