profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/marcan/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.
Hector Martin marcan Tokyo, Japan https://marcan.st If it ain't broke, I'll fix it!

marcan/cl-waifu2x 134

OpenCL implementation of waifu2x image upscaling

marcan/blitzloop 131

Open source karaoke software

marcan/deresuteme 72

The code behind https://deresute.me

marcan/certbot-external 52

Certbot plugin that uses an external shell script for domain validation

marcan/asbestos 40

PS3 Linux bootloader

marcan/hackit 25

Hack It Framework

marcan/cd-analysis 21

Scripts to analyze audio CD data quality

marcan/circa 12

Consumer IR encoding/decoding/format conversion

marcan/esp-sensornode 10

ESP8266 + BME280 sensor node

issue commentlibimobiledevice/idevicerestore

AS/M1: Support the iMac M1 (AppleTCON firmware)

Sure, how do you usually do that? Is it just plaintext HTTP?

The other issue I had was a missing USB PID in usbmuxd, but you saw that PR already.

marcan

comment created time in 17 hours

issue commentlibimobiledevice/idevicerestore

AS/M1: Support Mac OS Monterey (12.0)

FWIW, BootabilityBundle is a new Preboot component that is used as a shared library so that newer OSes can get correct BootPolicy and bootability configuration even when the OS doing so (via the boot picker) is older. Presumably idevicerestore needs support for sending this (it's just a directory in the ipsw).

marcan

comment created time in a day

issue openedlibimobiledevice/idevicerestore

AS/M1: Support the iMac M1

Trying to restore 11.4 (UniversalMac_11.4_20F71_Restore.ipsw) fails with:

Checkpoint 4873 complete with code 0
Updating Veridian (66)
Checkpoint 4881 complete with code 0
Unknown operation (81)
ERROR: restore_send_firmware_updater_data: Got unknown updater name 'AppleTCON'.

TCON is the timing controller for the display AIUI, so it's unsurprising that this would be unique to the iMac or to Macs with a built-in screen in general.

created time in a day

issue openedlibimobiledevice/idevicerestore

AS/M1: Support Mac OS Monterey (12.0)

Restoring the latest 12.0 beta7 (UniversalMac_12.0_21A5522h_Restore.ipsw) fails with the following:

restore-step-ids = {0x11030655:12}
restore-step-names = {0x11030655:is_host_compatible}
restore-step-uptime = 6
restore-step-user-progress = 0
host/device compatibility check found that a message is missing or is not supported (optional): FirmwareUpdaterPreflight

host/device compatibility check found that a message is missing or is not supported (required): BootabilityBundle

host/device compatibility check found that a message is missing or is not supported (optional): RestoreLocalPolicy

host/device compatibility check found that a message is missing or is not supported (optional): AuthInstallCACert

Detected a mismatch between the messages the device can send and what the host can support.
Checkpoint engine recorder path set to /mnt5
[16:27:44.0228-GMT]{3>7} CHECKPOINT FAILURE:(FAILURE:10) RESTORED:[0x0655] is_host_compatible [0]D(This host version is unsupported. You may need to update your host tools.)

created time in a day

PR opened libimobiledevice/usbmuxd

usb: Add more Apple Silicon retstore PIDs

1904 is the M1 iMac; presumably this is now a range

This isn't enough to get idevicerestore to work with recent versions of macOS, but it's a start.

Signed-off-by: Hector Martin marcan@marcan.st

+8 -6

0 comment

3 changed files

pr created time in a day

create barnchmarcan/usbmuxd

branch : as-pids

created branch time in a day

fork marcan/usbmuxd

A socket daemon to multiplex connections from and to iOS devices

https://libimobiledevice.org

fork in a day

push eventAsahiLinux/linux

Hector Martin

commit sha 6012693c4203d53e54aeed744287c3a5e8c8c238

t8103.dtsi: remove ps_ans2 from the parents of ps_apcie_st Both ps_apcie_st and ps_ans2 should supply the actual ANS device node instead. This is likely closer to what the hardware does. Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in a day

push eventAsahiLinux/m1n1

Hector Martin

commit sha 2736f0dd0751592a5c31ba6d1ba3f6958997c177

hv: Transfer ACTLR_EL1 (not just EL2) to secondaries Also capture config at cpu0 guest entry time, to make sure we don't carry over guest changes to EL1 regs after that. Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 2 days

push eventAsahiLinux/m1n1

Hector Martin

commit sha 3a4b2870a820774b8648aea009e67c7ec1317b93

experiments/cpu_pstates.py: Only set DVMR if necessary Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 2 days

push eventAsahiLinux/m1n1

Hector Martin

commit sha febf83a981f28301914a7a42d96636487ecd0488

experiments/cpu_pstates.py: Enable DVMR unconditionally Seems this is not always enabled by the bootloader on some systems Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 2 days

push eventAsahiLinux/m1n1

Hector Martin

commit sha 9ed9c0d35da4a1916433913f5b8417d264848159

experiments/cpu_pstates.py: Enable DVMR unconditionally Seems this is not always enabled by the bootloader on some systems Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 2 days

push eventAsahiLinux/m1n1

Hector Martin

commit sha 698205d6427418dde3ff988a0500eac015649854

cpu_pstates: Enable DVMR unconditionally Seems this is not always enabled by the bootloader on some systems Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 2 days

push eventAsahiLinux/AsahiLinux.github.io

TophEvich

commit sha 2167a13084aa478ef1e1e30fd13ff828dd15d5fa

Progress Report Aug'21: pin referenced source (#16) So that that referenced source code on github stays in sync with the explanations of the 2021-08-14 progress report, all links have been changed from "main" branch to commit: c2c6da3df25c0605894244b4ea9387e882321efc which is the current version of the day. Signed-off-by: TophEvich <84676511+TophEvich@users.noreply.github.com>

view details

push time in 3 days

PR merged AsahiLinux/AsahiLinux.github.io

Progress Report Aug'21: pin referenced source

So that that referenced source code on github stays in sync with the explanations of the 2021-08-14 progress report, all links have been changed from "main" branch to commit: c2c6da3df25c0605894244b4ea9387e882321efc which is the current version of the day.

Signed-off-by: TophEvich 84676511+TophEvich@users.noreply.github.com

+5 -5

0 comment

1 changed file

TophEvich

pr closed time in 3 days

push eventTophEvich/AsahiLinux.github.io

Hector Martin

commit sha 6ad7a76a28ff25450132cadd0048997772a094f9

blog/2021/08/14-progress-report.md: s/Apple silicon/Apple Silicon/ Signed-off-by: Hector Martin <marcan@marcan.st>

view details

Hector Martin

commit sha 6fd48e13229966a799dbaba4e93b313ed137f2ef

blog/2021/08/14-progress-report.md: Add link to HV recap stream Signed-off-by: Hector Martin <marcan@marcan.st>

view details

Hector Martin

commit sha acc6cceaf0e8cd13730dc8c71306927aa0907a05

Merge branch 'main' into august_report_pin_hyperlinks_to_current_version

view details

push time in 3 days

push eventAsahiLinux/m1n1

Sven Peter

commit sha 0b4d5bd7937b9426e35e773d82a9beeb7500c463

proxyclient: hv: Add simple DWC3/XHCI/ATCPHY tracer Signed-off-by: Sven Peter <sven@svenpeter.dev>

view details

push time in 3 days

PR merged AsahiLinux/m1n1

proxyclient: hv: Add simple DWC3/XHCI/ATCPHY tracer

Signed-off-by: Sven Peter sven@svenpeter.dev

+298 -0

0 comment

1 changed file

svenpeter42

pr closed time in 3 days

push eventAsahiLinux/m1n1

TophEvich

commit sha 913c775165d74c9479290533265cb91c3515c770

proxyclient: removed unused code Little clean-up of unused proxyclient code. Code used/referenced by commented out sections were taken into account and left behind to be used by it at a later point. Signed-off-by: TophEvich <84676511+TophEvich@users.noreply.github.com>

view details

push time in 3 days

PR merged AsahiLinux/m1n1

proxyclient: removed unused code

Little clean-up of unused proxyclient code. Code used/referenced by commented out sections were taken into account and left behind to be used by it at a later point.

Signed-off-by: TophEvich 84676511+TophEvich@users.noreply.github.com

+3 -14

4 comments

7 changed files

TophEvich

pr closed time in 3 days

PullRequestReviewEvent

issue commentAsahiLinux/m1n1

kernel image not aligned on 64K boundary (uboot via m1n1)

I'm confused; you say you're using grub, but that screenshot shows the EFI stub. Are you booting the kernel as an EFI application instead of as a native ARM64 kernel from GRUB?

We aren't really testing with GRUB yet, so various things might be broken... though this sounds like a bug in downstream components, not m1n1.

juliosueiras

comment created time in 3 days

issue commentAsahiLinux/m1n1

trace_all.py fails while tracing with "Event checksum error"

That normally happens if some error happened that triggered an exception message that stepped over the MMIOTrace code. Yesterday I just pushed a rework that finally makes handling such exceptions properly possible, can you try again? (remember to always chainload m1n1 before using the hypervisor; we do not support mixing versions).

FWIW, I tried your command line and it works fine for me, on a Mac Mini with the same kernel (though of course it panics due to not being able to mount the ramdisk).

Also, hint: you probably want to pass serial=3 and cat the secondary m1n1 serial device to get the kernel log via serial instead of on the framebuffer. At least on my setup, XNU borks the HDMI output during the boot process so the single-user shell doesn't show up on there properly (not a hypervisor thing, it happens on bare metal too), but it works fine over serial. I think it's probably switching to a display mode that my capture card does not support or something like that.

zhuowei

comment created time in 3 days

push eventAsahiLinux/m1n1

Martin Povišer

commit sha d25581ddb39a7396a846f8684be63258e660634c

Fill in m1n1.hw.i2c Signed-off-by: Martin Povišer <povik@protonmail.com>

view details

Martin Povišer

commit sha 18bc2c7db11e403aa706e4a7f6c6ef4bd2ef4f20

Add m1n1.hw.admac, experiments/speaker_amp.py Add initial code for driving the ADMAC hw blocks, also add a script which shows it in action by streaming audio to the Mac mini's embedded speaker. Signed-off-by: Martin Povišer <povik@protonmail.com>

view details

push time in 3 days

PR merged AsahiLinux/m1n1

Bit of audio

This is the code from a paste I posted earlier (https://gist.github.com/povik/66f69f539905bcde6f4cacbae1d24ad6) after some additional care.

(I will be happy to add sign offs once it's specified somewhere what those sign offs mean. :) )

+567 -0

3 comments

3 changed files

povik

pr closed time in 3 days

pull request commentAsahiLinux/m1n1

Bit of audio

Thanks! :)

I'm curious, are you interested in looking into writing a Linux driver for this? Audio has been on my list for right after the GPU gets on track, but with this much already figured out making an ASoC driver should be fairly easy. I'm happy to take care of it, but if you're interested feel absolutely free to go ahead too! It would be really cool to have basic audio working in the following month or two.

povik

comment created time in 3 days

pull request commentAsahiLinux/m1n1

Bit of audio

Indeed we use the same DCO as Linux. We should probably add that to the README...

povik

comment created time in 3 days

push eventAsahiLinux/m1n1

Hector Martin

commit sha 9dc332b66031da29fd57b20ec1d43c7ece6a89eb

tools/dump_pmgr.py: Show fixed clock usages from DT Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 4 days

push eventAsahiLinux/linux

Hector Martin

commit sha 83f87c34531ec8bf83cc234ee3c48ce9cdd157bb

apple-pmgr-pstate: Fix OF refcounting bug Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 4 days

push eventAsahiLinux/linux

Hector Martin

commit sha 6e7a19be1a5e4496a87956089d4172e92c085123

apple-pmgr-pstate: Fix OF refcounting bug Signed-off-by: Hector Martin <marcan@marcan.st>

view details

push time in 4 days