I did build 8.2.5 with kernel 4.16 and Mesa 18.0.3, but I don't know if it is enough for new AMD Apu. You could try my 8.2.5-6.42.1 from here
LE 8.2.5 with UHD 630 / Coffee Lake support and luks, lvm2, dm-raid
I did build 8.2.5 with kernel 4.16 and Mesa 18.0.3, but I don't know if it is enough for new AMD Apu. You could try my 8.2.5-6.42.1 from here
LE 8.2.5 with UHD 630 / Coffee Lake support and luks, lvm2, dm-raid
For adventures natures i just build a 8.2.5-6.42.1 with kernel 4.17-rc5. I did not try it on my own box, because i need to go to work.
@piotrasd since now 4.17-rc5 is building (rc4 would not) i can also try to make a drm-tip verion. I think that will be a extra version like 8.2.5-9.42.0.
8.2.5-5.42.1 with kernel 4.14.40 and Coffee Lake support
...
- libva-intel-driver disabled and replaced by intel-vaapi-driver 2.1.0 (this could break tvheadend42 plugin VAAPI support)
...
Sorry @psycmos i needed to update the intel lib and my fear was that will break tvheadend and it did. I just did a short try to build tvheadend42 in my 8.2.5-6.42.1 build env, but it wont build with or without transcoding.
The last 2 days i rebuild all my image files and began from scratch for more order in my configs and patches.
All the new versions are now with a .1 like 8.2.5-X.42.1.
what was done
- rebuild all kernel config from scratch and used menuconfig to switch on what is needed, after that diff and applied this to the kernel config from git
- added to all versions PR init: add sky42 enhancements/bug fixes by MilhouseVH · Pull Request #2689 · LibreELEC/LibreELEC.tv · GitHub
- removed all config/patches from 3.42, 5.42 and 6.42 for architektures that are not build
- added Wetek_Play to 1.42
- added RPi to 1.42, but i dont have one to test
- wrote a small readme and put it in all the 8.2.5 download folders
So if your setup is running fine you dont need to do anything, but you can.
I see a potetial problem or two.
A old installation is updated with a image with PR 2698. After that the user does getedid which changes syslinux.cfg, but it is a UEFI system with a existing /flash/EFI/BOOT/syslinux.cfg and getedid only changes /flash/syslinux.cfg. So the UEFI system will still use the /flash/EFI/BOOT/syslinux.cfg which is not changed.
Move the /flash/EFI/BOOT/syslinux.cfg to /flash/EFI/BOOT/syslinux.cfg.off will solve that. Deleting it may destroy work in syslinux.cfg what somebody did on his box.
@milhouse and chewitt thanks for the explanation. I did read all the old requests and expected such answer and understand it too. @milhouse you did kind of ask "is that all you need for encryption" and i could not resist to ask for my cryptsetup feature request and also practicly knowing what the answer will be. Sorry.
I do like less is more too. At work i do whenever possible working by the KISS principal.
How about giving the community the change to build addons with the encryption, lvm2, mdadm tools, but for that we need support in the kernel. Then i will lern to build addons. And there is less discussion why this tool gets into LE and that not. Most of them need kernel support.
I checked out le master and build 5 differen le versions in 2 flavors
PROJECT=Generic ARCH=x86_64
PROJECT=RPi DEVICE=RPi ARCH=arm
PROJECT=RPi DEVICE=RPi2 ARCH=arm
PROJECT=Rockchip DEVICE=RK3399 ARCH=arm UBOOT_SYSTEM=rockpro64
PROJECT=Amlogic DEVICE=S912 ARCH=arm
1st full build with IMAGE_SUFFIX=git (fresh and clean from git)
then i changed the kernel config for encryption, lvm2, raid
2nd just make again with IMAGE_SUFFIX=git+md+crypt+lvm+raid
2 concerns i did read about the kernel
- size for the kernel: that changes less than 10KB and the system around 1MB
- i am pretty sure there is no mesurable performance impact as long as you dont use the new modules
You can see for your self here HiDrive all images i build and the git diff for the kernel changes
ls -g kernel/system sizes le-9.0-testing-images-ls.txt
kernel config patch le-9.0-kernel-config-git+md+crypt+lvm+raid.patch.txt
Please consider the support in the kernel. I will build all the kernel config diff or more images for testing. Just ask and say what to build. Sorry for the patches again, i promise i will learn git.
Thanks for reading
sky42
new 8.2.5-5.42 here HiDrive with the following changes
- kernel 4.14.40
I did rebuild my 8.2.5 with init: add sky42 enhancements/bug fixes by MilhouseVH · Pull Request #2689 · LibreELEC/LibreELEC.tv · GitHub and it applied except one hunk for scripts/mkimageand syslinux.cfg changes. That one i did by hand.
This PR solves the problem of this thread. I just tested Generic.
Thanks
The 8.2.5-6.42 build is now done. Please read carefull because i needed to disbale some WLAN drivers to build 4.16
8.2.5-6.42 with kernel 4.16.8 and Coffee Lake support
- based on 8.2.5-5.42
- Coffee Lake Support without i915 alpha features
- disabled additional driver RTL8192CU RTL8192DU RTL8192EU RTL8188EU RTL8812AU bcm_sta
- enabled kernel driver RTL8XXXU RTL8192CU
- removed kernel patch linux-051-ouya_controller_support.patch and linux-053-spinelplus-remote-0.2.patch
- added openssl to linux kernel PKG_DEPENDS_TARGET
- Mesa 3D update 18.0.3
NINjak I be very interessted if the 4.16 solves your LG problem. I had also a Problem wirh my LG 55E6D and that was gone with 4.14
tomtomclub dont use a testing version if you dont have to
so here it is 8.2.5-6.42 with kernel 4.16.8 HiDrive
5schatten i had trouble with 4.11.12. Then moved on to 4.13 and not really better. Found a patch for 4.13 that helped. And since 4.14 all ist fine and all the LSPCon fixing was already in the kernel.
One big issue is the Firmware in the LSPCon. For Parade PS176 it should be 7.85 and for MDCP2800 for ASrock 1.66 for some Intel NUCs 1.72 and i did see also 1.71. From your HTPC setup description i would guess you nee MDCP2800 Firmware and it must be at least 1.66.
The description says LSPCon fix for my 8.2.5-3.42 beause the kernel 4.14. fixes it. Sorry that you searched in my patches, because i did not state it clear.
You can check what you have with
cat /sys/kernel/debug/dri/0/i915_display_info
NINjak try this one HiDrive with kernel 4.16.8 build just now and still with a wrong version numnber 8.2.5-5.42. I am just doing a full rebuild and that takes about 90min. On my i3-8100 the testing version bootet just now, but no testing at all so far.
The real 8.2.5-6.42 wil follow and here are the changes so far
8.2.5-6.42 with kernel 4.16.8 and Coffee Lake support
- based on 8.2.5-5.42
- Coffee Lake Support without i915 alpha features
- additional driver disabled bcm_sta RTL8192CU RTL8192DU RTL8192EU RTL8188EU RTL8812AU
- kernel driver enabled RTL8XXXU RTL8192CU
- removed kernel patch linux-051-ouya_controller_support.patch and linux-053-spinelplus-remote-0.2.patch
Thanks!
I will apply and check it with my comminity build.
Edit: 2018-05-09 Sorry erro corrected: i did see cryptsetup needs lvm2 too
Interessting change. I will try to incorporate the PR in my community build LE 8.2.5 with UHD 630 / Coffee Lake support and luks, lvm2, dm-raid and test it. When i checked all the changes i will report back.
But the encryption needs one more feature request cryptsetup luks/truecrypt/veracrypt/loopaes, lvm2 and probably dm-raid too (patches included)
There is a small kernel config change and the cryptsetup, lvm2 packages are needed. The package mdadm is in there too, because the kernel has all what is needs to do dm-raid.
I am fairly familiar with encryption and hope to make it easy enough with some dokumentation and some scripts. With the help of a developer it should be not that hard to mount encrypted usb drives as they are attached. That will work too for e.g. veracrypt encryption and a exfat fs on there (created and useable on windows).
This commit should do the trick. I will try changing my community build and check. Tomorrow i am on vacation for 5 days and will have time to check it.
Sorry i am not yet familiar how to quote in the forum. I try to answer in order.
What do i want to do with these hooks:
- probably tweak the update process some how after flash is mounted, but no plans for that so far
- when storage is mounted i will try to mount .kodi/userdata encrypted on a different partition (working on that), at the moment i have a solution to do that as systemd unit before kodi starts
- my first PoC was getting a password from the user in the boot process (if the user wants) and post-storage.sh calls get-pw-init.sh.txt and then from autostart.sh do this mount-all.sh.txt done with OE 5 and truecrypt
- the real goal is when there is encrypted storage (partition) and i can open it umount unecrypted storage and mount the encrypted one instead
- after that mounting encrypted data disks with the help of a crypttab, that is mainly for offline usage on the road with no server to get the keys/data from (NUC and disks with some films in my luggage)
- this need the password from the boot process
- when all i need is in the LE image and not a addon (like truecrypt in OE 5) i can do it earlier and since there is no truecrypt anymore (and it is old and needs replacing) i use cryptsetup, that can mount also truecrypt/veracrypt volumes
I am fine with 1- 3 and the commit looks fine
And yes i know i need to lern git and start to use it. I still use cvs (dont laugh) for versioning my server configs/scripts. I need to sit down with one of our developers at work and learn how to use git, have my own github repo and doing pull requests. Sorry for the inconvenience with the patches.
For grub2 that depends on your grub version. On CentOS and Ubuntu they are diffrent.
On my pxe server at home i use one /uefi/grub.cfg and in all the other places is a grub.cfg with "configfile /uefi/grub.cfg", because for secure boot i need to use the grub from the distro.
So yes it should be possible and the content of /flash/EFI/BOOT/grub.cfg must be "configfile /grub.cfg"
There is also a way to write one grub.cfg which does different things if it is uefi or not. I had my guys at work build one with this differentiation to start the correct memtest. I dont thing that is needed for LE, but i can get these configs.
Thanks for the informations. The UID feature is cool. And with PXE i did full diskless with OE 4 (i think) and my old Zotac min PC with Nvidia GPU.
And now as you say it: yes the last sd card problem was a long time ago. I dont hang on the SYSTEM_TORAM_LIMIT feature. It was just there and i used it (i think) some years sind OE 5.
I was just excided that i finally did take the time to share my work in the forum and i like giving it back. And sometimes a nice sideeffect is that i have something less to maintain.