#asahi-dev · 2022-09-25 18:42
jannauplatform 38bc00000.dcp:piodma: DMA addr 0x000001002e268000+32768 overflow (mask ffffffff, bus limit 0).#asahi-dev · 2022-09-25 18:42
jannaulooks like the unsafe sg_table usage with dcp/piodma caught up with us. although it looks like it could be another instance of physical memory must start below 1<<32#asahi-dev · 2022-09-25 11:39
jannaucpufreq is broken on the m1 ultra in asahi-dev. Can't access the cpu registers on the second die now that they are out of "ranges". making cpufreq_hw a multi die node seems to work (working on dart changes which currently crash so I can't say yet)#asahi-dev · 2022-09-24 13:18
jannauchadmed: I would have tested a local kmod build and not the system one#asahi-dev · 2022-09-24 13:07
jannauwhat's the smallest and easiest to reproduce example? kmod with zstd compressed modules?#asahi-dev · 2022-09-24 12:54
jannauno AMX in non-Apple llvm, see https://github.com/llvm/llvm-project/blob/main/llvm/lib/Target/AArch64/AArch64.td for llvm's model of the cpu. CycloneModel, ProcessorFeatures.AppleA14 and TuneAppleA14#asahi-dev · 2022-09-22 21:28
jannaunice, module and firmware loading working in the initramfs with tools/run_guest_kernel.sh via multiple (3) cpio archives#asahi-dev · 2022-09-22 18:12
jannau12.3 is "system-firmware-version = iBoot-7459.101.2"#asahi-dev · 2022-09-22 18:07
jannauthat could be a hint that your installed m1n1 is too old#asahi-dev · 2022-09-22 18:04
jannaudcp still doesn't work for you? did you revert the changes to rt_bandwidth_setup_ap for dcp\#asahi-dev · 2022-09-22 17:59
jannausven: https://www.jannau.net/asahi/001_dcp.py_start.log.txt#asahi-dev · 2022-09-22 17:57
jannauextended logging works now and I get "_pmgrConfigVideoClock Failed to configure video clock freq= 74250000" but it still works, even for an uninitilized display#asahi-dev · 2022-09-22 17:52
jannauok, I'm not#asahi-dev · 2022-09-22 17:52
jannauI don't know why I thought it was already merged#asahi-dev · 2022-09-22 17:51
jannausven: argh! you should use https://github.com/jannau/m1n1/commits/experiments_dcp_set_mode#asahi-dev · 2022-09-22 17:45
jannau"sr_setProperty(IOMF/M3TimingParameters = {'subframe-duration-nclks': 0 ..." suggests it's not#asahi-dev · 2022-09-22 17:45
jannauis your display on and displaying something when you run dcp.py?#asahi-dev · 2022-09-22 17:20
jannauextended log works with dcpext_iboot.py#asahi-dev · 2022-09-22 17:15
jannauI don't see "[system] Ignoring report on channel 3" in your output#asahi-dev · 2022-09-22 17:07
jannauis what I have#asahi-dev · 2022-09-22 17:07
jannauand dcp instead of dcp_iboot and adding those endpoints to DCPClient#asahi-dev · 2022-09-22 17:01
jannauI have that before starting dcpep#asahi-dev · 2022-09-22 17:00
jannauyou still get more syslog messages#asahi-dev · 2022-09-22 16:57
jannausven: https://www.jannau.net/asahi/dcp.py_log.txt#asahi-dev · 2022-09-22 16:52
jannaustill no "Failed to configure video clock freq=" messages#asahi-dev · 2022-09-22 16:45
jannauworks without hv#asahi-dev · 2022-09-22 16:44
jannauany fixes on dcp tracer side?#asahi-dev · 2022-09-22 16:36
jannauor does it depend on statrting the other endpoints as well?#asahi-dev · 2022-09-22 16:35
jannaudoes that depend on your EPIC fixes? I'm getting an error after a few system messages in "EPICCmd.parse_stream(fd)"#asahi-dev · 2022-09-22 16:18
jannauno "Failed to configure video clock freq=" here for dcp#asahi-dev · 2022-09-22 16:14
jannaudoes the installed m1n1 already shuts dcp down after configuring the display?#asahi-dev · 2022-09-22 16:13
jannauah, I overread the dcpext#asahi-dev · 2022-09-22 16:12
jannauI guess it crashes because we continue after the set_digital_out_mode() error#asahi-dev · 2022-09-22 16:11
jannauworks for me with chainloaded m1n1 but not with the not that old installed m1n1#asahi-dev · 2022-09-22 06:03
jannauprobably easier as well as it then only depends on a single other tree (sound/for-6.1, https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git/log/?h=for-6.1)#asahi-dev · 2022-09-22 05:57
jannaumarcan: I've still haven't heard anything from Joerg. he handled some iommu fixes yesterday. I would take the dart t6k dt binding commit and reply to it that you will merge since you have dt changes which depend on it for verification (if you want to prepare/send) your pull request today#asahi-dev · 2022-09-21 19:32
jannauhmon[m]: what do you want to know? I have an ioreg dump and device dumped ADTs. those might be more detailed than the temlate ADTs#asahi-dev · 2022-09-21 06:03
jannauyes, that was more a comment that marcan's hardcoded mode IDs are gone#asahi-dev · 2022-09-21 05:19
jannauI've changed experiments/dcp.py to select the mode based on the score a couple of weeks ago#asahi-dev · 2022-09-20 20:17
jannauI was more thinking of comparing dcpext macos / your m1n1 scripts#asahi-dev · 2022-09-20 20:10
jannauthat's easier to compare than the full output. it's only dcpep though#asahi-dev · 2022-09-20 20:09
jannauthe dump can be analysed with python -m m1n1.fw.dcp.parse_log dcp.log#asahi-dev · 2022-09-20 20:09
jannauhave you already found trace_dcp dump? uncomment/copy the last line in hv/trace_dcp.py#asahi-dev · 2022-09-20 19:57
jannau0x23bc3c000 is pmp and 0x23b738018 is inside pmgr, the last 3 regs of disp0 and dispext0 are identical#asahi-dev · 2022-09-20 19:46
jannaui.e. wrong values should try to set the wrong clock#asahi-dev · 2022-09-20 19:46
jannauare you sure that the values in rt_bandwidth_setup_ap don't make a difference? that are offsets from the pmgr reg in dcp(ext)'s regs#asahi-dev · 2022-09-20 19:39
jannauis the DISPEXT_CPU0 power-domain enabled? I doubt you would get that far if it wasn't#asahi-dev · 2022-09-20 19:37
jannauor the video clock is not in the adt since dcp manages it directly#asahi-dev · 2022-09-20 19:35
jannaumaybe, for disp0 it none of the 2 clocks is the pixel or video clock though#asahi-dev · 2022-09-20 19:20
jannauhave you checked pmgr with tools/dump_pmgr.py?