x86: mm: early boot problem on i386 with KPTI enabled

Bug #1827884 reported by Michael on 2019-05-06
100
This bug affects 15 people
Affects Status Importance Assigned to Milestone
linux (Ubuntu)
Undecided
Unassigned
Xenial
High
Unassigned
Bionic
High
Andrea Righi

Bug Description

[Impact]

Commit d653420532d580156c8486686899ea6a9eeb7bf0 in bionic enabled kernel page table isolation for x86_32, but also introduced regressions. One of them ("BUG_ON() condition in vmalloc_sync_one()") has been addressed by bug 1830433, but there are other issues reported on i386.

Specifically on some i386 systems the kernel seems to fail in the early stage of boot (black screen and frozen keyboard) with no error reported on the console.

If the kernel is booted with "mitigations=off" and "nopti" the problem doesn't happen (that is a clear indication of being a kernel page table isolation issue).

However, users have been reported positive results with the following upstream fixes applied (all clean cherry picks), even with mitigations *and* kernel page table isolation enabled.

[Test Case]

Unfortuantely this problem is not easily reproducible, the kernel simply fails to boot (black screen and frozen keyboard) after the GRUB prompt, so we don't have a real test case (except asking the bug reporters to boot the kernel and see if it works).

[Fix]

The following upstream fix seems to resolve (prevent) the problem:

1d8ca3be86ebc6a38dad8236f45c7a9c61681e78 x86/mm/fault: Allow stack access below %rsp
aa37c51b9421d66f7931c5fdcb9ce80c450974be x86/mm: Break out user address space handling
8fed62000039058adfd8b663344e2f448aed1e7a x86/mm: Break out kernel address space handling
164477c2331be75d9bd57fb76704e676b2bcd1cd x86/mm: Clarify hardware vs. software "error_code"
0e664eee65337082be49fbbd2ee24aa0d111d0f2 Revert "perf/core: Make sure the ring-buffer is mapped in all page-tables"

[Regression Potential]

All upstream fixes, tested on the affected platform, backport changes are minimal.

[Original bug report]

NOTE: Boot problems on TWO different laptops after Software Updater updated to kernel 4.15.0-48

Both laptops running Ubuntu 18.04.2LTS i386/32bit with Xubuntu desktop. Both fresh installs from minimal CD in early April, originally kernel 0.4.15.0-47, both working fine.

Bug report filed on THIS MACHINE:
HP Compag 6710b, Intel Core2 Duo T7100 @ 1.8GHz, RAM 4GB, with Intel integrated graphics controller GM965/GL960 (with kernel command line parameter ¨video=SVIDEO-1:d¨)

SYMPTOM:
4.15.0-48 requires a second ´recovery´ boot to get it up and running.
Will not boot properly after a normal shutdown - screen goes funny and it hangs
Does boot successfully after a hard shutdown - detects improper shutdown, loads grub menu, then boot successfully

Boot problem appeared after update from kernel version 4.15.0-47 to 4.15.0-48
Problem occurs in 4.15.0-48 kernel, but not in 4.15.0-47 (which still boots normally after a normal shutdown)

OTHER MACHINE:
Even older!!! ASUS W1N Pentium M banias 1.7gHz, 1.5GB RAM, ATI Mobility Radeon 9600 GPU. Running with command line parameter "forcepae".
SYMPTOM:
Won't boot at all.
Was working fine under 4.15.0-47 as originally installed, until software updater installed 4.15.0-48. Now it won't boot (hangs, black screen), and I can't even get into the grub menu to try the older kernel.

More Info on what I've already tried for troubleshooting (on the HP6710b) is outlined in https://answers.launchpad.net/ubuntu/+question/680486

Happy to provide more info if needed, but I'm just an end user user and new to Linux, so will be on a learning curve at this end. (and apologies in advance if this a user problem, and not an actual bug). Thanks

ProblemType: Bug
DistroRelease: Ubuntu 18.04
Package: linux-image-4.15.0-48-generic 4.15.0-48.51
ProcVersionSignature: Ubuntu 4.15.0-48.51-generic 4.15.18
Uname: Linux 4.15.0-48-generic i686
ApportVersion: 2.20.9-0ubuntu7.6
Architecture: i386
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: michael 1681 F.... pulseaudio
CurrentDesktop: XFCE
Date: Mon May 6 21:16:34 2019
MachineType: Hewlett-Packard HP Compaq 6710b
PccardctlIdent:
 Socket 0:
   no product info available
PccardctlStatus:
 Socket 0:
   no card
ProcFB: 0 inteldrmfb
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-4.15.0-48-generic root=UUID=1f4f1f23-b034-4931-af71-c51eba39fbd3 ro video=SVIDEO-1:d
RelatedPackageVersions:
 linux-restricted-modules-4.15.0-48-generic N/A
 linux-backports-modules-4.15.0-48-generic N/A
 linux-firmware 1.173.5
SourcePackage: linux
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 08/18/2008
dmi.bios.vendor: Hewlett-Packard
dmi.bios.version: 68DDU Ver. F.13
dmi.board.name: 30C0
dmi.board.vendor: Hewlett-Packard
dmi.board.version: KBC Version 71.2E
dmi.chassis.asset.tag: SGH8080272
dmi.chassis.type: 10
dmi.chassis.vendor: Hewlett-Packard
dmi.modalias: dmi:bvnHewlett-Packard:bvr68DDUVer.F.13:bd08/18/2008:svnHewlett-Packard:pnHPCompaq6710b:pvrF.13:rvnHewlett-Packard:rn30C0:rvrKBCVersion71.2E:cvnHewlett-Packard:ct10:cvr:
dmi.product.family: 103C_5336AN
dmi.product.name: HP Compaq 6710b
dmi.product.version: F.13
dmi.sys.vendor: Hewlett-Packard
---
ProblemType: Bug
ApportVersion: 2.20.9-0ubuntu7.6
Architecture: i386
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC1: michael 1034 F.... pulseaudio
 /dev/snd/controlC0: michael 1034 F.... pulseaudio
CurrentDesktop: XFCE
DistroRelease: Ubuntu 18.04
Lsusb:
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
 Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 002 Device 002: ID 046d:c00f Logitech, Inc. MouseMan Traveler/Mobile
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
MachineType: ASUSTeK Computer Inc. W1N
Package: linux (not installed)
PccardctlIdent:
 Socket 0:
   no product info available
 Socket 1:
   no product info available
PccardctlStatus:
 Socket 0:
   no card
 Socket 1:
   no card
ProcFB: 0 radeondrmfb
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-4.15.0-48-generic root=UUID=5e6f46b1-73a8-4779-bc20-414a46286723 ro forcepae quiet splash
ProcVersionSignature: Ubuntu 4.15.0-48.51-generic 4.15.18
RelatedPackageVersions:
 linux-restricted-modules-4.15.0-48-generic N/A
 linux-backports-modules-4.15.0-48-generic N/A
 linux-firmware 1.173.5
RfKill:
 0: phy0: Wireless LAN
  Soft blocked: no
  Hard blocked: no
Tags: bionic
Uname: Linux 4.15.0-48-generic i686
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip lpadmin plugdev sambashare sudo
_MarkForUpload: True
dmi.bios.date: 05/26/2004
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: 0203
dmi.board.name: W1N
dmi.board.vendor: ASUSTeK Computer Inc.
dmi.board.version: 1.0
dmi.chassis.asset.tag: ATN12345678901234567
dmi.chassis.type: 10
dmi.chassis.vendor: ASUSTeK Computer Inc.
dmi.chassis.version: 1.0
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvr0203:bd05/26/2004:svnASUSTeKComputerInc.:pnW1N:pvr1.0:rvnASUSTeKComputerInc.:rnW1N:rvr1.0:cvnASUSTeKComputerInc.:ct10:cvr1.0:
dmi.product.name: W1N
dmi.product.version: 1.0
dmi.sys.vendor: ASUSTeK Computer Inc.
---
ProblemType: Bug
ApportVersion: 2.20.9-0ubuntu7.6
Architecture: i386
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC1: michael 1034 F.... pulseaudio
 /dev/snd/controlC0: michael 1034 F.... pulseaudio
CurrentDesktop: XFCE
DistroRelease: Ubuntu 18.04
Lsusb:
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
 Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 002 Device 002: ID 046d:c00f Logitech, Inc. MouseMan Traveler/Mobile
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
MachineType: ASUSTeK Computer Inc. W1N
Package: linux (not installed)
PccardctlIdent:
 Socket 0:
   no product info available
 Socket 1:
   no product info available
PccardctlStatus:
 Socket 0:
   no card
 Socket 1:
   no card
ProcFB: 0 radeondrmfb
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-4.15.0-48-generic root=UUID=5e6f46b1-73a8-4779-bc20-414a46286723 ro forcepae quiet splash
ProcVersionSignature: Ubuntu 4.15.0-48.51-generic 4.15.18
RelatedPackageVersions:
 linux-restricted-modules-4.15.0-48-generic N/A
 linux-backports-modules-4.15.0-48-generic N/A
 linux-firmware 1.173.5
RfKill:
 0: phy0: Wireless LAN
  Soft blocked: no
  Hard blocked: no
Tags: bionic
Uname: Linux 4.15.0-48-generic i686
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip lpadmin plugdev sambashare sudo
_MarkForUpload: True
dmi.bios.date: 05/26/2004
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: 0203
dmi.board.name: W1N
dmi.board.vendor: ASUSTeK Computer Inc.
dmi.board.version: 1.0
dmi.chassis.asset.tag: ATN12345678901234567
dmi.chassis.type: 10
dmi.chassis.vendor: ASUSTeK Computer Inc.
dmi.chassis.version: 1.0
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvr0203:bd05/26/2004:svnASUSTeKComputerInc.:pnW1N:pvr1.0:rvnASUSTeKComputerInc.:rnW1N:rvr1.0:cvnASUSTeKComputerInc.:ct10:cvr1.0:
dmi.product.name: W1N
dmi.product.version: 1.0
dmi.sys.vendor: ASUSTeK Computer Inc.

Michael (perthmichael) wrote :

This change was made by a bot.

Changed in linux (Ubuntu):
status: New → Confirmed

You'll probably want the boot.log?

Brad Evans (brad-of-ottawa) wrote :

I have two old Dell computers which experience a problem similar to what Michael describes. They work fine at Linux 4.15.0-47 but at 4.15.0-48 they show a corrupted display while booting. I can revert to 4.15.0-47 and await a fix.

I will attach the output file of ubuntu-bug made in 4.15.0-48 recovery mode.

Brad Evans (brad-of-ottawa) wrote :

Here is the Ubuntu-bug data collection for my other Dell the exibits the same problem.

apport information

tags: added: apport-collected
description: updated

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

description: updated

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

Update:
HP Laptop: I have upgraded BIOS on the HP6710b laptop to F16, which included upgrade of video BIOS. It had no effect on the boot behaviour.

ASUS Laptop: I have realised that I CAN get the grub menu - its just that the screen is black and I can´t see it. (ie: I can navigate the grub menu using arrow/enter keys and select various options from memory of where they are in the menu tree). Behaviour is essentially the same as HP laptop - it will only boot successfully into 4.15.0-48 after a previous failed boot. A first boot fails.

Note that I tried a fresh install from the mini.iso on the asus laptop. Boot problem is the same on a fresh install of 4.15.0-48, not just upgrading from earlier version through software updater.

Ubunutu-bug output of the asus filed a new bug report Bug #1828764

This problem affected all Intel CPUs we have, except for Atoms

Valentyna (valia0906) wrote :

This problem also exists in kernel 4.15.0-50 version on Ubuntu 18.04.

Mark Haiman (mhaiman) wrote :

I am seeing a similar problem.

Linux Mint 18.3 XFCE, Lenovo Thinkpad T60 i686 / ATI Radeon RV515 GPU

After upgrade to kernel 4.15.0-48, I get a black screen and frozen keyboard once the boot splash screen exits, when the login screen should appear. There is no mouse pointer, and I can't escape with ctrl-alt-f1, ctrl-alt-backspace, etc. If I leave the system running for a while, there is occasional disk activity, and syslog entries with later time stamps, indicating that the kernel is still running, but the X server is frozen.

Same thing with latest kernel 4.15.0-50.

I can boot into the previous kernel 4.15.0-47 with no problems.

If I boot into run level 3 (terminal only, no GUI) by editing the Grub kernel command line, the system will boot, but enters the same frozen state if I then start the GUI with 'sudo service lightdm start' (which works fine with the previous 4.15.0-47 kernel).

If I add "nomodeset" to the kernel command line, the system will boot into the GUI. In this case Xorg.0.log shows that the X server unloads the Radeon driver (which will not run without KMS) and loads the fallback VESA driver instead. I.e., on my computer the bug affects the Radeon driver but not the VESA driver. (However, this is not a solution for me because the VESA driver doesn't handle my 1400x1050 screen resolution correctly.)

One other thing - the bug happens maybe 9 times out of 10 but not every time. On the occasional successful boot, once the GUI comes up without freezing, everything seems to work OK.

Stefano Bandini (bando85) wrote :

I have three different machines running Xubuntu 18.04 i386. All are affected.

After upgrade to kernel 4.15.0-48 the boot process will stop to an almost black screen with a few multi-coloured lines.
Problem still persists with kernel 4.15.0-50.

Everything works fine with kernel 4.15.0-47.

Jeff (jharan) wrote :

I have a Lenovo G500 (32 bit laptop about 5 years old).

I am seeing the same behavior on my system. Booting 4.15.0-47-generic works fine, but both
4.15.0-48-generic and 4.15.0-50-generic usually hang at boot time. If I enter grub at boot and select 4.15.0-50-generic, it usually hangs at this:

Loading Linux 4.15.0-50-generic ...
Loading initial ramdisk ...

If I let the system just boot up w/o going into grub I don't see the above. I usually either get a solid blank purple screen or a blank purple screen with a bar of random colored pixels across the middle of the screen.

Now yesterday I was able to get 4.15.0-50-generic to boot all they way into the GUI after booting it from the grub menu and was able to use the system for the rest of the day, no problems. So it seems to be a startup only problem.

When in the hung state, I can't get to the console screen via the usual Fn-Ctrl-Alt-F3.

Matthias Homann (palto) wrote :

I have the same issue on 3 Laptops (2 HP nc6230 and 1 Asus, all 32bit systems), only solution so far is to revert to Linux 4.15.0-47. I could not find any useful hint in any of the logs.
One strange thing is that only every second boot fails. After hard power-off a hanging boot the next boot works fine one time, but the next one again hangs.
I could not even get a text console with Ctrl-Alt-F1 in case it hangs.

Norman Price (norman-price) wrote :

I have precisely the same issue:
Dell Latitude D530
Intel graphics GM965/GL960
Lubuntu 18.04 (32 bit)
Kernel 4.15.0-50

Valentyna (valia0906) wrote :

This problem also exists in kernel 4.15.0-51 version on Ubuntu 18.04 (i386).

Treno70 (atgm) wrote :

I confirm.
This bug affects all my intel i386 machines (three with intel graphics and one with amd graphics).
The 64 bit systems are all working fine.
Latest working kernel was 4.15.0-47

This problem does not exist in kernel 4.18.0-21 (linux-generic-hwe-18.04),
but exists in kernel 5.0.0-16 (linux-generic-hwe-18.04-edge)

Kai-Heng Feng (kaihengfeng) wrote :

Does kernel parameter "mitigations=off" help?

Brad Evans (brad-of-ottawa) wrote :

Setting mitigations=off did not make any difference for me.

Treno70 (atgm) wrote :

On Celeron M (Dothan) + i915 graphics setting "mitigations=off" seems to work.

Treno70 (atgm) wrote :

Q: does adding "mitigations=off" as kernel parameter put hardware at risk?

Brad Evans (brad-of-ottawa) wrote :

I tried again to set mitigations=off for kernel 4.15.0-48 after reading off Treno70's success. I had one successful boot out of six.

I have read of security concerns in using this setting.

Treno70 (atgm) wrote :

I've tried on a Pentium M (Dothan) + Mobility Radeon X600 graphics: it works.
I'm still concerned about security

Treno70 (atgm) wrote :

P.S. Kernel 4.15.0-51

Valentyna (valia0906) wrote :

When I add kernel parameter "mitigations=off" the problem disappears. I tested kernel version 4.15.0.47.49 and 4.15.0.51.53, all boots were successful.

Treno70 (atgm) wrote :

Also tried on very old Thinkpad Pentium M (Banias) Mobility Radeon 7500, Xubuntu 18.04, Kernel 4.15.0-51.
It Works

Michael Kozma (myckeul) wrote :

Hi,

Adding "mitigations=off" to kernel parameters solved issue for me too.

Thanks

Michael (perthmichael) wrote :

Hi @kaihengfeng

Yes, mitigations=off works on my ASUS pentium M (banias) machine with with ATI Mobility Radeon graphics, for all of 4.15.0-48, 4.15.0-50 and 4.15.0-51.
At the moment, unable to check on the HP6710b laptop this bug was originally reported on, as I've loaned it to my brother.

Q: Were you suggesting mitigations=off as a solution (given the theoretical security vulnerabilities of spectre/meltdown), or just as a troubleshooting/diagnostic measure?

I'm wondering if I would be better off sticking to 4.15.0-47 with mitigations on, or allowing other kernel patches with mitigations=off? Or 64bit linux may be the better solution, where possible/practicable.

Thanks for suggesting the mitigations workaround!!!

Tyler Hicks (tyhicks) wrote :

Hello - The 'mitigations=off' option is not the best workaround as it disables all CPU mitigations.

Since this bug is about 4.15.0-48.51 and i386 installations, the bug is most likely caused by the addition of Page Table Isolation (PTI) for i386. It landed in the aforementioned kernel release.

To be selective and only disable PTI, please use the 'nopti' boot option rather than 'mitigations=off'.

In the meantime, we'll start working to figure out the cause of this problem. Thanks for the bug report.

Brad Evans (brad-of-ottawa) wrote :

The nopti switch is working for me with 4.15.0-51-generic.

I think possibly the mitigations=off was not working for me because I positioned it at the wrong part (the end) of the linux arguments list. I put nopti closer to the front of the arguments list.

Brad Evans (brad-of-ottawa) wrote :

Thank you

On 2019-06-07 16:11:07, Brad Evans wrote:
> The nopti switch is working for me with 4.15.0-51-generic.
>
> I think possibly the mitigations=off was not working for me because I
> positioned it at the wrong part (the end) of the linux arguments list.

The 'mitigations=off' option wasn't working for you because the
4.15.0-48.51 kernel didn't know about that option. It was first
backported in 4.15.0-50.54. Sorry for the confusion.

Similar/same problem affects me on
Asus X51R (Laptop).

OS: Lubuntu 18.04. 32-Bit

Occured after automatic update including Kernel probably around 4.15.0-48

Ubuntu seems to boot up, switches on and off screen several times.
Logon screen does NOT appear.
Instead, it restarts itself. After restarting itself it gets stuck onto a black screeen showing a very big mouse-pointer. (Mouse-Pointer can be moved using the track-pad. However not further interaction with the system is possible.)

Booting up the system in recovery mode is possible. However, screen resolution is restricted.
Booting up the system using an older kernel (4.09.xxx) is possible. Screen resolution is OK, but WIFI is not available, then.

Thanks a lot,
Thomas

Thomas Lenarz (tommesml) wrote :

Boot switch "nopti" solves the problem as well on Asus X51R (Laptop).
However, I think it is not a final solution, cause its Processor Celeron M 520 is on the vulnerability list.

On 2019-06-07 17:32:04, Thomas Lenarz wrote:
> Boot switch "nopti" solves the problem as well on Asus X51R (Laptop).
> However, I think it is not a final solution, cause its Processor Celeron M 520 is on the vulnerability list.

It most definitely isn't a final solution. However, the i386 PTI changes
are what's causing you problems and it is a temporary workaround to
disable it in the meantime until we can identify the proper fix. Once
the bug is fixed and you install the updated kernel with the bug fix,
you should remember to remove the 'nopti' boot option.

Thanks Tyler, greatly appreciated
nopti workaround all good on Asus W1N Pentium M banias also.

Thomas Lenarz (tommesml) wrote :

@Tyler: Thanks a lot for the great help! Please tell, if I can help by providing logs or anything else.

Valentyna (valia0906) wrote :

When I add kernel parameter "nopti" the problem also disappears. I tested kernel version 4.15.0.50.52 and 4.15.0.51.53, all boots were successful.

Treno70 (atgm) wrote :

Changed kernel parameter to "nopti" on all my machines. It worked. Thanks.

Takahe (pukemanu) wrote :

Changed kernel parameter on Dell D620 and Dell 530s - both booting properly first time since upgrade from kernel 47. Thanks

Valentyna (valia0906) wrote :

The problem also exists in kernel version 4.15.0.52.54.

Dima (dima2017) wrote :

It still present in 4.15.0-54.58 (Tested with lowlatency kernel version on Toshiba Sattelite L300-11Q). "nopti" helps.

Valentyna (valia0906) wrote :

I also tested kernel version 4.15.0-54.58, problem still exists.

Andrea Righi (arighi) wrote :

This might be a duplicate of 1830433, could you give it a try with this test kernel and see if the problem is still happening?

https://kernel.ubuntu.com/~arighi/LP-1827884/

Thanks!

On 2019-07-02 16:36:27, Andrea Righi wrote:
> This might be a duplicate of 1830433, could you give it a try with this
> test kernel and see if the problem is still happening?
>
> https://kernel.ubuntu.com/~arighi/LP-1827884/

For everyone who is able to try out that test build, please remember to
remove the "mitigations=off" and "nopti" kernel command line
parameters that you've added to work around this bug. We want you to
test that kernel without using the workaround.

If you report successful results when trying the test kernel, please
include the output from the following commands:

 $ cat /proc/cmdline

 $ cat /sys/devices/system/cpu/vulnerabilities/meltdown

The output from those commands will help us make sure that the bug is
fixed rather than a workaround giving us false hope.

My hope is that Andrea's work on bug 1830433 ends up fixing the problems
that you all are seeing in this bug. Thanks in advance for any testing
that you can provide.

Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in linux (Ubuntu Bionic):
status: New → Confirmed
Valentyna (valia0906) wrote :

I`ve tried kernel test version, the problem is still happening.

Andrea Righi (arighi) wrote :

So, it looks like bug 1830433 didn't fix this problem... assuming it's still an x86/mm issue, it'd be interesting to test another kernel (I've applied a few more x86/mm-related fixes).

This new test kernel is available at the same place:

https://kernel.ubuntu.com/~arighi/LP-1827884/

As correctly pointed out by Tyler, for the test you should remove "mitigations=off" and "nopti" and in case of success, please report the output of:

 $ cat /proc/cmdline

 $ cat /sys/devices/system/cpu/vulnerabilities/meltdown

Thanks again for all the tests!

Dima (dima2017) wrote :

Seems I used wrong kernel for previous test. Sorry.
This new kernel works, but xorg doesn't start.
I'm testing on Lubuntu xenial. How to add "Xenial" under the "Bionic" on the top of this thread?

Valentyna (valia0906) wrote :

I`ve tried second kernel test version on ubuntu 18.04(bionic), the problem still exists.

Dima (dima2017) wrote :

I've rechecked the last test version. Previous problem is gone, but xorg doesn't start. Xorg doesn't start even with the "nopti" option.

BOOT_IMAGE=/@/boot/vmlinuz-4.15.0-54-generic root=UUID=50c3a930-a3e9-4444-b4ea-1646d276c7c6 ro rootflags=subvol=@ ipv6.disable=1 zswap.enabled=0 raid=noautodetect priority=low video=SVIDEO-1:d module_blacklist=r8169,mii,msr,jfs,xfs,bluetooth,hfs,hfsplus,ufs,minix,ntfs,joydev,mac_hid

Mitigation: PTI

Dima (dima2017) wrote :

But I use Lubuntu Xenial. Probably this is why xorg doesn't start with the last test version. I can upload dmesg or something else if it is needed.

Andrea Righi (arighi) wrote :

Thanks for the tests and the feedback @dima2017 and @valia0906! Could you also post the output of the following commands?

$ uname -r

$ cat /proc/cmdline

$ cat /sys/devices/system/cpu/vulnerabilities/meltdown

Valentyna (valia0906) wrote :

My output:
uname -a
Linux TEPL-ET2013 4.15.0-54-generic #58+lp1827884 SMP Wed Jul 3 10:05:45 UTC 2019 i686 i686 i686 GNU/Linux

cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-4.15.0-54-generic root=UUID=63b3d2c0-b9a9-4112-97eb-de501efd9205 ro quiet splash

cat /sys/devices/system/cpu/vulnerabilities/meltdown
Mitigation: PTI

Michael (perthmichael) wrote :

Thanks Andrea, much appreciated.

Have installed your latest test kernel on my ASUS W1N Pentium M (banias) machine running 18.04 Xubuntu, and it appears to work! Have shutdown normally and rebooted several times with no problems. (Note: unable to test on the HO6710b machine that originally reported the bug - its still on long term loan to my brother)

Output of commands...

uname -a
Linux asusw1na 4.15.0-54-generic #58+lp1827884 SMP Wed Jul 3 10:05:45 UTC 2019 i686 i686 i686 GNU/Linux

cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-4.15.0-54-generic root=UUID=9e042291-28d3-4f6b-a455-488355b84473 ro forcepae

cat /sys/devices/system/cpu/vulnerabilities/meltdown
Mitigation: PTI

______

NOTE: I'm still a Linux newbie. I have a couple of questions about what happens next, and grateful if anyone could offer some guidance...

I installed 3 of the test kernel deb packages:
   linux-headers-4.15.0-54_4.15.0-54.58+lp1827884_all.deb
   linux-headers-4.15.0-54-generic_4.15.0-54.58+lp1827884_i386.deb
   linux-image-4.15.0-54-generic_4.15.0-54.58+lp1827884_i386.deb
Should I be installing any others to make sure everything else and all my software applications etc still work? (Just a basic end user of the generic version)

And what happens next once the bug fix is confirmed? Will it roll into the automatic updates? Do I have to uninstall the test kernel to still get those updates?

Thanks for any advice on the newbie questions. And Thanks again Andrea for the fix!

Andrea Righi (arighi) wrote :

Michael, first of all thanks for testing. It seems that (in your case at least) the problem was fixed by the extra commits that I backported into 4.15.0-54, so I think it's worth to have them applied.

To answer your questions, 1) you may want to install also linux-modules-4.15.0-54-generic_4.15.0-54.58+lp1827884_i386.deb and 2) the next step is to send an SRU email to the kernel-team mailing list (I'll do that) requesting to include these fixes. If the request is accepted, during the next SRU cycle these fixes will be applied to the "official" kernel.

In the meantime you can safely use the test kernel if it works and, as soon as the new kernel comes out, you'll get it via the regular automatic updates.

Andrea Righi (arighi) on 2019-07-04
Changed in linux (Ubuntu Bionic):
assignee: nobody → Andrea Righi (arighi)
Andrea Righi (arighi) on 2019-07-04
description: updated
summary: - Boot problems after upgrade to kernel 4.15.0-48 (18.04.2LTS/i386)
+ x86: mm: early boot problem on i386 with KPTI enabled
Thomas Lenarz (tommesml) wrote :

Hi Andrea,

Thanks a lot for the test kernel.

https://kernel.ubuntu.com/~arighi/LP-1827884/

I tried it on Laptop ASUS X51 R.

Results:

1. It boots just fine (without nopti boot option.)
2. However, after boot Screen Resolution is fixed to 1024x768.
   Native LCD-Resolution with working ATI Radeon Graphics would be 1280x800 (and adjustable)
   This is the same effect to when I use my current kernel 4.15.0-51 in recovery mode.

   Boot with 4.15.0.51 using nopti-Kernel-Option results in graphics OK.

Thanks a lot,
Thomas

Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in linux (Ubuntu Xenial):
status: New → Confirmed
Stefan Bader (smb) on 2019-07-10
Changed in linux (Ubuntu Xenial):
importance: Undecided → High
Changed in linux (Ubuntu Bionic):
importance: Undecided → High
Changed in linux (Ubuntu Bionic):
status: Confirmed → Fix Committed
Dima (dima2017) wrote :

It flooded this before reboot:

IP: print_modules+0x40/0xbd
*pdpt = 0000000036445001 *pde = 000000000de36063 *pte = 0000000000000000
Thread overran stack, or stack corrupted
Oops: 000 [a1672] PREEMPI SMP PTI
Modules linked in:
BUG: unable to handle kernel paging request at fffffffc

I don't know which string is first.
I can see some video artifacts before rebooting.

(I just copied this message from #1827725)

Dima (dima2017) wrote :

I could see that without "quiet" boot option.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers