#asahi-dev · 2022-10-23 11:00
jannauit's only a problem for locked darts and how they are handled right now#asahi-dev · 2022-10-23 11:00
jannauI don't see how we can enforce that if we map the L1 table for locked darts and the devices is not detached from the default domain#asahi-dev · 2022-10-23 10:55
jannauit ends up in the dart pagetable, not sure how it gets there but the default domain and the unmanaged can write both there#asahi-dev · 2022-10-23 10:45
jannauI think everything is working more or less as expected, the default_domain handling is a little surprising and that dma_map_sgtable() still works with the unmanaged domain#asahi-dev · 2022-10-23 10:38
jannausysfs shows the type of the default domain#asahi-dev · 2022-10-23 10:31
jannauapple_dart_hw_set_ttbr() is only called from attach_dev path#asahi-dev · 2022-10-23 10:30
jannauexcept there is nothing in in apple-dart which can switch between two pagetables#asahi-dev · 2022-10-23 10:29
jannauthere is a check in iommu.c __iommu_attach_group(), the default and blocking domains can be replaced this way#asahi-dev · 2022-10-23 10:21
jannauI guess it works as long as just the last domain which has device attached is used#asahi-dev · 2022-10-23 10:12
jannauI think I spotted in the path in iommu.c just a check if there's just a single device in the group#asahi-dev · 2022-10-23 10:09
jannauit doesn't prevent iommu_attach_device() from succeeding#asahi-dev · 2022-10-23 10:07
jannauso I think the default domain is still there and keeps working#asahi-dev · 2022-10-23 10:06
jannauthe group in sysfs still says DMA as type but I end up apple_dart_attach_dev() with domain->type IOMMU_DOMAIN_UNMANAGED#asahi-dev · 2022-10-23 10:01
jannauI started looking into switching piodma to an unmanaged domain. I've currently just create the domain and attach piodma to it. everything still works since the dart is locked but if it wasn't it would be completely broken#asahi-dev · 2022-10-23 09:58
jannausven: does apple-dart miss a check to prevent .domain_alloc or at least .attach_dev if there's already a device attached to a specific dart sid?#asahi-dev · 2022-10-22 18:33
jannausven: includes t8103 dcpext and your tracepont commit#asahi-dev · 2022-10-22 18:33
jannaupushed dcp which replaces the L2 tables atomically for the premapped regions for lockked DARTs#asahi-dev · 2022-10-22 18:30
jannauwe are planning to make this easier by storing a firmware.cpio there. that one can simply appended to the initramfs. not sure what we need to do for compression. I think the cpio archives have all use the same/none compression#asahi-dev · 2022-10-22 18:27
jannauuse $ESP/vendorfw/firmware.tar#asahi-dev · 2022-10-22 17:07
jannauit was tasksetted to the 1st efficiency cluster#asahi-dev · 2022-10-22 16:11
jannausven: ./modetest -M apple#asahi-dev · 2022-10-22 16:02
jannaualso \o/!!!!#asahi-dev · 2022-10-22 16:02
jannaufb emulation / fbcon starts automatically for dcp#asahi-dev · 2022-10-22 16:00
jannausven: libdrm has a modeset test utility#asahi-dev · 2022-10-22 14:42
jannaumarcan: cpufreq seems to work. I don't see the first perf cluster drop below 1.98 GHz but I guess that's a fight against schedutil#asahi-dev · 2022-10-22 13:43
jannauatomic lcoked dart pgtable updates done, certainly not upstreamable but the previous code wasn't better in that regard. as bonus this get's rid of a bunch of custom of parsing code in apple-dart#asahi-dev · 2022-10-22 12:04
jannauFoxboron: I'll test (and close my PR) but I don't think we will use it. We use uncompressed kernels for sd-boot compatibility. if we use compressed kernels it's probably EFI_ZBOOT only#asahi-dev · 2022-10-22 09:36
jannauplease stop wasting my time#asahi-dev · 2022-10-22 09:36
jannauit was merged this wednesday#asahi-dev · 2022-10-22 09:34
jannaumps: "10:53:21 < jannau> mps: it works unmodiefied m1n1 from AsahiLinux but the necessary changes are not yet in any tag"#asahi-dev · 2022-10-22 09:30
jannaumps: which device? you most likely crashed dcp. please search in the output of 'dtc -I fs -O dts -o - /proc/device-tree' for reserved-memory#asahi-dev · 2022-10-22 09:29
jannauthe probing is still annoying since it breaks completely as soon as a single dcp doesn't probe/start#asahi-dev · 2022-10-22 09:14
jannauif it doesn't appear in dmesg at the problem is not that simpledrm is present#asahi-dev · 2022-10-22 09:13
jannaumake sure you use the dtb from the kernel. it works built-in as well along with simpledrm#asahi-dev · 2022-10-22 09:11
jannaulook for dcp in dmesg#asahi-dev · 2022-10-22 09:10
jannauso you can boot with simpledrm built-in and have dcp as module. during boot simpledrm is used until the dcp modules are loaded#asahi-dev · 2022-10-22 09:08
jannaumps: dcp will take over the framebuffer from simpledrm#asahi-dev · 2022-10-22 09:07
jannausven: let's use dcp. I just used iomfb since that was there all my tracepoints are#asahi-dev · 2022-10-22 09:06
jannaunot sure if that's caused by the badly emulated vblank handling, llvmpipe rendering times excceding 1/60 s with high resolution displays or something else#asahi-dev · 2022-10-22 09:02
jannaumps: don't expect much change especially on laptops/imacs. performance on wayland (compositors using atomic_swap) is bad#asahi-dev · 2022-10-22 08:58
jannausven: dts with dcpext is in https://github.com/jannau/linux/tree/asahi-dcpext-v6.1-rc1-1 mostly because dcpext0 on t6002 doesn't want to boot#asahi-dev · 2022-10-22 08:54
jannaumarcan: cpufreq boots, I haven't noticed anything but I wasn't looking. main system is not yet yet updated#asahi-dev · 2022-10-22 08:53
jannaumps: it works unmodiefied m1n1 from AsahiLinux but the necessary changes are not yet in any tag#asahi-dev · 2022-10-22 08:34
jannaumarcan: yes, it works without additional m1n1 changes#asahi-dev · 2022-10-22 08:34
jannausven: there's aslo a fix for the probe race#asahi-dev · 2022-10-22 08:01
jannaudcp rebased on asahi-wip#asahi-dev · 2022-10-22 08:01
jannausven: you want https://github.com/jannau/linux/commit/9bec832ab04f90a250013079fd8ea0f37a2f3ed0 if you haven't fixed that yourself yet#asahi-dev · 2022-10-21 20:30
jannauworks in X and on other ttys with login#asahi-dev · 2022-10-21 20:30
jannauhmm, it might work and just tty0 is not active after booting with under the HV with serial port#asahi-dev · 2022-10-21 20:25
jannauafter fixing code rot due to not compiling with CONFIG_PM_SLEEP