* fix permisson denied error happening in line 79
`mktemp` creates files with `-rw-------` (on Ubuntu bionic at least), which causes `line 79` to throw a permission denied error if the file isn't world readable.
* fixes umount error caused by single quotes
* force metadata duplication for btrfs root
Without this change metadata duplication is automatically turned off for `mkfs.btrfs` when the **build system** has a SSD drive.
However, since Armbian image is most likely flashed to a SD card afterwards, the drive type of the build host system should not be the deciding factor.
Metadata duplication is a good idea (better reliability) to have on a SD card.
From `man mkfs.btrfs`:
```
-m|--metadata <profile>
Specify the profile for the metadata block groups. Valid values are raid0, raid1, raid5, raid6, raid10, single or dup, (case does not matter).
A single device filesystem will default to DUP, unless a SSD is detected. Then it will default to single.
```
* armbian-resize-filesystem fails on first run due to missing fdisk
The `fdisk` package is missing in the `minimal` image, in bullseye at leat. Maybe other, future minimal images as well.
This will cause the `armbian-resize-filesystem` script to fail on first boot, which contains several
`fdisk` references. This could be fixed in the `bullseye` specific section, but since fdisk is so
essential for Armbian, it's probably best to list it in the general, mandatory packages to make sure it's available in every image in any case.
* fix for https://github.com/armbian/build/issues/1584
Since Debian buster the final `update-initramfs` call must be made
on the finished image where the final root device is mounted, else
`cryptsetup-initramfs` hooks fail to detect the root device.
* Merge remote-tracking branch 'upstream/master'
* make sure cryptsetup-initramfs is installed in any case if CRYPTROOT_SSH_UNLOCK is configured
Before Debian **bullseye**, `cryptsetup-initramfs` was a hard dependency of the `cryptsetup` package. Not anymore.
Therefore we need to install `cryptsetup-initramfs` explicitly, else CRYPTROOT_SSH_UNLOCK won't work.
NanoPi R2S has the same issue as ROC-RK3328-CC.
With working GPIO, during init the GPIO state s reset.
This causes the sdmmc regulator to shut down, preventing detection.
Removing and replacing the card will allow it to be detected, but that should not be necessary.
Fix this by setting the regulator on at boot.
Signed-off-by: Peter Geis <pgwipeout@gmail.com>
Link: 75aa567803
Signed-off-by: Jayantajit Gogoi <jayanta.gogoi525@gmail.com>
* patch: kernel: rockchip-dev: Handle more resolutions with HDMI (β)
The added patch add more PLL configurations, in order to satisfy
more HDMI frequencies requirements.
This should allow users to benefit from more resolutions.
However, this is fairly untested. I only tested it on my 1080p
screen and, yeah, it works but so does the kernel without this
patch.
Signed-off-by: Miouyouyou (Myy) <myy@miouyouyou.fr>
* patch: kernel: rockchip-dev: Handle more resolutions with HDMI (β)
Now with proper patches names.
Anyway :
The added patch add more PLL configurations, in order to satisfy
more HDMI frequencies requirements.
This should allow users to benefit from more resolutions.
However, this is fairly untested. I only tested it on my 1080p
screen and, yeah, it works but it also does without this
patch.
So could anyone test this with the following configurations ?
* [ ] A HDMI 4K screen
* [ ] A HDMI 1366x768 screen resolution
* [ ] A VGA screen with a VGA to HDMI adapter
* [ ] A DVI screen with a DVI to HDMI adapter
Signed-off-by: Miouyouyou (Myy) <myy@miouyouyou.fr>
* Using the patches provided by @Kwiboo for HDMI
This provides me way more resolutions, including interlaced modes.
I don't see 4K modes, though the screen attached can't do it,
so I still don't know if that can help provided 4K modes on mainline
kernels.
Signed-off-by: Miouyouyou (Myy) <myy@miouyouyou.fr>
* Support for 4K screens
Using fixes provided by @czak
Tested and approved by @czak too !
Signed-off-by: Miouyouyou (Myy) <myy@miouyouyou.fr>
There have been intermittent issues with SPDIF output on H3 & H2+ devices.
Digging into the H2+ & H3 datasheets I encountered an important clue on
table 8.7.2.2: s_clk needs to be 4x24.576 MHz or 4x22.5782 Mhz.
Probably due to how SPDIF is a huge oversample of the audio frequency.
Co-authored-by: George Lander <lander@jagmn.com>
* Attach Meson64 CURRENT to 5.6.y and make DEV = CURRENT at this point.
There is a lot of changes to 5.7.y and can be done after release or by someone that feels a need for this right now.
* Delete meson64_fclk_div3.patch
this fix has been upstream for some time, I had issues with it on 5.4, which is why it had been removed there.
* [ meson64 current ] kconfig tweak
disable Rockchip SoC drivers
* [ meson64 current ] remove rockchip patches
* [ meson64 current ] disable CMA patch GX
* [ meson64 current] add libretech cc audio patch
This brings the dts even with the khilman 5.8/integ branch which includes all of the audio changes. The kernel is registering audio devices, however nothing comes out. committed in case someone wants to spend time debugging/testing theories.
* fix permissions
was editing from another machine, accidental change of permissions in the patch
* Set default mixer settings
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* [ meson64 current ] GXL audio commit
This moves to the mainline patches and covers le potato and la frite. Adjusted asound config to handle a commonized sound card name.
* [ meson64 current ] add gxbb audio WIP
Playback is too fast, a clock setting is off somewhere.
* Update kernel configs
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
Co-authored-by: Tony <tonymckahan@gmail.com>
* Switching Odroid C1 boot partition to FAT
U-boot FAT driver is very unstable and there is no wish to dig into u-boot. Seems like a sane workaround @plntyk
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* Add notes for separate boot partitions
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* Typo
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* Bugfix: use zImage for converting to uImage since on FAT partitions we use move
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* Cleanup C1 boot scripts
We have no plans to ever support legacy kernel again. Removing related things
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* A bit more radical change. Moving meson current to (new) legacy, current to 5.6.y and dev to some development branch / commit.
* Remove unneded patch
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
Kernel install image is broken on VFAT images since it unmounts directory in preinstall and doesn't mount back. We used to have this for NAND / FAT install which fixes this in the postinst script. Removing since we have no more support for NAND anyway. Tested on Odroid C1. @the-going
Signed-off-by: Igor Pecovnik <igor.pecovnik@gmail.com>
