[fritsch]

It won't - this is kernel level.

[alxtr1bal]

@alxtr1bal: Include [pkg] patch: xf86-video-amdgpu: Latest commits (6a1ba04) adding Polaris 10 & 11 support (xf86-video-amdgpu) <- such commits are highly relevant for you.

Next time, please gather logfiles. Xorg.0.log and dmesg | pastebinit

With these we can file bugs at bugs.freedesktop.org so that this gets fixed. It won't happen magically.

Ok, last update not working :

Xorg log : http://sprunge.us/VcJK
dmesg : http://sprunge.us/XcBW

here's what i did to install :
First , installed the 7.90.005, with the processor GPU as primary.
Then , put the latest update in /Updates, change the primary video card as the RX 460, and boot.
Update is going ok, reboot, and then the Xorg won't start.

Not sure if this is the right way...

Thanks for the logs. It looks like the "ATI Radeon" driver is being loaded and not the updated "amdgpu" driver. Not sure how easy it is to work around that. Do you have two GPUs in your system, a Kaveri and Polaris? Can you disable the Kaveri?

For those logs , the Kaveri (AMD A10 GPU) was disabled in the BIOS.

Question is : how can i do fresh install of LE with the new drivers ?

Because what i did is i had an installed LE with Kaveri enabled, updated to #0908, then installed this new RX 460 card and disabled the Kaveri in the bios. Then Xorg does not start... So may be it's because i did not install with this card enabled at the first time ... but this is not possible.

[madmarkym]

Ok, last update not working :

Xorg log : http://sprunge.us/VcJK
dmesg : http://sprunge.us/XcBW

here's what i did to install :
First , installed the 7.90.005, with the processor GPU as primary.
Then , put the latest update in /Updates, change the primary video card as the RX 460, and boot.
Update is going ok, reboot, and then the Xorg won't start.

Not sure if this is the right way...

Thanks for the logs. It looks like the "ATI Radeon" driver is being loaded and not the updated "amdgpu" driver. Not sure how easy it is to work around that. Do you have two GPUs in your system, a Kaveri and Polaris? Can you disable the Kaveri?

For those logs , the Kaveri (AMD A10 GPU) was disabled in the BIOS.

Question is : how can i do fresh install of LE with the new drivers ?

Because what i did is i had an installed LE with Kaveri enabled, updated to #0908, then installed this new RX 460 card and disabled the Kaveri in the bios. Then Xorg does not start... So may be it's because i did not install with this card enabled at the first time ... but this is not possible.

I tried installing Ubuntu 14.04 LTS server then installing Kodi nightly build then installing the amd drivers on top of that. Kodi still wouldnt work even though everything installed fine.
The drivers for this card are junk it wont work in Windows properly either.

[Milhouse]

You are the only one to find this out, what happend between the last version that was ok and the follwing where power on is broken.

It's all there. In the changelogs.

If #0420 works, as claimed, but the next build (#0423 in this case) doesn't work, then look at what was added or changed in #0423 - it's all listed, nothing is hidden.

In this instance the most likely cause is the change of kernel from 4.4.7 to 4.6.0-rc4. This means there is something in the newer kernels that breaks resume for your hardware. This is hard to diagnose, and might even be a manufacturer issue (BIOS), so unfortunately there's little I can do about that - there is no guarantee that resume will work for everyone, it's flaky at best.

To confirm if it is the kernel change, you can try builds #0509 (4.6.0-rc7) then #0509x (4.4.7) - the only difference between these two builds is the kernel. If resume DOES NOT work with #0509, but DOES work with #0509x, then we know it's the kernel change that is responsible.

[Milhouse]

Not sure,
@MilhouseVH you might want to try just updating xf86-video-amdgpu to master and see if that works.

That's pretty much what I did, albeit by patching the base package to bring it up to master. In tonight's build I'll bump the package instead, and pull the latest source with no patching. Net result should be the same, however.

[Milhouse]

Question is : how can i do fresh install of LE with the new drivers ?

Because what i did is i had an installed LE with Kaveri enabled, updated to #0908, then installed this new RX 460 card and disabled the Kaveri in the bios. Then Xorg does not start... So may be it's because i did not install with this card enabled at the first time ... but this is not possible.

Installing with 7.90.005 and upgrading to #0908, then disabling Kaveri and installing RX460, should work - assuming the RX460 drivers are working. It does sound like the drivers may not yet be ready for prime time, or the 4.7.x kernel may not like your card (or we're hitting a configuration issue we've not seen before).

Next week I may give 4.8-rcX a try... I might even throw up some early experimental builds, but unfortunately I'm having a few build issues (build box hardware failure).

[R.O.H.3000]

is it normal I still have high CPU usage with the latest builds?

[Hanxi]

Hi,

I had updated my Zotac Zbox ci321 nano from libreelec v7.02 to libreelec v7.90.005 and now I have issues to wake up the device via IR-remote. The Power.Off-button and all others works fine, only Power.On works no more. With v7.02 there was no problems.
Then I tested the milhous-rls and found out, that the releases up to 'LibreELEC-Generic.x86_64-8.0-Milhouse-20160421040821-#0420-gfee76be' works with rc-power.on, later releases up to the current version #907 not.

Is there any chance for a fix?

This is likely an issue between your BIOS and the more recent kernels - see if you have any BIOS updates available. Unfortunately there's little that anyone can do about this kind of issue, as it's likely to be a hardware specific issue (ie. buggy BIOS).

Same problem here, same ZBox. Bios is up to date.
@Milhouse:
You are the only one to find this out, what happend between the last version that was ok and the follwing where power on is broken.
Maybe in this stadium, there are not so many users, who tried the alpha versions, but when its final and there is this this IR wake up issue, there will be many more.
Please help us

Hi,

I mean, it can't be the BIOS.

I make a UEFI dual-boot system. first I install libreelec, then Win10 with rEFind bootmanager. Basically, the whole works, but after libreelec start and shutdown, the rc-button Power.On don't work. On the other hand, when I starts Win10 and shutdown, Power.On works flawlessly, until I boot libreelec again... like a flip-flop.
There must be an error in the current kernel, because with windows there is no issue.

[Milhouse]

is it normal I still have high CPU usage with the latest builds?

Depends if what you are doing is normal. Maybe explain what it is you are doing, or tell us about your hardware, or even when this problem started. Test with a "clean" .kodi to eliminate any rogue add-ons as a cause.

[Milhouse]

I mean, it can't be the BIOS.

I make a UEFI dual-boot system. first I install libreelec, then Win10 with rEFind bootmanager. Basically, the whole works, but after libreelec start and shutdown, the rc-button Power.On don't work. On the other hand, when I starts Win10 and shutdown, Power.On works flawlessly, until I boot libreelec again... like a flip-flop.
There must be an error in the current kernel, because with windows there is no issue.

Test with #0509 and #0509x.

[Hanxi]

I mean, it can't be the BIOS.

I make a UEFI dual-boot system. first I install libreelec, then Win10 with rEFind bootmanager. Basically, the whole works, but after libreelec start and shutdown, the rc-button Power.On don't work. On the other hand, when I starts Win10 and shutdown, Power.On works flawlessly, until I boot libreelec again... like a flip-flop.
There must be an error in the current kernel, because with windows there is no issue.

Test with #0509 and #0509x.

Testet. #509 dosen't work, #509x works like libreelec7.0_Jarvis without issues.
Ergo kernel issue.

[tinzarian]

To confirm if it is the kernel change, you can try builds #0509 (4.6.0-rc7) then #0509x (4.4.7) - the only difference between these two builds is the kernel. If resume DOES NOT work with #0509, but DOES work with #0509x, then we know it's the kernel change that is responsible.

No shit, Sherlock.It's been confirmed that it is a kernel issue months ago. But with the huge benefits the new kernels bring it just makes sense to ignore this and use any new kernel that's available.

[Milhouse]

No shit, Sherlock.It's been confirmed that it is a kernel issue months ago. But with the huge benefits the new kernels bring it just makes sense to ignore this and use any new kernel that's available.

Then find yourself another build/kernel/platform/whatever.

[Brian B]

To confirm if it is the kernel change, you can try builds #0509 (4.6.0-rc7) then #0509x (4.4.7) - the only difference between these two builds is the kernel. If resume DOES NOT work with #0509, but DOES work with #0509x, then we know it's the kernel change that is responsible.

No shit, Sherlock.It's been confirmed that it is a kernel issue months ago. But with the huge benefits the new kernels bring it just makes sense to ignore this and use any new kernel that's available.

What huge benefits or are you being facetious?

[Milhouse]

What huge benefits or are you being facetious?

I only read it as the latter, unfortunately.

In terms of benefits, we'd certainly have a lot less Intel users if we remained on the older kernel as those users with modern Intel CPUs wouldn't be able to boot the older 4.4.x kernels. I'm pretty sure I couldn't sell them on the idea that we can't support their new CPU because @tinzarian won't be able to resume his old system.

@tinzarian: this is a bleeding edge testing thread - it even says so at the top of the first post. If you don't like the results then please quit whining and find a stable build to your liking.