#asahi-dev · 2022-05-12 21:28
jannauthe default code is just of_iommu_get_resv_regions(), the actual mapping would have to be implemented in the iommu driver#asahi-dev · 2022-05-12 21:26
jannaudcp/disp0 use only non-identity mappings, everything is mapped starting at 0 in the IOVA space#asahi-dev · 2022-05-12 21:22
jannauthe common parsing code is not helpful for us though as supports only reservations an identity mappings#asahi-dev · 2022-05-12 21:21
jannauyes, except for a property name change#asahi-dev · 2022-05-12 21:15
jannauThierry posted a patchset for devicetree iommu mapping https://lore.kernel.org/linux-iommu/20220512190052.1152377-1-thierry.reding@gmail.com/#asahi-dev · 2022-05-11 14:35
jannauyes, pcie xhci controller is still ASM3142#asahi-dev · 2022-05-11 14:33
jannauspi2 has a node usb3hub-fw with 'compatible = [usb3hub-fw-gl3590,spi]'#asahi-dev · 2022-05-08 19:44
jannauommitting the '_d0' suffix for die 0 though#asahi-dev · 2022-05-08 19:43
jannaunow with die 0 and 1 on the ultra https://github.com/jannau/linux/tree/asahi-t6002#asahi-dev · 2022-05-08 06:42
jannauit seems to work without problems in u-boot / linux. u-boot reports the usb devices by the label #asahi-dev · 2022-05-07 20:00
jannauonly annoyance is that we end up with duplicate nodes like usb@702280000. not sure if that is a problem if they live in different sub-trees#asahi-dev · 2022-05-07 19:57
jannaumarcan's idea. it solves the part of the problem which would have been annoying to solve in C preprocessor#asahi-dev · 2022-05-07 15:55
jannauhttps://github.com/jannau/linux/tree/asahi-t6002 uses now ranges to translate device addresses on the second die for the m1 ultra#asahi-dev · 2022-05-03 20:42
jannauyes, even for the m1 ultra Linux works without code changes#asahi-dev · 2022-05-03 20:30
jannauI've also added the device tree for the M1 Max Mac Studio in case someone can test it#asahi-dev · 2022-05-03 20:29
jannaupushed to https://github.com/jannau/linux/tree/asahi-t6002, rebased onto current asahi/asahi#asahi-dev · 2022-05-03 19:56
jannauat least for the power domains per die includes with C preprocessor macros works. avoids duplicating the 2000 line t6001-pmgr.dtsi#asahi-dev · 2022-05-02 07:05
jannausven: I intend to send https://github.com/jannau/linux/tree/dart-t6000 tonight. I kept you as commit author and added myself as "Co-Developed-by:" for the two commits I changed. Is that ok?#asahi-dev · 2022-04-30 08:29
jannausven: for keygen=100000000 all keys differ, see aes*janne2*.txt in https://www.jannau.net/aes/#asahi-dev · 2022-04-29 18:02
jannauseems to apply. key 3 is the same on the m1 max and the 1st die of the m1 ultra#asahi-dev · 2022-04-29 17:55
jannausven: https://www.jannau.net/aes/ key 1 differs and key 3 is identical on my mini, m1 max has different keys#asahi-dev · 2022-04-26 19:59
jannauread and write protect bit on dart-t6000 verified#asahi-dev · 2022-04-26 19:48
jannauworks on t6000 as well#asahi-dev · 2022-04-26 19:40
jannauworks now after a reboot, seems broke it#asahi-dev · 2022-04-26 19:24
jannauI switched to dart.py to test it on t6k so that should take care of that and it enables stream_id 1#asahi-dev · 2022-04-26 19:19
jannausven: was your aes experiment working without dart bypass?#asahi-dev · 2022-04-26 18:01
jannau12.4 I pressume#asahi-dev · 2022-04-26 17:50
jannausven: boots and mount root fs from nvme, (asahi branch with bits/050-nvme replaced by nvme-v3, smc-rtkit disabled, no dcp)#asahi-dev · 2022-04-26 16:11
jannaumillenialhacker: judging by "dart-tunables-instance-X" in the adt there are 3 dart instances for isp. compare with 2 instances for the usb-c ports#asahi-dev · 2022-04-26 16:09
jannaumaz: doesn't the PCIe driver looks it up in iommu-map#asahi-dev · 2022-04-24 14:09
jannaupcie sdhci is broken as well and has no allow-subpage-mapping#asahi-dev · 2022-04-24 13:53
jannaumaybe related to "allow-subpage-mapping = None" in the mapper subnodes#asahi-dev · 2022-04-24 13:52
jannauusb is broken though#asahi-dev · 2022-04-24 13:52
jannaudart-dcp/disp0 works with subpage end set to 0#asahi-dev · 2022-04-24 13:51
jannauit doesn't seem to be required/supported for all darts though#asahi-dev · 2022-04-24 13:39
jannauno, I'm just confused by differences between asahi and dart/t6000. subpage protection in asahi is setup in paddr_to_iopte() and in dart/t6000 it's in __arm_lpae_init_pte()#asahi-dev · 2022-04-24 11:29
jannauoops, I forgot to setup subpage protection and nothing broke#asahi-dev · 2022-04-24 09:41
jannau+me#asahi-dev · 2022-04-24 09:41
jannauI wonder how much Robin will hate for this https://github.com/jannau/linux/commit/4aeb93dac0857a027a1816be1bda9a663b6e71bd#asahi-dev · 2022-04-23 16:38
jannaushouldn't it make a difference for iommu_dma_sync_*#asahi-dev · 2022-04-23 16:35
jannauyes, I'm want to use IOMMU_CACHE to control the uncached bit. As far as I can see IOMMU_CACHE is controlled by the dev->dma_coherent which is set by the dma-coherent dt property#asahi-dev · 2022-04-23 16:30
jannauI'm looking into using it for the t6000 dart uncached PTE bit#asahi-dev · 2022-04-23 16:29
jannausven: are you sure that the missing "dma-coherent" makes no difference? casual reading of dma-iommu suggests it might.#asahi-dev · 2022-04-19 18:35
jannauindeed, the last thing they added#asahi-dev · 2022-04-19 18:34
jannaumaz: Corellium had/has usb4/thunderbolt support?#asahi-dev · 2022-04-19 18:33
jannauis there even usb4/thunderbolt HW beside Intel (and apple)? zen4 will be pcie based too#asahi-dev · 2022-04-19 15:42
jannaushutdown/restart is slower but still reasonable fast#asahi-dev · 2022-04-19 15:32
jannauonly annoying case was window resizing/moving which could have been caused by dcp tracing#asahi-dev · 2022-04-19 15:31
jannaumarcan: yes, macos HV is much faster. linux seems to be faster as well although it was less noticeable#asahi-dev · 2022-04-19 14:50
jannaumarcan: thanks. format check failed