English Community

Linux Operating SystemsOther Linux Discussions
All Forum Topics
Options

25 Posts

01-03-2019

AU

43 Signins

300 Page Views

  • Posts: 25
  • Registered: ‎01-03-2019
  • Location: AU
  • Views: 300
  • Message 81 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 9:17 AM

What parameters are you using in grub?
On kernel 5.0(rc2) i need iommu=soft, with that set i have issues with resume from standby.

I find kernel 4.20.3 most stable so far.

Reply
Options

25 Posts

01-03-2019

AU

43 Signins

300 Page Views

  • Posts: 25
  • Registered: ‎01-03-2019
  • Location: AU
  • Views: 300
  • Message 82 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 11:17 AM

I found this

In the German speaking Thinkpad forum someone posted a message with options that make the E585 run with kernel 4.20 and make X run with kernel 4.20 and fixex resume from suspend. No need to understand the whole posting, the important stuff is only the boot parameters:

ivrs_ioapic[32]=00:14.0 ivrs_ioapic[33]=00:00.1 clocksource=hpet libata.force=1:nohrst iommu=pt

 

https://thinkpad-forum.de/threads/217551-Boot-Parameter-f%FCr-E585-bzw-f%FCr-AMD-Ryzen-allgemein-besonders-ab-Kernel-4-20-x?p=2176002#post2176002

Reply
Options

27 Posts

08-31-2018

US

207 Signins

1152 Page Views

  • Posts: 27
  • Registered: ‎08-31-2018
  • Location: US
  • Views: 1152
  • Message 83 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 14:24 PM

^That was posted a couple of pages back by @Fernseher. Did it work for you?  Didn't for me.  So, I've been at this for a few days now.  I can boot into any bios with 4.20 and have no issues as long as I have the ivrs kernel parm attached and have no issues whatsoever.  I can boot into any bios with kernel 5.0rc2 without the ivrs kernel parm except I have the suspend problem.  I don't use iommu=soft/pt or whatever combination at all.

 

My matrix is as follows:

Kernel: 4.20 > any bios works with ivrs kernel parameters.

Kernel: 5.0rc2 > any bios works withOUT ivrs kernel parameters, however suspend is broken.

 

So, for me it's about the kernel, not the bios.  What I don't understand is how some are having the issue with 4.20 unless the distro maintainers are backporting the issues. 

 

I have recompiled my kernel many times with different configurations based on a vimdiff of the working 4.20 kernel and a new one with 5.0 merged in.  I have found two things that seem like it could be the culprit.  According to this article from the red hat forums when ASPM isn't being handled by the kernel, it's defaulted to the bios.  If I run:

 

┌───[ridobe]@[archpad]──[~]──[12 files]
└──╼ journalctl -b | grep ASPM
Jan 18 11:27:37 archlinux kernel: ACPI FADT declares the system doesn't support PCIe ASPM, so disable it
Jan 18 11:27:37 archlinux kernel: acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
Jan 18 11:27:37 archlinux kernel: acpi PNP0A08:00: FADT indicates ASPM is unsupported, using BIOS configuration

We see that is exactly what is happening.  So, the bios is messing it up right?  Except I get the EXACT same message when running 4.20 and there are no issues.  So, now what?

 

Well, doing a vimdiff of the 4.20 > 5.0rc2 kernel reveals that under "Bus options (PCI etc.) CONFIG_HOTPLUG_PCI_PCIE has been removed.  Additionally, the entire sub selections have also been removed.  CONFIG_PCIEASPM, CONFIG_PCIEASPM_DEFAULT, and a bunch of others are gone.  If this is on purpose then how does anything awake from sleep on the 5.0 kernel that needs to bypass the bios?

 

Holy crap I could be on the wrong path.  If so, I'm really wasting my time and everyone elses too.  If so, I apologize.

 

edit: I guess I'm already finding out I'm wasting my/our time.  The HOTPLUG configs have moved from Bus options to Device Drivers and all appear to be the same.

Reply
Options

27 Posts

08-31-2018

US

207 Signins

1152 Page Views

  • Posts: 27
  • Registered: ‎08-31-2018
  • Location: US
  • Views: 1152
  • Message 84 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 17:00 PM

Well, I got it fixed.  I don't know if it was any number of commits since 5.0rc2 or if it was the latest change I made.  Based on the artcle I linked above (Red Hat power-management) I tried using the 'pcie_aspm.policy=performance' kernel parm to force kernel ASPM instead of the bios.  It didn't work.  I git-pulled everything new and recompiled with the "performance" parm built-in and it now works.  

 

┌───[ridobe]@[archpad]──[~]──[12 files]
└──╼ cat /proc/cmdline
initrd=\amd-ucode.img initrd=\initramfs-5.0.0-rc2-ridobe-rz-00281-g2339e91d0e66.img root=UUID=[redacted] rw quiet amdgpu.dc=1 loglevel=0 rd.systemd.show_status=auto rd.udev.log_priority=0

No additional kernel parameters needed beyond personal preferences.  

 

┌───[ridobe]@[archpad]──[~]──[12 files]
└──╼ sudo dmidecode -t 0 | grep 'Version'
        Version: R0UET68W (1.48 )

 edit: spelling

Reply
Options

14 Posts

09-14-2018

DE

21 Signins

127 Page Views

  • Posts: 14
  • Registered: ‎09-14-2018
  • Location: DE
  • Views: 127
  • Message 85 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 18:11 PM

000000:

 

These seem to be very many commands for me. I would not apply them if I don't know what these exactly do. Maybe they bring other limitations to the system.

 

I only see this bug finally fixed if I do not need any command, and the system runs fine including working suspend.

Reply
Options

25 Posts

01-03-2019

AU

43 Signins

300 Page Views

  • Posts: 25
  • Registered: ‎01-03-2019
  • Location: AU
  • Views: 300
  • Message 86 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 21:07 PM

ridobe;

 

Thats fantastic work, well done. I'm not as skilled at linux but i did try kernel 5.0(rc2) on ubuntu 18.10 and wasn't able to boot. Like you said i suspect distro maintainers are back porting the issue.

 

I think ill go read on how to compile my own kernel.

 

Ziniral:

I have tried those parameters, while excessive they do seems stable. I have gone back to just using iommu=soft with kernel 4.20.3.

 

Reply
Options

27 Posts

08-31-2018

US

207 Signins

1152 Page Views

  • Posts: 27
  • Registered: ‎08-31-2018
  • Location: US
  • Views: 1152
  • Message 87 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 21:21 PM

@000000, I'd wait until tomorrow.  5.0.0rc3 will be out and you or anyone on this thread can reply and see if the commits fixed the issue or if my edits did.  I'd certainly like to know.  Honestly, since the kernel parm didn't work but the addition to the kernel did, leads me to believe that one of the kernel commits actually fixed it.

Reply
Options

14 Posts

01-10-2019

US

125 Signins

563 Page Views

  • Posts: 14
  • Registered: ‎01-10-2019
  • Location: US
  • Views: 563
  • Message 88 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 21:22 PM

@ridobe - to be clear:

 

You can use 4.20 with the ivrs overrides (but without iommu=pt/soft) on bios 1.48 and resume-from-suspend always works.

  • Any additional kernel boot params needed?
  • Nothing concerning in the boot logs at all?

You can use 5.00 git-head, hard-compiling in the pcie_aspm.policy=performance config, without needing ivrs overrides, and resume-from-suspend always works on bios 1.48.

  • Any kernel boot params needed at all?

 

Can you give some more details about your environment?

  • Are you building the kernel without any vendor patches straight from kernel.org?
  • What is your kernel config for 4.20? Would you mind sharing?
  • Latest firmwares from kernel.org?

 

I am using Ubuntu 18.10 and the Ubuntu mainline 4.20 build and I have to use bios 1.32 for resume-from-suspend to work. Ubuntu has some fairly large patches they apply.

Reply
Options

25 Posts

01-03-2019

AU

43 Signins

300 Page Views

  • Posts: 25
  • Registered: ‎01-03-2019
  • Location: AU
  • Views: 300
  • Message 89 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 21:24 PM

Thanks Ridobe

 

I will test and let you know. Appreciate your work.

Reply
Options

27 Posts

08-31-2018

US

207 Signins

1152 Page Views

  • Posts: 27
  • Registered: ‎08-31-2018
  • Location: US
  • Views: 1152
  • Message 90 of 310

Re: ThinkPad E485/E585 - Firmware bug ACPI IVRS table

2019-01-19, 22:42 PM

You can use 4.20 with the ivrs overrides (but without iommu=pt/soft) on bios 1.48 and resume-from-suspend always works.

  • Any additional kernel boot params needed?
  • Nothing concerning in the boot logs at all?

By overrides I assume you mean both.  I've only ever used ivrs_ioapic[32]=00:14.0.  But yes, on 1.48 I have no issues or error messages. No other boot parms.  Mind you, this is the clean 4.20 kernel, not from the Arch repos.

 

You can use 5.00 git-head, hard-compiling in the pcie_aspm.policy=performance config, without needing ivrs overrides, and resume-from-suspend always works on bios 1.48.

  • Any kernel boot params needed at all?

 No other boot parms at all.  See my boot line in previous posts.  Correct, straight git pull from mainline.  And by always, I mean I've successfully gotten back into my laptop ~3 times today, lol.  Also, see more about the .config below.

 

Can you give some more details about your environment?

  • Are you building the kernel without any vendor patches straight from kernel.org?
  • What is your kernel config for 4.20? Would you mind sharing?
  • Latest firmwares from kernel.org?
┌───[ridobe]@[archpad]──[~]──[12 files]
└──╼ inxi -F
System:    Host: archpad Kernel: 5.0.0-rc2-ridobe-rz-00281-g2339e91d0e66 x86_64 bits: 64 Desktop: Xfce 4.12.4 
           Distro: Arch Linux 
Machine:  UEFI: LENOVO v: R0UET68W (1.48 )

No patches whatsoever.  Straight from kernel.org.  Not sure what you mean by "latest firmwares from kernel.org".

 

I am using Ubuntu 18.10 and the Ubuntu mainline 4.20 build and I have to use bios 1.32 for resume-from-suspend to work. Ubuntu has some fairly large patches they apply.

 

I'll add my 4.20 kernel .config.  (It's acutally 4.20rc7 but I saved it off prior to pulling in 5.0)  Here is what what is pertinent to our conversation in 4.20:

 

CONFIG_PCIEASPM_DEFAULT=y
# CONFIG_PCIEASPM_POWERSAVE is not set
# CONFIG_PCIEASPM_POWER_SUPERSAVE is not set
# CONFIG_PCIEASPM_PERFORMANCE is not set

Here is what I have changed today to alleviate the suspend issues. **

 

# CONFIG_PCIEASPM_DEFAULT is not set
# CONFIG_PCIEASPM_POWERSAVE is not set
# CONFIG_PCIEASPM_POWER_SUPERSAVE is not set
CONFIG_PCIEASPM_PERFORMANCE=y

** Look, I'll be honest.  I think this MAY have been the problem.  Thing is, there are dozens of ASPM fixes already pulled into mainline.  I've included them so who knows.  We won't know until someone boots into the clean rc3 tomorrow and lets us know. 

Reply
Forum Home

Community Guidelines

Please review our Guidelines before posting.

Learn More

Check out current deals!

Go Shop
X

Save

X

Delete