System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)

[Maciej Dziardziel]

[Impact]
Gigabyte AM4 boards users cannot boot Ubuntu successfully.
Commit linux-gpio/fixes babdc22b0ccf4ef5a3075ce6e4afc26b7a279faf "pinctrl/amd: Use regular interrupt instead of chained" can fix the issue.

[Test Case]
All Gigabyte AM4 boards can reproduce the issue.
With the patch, the issue is resolved, per comment #170.

[Regression Potential]
Regression Potential is low. It limits to rather new AMD platform which has pinctrl-amd.
As the commit log says, use chained interrupt is not a good idea. Use regular interrupt is the correct way.

I also test the patch on an AMD laptop, where its touchpad depends on pinctrl-amd. No regression found.

Original bug report:
I'm trying to run ubuntu on Ryzen 1700x with Gigabyte GA-AB350-gaming-3 motherboard,
and it has a load of problems, starting with not being able to boot normally.

During normal boot, on 16.10 as well as 17.04 beta:
system doesn't boot normally, hangs with a lot of "unexpected irq trap at vector 07"
messages displayed.

Following advice from various places, I've tried:disable cpu freq governor and cpu handling in acpi settings

1. add "acpi=off" to boot params

That helps, allowing me to boot into recovery mode, though it leaves me with system seeing only one core, is really slow and still only boots in recovery mode.

2. Compile own kernel using 4.11.rc1 and disabling cpu freq governor and cpu handling in acpi settings. Boot with "quiet loglevel=3" option.

That gets me even further - system sees all cores now. Still only recovery mode though,
but its enough to get info for this bug report.

Some observed problems:

1. dmesg reports *a lot* of messages like this all the time:

[ 163.362068] ->handle_irq(): ffffffff87a7e090,
[ 163.362081] bad_chained_irq+0x0/0x40
[ 163.362089] ->handle_irq(): ffffffff87a7e090,
[ 163.362090] amd_gpio_irq_handler+0x0/0x200
[ 163.362090] ->irq_data.chip(): ffffffff88587e20,
[ 163.362090] ioapic_ir_chip+0x0/0x120
[ 163.362090] ->action(): ffffffff884601c0
[ 163.362091] IRQ_NOPROBE set
[ 163.362099] ->handle_irq(): ffffffff87a7e090,
[ 163.362099] amd_gpio_irq_handler+0x0/0x200
[ 163.362100] ->irq_data.chip(): ffffffff88587e20,
[ 163.362100] ioapic_ir_chip+0x0/0x120
[ 163.362101] ->action(): ffffffff884601c0

I've tried to redirect dmesg to a file, stopped after a short while, it generated 400M of those.

2. Systemd cannot start journald. Perhaps because it cannot cope with amount of kernel logs?

3. Looking at pci, I've noticed something called AMDI0040 (/sys/bus/acpi/devices/AMDI0040, path=_SB_.EMMC), among AMDI0010, AMDI0020, AMDI0030. Those however are mentioned in kernel source, kernel and google are completely silent about AMDI0040.

Phoronix tested ryzen using different motherboard, and it worked better (though not well),
so I suspect it is an issue with motherboard.
---
ApportVersion: 2.20.4-0ubuntu2
Architecture: amd64
DistroRelease: Ubuntu 17.04
InstallationDate: Installed on 2015-08-06 (581 days ago)
InstallationMedia: Kubuntu 15.10 "Wily Werewolf" - Alpha amd64 (20150728.1)
Package: linux (not installed)
ProcEnviron:
 LANGUAGE=en_US:en
 TERM=linux
 PATH=(custom, no user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
Tags: zesty
Uname: Linux 4.11.0-rc1-custom x86_64
UnreportableReason: The running kernel is not an Ubuntu kernel
UpgradeStatus: Upgraded to zesty on 2017-03-03 (6 days ago)
UserGroups:

_MarkForUpload: True

[Brad Figg (brad-figg)]

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1671360

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[fiedzia (fiedzia-linux-kernel-bugs)]

Created attachment 255151
kernel config

I'm trying to run ubuntu on Ryzen 1700x with Gigabyte GA-AB350-gaming-3 motherboard,
and it has a load of problems, starting with not being able to boot normally.

During normal boot, on 16.10 as well as 17.04 beta:
system doesn't boot normally, hangs with a lot of "unexpected irq trap at vector 07"
messages displayed.

Following advice from various places, I've tried to:

1. add "acpi=off" to boot params

That helps, allowing me to boot into recovery mode, though it leaves me with system seeing only one core, is really slow and still only boots in recovery mode.

2. Compile own kernel using 4.11.rc1 and disabling cpu freq governor and cpu handling in acpi settings. Boot with "quiet loglevel=3" option.

That gets me even further - system sees all cores now. Still only recovery mode though,
but its enough to get info for this bug report.

Some observed problems:

1. dmesg reports *a lot* of messages like this all the time:

[ 163.362068] ->handle_irq(): ffffffff87a7e090,
[ 163.362081] bad_chained_irq+0x0/0x40
[ 163.362089] ->handle_irq(): ffffffff87a7e090,
[ 163.362090] amd_gpio_irq_handler+0x0/0x200
[ 163.362090] ->irq_data.chip(): ffffffff88587e20,
[ 163.362090] ioapic_ir_chip+0x0/0x120
[ 163.362090] ->action(): ffffffff884601c0
[ 163.362091] IRQ_NOPROBE set
[ 163.362099] ->handle_irq(): ffffffff87a7e090,
[ 163.362099] amd_gpio_irq_handler+0x0/0x200
[ 163.362100] ->irq_data.chip(): ffffffff88587e20,
[ 163.362100] ioapic_ir_chip+0x0/0x120
[ 163.362101] ->action(): ffffffff884601c0

I've tried to redirect dmesg to a file, stopped after a short while, it generated 400M of those.

2. Systemd cannot start journald. Perhaps because it cannot cope with amount of kernel logs?

3. Looking at pci, I've noticed something called AMDI0040 (/sys/bus/acpi/devices/AMDI0040, path=_SB_.EMMC), among AMDI0010, AMDI0020, AMDI0030. Those however are mentioned in kernel source, kernel and google are completely silent about AMDI0040.

Phoronix tested ryzen using different motherboard, and it worked better (though not well),
so I suspect it is an issue with motherboard.

[fiedzia (fiedzia-linux-kernel-bugs)]

Created attachment 255153
lspci_vv_nn

[fiedzia (fiedzia-linux-kernel-bugs)]

Created attachment 255155
dmidecode

[fiedzia (fiedzia-linux-kernel-bugs)]

Created attachment 255157
find_sys_bus_acpi

[fiedzia (fiedzia-linux-kernel-bugs)]

Created attachment 255159
content of /sys/bus/acpi/devices/AMDI0040

[Maciej Dziardziel (fiedzia)]

apport information

[Maciej Dziardziel (fiedzia)]

I've added appport info

[Joseph Salisbury (jsalisbury)]

Did this issue start happening after an update/upgrade? Was there a prior kernel version where you were not having this particular problem?

Would it be possible for you to test the latest upstream kernel? Refer to https://wiki.ubuntu.com/KernelMainlineBuilds . Please test the latest v4.11 kernel[0].

If this bug is fixed in the mainline kernel, please add the following tag 'kernel-fixed-upstream'.

If the mainline kernel does not fix this bug, please add the tag: 'kernel-bug-exists-upstream'.

Once testing of the upstream kernel is complete, please mark this bug as "Confirmed".

Thanks in advance.

[0] http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.11-rc1/

[Maciej Dziardziel (fiedzia)]

The issue had nothing to do with upgrades, its the same problem on 16.10 as well as on 17.04.
I've tested it on my system and on live ubuntu on usb.

I've tried to use mainline kernel and the problem is still there.

[Maciej Dziardziel (fiedzia)]

I've found a workaround (requires compiling own kernel, tested on 4.11 and r4.11rc1):

When configuring kernel (I used make menuconfig) disable following things:

Everything under device drivers/gpio support, especially:
memory mapped files/amd promontory support (GPIO_AMDPT)
and pci gpio expanders / amd 8111 gpio driver (GPIO_AMD8111)
but I've turned of everything to make sure nothing is using gpio.

device drivers / pin controllers / amd gpio pin control (PINCTRL_AMD)

Perhaps only one of them could be really necessary, but I didn't tested that.

System boots normally, no problems with journald and no weird messages from dmesg.

[fiedzia (fiedzia-linux-kernel-bugs)]

I've found a workaround (requires compiling own kernel, tested on 4.11 and r4.11rc1):

When configuring kernel (I used make menuconfig) disable following things:

Everything under device drivers/gpio support, especially:
memory mapped files/amd promontory support (GPIO_AMDPT)
and pci gpio expanders / amd 8111 gpio driver (GPIO_AMD8111)
but I've turned of everything to make sure nothing is using gpio.

device drivers / pin controllers / amd gpio pin control (PINCTRL_AMD)

Perhaps only one of them could be really necessary, but I didn't tested that.

System boots normally, no problems with journald and no weird messages from dmesg.

[Robert Sandru (rsandru)]

Maciej, do you have any suggestions if there is no previous OS installed? Any way for to fix the installation image so that I can at least get a base installation going?

[Eric Hartmann (hartmann-eric)]

Guys, I've updated the firmware of my GA-X370-gaming-5 to F4 and with kernel 4.10.1 and acpi=off it's almost working. My last issue is that only one core is available.

I've tested with Fedora 25 (4.8 kernel with backports) and everything is working perfectly : all cores are available so it may due to kernel configuration, but I did not find where the issue is.

[Maciej Dziardziel (fiedzia)]

> Maciej, do you have any suggestions if there is no previous OS installed?

You may try to boot with those options:

acpi=off quiet loglevel=3

(just acpi=off might be enough). That should get you through installation.

[Cosmin Mutu (cosmin-mutu)]

Getting same problem on AMD Ryzen X18700 with GIGABYTE AB350-GAMING3 motherboard (F5 - latest bios version)

Setting acpi=off gets me into login screen but fails to authenticate me .. so, I`m stuck there.

This occured when I tried to upgrade from 16.04 to 16.10 ... then I tried installing the 17.04 with same problems ... instal was hitting the IRQ vector 07 problem.

If you guys need any other info, please let me know and I can collect some more.

[Cosmin Mutu (cosmin-mutu)]

AMD Ryzen X1800 (don`t know how a 7 got in there)

[Cosmin Mutu (cosmin-mutu)]

authentication does work from terminal ( ctrl-alt-f1 )

[Cosmin Mutu (cosmin-mutu)]

Just booted 16.04 from disk (try mode) and it works without problems ... attached a screenshot also

[Heiko Hartmann (yrwyddfa)]

I can confirm that for Ryzen X1700 and same Mainboard as mentioned by Cosmin. Ubuntu Studio 16.04 is working without any problems, Live version of 16.10 messes up with thousands of IRQ traps.

[Robert Sandru (rsandru)]

Still no dice.

Tried another path:
- Installed on a spare HDD using VMware in Windows (assigned full disk to VMware), no additional drivers or anything changed
- Updated kernel to 4.11rc1 (using prebuild binaries)
- Updated grub parameters: acpi=off quiet loglevel=3
- Rebooted computer with that drive as the boot disk

Still gets stuck during startup. Removing the acpi=off parameter spews again the irq trap messages so same situation.

Setup is 1700X on Gigabyte GA-AX370-Gaming 5 motherboard
For all it's worth I'm also using an NVMe bootdrive - not sure it makes a difference.

Giving up until a kernel dev with the appropriate knowledge looks into this.

[Cosmin Mutu (cosmin-mutu)]

Installed 16.04 - runs just fine (just like Heiko mentioned above). All 16 cpu`s are visible, however, no way to get temp information (sensors can`t find anything).

[Robert Sandru (rsandru)]

Cosmin, when you say you installed 16.04, do you mean Ubuntu Studio 16.04 or the standard distribution?

Did you install the OS from scratch using an USB stick of some other media? You were trying to do an update to 17.04 so did 16.04 boot initially correctly? Please provide details, just saying it works isn't helping to pinpoint the issue.

[Eric Hartmann (hartmann-eric)]

As Maciej Dziardziel suggest, I've rebuild 4.10.1 kernel and deactivated GPIO_AMDPT, GPIO_AMD8111 and PINCTRL_AMD.
The system was booting correctly, however I cannot use any USB keyboard.
So I've rebuild the kernel, deactivating only PINCTRL_AMD and everything is working fine now.

Here is my setup : Aorus GA-X370-Gaming-5 with Ryzen 1800X.

And here is the link of the builds :
* https://mega.nz/#!R493SRIb!UgMpquYIqQuJkmHoY1kgp5rtVSMK7yo7tw_KP3pUKzw
* https://mega.nz/#!YoMmHBRQ!plT0X61pQJWz00rkZ5ANZC50rjNHKID3BZ4v-wia2mE
* https://mega.nz/#!F8801QKD!pRSay0_qWhKqvfl6MKOJSpGSBxf1t-NMspdsYPM5eDQ

[Huang YangWen (yangwen5301)]

Works find with Ubuntu 16.04 + 4.10.1-041001-generic

The kernel is downloaded from Ubuntu website

motherboard ASUS PRIME B350-PLUS.

[Huang YangWen (yangwen5301)]

CPU is Ryzen 1700x

However lm-sensor is not working properly.

[fiedzia (fiedzia-linux-kernel-bugs)]

>tested on 4.11 and r4.11rc1

Should be "tested on 4.10 and r4.11rc"

[Maciej Dziardziel (fiedzia)]

Huang YangWen: can you attach result of those command:

ls -l /sys/bus/acpi/devices/
sudo lspci -vvnn

I wonder what the differences between motherboards are. I suspect Gigabyte has something other boards don't have.

[Cosmin Mutu (cosmin-mutu)]

@Robert ... sure, let me clarify the events.

1. I had Ubuntu 16.04 LTS for some time already running on my system.
2. I bought AMD Ryzen 1800 + Gigabyte AB350 Gaming 3 and I took the SSD from the old system and pushed it into the new one
3. Everything worked fine, except I had to install NVIDIA drivers (I also bought a better video card)
4. I noticed I couldn`t get CPU temperature, so I tried my luck by upgrading to 16.10 (BAD DECISION)
----> during OS boot the IRQ VECTOR 07 error is repeated over and over
5. I then downloaded the 17.04 on a disk, and tried to install it
----> ended up with IRQ VECTOR 07 error during install (so, no go)
6. I tried the Ubuntu 16.04 LTS from a disk (which was the origin of the OS at step 1 ...) --> worked fine
7. I installed Ubuntu 16.04 LTS and here I am .. back at step 3 :)

I`ve never installed STUDIO, only DESKTOP as in this : http://releases.ubuntu.com/16.04.2/ubuntu-16.04.2-desktop-amd64.iso?_ga=1.156896511.598999195.1489430754

[Cosmin Mutu (cosmin-mutu)]

"on a disk" I mean on a DVD :)

[Cosmin Mutu (cosmin-mutu)]

@Maciej : I think you missunderstood ... both Gigabyte (mine) and Asus boards (Huang`s) work fine with 16.04 LTS ... what you need to do .. is ask Huang to try his luck with 16.10 :) or 17.04

So, Huang, can you make 2 DVD`s .. one with 16.10 and another with 17.04 (or 2 bootable flashes) and just "TRY" both of them, without installing ... I wonder if that works for you.

[Robert Sandru (rsandru)]

Hi Cosmin, thanks a lot for the details.

I can see a couple of differences that I need to look into:
 - Chipset is different B350 vs. X370
 - I attempted the fresh 16.04 install from a USB stick instead of a DVD (I've got no optical drive at all, so can't test that) -> IRQ TRAP.... (same with 16.10 and 17.04 beta)

So my next tentative will be to take the drive out from that computer, install in a different one and perform the 16.04 install and move it back to the Ryzen system. That would be similar to your initial state with a working 16.04.

One last question (:-), was your existing 16.04 running on a special kernel version before you moved your SSD to your new Ryzen system?

Thanks!

[Robert Sandru (rsandru)]

@Eric: I've managed to boot the system using your custom built kernel!

Not usable really now as the nvidia driver installation is failing but much better...

So clearly the code around PINCTRL_AMD seems to be related.

[Daniele (protomucca)]

Dear All,

I had exactly the same situation of post #23. Coming from Kubuntu 16.04, bought a new system Ryzen 1800x+GA-AB350 Gaming 3.
First boot, using the "old" SSD worked fine, so I decided to upgrade to version 16.10. The procedure goes well, but after the reboot got plenty of error message "IRQ VECTOR 07".
So I tried with a USB stick and Kubuntu live version 17.04 both beta 1 and Nightly but got the same error.
The only way to make the system working, is to manually select kernel 4.4.0-64-generic from grub.

I've also tried to install ubuntu kernel 4.10 and 4.11rc1, but I always get the same error.

Hope this helps

[Eric Hartmann (hartmann-eric)]

@Robert, on Ubuntu 16.10 you can install the nvidia-378 drivers that contains the fix for the kernel 4.10 (I'm using it).

The patches are available here also : https://devtalk.nvidia.com/default/topic/995636/linux/-patches-378-13-4-10-and-4-11-rc1/

[Eric Hartmann (hartmann-eric)]

@Marciej here are the results for GA-X370-Gaming-5 board (with 1800X processor).
Please note that it's the result on 4.10.1 with PINCTRL_AMD removed.

[Eric Hartmann (hartmann-eric)]

And lspci

[Cosmin Mutu (cosmin-mutu)]

@Robert : what I have on DVD is an image of 16.04.01 LTS : http://old-releases.ubuntu.com/releases/xenial/ubuntu-16.04.1-desktop-amd64.iso

So, I would suggest you "try / install" this first, and then update to 16.04.02
At least that`s how it worked for me.

Current kernel that I have (didn`t performed any tricks to get this in place, it`s just from installing 16.04.01 and upgrading to 16.04.02 via normal upgrade procedure) :

$ uname -a
Linux alpha-desktop 4.4.0-66-generic #87-Ubuntu SMP Fri Mar 3 15:29:05 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

[Cosmin Mutu (cosmin-mutu)]

By normal upgrade procedure I mean :

1. Right Click on "Power Icon" (top right corner)
2. Select "About this computer"
3. On opened pop-up in the bottom right corner there should be a button which either says "Update" or "System Up-To-Date"

Let me know if the 16.04.01 install worked ... hope it does!

[anders_c_ (anders-c-)]

Ryzen 1700
Gigabyte ab350 Gaming 3

I created a usb with 16.04.1 and it starts without problems. When i tried to install it asked if I wanted UEFI or bios mode, UEFI failed to install but bios worked.

[blarg (blargblarg)]

Disable the CPU based USB ports - That's all of the version 3.0 ones. Probably most of the rear and front ports.

Only use the 3.1 or 2.0 USB ports associated with the B350 southbridge.

[anders_c_ (anders-c-)]

@blarg
Looked around in bios but found no way to disable the CPU USBs.
Tried to boot with nothing connected to them but it failed with same error messages as before.

[Marc Singer (nextized)]

I can replicate the issue on a Gigabyte GA370x Gaming 5 Mainboard with AMD 1700x. After I upgraded Mint to Linux 4.10.3 the bug even happened on older kernel versions installed with Ukuu (I could only boot on 4.4.0 used as default on Mint 18.2)

[James Willcox (snorp)]

I have this issue with a Ryzen 1800X and a Gigabyte GA-AB350 Gaming (not 3) mainboard. I can boot the live desktop from USB if I pass acpi=off, but /proc/cpuinfo lists only one core.

[Eric Joslyn (mirth99)]

I have exactly the same problem with a Gigabyte AB350-Gaming 3 motherboard and a Ryzen 1800+.

- Ubuntu variations (Ubuntu, Neon, Kubuntu, Budgie) and openSUSE do not work out of the box. They stall with a bunch of 'irq trap at vector 07' messages.
- For all of these, after adding acpi=off to the kernel parameters, it runs/installs, but runs on only one core. The system locks up when you try to shutdown.
- I tried other acpi parameters but nothing seems to make a difference. Only acpi=off has an effect.
- Fedora and KaOSx run and install with no obvious problem.

[Roland Kaiser (roland-kaiser666)]

Here is some additional information. I built upstream kernel 4.11-rc5 with

    CONFIG_PINCTRL_AMD=m
    CONFIG_DEBUG_PINCTRL=y
    CONFIG_DEBUG_GPIO=y

and blacklisted pinctrl-amd so as to able to boot that kernel and set up netconsole. Doing a

    modprobe pinctrl-amd

then results in a crash (obviously) and output along the lines of what you can see in the attached files. Case A seems to be more common than B. In case B, the system can still be used for a few seconds before becoming unresponsive.

[Marc Singer (nextized)]

Hey Roland

Your informations are a breakthrough in this case, since Case A seems to be a NullPointerException inside the Kernel (maybe amd_gpio is trying to access an object that has been removed in newer kernel versions?).

Case B was almost 100% for myself but is also very interesting.
It seems like when amd_gpio wants to register its interrupts / irq, it fails on Vector 07 and then does not advance the loop (it seems to be stuck in a loop after). Could be that Gigabyte changed something about their IO on the mainboards and the standard operation does not work on that. I'll forward the informations to an engineer from amd to possibly create an update that will hopefully work with our Gigabyte Mainboards.

[Roland Kaiser (roland-kaiser666)]

Yeah, it looks like the call to irq_desc_get_handler_data() inside that irq handler is returning null. But I don't understand most of that code, so no idea what might be causing that.

As a side note: From the looks of it it's probably not going to be relevant, but just in case it is I have dumped the contents of the memory region which is returned by the _CRS method of the AMDI0030 ACPI device.

[Richard (rs-workmail)]

I can confirm this issue on a Gigabyte GA-AX370-Gaming 5 (X370 chipset) with F4 bios and an AMD Ryzen 7 1700 CPU.

I moved my disks from an sandy bridge system with Kubuntu 16.04.2 installed and kernel 4.4.0-71-generic which boots fine on the new hardware.

When I upgrade the kernel to 4.10.0-14-generic I get the handle_irq() spam at boot.
With acpi=off I am able to boot but I get USB enumeration errors during boot and my wifi usb stick (AR9271 802.11n) stops working.

I have only a GTX 770 installed in the PCI-E slots (using the nvidia 375.39 driver from apt).

[Roland Kaiser (roland-kaiser666)]

I have found that kernel 4.1 actually seems to work fine on my board (AX370-GK7, F2 BIOS), even when compiled with CONFIG_PINCTRL_AMD=y. Note that to properly test kernels that old, the correct ACPI device ID needs to be added as seen in commit

    42a4440 pinctrl: amd:Add device HID for future AMD GPIO controller

since otherwise the driver won't find the device.

Anyway, I did that, and there is now a /sys/devices/platform/AMDI0030:00 directory and no obvious complaints in dmesg, so everything seems to be working as intended as far as I can tell.

I did some bisecting and I am now fairly certain that a regression was introduced early in the 4.2 cycle with the commit

    0be275e x86/irq: Use cached IOAPIC entry instead of reading from hardware.

I need to do some more testing to make sure I didn't mess up somewhere, will report back when/if I know more.

[Roland Kaiser (roland-kaiser666)]

Bad news.

Using 4.11-rc5 as a base and reverting the commit I previously mentioned (0be275e3a5607b23f5132121bca22a10ee23aa99) does yield a mostly functional kernel.

I say "mostly" because there is still a lone "unexpected IRQ trap" in dmesg when the pinctrl driver loads, though the system boots successfully... most of the time. Also, configuring pinctrl-amd as a module instead of built-in and loading it later on will still result in the same crash scenarios as before.

So the code that the commit touches may be vaguely related to the problem, but it's certainly not the root cause.

Still kinda seems like a BIOS bug to me, particularly since other motherboard vendors are not affected. Let's hope someone at Gigabyte does something about this soon. I just tested the F3e beta BIOS, made no difference.

[Roland Kaiser (roland-kaiser666)]

In the meantime, is there any particular reason Ubuntu kernels need to be shipping with pinctrl-amd built-in?

If not, I suggest building it as a module from now on, so that users affected by this issue don't have to resort to custom built kernels, but can instead just blacklist the module.

Note that this issue is present on xenial and yakkety as well, not just zesty.

[Marc Singer (nextized)]

Hey Roland

You are surely right its a Gigabyte related issue, allthough it's amost surely a change the engineers from Gigabyte made on either UEFI or even Hardware level. Its pretty common for mainboard manufacturers to add custom stuff to their mainboards as additional selling points.

The question is more if they know about this issue or if they even care about it, since the linux community is very small compared to windows.

[Robert Sandru (rsandru)]

In any case if this is confirmed we need to raise the issue with them. In the past they've been "hiding" behind the fact that the motherboard specification page says:

Operating System

    Support for Windows 10 64-bit
    Support for Windows 7 64-bit
    * Please download the "Windows USB Installation Tool" from GIGABYTE's website and install it before installing Windows 7.

Given recent Microsoft announcements that the most recent silicon will only be supported under Windows 10, this makes all Ryzen Gigabyte boards pretty much mono OS... I'm hoping that I'm not the only one to find it unacceptable that consumer choice has narrowed down to such an extent :-)

[Marc Singer (nextized)]

I've created a post in the Gigabyte Forums AM4 Beta BIOS Thread.

http://forum.gigabyte.us/thread/886/am4-beta-bios-thread?page=21

Maybe they can figure out what is causing this issue to happen.

[DIONYSIS KAPATSORIS (dfk7677)]

I have the same problem with Gigabyte AB350M-Gaming 3, latest BIOS(F4) and ubuntu 16.10. I can only boot with kernel 3.8, but get in an infinite log in loop.

[Eric Joslyn (mirth99)]

OK - I found a good mention of the problem here. He says “There's a bug relating to the Gaming 5 board where dmesg gets flooded with errors about interrupts unless you disable CONFIG_PINCTRL_AMD.”
https://www.reddit.com/r/VFIO/comments/63igfa/horrible_amd_ryzen_performance/

That prompted me to compile a new kernel without the pinctrl-amd option. That option is found when in the kernel configuration utility makeconfig, in the section called Device Drivers/Pin control/AMD GPIO pin control.

I used the compiling instructions found here:
http://askubuntu.com/questions/163298/whats-a-simple-way-to-recompile-the-kernel

I used GIT to download the latest kernel source (4.11.0-rc6) I made no changes to the default kernel configuration other than to check off the pinctrl-amd option. Compiling took over 2 hours, running on only one core!

The new kernel unstalled with no problems. acpi=off is no longer necessary to boot, and 16 cores are visible and active in KDE sysguard. No crashes.

After installing the new kernel, I tried compiling with the multicore switch, 'make -j 16'. It's pretty impressive to see 16 threads running at 97%.

Of course, this is just a temporary fix.

[Robert Sandru (rsandru)]

@Eric Joslyn: you can do the same with a stable kernel from the 4.10 series, no need to be on the bleeding edge with release candidates. (I see it's been backported to 4.9 too)

The key patch you really want to get in is the one that correctly identifies the hyper-threading capabilities of the Ryzen which came in 4.10 so anything from 4.10.0 upwards should give you the full performance. More info over there: https://www.phoronix.com/scan.php?page=news_item&px=AMD-Ryzen-Newer-Kernel

[Marc Singer (nextized)]

Good news:

For the people that are not able to compile a custom kernel, I provide the download links of the binarys for you:

- Linux 4.10.3 AMD64 with Ubuntu patches
- Compiled from official ubuntu "linux-source-4.10.0" package
- AMD GPIO PIN Control removed
- No other changes from default configuration
- Compiled on Linux Mint 18.1 (Ubuntu 16.04 LTS)

Should work for 16.04 and has been tested in a VM. I will test it tomorrow on a Gigabyte X370-G5 System.
Could you test it and provide me with further feedback?

https://www.nextized.net/repo/linux/ubuntu-linux-4.10.3_ryzen01_amd64_generic.tar.gz

Thanks you

nextized

[Marc Singer (nextized)]

Ups I posted the wrong URL to my custom kernel:

http://www.nextized.net/repo/linux/ubuntu-linux-4.10.3_ryzen01_amd64_generic.tar.gz

- Linux 4.10.3 AMD64 with Ubuntu patches
- Compiled from official ubuntu "linux-source-4.10.0" package
- AMD GPIO PIN Control removed
- No other changes from default configuration
- Compiled on Linux Mint 18.1 (Ubuntu 16.04 LTS)

Please test it and give me feedback.

[Leonardo Varuzza (varuzza)]

I tested the kernel from Marc Singer in my Gigabyte X370 Gamming K7 with Ryzen 1700 and everything is working (typing it right now from the system).

[Julio Alves (scalv)]

Didn't work in my AB350 Gaming 3 with Ubuntu 16.04, I get a lot of "unexpected irq trap at vector 07" without acpi=off before system freezes.

I've also tried other kernels, from 4.8 to 4.10, to no avail.

[Eric Joslyn (mirth99)]

@Marc Singer. The new 4.10.3 kernel seems to work ok in my Gigabyte AB350 Gaming 3. No need for acpi=off, and all 16 cores show up in the system monitor. Is there anything specific that I can test?

I originally compiled the latest 4.11 rc6 kernel out of sheer curiosity. I wanted to see what was default for the amd gpio option (it was active). Once I had the code, I thought what the heck, might as well compile it.

[Marc Singer (nextized)]

What you guys could test with the new kernel are mainboard related stuff like sound, usb and so on. I would like to know what effects the disabling AMD Pin CTRL has on other stuff.

[Robert Sandru (rsandru)]

Some inputs for what it's worth... with a self built 4.10.4 from a few weeks back but same config as yours (AMD pinctrl disabled) - no other changes:

- USB works fine on all the ports I tried (several devices)
- Network works fine (both ports Intel and Killer Networks)
- NVMe drive works fine
- SATA drive works fine
- NVidia closed source drivers work fine with the patch (see above in this thread for pointers to make the nvidia driver work on 4.10.x)

- I haven't tested audio but it's displayed correctly in the settings

I haven't yet noticed something that doesn't work because of that change.

I'm tempted to think that the gpio ports are possibly used to control the onboard LED lighting and stuff like that...

Board is GA-AX370-Gaming5

[Tekkla (tekkla)]

@Marc Singer Audio is very rudimental on my GA-AB350-Gaming 3, only a simple line-out, but I can hear sound. Everything else works. Thank you very much!

[Marc Singer (nextized)]

Thanks you guys for your early feedback.

I'll release a binary 4.11 with fixed audio for Gigabyte Ryzen Mainboards as soon as 4.11 is stable (end of april). The release candidates are mostly stable but not recommended to use on productive systems.

If its just the LED Control that doesn't even matter for most users. It seems like just Gigabyte Mainboards are issued here, so I guess this will not be fixed in the official channels.

[Lei Ni (nilei81)]

@Marc Singer, thanks for the 4.10 kernel! It works fine on my Gigabyte AB350M Gaming 3 + Ryzen 1700 system, SATA, audio, networking and video (Nvidia 381.09 manually installed) all work fine.

[Julio Alves (scalv)]

It worked on my AB350 Gaming 3. The kernel was installed, but it wasn't running. I had to manually edit the loader. No problems so far.

Thank you, @Marc Singer.

[Gabi Dumi (djibrille)]

I got it to work too on my Gigabyte AB350M Gaming 3 + Ryzen 1700 +nvidia 750.
Installed Ubuntu Mate 16.04.01 (that was on kernel 4.4.0) - this worked reasonably well, some errors popped up from time to time (eg. USB 3.0 stick was unmounting then auto-remounting from time to time, also compiling the kernel I got a seg fault from cc - don't know if it's related).
I downloaded the zesty kernel and used the instructions above to configure (disabled AMD GPIO pin control as mentioned above) compile and install 4.10.0. Compilation went ok (except the seg fault I mentioned and some packages I had to manually install) all 16 threads where at 60-70% when compiling.
Now I'm running 4.10.0-19 with seemingly no problems, everything seems to work including nvidia drivers 375.39, sound through hdmi, network, usb. No errors yet.
I'm not a linux expert by any means so there might still be problems I haven't discovered.

[Dominique Joye (djoye)]

Thanks to Marc, the sequence: install 16.04.1, install the kernel update and then nvidia drivers, patched of course seems to work on Gigabyte 350 Gaming 3.

[Stefan (groaznic)]

Bug confirmed on Gigabyte GA-AB350-Gaming (non-X version) @ BIOS version F2 + Ryzen 1800X when trying to boot installation media of Ubuntu 17.04 final release.

[Milosz Tanski (mtanski)]

There's a SEMI official update here on the Gigabyte forums from an employee (Matt). http://forum.gigabyte.us/thread/886/am4-beta-bios-thread?page=32&scrollTo=3601 Does anybody know if there's anything queued up from AMD for 4.11 for GPIO?

[Joseph Salisbury (jsalisbury)]

I built a Zesty test kernel, which can be downloaded from:
http://kernel.ubuntu.com/~jsalisbury/lp1671360/

Can you test this kernel and see if it resolves this bug?

Note, with this test kernel you need to install both the linux-image and linux-image-extra .deb packages.

Thanks in advance!

[Roland Kaiser (roland-kaiser666)]

I tested the kernel on an AX370 Gaming K7 with F3 BIOS, it does seem to resolve the issue for me.

[Dominique Joye (djoye)]

I tested Joseph solution on B350 Gaming 3 (bios F6) without success (stop in the boot without message). I can attach the syslog part if it helps
Many thanks in any case

[Leonardo Varuzza (varuzza)]

It also didn't work for me on X370 Gaming K7 (bios F3). But the 4.10-3
works fine.

On Wed, Apr 19, 2017 at 8:06 AM Dominique Joye <email address hidden> wrote:

> I tested Joseph solution on B350 Gaming 3 (bios F6) without success (stop
> in the boot without message). I can attach the syslog part if it helps
> Many thanks in any case
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1671360/+subscriptions
>

[Milosz Tanski (mtanski)]

Roland, you sure there isn't another BIOS toggle you managed to find? On the Gaming K7 with F3 BIOS and this still happens unless the offending code is compiled out.

[Allingo.M (allingo)]

I tested Joseph solution on Gigabyte AB350M(bios F4) with success.

[Roland Kaiser (roland-kaiser666)]

I double checked a few times. Joseph's test kernel boots fine for me, while the default kernel from the repositories still hangs with the same BIOS settings.

Are you guys sure it's still the same issue that's preventing a successful boot on your systems?

[Marc Singer (nextized)]

If you want I could power up a Zesty VM to compile a new kernel with same settings as my 4.10.3 kernel and provide it for you guys.

[Tekkla (tekkla)]

I would appreciate that very much :)

[Tekkla (tekkla)]

@Joseph I just installed your kernel an everythings working with my GA-AB350-Gaming3 board like the kernel from Marc.

[Dominique Joye (djoye)]

Just a stupid question:on my AB350 I have installed in the order: 1) 16.04.1, 2) Marc 10.3 Kernel, 3) Last version of NVidia driver for my 1050ti. That was OK

When I tried Joseph's kernel, it was after the installation of the Nvidia drivers. Do I have to remove them in order to test that properly?

[Gabi Dumi (djibrille)]

I didn't remove the Nvidia drivers when I upgraded the kernel, and everything worked fine after that, including the drivers, no reinstall necessary either. I compiled the kernel from sources though, don't know if that matters.

[edward (deltorodata)]

hi guys,

maybe my information helps a bit (at least a hope)

my PC is a 1700x with the gaming aorus 5 from gigabyte (GA-AX370-Gaming K5), got 32 GB RAM but i dont think that it has much to do with all the problems metioned above.

i got all those errors, and i can only use my computer with my new graphic card (all cores and stuff) using the kernel "4.7.8-040708-generic"... there is no other kernel that works with that graphic card (rx 460) for that reason im using it... and other newer kernel just failed... i have like 24 newer kernels installed... all of em come with this IRQ erros and stuff...

but i did some check, i changed the default grub, i wrote (if am not mistaken) acpi=off noacpi or something like that (sorry i cant remember it right now, im not at my pc) and i got this irq 7 issue even with the kernel im using... so i had to changed the default grub... and i could work again with the 4.7.8... kernel

i hope this information can help you somehow.

[edward (deltorodata)]

i forgot to mention that i use the f4 version of the bios

[Milosz Tanski (mtanski)]

edward,

If you have ACPI off, you'll end up running with only 1 core enabled. Just take a look at number of cores in top (press 1 to list cores/threads).

[Kevin Norman (kn100)]

I'd like to add that I have a Gigabyte AB350m Gaming 3 motherboard along with a Ryzen 1600 CPU, on the F4 (newest) BIOS.

I have found an workaround, which can be found here: https://www.reddit.com/r/openSUSE/comments/65ffc5/kernel_panic_on_boot_related_to_ryzengigabyte/ (Relates to OpenSuSE Tumbleweed but the issue is the same on Ubuntu). This workaround allows the system to boot, however it is unstable.

If I can supply any information that will be helpful to the developers, please just let me know and I'll supply what I can!

[edward (deltorodata)]

Thanks Milosz!

actually i couldnt start the pc like that, im still using the kernel 4.7.8, the good thing is that all cores wore good, like i said... the problem is that the Network cards are totally crazy and the soundcard just does not work anylonger since a couple of weeks.

[edward (deltorodata)]

Hi again guys,

ive tried Joseph's kernel but this is what i got...

linux-headers-4.10.0-19-generic (4.10.0-19.21~lp1671360) wird eingerichtet ...
Examining /etc/kernel/header_postinst.d.
run-parts: executing /etc/kernel/header_postinst.d/dkms 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
Error! Bad return status for module build on kernel: 4.10.0-19-generic (x86_64)
Consult /var/lib/dkms/rtl8812au/4.3.8.12175.20140902+dfsg/build/make.log for more information.
linux-image-4.10.0-19-generic (4.10.0-19.21~lp1671360) wird eingerichtet ...
Running depmod.
update-initramfs: deferring update (hook will be called later)
Examining /etc/kernel/postinst.d.
run-parts: executing /etc/kernel/postinst.d/apt-auto-removal 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
run-parts: executing /etc/kernel/postinst.d/dkms 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
ERROR: Cannot create report: [Errno 17] File exists: '/var/crash/rtl8812au-dkms.0.crash'
Error! Bad return status for module build on kernel: 4.10.0-19-generic (x86_64)
Consult /var/lib/dkms/rtl8812au/4.3.8.12175.20140902+dfsg/build/make.log for more information.
run-parts: executing /etc/kernel/postinst.d/initramfs-tools 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
update-initramfs: Generating /boot/initrd.img-4.10.0-19-generic
run-parts: executing /etc/kernel/postinst.d/pm-utils 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
run-parts: executing /etc/kernel/postinst.d/unattended-upgrades 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
run-parts: executing /etc/kernel/postinst.d/update-notifier 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic
run-parts: executing /etc/kernel/postinst.d/zz-update-grub 4.10.0-19-generic /boot/vmlinuz-4.10.0-19-generic

it looks like the driver of one of my cards does not like this kernel :S

[edward (deltorodata)]

i treid to boot with Joseph's kernel and the machine booted, but there is no way for me to use my keyboard :S

[Marc Singer (nextized)]

@Edward: It seems like the kernel fails to build a module for a realtek usb wlan card (https://github.com/gnab/rtl8812au) from kernel 4.3.8 - could you remove this module or the kernel 4.3.8?
What version of ubuntu are you using? and could you reinstall it when its necessary?

[edward (deltorodata)]

Hi Marc,

yep i got that.. i desinstalled it so that it would work better.

im running 16.04, i dont know actually what to do, cuz i fond now way to boot using the keyboard with that kernel, and also both monitors (i might solve that with the graphic card, but i need to use the keyboard for that.)

[edward (deltorodata)]

i need to find* sorry, i am writing from my Android device.

[Woody (wo0dy)]

I'm experiencing the same issues. I've tried the blacklist pinctrl-amd workaround and it still won't work. This board won't even let me adjust memory or cpu clocks or voltages which is advertised! I will be returning my gigabyte AB350m HD3 board.

To everyone not seeing cpu temps or fan speeds in lm-sensors, install the it87 module. All temperatures and fan speeds become readable afterwards: https://github.com/groeck/it87 You just need to sudo make and make install it.

[Woody (wo0dy)]

Turns out the voltage and clock adjustments can be made (I reported earlier, directly above this comment, that it was not possible which is wrong) it can be done, its interface is .. unintuitive. (hint: use the + and - keys when item is highlighted.

[Woody (wo0dy)]

I tested joseph's kernel (comment no. 68). For me I just get a black screen. no video output at all on mobo gpu or discreet.
My hardware is: gigabyte AB350m HD3 with bios release F2. GPU is Nvidia gtx1060. Ryzen 1600.

[Woody (wo0dy)]

Just to add to my last comment above ^, Josephs' kernel in comment no. 68 does work for me. the installation of the kernel did not complete properly on my first attempt, hence I had a black screen.

Thanks joseph!!!
If this fix gets mainlined, that'd be wonderful.

[Marc Singer (nextized)]

Hello Guys

Seems AMD_PINCTRL has been excluded from official sources; I've built 2 new kernel versions today and both of them had AMD_PINCTRL disabled, even though I applied all the patches from kernel.ubuntu.org.

Could someone please check if the newest stuff works without compiling a custom kernel.

If not you could download the newest compiled kernel from my repo:

4.11 RC8 : http://www.nextized.net/repo/linux/ubuntu-linux-4.11.rc8_ryzen01_amd64_generic.tar.gz

[Woody (wo0dy)]

For joseph's kernel: Install it, then proprietary drivers. It works a charm! New kernel also now supports my Gigabyte gc-wb867d-i wifi!

Thank you so much! Sorry folks for so the number of comments in a row, if there were an edit button I'd be using it. :-)

[Daniel Harvison (sirbubbles01-x)]

Marc Singer, could you please tell me which kernel version this is? And you say it's the official ubuntu repos, not the custom kernel ppa?
I've managed to compile and install 4.11-rc5 without too many issues, but an official fix would be wonderful.

[Marc Singer (nextized)]

Its version 4.11 RC8

I took the source from kernel.org, applied ubuntu fixes (0001 - 0006) from kernel.ubuntu.org (so it's basicly a ubuntu kernel), then configured it and found out that AMD_PINCTRL was disabled from the beginning.
I'm not sure if ubuntu finally updated the fixes to disable it or if its just because I took the original sources from kernel.org as base.

Could someone please check for me? Thanks.

[edward (deltorodata)]

i dont understand why i the Joseph's kernel wont let me type (kernel is on, but aint gonna work) anything :( it looks like the solution for me for, at least temporal... :(

[Chris Reid (chrinor2002)]

Tried the kernel from #94 on the following:

17.04 desktop
AMD Ryzen 5 1500X
Gigabyte AORUS AX370-GAMING K7 (F3 bios update)
Gigabyte GeForce GTX 1050 Ti (nvidia-378)

Booted fine, but I was unable to login. kern.log is showing:
Apr 25 19:30:41 spartan kernel: [ 7.634267] compiz[1649]: segfault at 38 ip 00007ff61b0f298e sp 00007fff8e991350 error 4 in libmove.so[7ff61b0e8000+15000]

I also experienced this with the other 4.11.0 rc1 and rc7. However 4.10.3 from #54 works with zero problems. Audio over HDMI even works, and I have access to all my cores.

[Roland Kaiser (roland-kaiser666)]

Marc, from what I can tell, pinctrl-amd is still enabled on all the Ubuntu kernels, including the sources in the various git repos (git://kernel.ubuntu.com/ubuntu/ubuntu-zesty.git etc.)

And I don't think that will change. It is needed for things like trackpads on some laptops, I believe. Meaning, completely disabling it may be a fix for us, but it would break stuff for other people.

[Roland Kaiser (roland-kaiser666)]

I still think however that shipping Ubuntu kernels with pinctrl-amd built as a blacklistable module would be an acceptable interim solution until someone of the AMD/GIGABYTE dream team that's responsible for this mess can deliver an actual fix.

[edward (deltorodata)]

i tried the wrong kernel, now i am using that 4.10.3 (ryzen) kernel, it works fine

[Klaus Gebert (klgebert)]

Hello

I installed 17.04 on my Gigybyte 370 Gaming 5/Ryzen 1700x with acpi=off und integrated the the 4.11RC8 Kernal into that installled Version - but without apcpi=off it still crashes like before.

[Marc Singer (nextized)]

Hello Guys

I just compiled 4.11 final without AMD PINCTRL this morning. Could someone please test it?

http://www.nextized.net/repo/linux/ubuntu-linux-4.11.rc8_ryzen01_amd64_generic.tar.gz

Thanks.

Marc

[Ying (yingternet)]

Hi Marc, just tried the following

http://www.nextized.net/repo/linux/ubuntu-linux-4.11.rc8_ryzen01_amd64_generic.tar.gz

systems boots fine, ssh is responding, but GUI (Gnome) won't let me login (ubuntu 16.04)

I am currently running kernel 4.10.3_ryzen

http://www.nextized.net/repo/linux/ubuntu-linux-4.10.3_ryzen01_amd64_generic.tar.gz

Thanks,

Ying

[Marc Singer (nextized)]

Hello Guys

Posted the wrong link again.

http://www.nextized.net/repo/linux/ubuntu-linux-4.11.0_ryzen01_amd64_generic.tar.gz

This is the new 4.11.0 Final without AMD_PINCTRL.

Thanks you for testing it.

[Ying (yingternet)]

I just tried with newest

http://www.nextized.net/repo/linux/ubuntu-linux-4.11.0_ryzen01_amd64_generic.tar.gz

same behavior

systems boots fine, ssh is responding, but GUI (Gnome) won't let me login (ubuntu 16.04)

I noticed the following error (not sure if it is related) while installing kernel (I have nvidia 1070 and running proprietary driver)

ERROR (dkms apport): kernel package linux-headers-4.11.0 is not supported
Error! Bad return status for module build on kernel: 4.11.0 (x86_64)
Consult /var/lib/dkms/nvidia-375/375.39/build/make.log for more information.
update-initramfs: Generating /boot/initrd.img-4.11.0

Thanks,

Ying

[Leonardo Varuzza (varuzza)]

Just booted this 4.11 and it is working on my Gigabyte Aorus K7 with Ryzen
1700. Lets see if the stability will improve on the 4.10

On Tue, May 2, 2017 at 7:06 AM Ying <email address hidden> wrote:

> I just tried with newest
>
> http://www.nextized.net/repo/linux/ubuntu-
> linux-4.11.0_ryzen01_amd64_generic.tar.gz
>
> same behavior
>
> systems boots fine, ssh is responding, but GUI (Gnome) won't let me
> login (ubuntu 16.04)
>
> I noticed the following error (not sure if it is related) while
> installing kernel (I have nvidia 1070 and running proprietary driver)
>
> ERROR (dkms apport): kernel package linux-headers-4.11.0 is not supported
> Error! Bad return status for module build on kernel: 4.11.0 (x86_64)
> Consult /var/lib/dkms/nvidia-375/375.39/build/make.log for more
> information.
> update-initramfs: Generating /boot/initrd.img-4.11.0
>
>
> Thanks,
>
> Ying
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1671360/+subscriptions
>

[RussianNeuroMancer (russianneuromancer)]

How stability with 4.11 posted above overall?

[Kai-Heng Feng (kaihengfeng)]

I guess it's better not to load pinctrl_amd until AMD & Gigabyte solve the issue.

What's the name of the board do you guys have?

You can get the name via `cat /sys/class/dmi/id/board_name`.

[Woody (wo0dy)]

cat /sys/class/dmi/id/board_name
gives me:
AB350M-HD3-CF
Yes, it is an Gigabyte AB350m HD3 that I own.

[Tekkla (tekkla)]

@Nextized Your kernel 4.11 works for me like a charm. No instabilities and sound is fully working.

[belgianguy (wietse-jorissen)]

@Nextized: AX370-Gaming K7 reporting in, after installing your kernel, I can now boot into 17.04 without any hacks, all seems to work (haven't been able to test much yet though, need to move some stuff around). Sound is working (Logitech USB headset)

[Ying (yingternet)]

@Nextized: My problem is simply because nvidia proprietary driver is not compatible with linux 4.11 kernel, so when I use open source version it works. so your kernel works for me

[Kai-Heng Feng (kaihengfeng)]

Please try http://people.canonical.com/~khfeng/lp1671360/
It should work on "AB350-Gaming 3-CF" and "AB350M-HD3-CF".

[belgianguy (wietse-jorissen)]

cat /sys/class/dmi/id/board_name
gives me:
AX370-Gaming K7

(I didn't mention the command in my previous post)

[Kai-Heng Feng (kaihengfeng)]

So it happens to all Gigabyte boards? A320 & B350 & X370?

[belgianguy (wietse-jorissen)]

@kaihengfeng: that seems to be very likely. (Aorus is just a Gigabyte branding name)

[Tekkla (tekkla)]

@kaihengfeng: GA-AB350-Gaming3 is affected too.

[Stefan (groaznic)]

For the people who aren't boss enough and just want to install Ubuntu once this is sorted out, will the fix/work-around be rolled into the iso as soon as it's confirmed fixed?

[Leonardo Varuzza (varuzza)]

I can boot 4.11, but it freezes randomly, I don't if the problem is in the
NVIDIA driver, I'm using the 381.09 with the patch to compile on 4.11

On Wed, May 3, 2017 at 3:16 PM Tekkla <email address hidden> wrote:

> @Nextized Your kernel 4.11 works for me like a charm. No instabilities
> and sound is fully working.
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1671360/+subscriptions
>

[Ondrej Masek (doublefacepalm)]

Same issue here on Ryzen 5 1600 + Gigabyte AB350M Gaming 3 when fresh-installing Kubuntu 17.10. Shouldn't the importance of this issue be at least "High"? What can be worse than if you can't even install the OS on your PC? In the meantime as a workaround, Fedora 26 seems to work fine.

[Kai-Heng Feng (kaihengfeng)]

Please try Linux kernel here: http://people.canonical.com/~khfeng/lp1671360-2/

Don't forget to report the result =)

This should make all Gigabytes board stop to load pinctrl_amd.

[belgianguy (wietse-jorissen)]

@kaihengfeng: I've been experiencing some system freezes. I have a clean 17.04 install except for the kernel (which is your 4.11 custom kernel). Any idea where I could look for what's causing it?

[Kai-Heng Feng (kaihengfeng)]

Probably LP: #1674838.
Can you check the /var/log/kern.log and see if it has the same error?

[Wenbin Leo (wenbinleo)]

Tested on AX370-Gaming K5-CF :
Default Ubuntu kernel do not work.
4.10.0-21 compiled by kaihengfeng [http://people.canonical.com/~khfeng/lp1671360/]
also shows "unexpected irq trap at vector 07" error message, can not boot to desktop.

Both kernel compiled by nextized worked:
http://www.nextized.net/repo/linux/ubuntu-linux-4.10.3_ryzen01_amd64_generic.tar.gz
http://www.nextized.net/repo/linux/ubuntu-linux-4.11.0_ryzen01_amd64_generic.tar.gz

[Woody (wo0dy)]

What will it take to get this bug fixed!?

Upstream most likely is (or will) point at the hardware vendors (I don't know, thats just me guessing), while the hardware vendor says 'not our problem, we don't support linux'. What are we to do?

[Wenbin Leo (wenbinleo)]

Testing on AX370-Gaming K5-CF :
4.10.0-21 compiled by kaihengfeng on http://people.canonical.com/~khfeng/lp1671360-2/ seems working, using it to post result.

[Milosz Tanski (mtanski)]

I'm trying to follow up with the hardware vendor (Gigabyte) to help us identify the root cause so we can fix it / work around it in the kernel. Here's my last conversation with Matt from Gigabyte: http://forum.gigabyte.us/thread/886/am4-beta-bios-thread?page=47&scrollTo=4429

[Ondrej Masek (doublefacepalm)]

Submitted a request with Gigabyte support just now. Can't believe this has been dragging on for two months. Who would have thought to check whether the chosen motherboard even boots up with the latest OS prior to purchasing :\ Extremely frustrating.

[Woody (wo0dy)]

Awesome. Thank you Milosz for following up there. Perhaps a patch can be created specifically for these boards, something along the lines of 'if board is type am4, vendor==gigabyte: disable pinctrl' Its messy, but it'd work.

[TheLXK (thelxk)]

4.10.0-21 compiled by kaihengfeng on http://people.canonical.com/~khfeng/lp1671360-2/ working for me without any additional kernel parameters on a GA-AB350 Gaming 3, with a Ryzen 1700X.

[Jaka Jaksic (jjaksic)]

If pinctrl causes issues with certain hardware, but is required for certain other hardware, and if it isn't possible to automatically determine which is which, then it would make sense to make it controllable through a boot flag (like acpi=off).

While having to specify a boot flag is still less than ideal, it is far preferable to having installing a different distro or build a custom kernel.

[Ondrej Masek (doublefacepalm)]

Just got this from Gigabyte Support:

"Hello,
Sorry but Gigabyte do not guarantee Linux Platform on the desktop motherboard.
It support for Windows 10/7."

Extremely helpful :/

In the meantime, I booted the Kubuntu install ISO with acpi=off setting. The same option was required for the first boot after install. Kernel version 4.10.0-21 compiled by kaihengfeng on http://people.canonical.com/~khfeng/lp1671360-2/ seems to work fine, the only thing to report is a long boot time.

Thanks Kai-Heng Feng!

[Leonardo Varuzza (varuzza)]

I just loaded the stock kernel with the Marc Singer configuration and
changed the preemption mode to low-latency. Running much smoothly now.

On Sun, May 7, 2017 at 9:30 AM Ondrej Masek <email address hidden>
wrote:

> Just got this from Gigabyte Support:
>
> "Hello,
> Sorry but Gigabyte do not guarantee Linux Platform on the desktop
> motherboard.
> It support for Windows 10/7."
>
> Extremely helpful :/
>
> In the meantime, I booted the Kubuntu install ISO with acpi=off setting.
> The same option was required for the first boot after install. Kernel
> version 4.10.0-21 compiled by kaihengfeng on
> http://people.canonical.com/~khfeng/lp1671360-2/ seems to work fine, the
> only thing to report is a long boot time.
>
> Thanks Kai-Heng Feng!
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1671360/+subscriptions
>

[Woody (wo0dy)]

Gigabyte says they don't support linux. End of story as far as they're concerned. I'd really just like to see something happen. A boot parameter would be fantastic. How would we go about notifying relevant parties to get something arranged/requested? Is here the right place?

Its been over 2 months since this platform released, would be nice to see a fix/proper workaround.

[jonnyboysmithy (jonnyboysmithy-linux-kernel-bugs)]

Link to what appears to be the same bug over at the ubuntu launchpad site: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1671360

(its linked there to here, but not vice versa, so doing it above^ )

[Nagy Ákos (e-contact-pecska-ro)]

I have a same issue with GA-AB350-Gaming3 and Ryzen 1600.
It's sad Gigabyte response, now I buy a new AsRock AB350M Pro4, and in the future I don't chose gigabyte.

[Gabi Dumi (djibrille)]

I don't believe that Gigabyte can just say they don't support Linux or even just Ubuntu for that matter. Linux is not that small a market anymore and I believe Linux as an SO is very well suited for a relatively cheap multi-core cpu like Ryzen. I myself bought a Ryzen with the intent of building a nice, low-cost server for productivity oriented applications. Yes, if you want a gaming platform, Windows is a better platform, but then an Intel I7 is also a better choice in that case.
So I think a big company like Gigabyte is not going to just let Linux users choose another manufacturer that supports them, it would not be good for business.
Support for temperatures and frequency monitoring tools would also be nice...
Anyone knows if there's progress about this? Since I run my CPU under the stock Spire cooler and in an media center case knowing what the temps are under load is not entirely unimportant.

[Woody (wo0dy)]

@Gabi. I hope you're right.

Regarding temperatures: https://github.com/groeck/it87
If you make and modprobe that module, that should get you all fan readings and temperature outputs.
Works a charm for me.

[Gabi Dumi (djibrille)]

Thanks @Woody!
I managed to install that module and I can now see temps, voltages and fan RPMs.
I tried to do the same a while ago but I could not get it to work. Maybe meanwhile something changed or maybe it was a PEBKAC error (problem exists between keyboard and chair).
Now, if I also find a method to label all those readings in XSensors I will stop tinkering with it(not). :)

[edward (deltorodata)]

Hi guys,

i think we should do something about this gigabyte-situation, kinda of writing then a bunch of letters from all of us, there is also a bunch of ppl on reddit having this same problem, and still, gigabyte is not giving a breath to address this situation... do we really have to accept this kinda answers "we do not support linux, see by yourself what u do"...

i think we should make us ourselves to a big voice of this thing, so that gigabyte stop threating our problem like nothing they have to do with...

any suggestions?

[Leonardo Varuzza (varuzza)]

Every linux user should put that linux is not support in their gigabyte
review.

I always bought ASUS motherboards, I gave gigabyte a try because of Ryzen
motherboards shortage, however I was my last gigabyte board.

On Sat, May 13, 2017 at 4:20 PM edward <email address hidden> wrote:

> Hi guys,
>
> i think we should do something about this gigabyte-situation, kinda of
> writing then a bunch of letters from all of us, there is also a bunch of
> ppl on reddit having this same problem, and still, gigabyte is not
> giving a breath to address this situation... do we really have to accept
> this kinda answers "we do not support linux, see by yourself what u
> do"...
>
> i think we should make us ourselves to a big voice of this thing, so
> that gigabyte stop threating our problem like nothing they have to do
> with...
>
> any suggestions?
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/linux/+bug/1671360/+subscriptions
>

[Daniel Harvison (sirbubbles01-x)]

My last gigabyte motherboard too, I'm pretty damn angry about their lack of any kind of action on this matter. In fact, I'm going to try to replace this mobo (GA-AB350 Gaming 3) at some point.

[RussianNeuroMancer (russianneuromancer)]

Drop letter to level1techs.com/contact-us
I guess their channel on YouTube is big enough to make vendor notice this issue.

[Gabi Dumi (djibrille)]

After a few struggles and Internet searches I found out that AMD-V also works with my current setup, only problem is it comes disabled by default in BIOS. It also doesn't help that it's named SVM and there are 2 related options (SVM support, and SVM mode - what the hell does SVM mean?). SVM mode was Disabled so I was having trouble launching a Virtualbox x64 mode client.
So I can confirm AMD-V and kernel 4.10.19 work OK.

[buzuk56 (buzuk56)]

Hi there,

I sent out yestaerday an email with the link to this bug report to Gigabyte team.
I received an email response today from Gigabyte team.
"Thank you for emailing GIGABYTE.
We are delighted with your interest in our products.

GIGABYTE does not support Linux OS.

Linux is open source, everyone can compile it´s own OS.

We can´t support , sorry.
Please check with the Linux community for newer kernel version..."

Same as Mazek Andrej reported above.
Although this issue is active for more than two month it sounds that there is no intend from Gigabyte to tackle it despite it is specific to Gigabyte Motherboard. I will send back my AB350 Gaming 3 to the reseller and definitely buy another brand. Likely to be an ASUS.

Regards.

[Ondrej Masek (doublefacepalm)]

Make sure you leave a review for your MB. I've already done that on Amazon UK.

[Woody (wo0dy)]

I'd return my board if I could, but I can't (had to wait 2+weeks for other parts to arrive before I started build).

Anyone have a good kernel compilation guide (or even just paste their commands I would be grateful for) so that I can build up-to-date official kernels with AMD_PINCTRL disabled? I tried various guides carefully, all failed. The problem always appears to be that this debian/ directory does not exist in my kernel source code folder.

Thanks,

[mtanski (mtanski-linux-kernel-bugs)]

This appears to impact all of Gigabyte's AMD AM4 motherboards. I have attempted to contact the vendor on their forums here: http://forum.gigabyte.us/thread/886/am4-beta-bios-thread?page=47&scrollTo=4429 . Promised to look into it, but haven't heard back. Offered to fix it myself in upstream in a PM if given enough info.

[Kai-Heng Feng (kaihengfeng)]

Please try [1].

This should prevent pintctrl-amd to be loaded on any A320/B350/X370.

[1] http://people.canonical.com/~khfeng/gigabyte-am4/

[buzuk56 (buzuk56)]

Here is a 4 steps procedures that worked on AB350 Gaming 3 Ryzen 1600 (F4 bios) to get 17.04 with all cores and threads

1)Ubuntu 17.04 install from usb bootable drive with acpi=off (press F6 when display menu)

2)log in and install sudo dpk -i *deb the kernel compiled by nextized worked:
http://www.nextized.net/repo/linux/ubuntu-linux-4.10.3_ryzen01_amd64_generic.tar.gz
3) push back acpi=on to your 17.04 system to get all core and threads from your Ryzen
sudo gedit /etc/default/grub
Change line GRUB_CMDLINE_LINUX_DEFAULT="quiet splash" to GRUB_CMDLINE_LINUX_DEFAULT="quiet splash acpi=on"
sudo update-grub
4) reboot and you should have all cores and threads on 17.04 new install

[Marc Singer (nextized)]

I would prefer 4.11.0 as it supports most of the mainboards newer ACL sound codecs:

http://www.nextized.net/repo/linux/ubuntu-linux-4.11.0_ryzen01_amd64_generic.tar.gz

As I can tell, Gigabyte is not really takling the issue because only a very small number of people are affected by it. Alternate Operating Systems are not worth investing time and money (you have to get someone with good know how in Linux Development to fix this error). Thats just a fact. If this was an Issue with ASUS or ASRock or whatever brand, we wouldn't get any support either.

The only thing we can do is
1) Install custom kernels or disable AMD_PINCTRL
2) Wait until AMD releases a fixed version of the Module;
It seems like an Issue with a NULLPOINTER on some of the higher PINS, then the Kernel gets stuck in a infinite loop, trying to get the Module to load, instead of just skipping it. If we get a release where the error is handled and loading of the module is fixed (by ignoring / aborting it), we can use Linux under Gigabyte Mainboards just fine.

But how do we get it fixed? Who is the maintainer of the Module? AMD itself? I dont really know.

[Woody (wo0dy)]

Excellent idea Marc Singer.

I did a quick google search. It turned up: https://www.kernel.org/doc/linux/MAINTAINER

There's 37 entries for pinctrl and none for amd_pinctrl

I think the most relevant one would be the entry for PIN CONTROL SUBSYSTEML

PIN CONTROL SUBSYSTEM
M: Linus Walleij <email address hidden>
L: <email address hidden>
T: git git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl.git
S: Maintained
F: Documentation/devicetree/bindings/pinctrl/
F: Documentation/pinctrl.txt
F: drivers/pinctrl/
F: include/linux/pinctrl/

I suggest we find the maintainer, contact, and attempt to make them aware of the issue?

[Woody (wo0dy)]

http://elixir.free-electrons.com/linux/latest/source/drivers/pinctrl/pinctrl-amd.c

Looks like I found the author? This is 2014-2015 so may have changed, but its probably worth a shot. Most likely he'd also know who is or how to find the current maintainer.

[dfk_7677 (dfk7677-linux-kernel-bugs)]

Knowing that kernel 4.4 boots normally (Ubuntu 16.04) and 4.8 is the first not booting (Ubuntu 16.10+), can't we assume that the problem is in the commits for 4.8 after 16/09/15?

https://github.com/torvalds/linux/commits/v4.8/drivers/pinctrl/pinctrl-amd.c
https://github.com/torvalds/linux/commits/v4.4/drivers/pinctrl/pinctrl-amd.c

[Kai-Heng Feng (kaihengfeng)]

Here's 4.11 artful linux kernel with the patch below:
http://people.canonical.com/~khfeng/lp1671360-artful/

I don't think upstream will be happy about this patch, but I'll try =)

But first, I need you guys test it for me.

diff --git a/drivers/pinctrl/pinctrl-amd.c b/drivers/pinctrl/pinctrl-amd.c
index d69e357a7a98..f2b3ed33c65a 100644
--- a/drivers/pinctrl/pinctrl-amd.c
+++ b/drivers/pinctrl/pinctrl-amd.c
@@ -29,6 +29,7 @@
 #include <linux/interrupt.h>
 #include <linux/list.h>
 #include <linux/bitops.h>
+#include <linux/dmi.h>
 #include <linux/pinctrl/pinconf.h>
 #include <linux/pinctrl/pinconf-generic.h>

@@ -739,6 +740,26 @@ static struct pinctrl_desc amd_pinctrl_desc = {
        .owner = THIS_MODULE,
 };

+static bool amd_gpio_is_gigabyte_am4(void)
+{
+ const char *board_name;
+
+ if (!dmi_match(DMI_BOARD_VENDOR, "Gigabyte Technology Co., Ltd."))
+ return false;
+
+ board_name = dmi_get_system_info(DMI_BOARD_NAME);
+
+ if (!board_name)
+ return false;
+
+ if (strstr(board_name, "A320") ||
+ strstr(board_name, "B350") ||
+ strstr(board_name, "X370"))
+ return true;
+
+ return false;
+}
+
 static int amd_gpio_probe(struct platform_device *pdev)
 {
        int ret = 0;
@@ -746,6 +767,10 @@ static int amd_gpio_probe(struct platform_device *pdev)
        struct resource *res;
        struct amd_gpio *gpio_dev;

+ /* Don't load this module if it's a Gigabyte AM4 board */
+ if (amd_gpio_is_gigabyte_am4())
+ return -ENODEV;
+
        gpio_dev = devm_kzalloc(&pdev->dev,
                                sizeof(struct amd_gpio), GFP_KERNEL);
        if (!gpio_dev)

[Adrian Bolden (adralien)]

My experience:
Ryzen 7 1700
Gigabyte AB350 Gaming Motherboard
NVIDIA gtx1070
Kubuntu 17.04 installer on USB stick.

Same install problems as others with IRQ mess.

Solved as buzuk56 and Nextized described, however the proprietary NVIDIA drivers do not work with 4.11 kernel, so used the 4.10.3 version. I installed both kernels, but had to modify Grub config and run update-grub. I added a 2 second Grub startup delay so you can get into Grub easily to mess with boot options.

I set 4.10.3 as default kernel, and have acpi=on

On install of Nvidia drivers, Linux will crash after install (can't shutdown, lots of application crashes). Forced reboot and it's OK.

System seems to work now, Nvidia drivers are working (GLXgears works), all cores are showing in system monitor. Screen locking/unlocking works, and system can restart itself with no problems.

Thank you for all your efforts on this, I've been a Linux user since 486's were a thing and I'm constantly amazed at how things have changed.

[Tekkla (tekkla)]

@kaihengfeng I just tested your latest kernel an ran into serveral problems. The IRQ error is gone but in exchange my NFS mounts in fstab won't get mounted and my monitor isn't recognized so I got only a 1024x768 resolution.

[Marc Singer (nextized)]

@kaihengfeng I don't think this will be accepted by upstream, since it's hardcoding hardware descriptors as magic numbers into the module.

@adrailien
Could you try the newest version of the nvidia drivers compatible with 4.11 http://www.askmetutorials.com/2017/04/install-uninstall-nvidia-driver-38109.html

Could someone from Canonical please answer, why this module is enabled by default on ubuntu? On other distributions it's disabled (maybe for stability / compatiblity reasons). And we still got no clue about what the functions of the GPIO Pins even are. I would like to get an answer to that.

[Wenbin Leo (wenbinleo)]

Test result of
http://people.canonical.com/~khfeng/lp1671360-artful/
on AX370-Gaming K5-CF:
Boots without IRQ error.
All devices work but ALC1220 audio codec, which is the same as previous 4.10 kernels.

[Kai-Heng Feng (kaihengfeng)]

@Marc,
Actually, there are lots of workaround/qurik table in linux kernel itself.

This module is essential for touchpad on AMD laptop. Any distro wants to have AMD laptop support should enable this.

[Marc Singer (nextized)]

Hello Guys

Just compiled a new version (4.11.2) for Ubuntu:

http://www.nextized.net/repo/linux/ubuntu-linux-4.11.2_ryzen01_amd64_generic.tar.gz

Please test it on your plattform. Confirmed works with nvidia 381

Answer from AMD developers who created this module:

Hi nextized,
This issue has been reported by a few of them, all on the Gigabyte Platform. We aren’t seeing this issue on our reference platform. We are working with other teams to resolve this issue.
Thanks,
Regards,
- Ram

[Woody (wo0dy)]

That is great news Marc, thank you for keeping us posted

[Kai-Heng Feng (kaihengfeng)]

Good to know AMD and Gigabyte are working on this issue.
So I don't need to send the patch to upstream, but as an Ubuntu sauce patch.

[DIONYSIS KAPATSORIS (dfk7677)]

@Marc
I confirm your kernel works with GA-AB350M-Gaming 3 and the nvidia-381 drivers.

[Tekkla (tekkla)]

@Marc

Your kernel works like a charm on my GA-AB350-Gaming3. No problems at all. NV 381 beta driver works for me too. Thanks for your work!

[Lei Ni (nilei81)]

@Marc, thanks again for the Kernel! It works great on my GA-AB350M-Gaming 3 + Ryzen 1700 + NVIDIA 381.22. Audio, SATA, NVME SSD all work fine with the new kernel.

[Hein (heinhuiz-deactivatedaccount)]

Will @Marc's kernel work with UEFI RAID? I'm trying to build a 2-disk system with root and boot on an SSD and swap, tmp and home on a RAID10 config consisting of 4 HDDs. I'd prefer the set-up to be UEFI based because this seems to be the most future proof.

I'm using a GA-AX370-Gaming K3 and an AORUS RX570 graphics card.

[Say-Tar Goh (saytargoh)]

Hi guys, I tried nextized's kernels above at Mint 18.1 and each of them worked (AX370-Gaming5). However I wanted to keep up with the latest kernels, and am now running Fedora without a hitch.

I then tried booting OpenSuSE Tumbleweed and got different error messages/kernel panic that seemed to be caused by the same problem, which somebody seemed to have fixed by blacklisting pinctrl-amd:

https://www.reddit.com/r/openSUSE/comments/5zu2mo/gigabyte_gaming_5/

Need my current machine for production so can't test this back in Ubuntu, but can somebody confirm? If so this will save a lot of people the hassle of eternally patching the kernel.

[Woody (wo0dy)]

I tested blacklisting as you described in the reddit post and it did not fix the vector 07 irq trap issue for me on ubuntu 16.04.

Mobo is gigabyte AB350m-HD3

[Kai-Heng Feng (kaihengfeng)]

Please try [1]. It has [2] which should solve the issue.

[1] http://people.canonical.com/~khfeng/lp1671360-regular-interrupt/
[2] https://patchwork.ozlabs.org/patch/766231/

[Wenbin Leo (wenbinleo)]

Tested on AX370-Gaming K5-CF:
http://people.canonical.com/~khfeng/lp1671360-regular-interrupt/
run flowlessly.
And previous report at #158 about ALC1220 audio codec is wrong. Analog audio did show up in the audio settings after speaker is plugged in, and 5.1 analog output works. Sorry for the misinformation.

[edward (deltorodata)]

@Marc,

thanks a lot mate! you are saving the day of all of us. Hope u are not from the NSA heheheh. thanks a lot mate!

[Gabi Dumi (djibrille)]

My Gigabyte AB350m G3 decided to play dead today. In the morning it started OK, and after a few light browsing and a little coding I shut it down. After about an hour, when I tried powering it on, it was dead. No fan spin, no LED-s, no sign of life whatsoever. After troubleshooting all day swapping parts, power supplies, removing RAM, re-seating the CPU, I was about ready to start a warranty claim. I still could not believe a month and a half old motherboard could just die without any prior symptoms. It wasn't even overclocked, even the ram was running at only 2133mhz - ultra durable my a**. As a last resort I tried removing the BIOS battery and shorting the clr_cmos jumper with the battery removed and it came back to life.
After all the problems I had with Ubuntu support, when things started to look like they were improving, this made me think seriously about ordering an Asus or even AsRock replacement board for when it decides to have some other annoying problem. I would not recommend this board in the current state, especially not for an Ubuntu Linux system.
Sorry for not being exactly on topic since this is probably not Linux or Ubuntu related. Unless Ubuntu does something strange on shutdown?

[OdelPasso (berillions)]

Hi,

I would like to buy the Gigabyte AB350M-gaming 3 + AMD Ryzen 5 1600X and do you know if this issue exist on Debian Stretch/Sid ?

Thanks for your help,

[Woody (wo0dy)]

@Gabi Its a known issue with the gigabyte boards. It is known as coldboot issue I believe. The reset procedure is to take the cmos battery out, leave the pc unplugged and batteryless for about 10 minutes and then it will work again. I had the same first reaction as you when I had the issue first time. Not fun.
 Typically it happens when the machine is left powered down for a while.

[buzuk56 (buzuk56)]

@Gabi,
I faced a similar issue with my Gigabyte mother board Z68 a couple of month ago and I sorted out the problem by applying the [2] below.

May be you could have a go with [1] or [2]

[1] https://www.youtube.com/watch?v=HhI-t5lh_38
[2] https://forums.tweaktown.com/gigabyte/57335-dual-bios-dead-corrupted-blue-screen-bios-freeze-lga-2011-ga-x79-ud3.html

Good luck.

B.

[Gabi Dumi (djibrille)]

@Woody
I also have a Gigabyte Z68-ud3h-b3 in another computer (i5 2500k) and event though it's approaching 10 years old, and it's overclocked, it never did that (there were times when it would not cold boot, but at least the fans moved and you could reset the bios via clr_cmos, no problem). My brother has a Gigabyte z97 board and this never happened to him. I will try replacing the power supply since it is quite old, maybe it's related? I measured all the rails with a DMM though and they were in spec but there could be something a multimeter can't detect like noise on the lines or too much delay on startup signal. Thanks for the info anyway!

@buzuk56
The board would not turn on at all, no fans ,no power supply start, nothing. And when it did start, it booted the normal bios no problem so it seems to me it wasn't bios corruption.

Anyway, it works OK now, hopefully it was just an isolated event.
While I was troubleshooting yesterday, I noticed on the back of the board there were very fine traces of silver colored residue. It looked like the board was soldered with solder paste and some of it spilled and flowed down on the PCB. I cleaned it as best as I could but I'm still worried about it. Any of you had this, or am I just being paranoid?

[Tekkla (tekkla)]

Yay!

[Woody (wo0dy)]

Will the fix get backported or become available on Xenial? (16.04)

[Kai-Heng Feng (kaihengfeng)]

I remember someone mentioned there's no such issue on 4.4 kernel?
Anyway, you can use Zesty's kernel in Xenial.
Install package "linux-generic-hwe-16.04-edge" will do.

[adebeus (adebeus-linux-kernel-bugs)]

I'm getting a similar error on an Acer E5-553G laptop (AMD Excavator, not Ryzen). It doesn't prevent the system from booting but does make the ELAN touchpad stop working shortly after boot, with similar dmesg output (though seemingly on a smaller volume - the bad_chained_irq errors take awhile to page through with less, but don't come anywhere near filling a 400MB file, and stop completely after the touchpad switches off). This occurs with all recent kernel versions I've tried (4.9.30, 4.10.17, 4.11.3) except 4.12-rc2, which doesn't get this error but also doesn't recognize the touchpad at all. With 4.12-rc3, it is back to the same behavior as the 4.11.3 kernel.

The only kernel I have found that works is the Ubuntu 4.4 kernel. The stock 4.4.70 kernel gets the same error, but if I replace the pinctrl-amd.c file with the version from the Ubuntu 16.04 kernel, it works fine. I haven't tried using the Ubuntu 4.4 pinctrl-amd.c file with newer kernels, but presume that it would be incompatible. I did look at the Ubuntu 17.04 kernel tree (4.10) on github and noticed that there's no difference between that pinctrl-amd.c and the upstream one, so it looks like Ubuntu has only fixed this issue in the 4.4 kernel series.

I have not yet tried updating the BIOS since Acer only provides the update as a Windows EXE.

[adebeus (adebeus-linux-kernel-bugs)]

I should also note that I'm using OpenRC and not systemd, which might be why I'm able to complete the boot process.

[Launchpad Janitor (janitor)]

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

[Leonardo Varuzza (varuzza)]

I had the same problem with my Aorus X370, changed for a ASUS prime,
everything is working, even the overclock.

On Wed, May 31, 2017 at 4:21 PM Gabi Dumi <email address hidden>
wrote:

> My Gigabyte AB350m G3 decided to play dead today. In the morning it
> started OK, and after a few light browsing and a little coding I shut it
> down. After about an hour, when I tried powering it on, it was dead. No fan
> spin, no LED-s, no sign of life whatsoever. After troubleshooting all day
> swapping parts, power supplies, removing RAM, re-seating the CPU, I was
> about ready to start a warranty claim. I still could not believe a month
> and a half old motherboard could just die without any prior symptoms. It
> wasn't even overclocked, even the ram was running at only 2133mhz - ultra
> durable my a**. As a last resort I tried removing the BIOS battery and
> shorting the clr_cmos jumper with the battery removed and it came back to
> life.
> After all the problems I had with Ubuntu support, when things started to
> look like they were improving, this made me think seriously about ordering
> an Asus or even AsRock replacement board for when it decides to have some
> other annoying problem. I would not recommend this board in the current
> state, especially not for an Ubuntu Linux system.
> Sorry for not being exactly on topic since this is probably not Linux or
> Ubuntu related. Unless Ubuntu does something strange on shutdown?
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/linux/+bug/1671360/+subscriptions
>

[Gabi Dumi (djibrille)]

I noticed a new beta BIOS ( version F5c) has appeared for AB350m G3, has anyone tried it yet?
Does it now properly boot from a live Ubuntu image?
I just got a new power supply, I think I will also try the new BIOS on the coming weekend.

[Woody (wo0dy)]

When should we expect the update with the fix to be available in Xenial? I have linux-generic-hwe-16.04-edge installed, but haven't seen any improvement.

[Kai-Heng Feng (kaihengfeng)]

@Woody

According to [1], it will be released to -updates at 26-June.

It will be at -proposed before that.

[1] https://wiki.ubuntu.com/KernelTeam/Newsletter/2017-05-31

[Tony Brobston (tbro)]

buzuk56's solution worked for me. I'm using Ryzen 5 1500X with Gigabyte GA-AB350-gaming-3.

Thanks! So glad to have Ubuntu on this machine.

[Navdeep Singh (nsingh-s)]

@kaihengfeng

Will your update be available on June26 on http://releases.ubuntu.com/17.04/ ? Those seems to be stuck on date when zesty was released. Do they files get updated as bugs get fixed?

e.g.
ubuntu-17.04-desktop-amd64.iso 2017-04-12 03:46

[Kai-Heng Feng (kaihengfeng)]

Unfortunately no. I don't think image for Zesty will get any updates.

If you want to do a fresh install, you need to use parameters like "apci=off" to install. You can remove "acpi=off" after upgrading the system.

[Kleber Sacilotto de Souza (kleber-souza)]

This bug is awaiting verification that the kernel in -proposed solves the problem. Please test the kernel and update this bug with the results. If the problem is solved, change the tag 'verification-needed-zesty' to 'verification-done-zesty'. If the problem still exists, change the tag 'verification-needed-zesty' to 'verification-failed-zesty'.

If verification is not done by 5 working days from today, this fix will be dropped from the source code, and this bug will be closed.

See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you!

[Woody (wo0dy)]

Must verification/testing be done in Zesty? I have Xenial on my machine.

[Kleber Sacilotto de Souza (kleber-souza)]

Hi @wo0dy,

You can verify the bug on Xenial by installing linux-generic-hwe-16.04-edge=4.10.0.23.16 which is currently on -proposed pocket. This kernel also have the fix for this bug.

Thank you.

[Tekkla (tekkla)]

Hi @all,

I tested it and I'm not sure that I did it right. The wiki link is somewhat outdated, because it relys on aptitude, which seems not to be installed with Zesty. Nevermind. I now have installed a kernel named

4.10.0-23-generic #25-Ubuntu SMP Fri Jun 9 09:39:09 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

It works flawless. Is this the proposed kernel to test? Any if yes, where do I have to add the needed 'verification-done-zesty' tag?

Thank you.

[Kleber Sacilotto de Souza (kleber-souza)]

Hi @tekkla,

Yes, kernel 4.10.0-23 is the one in -proposed for Zesty. So thank you for verifying it!

The tags are located just below the bug description. I will update it myself.

Thanks you.

[Woody (wo0dy)]

Hi I have installed the kernel, I'm running it on Xenial now and it works good. No IRQ07 issues what so ever. Thanks everyone involved!

[Richard (rs-workmail)]

I can confirm that 4.10.0-23-generic #25~16.04.1-Ubuntu SMP Fri Jun 9 10:45:17 UTC 2017 works
with a Gigabyte GA-AX370-Gaming 5 and UEFI version F5. (Ubuntu 16.04)

[Adie (a.k.)]

I confirm package linux-generic-hwe-16.04-edge=4.10.0.23.16.10.0 fix this bug for me on my gigabyte motherboard and Zesty

[nickleplated (nickleplated)]

Does the recent security related release (4.10.0-24) also include this fix, or should I skip it and get 4.10.0-25 from -proposed?

[John Kelly (khjaoibec7nn)]

I'm using: 4.10.0-24-generic and only able to use Ubuntu with acp=off nickleplated.

I'm unclear about how to fetch the versions that people are mentioning in the last few posts. Can someone clue me in on how to fetch `linux-generic-hwe-16.04-edge=4.10.0.23.16.10.0` or `4.10.0-25 from -proposed` with aptitude? Google is failing me. I'd be very happy to test them out and see if it solves the gigabyte motherboard issue.

[nickleplated (nickleplated)]

John, I used the link mentioned above (https://wiki.ubuntu.com/Testing/EnableProposed), but replaced xenial with zesty.

In aptitude, I looked through all the upgradable packages and only selected the ones with the new kernal version number.

[Christos (christosmichaelas)]

Hi,

My machine uses the Gigabyte X370 Gaming 5 with the 1800X. No OS on the SSD.

I'm unable to boot from USB using:

- 17.04 live usb
- 17.10 daily live usb

IRQ 07 errors and the system hangs without entering the installer.

I've tried the above grub parameters acpi=off with no luck.

What are the steps needed to boot into a live usb and get ubuntu installed to then enter -proposed and get the upgraded kernel packages for full usage?

Thanks!

[John Kelly (khjaoibec7nn)]

nickleplated, Thanks you I missed the link somehow. 4.10.0.-25 from proposed solved the issue for me with the Gigabyte AB350 Gaming 3 with bios 7.

[Kai-Heng Feng (kaihengfeng)]

Christos,

Try the "pending" one instead of "current".

[Christos (christosmichaelas)]

Hi Kai,

I've just tested pending 17.10 and have the same results. Are there any specific BIOS options required?

My system is currently using IOMMU for my 16GB ECC RAM. This is the only option I have changed in the BIOS.

Arch can successfully boot on the machine.

If you need anything testing, I can try it on my machine if needed, will use another SSD for Windows.

Thanks,

Christos

[Christos (christosmichaelas)]

BIOS version I have installed - F5
Latest release - F6

Have others installed F6 when testing?

[Kai-Heng Feng (kaihengfeng)]

What's the version of Arch's kernel?

[Christos (christosmichaelas)]

Hi Kai,

Sorry for the late reply.

Kernel v4.11.3

More info of the iso: https://www.archlinux.org/releng/releases/2017.06.01/

Hope this helps!

Christos

[Kai-Heng Feng (kaihengfeng)]

Christos,

What's the kernel version of 17.10 live USB?
Linux version 4.10.0-25 has the fix but 4.10.0-24 does not.

[Christos (christosmichaelas)]

Hi Kai,

Kernel version for the image I'm using is: 4.10.0-22-generic.

I downloaded the image from the following URL:

http://cdimage.ubuntu.com/daily-live/pending/

Should I download again as I see it was modified today, where as I downloaded it on the 26th.

Thanks

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 4.10.0-26.30

---------------
linux (4.10.0-26.30) zesty; urgency=low

  * linux: 4.10.0-26.30 -proposed tracker (LP: #1700528)

  * CVE-2017-1000364
    - Revert "UBUNTU: SAUCE: mm: Only expand stack if guard area is hit"
    - Revert "mm: do not collapse stack gap into THP"
    - Revert "mm: enlarge stack guard gap"
    - mm: larger stack guard gap, between vmas
    - mm: fix new crash in unmapped_area_topdown()
    - Allow stack to grow up to address space limit

linux (4.10.0-25.29) zesty; urgency=low

  * linux: 4.10.0-25.29 -proposed tracker (LP: #1699028)

  * CVE-2017-1000364
    - SAUCE: mm: Only expand stack if guard area is hit

  * CVE-2017-9074
    - ipv6: Prevent overrun when parsing v6 header options
    - ipv6: Check ip6_find_1stfragopt() return value properly.

  * [Zesty] QDF2400 ARM64 server - NMI watchdog: BUG: soft lockup - CPU#8 stuck
    for 22s! (LP: #1680549)
    - iommu/dma: Stop getting dma_32bit_pfn wrong
    - iommu/dma: Implement PCI allocation optimisation
    - iommu/dma: Convert to address-based allocation
    - iommu/dma: Clean up MSI IOVA allocation
    - iommu/dma: Plumb in the per-CPU IOVA caches
    - iommu/iova: Fix underflow bug in __alloc_and_insert_iova_range

  * Zesty update to 4.10.17 stable release (LP: #1692898)
    - xen: adjust early dom0 p2m handling to xen hypervisor behavior
    - target: Fix compare_and_write_callback handling for non GOOD status
    - target/fileio: Fix zero-length READ and WRITE handling
    - iscsi-target: Set session_fall_back_to_erl0 when forcing reinstatement
    - usb: xhci: bInterval quirk for TI TUSB73x0
    - usb: host: xhci: print correct command ring address
    - USB: serial: ftdi_sio: add device ID for Microsemi/Arrow SF2PLUS Dev Kit
    - USB: Proper handling of Race Condition when two USB class drivers try to
      call init_usb_class simultaneously
    - USB: Revert "cdc-wdm: fix "out-of-sync" due to missing notifications"
    - staging: vt6656: use off stack for in buffer USB transfers.
    - staging: vt6656: use off stack for out buffer USB transfers.
    - staging: gdm724x: gdm_mux: fix use-after-free on module unload
    - staging: wilc1000: Fix problem with wrong vif index
    - staging: comedi: jr3_pci: fix possible null pointer dereference
    - staging: comedi: jr3_pci: cope with jiffies wraparound
    - usb: misc: add missing continue in switch
    - usb: gadget: legacy gadgets are optional
    - usb: Make sure usb/phy/of gets built-in
    - usb: hub: Fix error loop seen after hub communication errors
    - usb: hub: Do not attempt to autosuspend disconnected devices
    - x86/boot: Fix BSS corruption/overwrite bug in early x86 kernel startup
    - selftests/x86/ldt_gdt_32: Work around a glibc sigaction() bug
    - x86, pmem: Fix cache flushing for iovec write < 8 bytes
    - um: Fix PTRACE_POKEUSER on x86_64
    - perf/x86: Fix Broadwell-EP DRAM RAPL events
    - KVM: x86: fix user triggerable warning in kvm_apic_accept_events()
    - KVM: arm/arm64: fix races in kvm_psci_vcpu_on
    - arm64: KVM: Fix decoding of Rt/Rt2 when trapping AArch32 CP accesses
    - block: fix blk_integrity_register to use templ...

[Christos (christosmichaelas)]

Will this fix be implemented in the downloadable daily images? I have downloaded the most recently modified (29/06/2017) and kernel 4.10.0-22 is still in use.

Thanks

[Kai-Heng Feng (kaihengfeng)]

When Artful (17.10) switch to 4.11, the daily image will also get new kernel.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 4.10.0-26.30

---------------
linux (4.10.0-26.30) zesty; urgency=low

  * linux: 4.10.0-26.30 -proposed tracker (LP: #1700528)

  * CVE-2017-1000364
    - Revert "UBUNTU: SAUCE: mm: Only expand stack if guard area is hit"
    - Revert "mm: do not collapse stack gap into THP"
    - Revert "mm: enlarge stack guard gap"
    - mm: larger stack guard gap, between vmas
    - mm: fix new crash in unmapped_area_topdown()
    - Allow stack to grow up to address space limit

linux (4.10.0-25.29) zesty; urgency=low

  * linux: 4.10.0-25.29 -proposed tracker (LP: #1699028)

  * CVE-2017-1000364
    - SAUCE: mm: Only expand stack if guard area is hit

  * CVE-2017-9074
    - ipv6: Prevent overrun when parsing v6 header options
    - ipv6: Check ip6_find_1stfragopt() return value properly.

  * [Zesty] QDF2400 ARM64 server - NMI watchdog: BUG: soft lockup - CPU#8 stuck
    for 22s! (LP: #1680549)
    - iommu/dma: Stop getting dma_32bit_pfn wrong
    - iommu/dma: Implement PCI allocation optimisation
    - iommu/dma: Convert to address-based allocation
    - iommu/dma: Clean up MSI IOVA allocation
    - iommu/dma: Plumb in the per-CPU IOVA caches
    - iommu/iova: Fix underflow bug in __alloc_and_insert_iova_range

  * Zesty update to 4.10.17 stable release (LP: #1692898)
    - xen: adjust early dom0 p2m handling to xen hypervisor behavior
    - target: Fix compare_and_write_callback handling for non GOOD status
    - target/fileio: Fix zero-length READ and WRITE handling
    - iscsi-target: Set session_fall_back_to_erl0 when forcing reinstatement
    - usb: xhci: bInterval quirk for TI TUSB73x0
    - usb: host: xhci: print correct command ring address
    - USB: serial: ftdi_sio: add device ID for Microsemi/Arrow SF2PLUS Dev Kit
    - USB: Proper handling of Race Condition when two USB class drivers try to
      call init_usb_class simultaneously
    - USB: Revert "cdc-wdm: fix "out-of-sync" due to missing notifications"
    - staging: vt6656: use off stack for in buffer USB transfers.
    - staging: vt6656: use off stack for out buffer USB transfers.
    - staging: gdm724x: gdm_mux: fix use-after-free on module unload
    - staging: wilc1000: Fix problem with wrong vif index
    - staging: comedi: jr3_pci: fix possible null pointer dereference
    - staging: comedi: jr3_pci: cope with jiffies wraparound
    - usb: misc: add missing continue in switch
    - usb: gadget: legacy gadgets are optional
    - usb: Make sure usb/phy/of gets built-in
    - usb: hub: Fix error loop seen after hub communication errors
    - usb: hub: Do not attempt to autosuspend disconnected devices
    - x86/boot: Fix BSS corruption/overwrite bug in early x86 kernel startup
    - selftests/x86/ldt_gdt_32: Work around a glibc sigaction() bug
    - x86, pmem: Fix cache flushing for iovec write < 8 bytes
    - um: Fix PTRACE_POKEUSER on x86_64
    - perf/x86: Fix Broadwell-EP DRAM RAPL events
    - KVM: x86: fix user triggerable warning in kvm_apic_accept_events()
    - KVM: arm/arm64: fix races in kvm_psci_vcpu_on
    - arm64: KVM: Fix decoding of Rt/Rt2 when trapping AArch32 CP accesses
    - block: fix blk_integrity_register to use templ...

[Kai-Heng Feng (kaihengfeng)]

Artful should switch to linux kernel 4.11 pretty soon.

[Kai-Heng Feng (kaihengfeng)]

Ok, Artful live image (the pending one) uses 4.10.0-26.30 now. If you still can't boot, then it should be another issue.

[rootexpression (rootexpression-linux-kernel-bugs)]

Using Gigabyte X370 K7 (this issue exists on all gigabyte B350 and X370 boards as far as I know).

I get this issue, and am completely unable to boot in Ubuntu 16.04, and have found reports of the same issue on 17.04. I can however install and boot in Manjaro (arch) just fine. The system runs fine mostly, but has what appears to be a memory leak related to vector 7 and xorg (a conflict possibly?). Issue persists in kernel 4.12 RC7 on arch base.

Found this: https://bugzilla.proxmox.com/show_bug.cgi?id=1366

[rootexpression (rootexpression-linux-kernel-bugs)]

I should note that at Gigabyte, they were able to get Ubuntu 16.04 to boot, and supposedly install on the F4 bios. I could not replicate their results.

[edisonalvaringo (edisonalvaringo-linux-kernel-bugs)]

I get this

irq 7: nobody cared (try booting with the "irqpoll" option)
Jul 15 19:17:29 linux-x5uw kernel: CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.12.0-2.g2399a91-default #1
Jul 15 19:17:29 linux-x5uw kernel: Hardware name: Gigabyte Technology Co., Ltd. AB350-Gaming/AB350-Gaming-CF, BIOS F5 06/19/2017
Jul 15 19:17:29 linux-x5uw kernel: Call Trace:
Jul 15 19:17:29 linux-x5uw kernel: <IRQ>
Jul 15 19:17:29 linux-x5uw kernel: dump_stack+0x5c/0x84
Jul 15 19:17:29 linux-x5uw kernel: __report_bad_irq+0x30/0xc0
Jul 15 19:17:29 linux-x5uw kernel: note_interrupt+0x23e/0x290
Jul 15 19:17:29 linux-x5uw kernel: handle_irq_event_percpu+0x41/0x50
Jul 15 19:17:29 linux-x5uw kernel: handle_irq_event+0x37/0x60
Jul 15 19:17:29 linux-x5uw kernel: handle_fasteoi_irq+0x95/0x160
Jul 15 19:17:29 linux-x5uw kernel: handle_irq+0x19/0x30
Jul 15 19:17:29 linux-x5uw kernel: do_IRQ+0x41/0xc0
Jul 15 19:17:29 linux-x5uw kernel: common_interrupt+0x8c/0x8c
Jul 15 19:17:29 linux-x5uw kernel: </IRQ>
Jul 15 19:17:29 linux-x5uw kernel: RIP: 0010:native_safe_halt+0x2/0x10
Jul 15 19:17:29 linux-x5uw kernel: RSP: 0018:ffffffffa8e03df8 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffc8
Jul 15 19:17:29 linux-x5uw kernel: RAX: 0000000000000000 RBX: ffff9cea5661a000 RCX: 0000000000000034
Jul 15 19:17:29 linux-x5uw kernel: RDX: 4ec4ec4ec4ec4ec5 RSI: ffffffffa8ed9cc0 RDI: ffff9cea5661a064
Jul 15 19:17:29 linux-x5uw kernel: RBP: ffff9cea5661a064 R08: cccccccccccccccd R09: 0000000000000008
Jul 15 19:17:29 linux-x5uw kernel: R10: 000000000000000a R11: 000000000000000a R12: 0000000000000001
Jul 15 19:17:29 linux-x5uw kernel: R13: 0000000000000001 R14: 0000000000000001 R15: 00000004118ee7d0
Jul 15 19:17:29 linux-x5uw kernel: acpi_safe_halt.part.5+0xa/0x20
Jul 15 19:17:29 linux-x5uw kernel: acpi_idle_enter+0xf6/0x290
Jul 15 19:17:29 linux-x5uw kernel: cpuidle_enter_state+0xef/0x2e0
Jul 15 19:17:29 linux-x5uw kernel: do_idle+0x184/0x1e0
Jul 15 19:17:29 linux-x5uw kernel: cpu_startup_entry+0x5d/0x60
Jul 15 19:17:29 linux-x5uw kernel: start_kernel+0x422/0x42a
Jul 15 19:17:29 linux-x5uw kernel: ? early_idt_handler_array+0x120/0x120
Jul 15 19:17:29 linux-x5uw kernel: x86_64_start_kernel+0x12c/0x13b
Jul 15 19:17:29 linux-x5uw kernel: secondary_startup_64+0x9f/0x9f
Jul 15 19:17:29 linux-x5uw kernel: handlers:
Jul 15 19:17:29 linux-x5uw kernel: [<ffffffffc0604030>] amd_gpio_irq_handler [pinctrl_amd]
Jul 15 19:17:29 linux-x5uw kernel: Disabling IRQ #7

[rootexpression (rootexpression-linux-kernel-bugs)]

(In reply to edisonalvaringo from comment #14)
> I get this
>
> irq 7: nobody cared (try booting with the "irqpoll" option)
> Jul 15 19:17:29 linux-x5uw kernel: CPU: 0 PID: 0 Comm: swapper/0 Not tainted
> 4.12.0-2.g2399a91-default #1
> Jul 15 19:17:29 linux-x5uw kernel: Hardware name: Gigabyte Technology Co.,
> Ltd. AB350-Gaming/AB350-Gaming-CF, BIOS F5 06/19/2017
> Jul 15 19:17:29 linux-x5uw kernel: Call Trace:
> Jul 15 19:17:29 linux-x5uw kernel: <IRQ>
> Jul 15 19:17:29 linux-x5uw kernel: dump_stack+0x5c/0x84
> Jul 15 19:17:29 linux-x5uw kernel: __report_bad_irq+0x30/0xc0
> Jul 15 19:17:29 linux-x5uw kernel: note_interrupt+0x23e/0x290
> Jul 15 19:17:29 linux-x5uw kernel: handle_irq_event_percpu+0x41/0x50
> Jul 15 19:17:29 linux-x5uw kernel: handle_irq_event+0x37/0x60
> Jul 15 19:17:29 linux-x5uw kernel: handle_fasteoi_irq+0x95/0x160
> Jul 15 19:17:29 linux-x5uw kernel: handle_irq+0x19/0x30
> Jul 15 19:17:29 linux-x5uw kernel: do_IRQ+0x41/0xc0
> Jul 15 19:17:29 linux-x5uw kernel: common_interrupt+0x8c/0x8c
> Jul 15 19:17:29 linux-x5uw kernel: </IRQ>
> Jul 15 19:17:29 linux-x5uw kernel: RIP: 0010:native_safe_halt+0x2/0x10
> Jul 15 19:17:29 linux-x5uw kernel: RSP: 0018:ffffffffa8e03df8 EFLAGS:
> 00000246 ORIG_RAX: ffffffffffffffc8
> Jul 15 19:17:29 linux-x5uw kernel: RAX: 0000000000000000 RBX:
> ffff9cea5661a000 RCX: 0000000000000034
> Jul 15 19:17:29 linux-x5uw kernel: RDX: 4ec4ec4ec4ec4ec5 RSI:
> ffffffffa8ed9cc0 RDI: ffff9cea5661a064
> Jul 15 19:17:29 linux-x5uw kernel: RBP: ffff9cea5661a064 R08:
> cccccccccccccccd R09: 0000000000000008
> Jul 15 19:17:29 linux-x5uw kernel: R10: 000000000000000a R11:
> 000000000000000a R12: 0000000000000001
> Jul 15 19:17:29 linux-x5uw kernel: R13: 0000000000000001 R14:
> 0000000000000001 R15: 00000004118ee7d0
> Jul 15 19:17:29 linux-x5uw kernel: acpi_safe_halt.part.5+0xa/0x20
> Jul 15 19:17:29 linux-x5uw kernel: acpi_idle_enter+0xf6/0x290
> Jul 15 19:17:29 linux-x5uw kernel: cpuidle_enter_state+0xef/0x2e0
> Jul 15 19:17:29 linux-x5uw kernel: do_idle+0x184/0x1e0
> Jul 15 19:17:29 linux-x5uw kernel: cpu_startup_entry+0x5d/0x60
> Jul 15 19:17:29 linux-x5uw kernel: start_kernel+0x422/0x42a
> Jul 15 19:17:29 linux-x5uw kernel: ? early_idt_handler_array+0x120/0x120
> Jul 15 19:17:29 linux-x5uw kernel: x86_64_start_kernel+0x12c/0x13b
> Jul 15 19:17:29 linux-x5uw kernel: secondary_startup_64+0x9f/0x9f
> Jul 15 19:17:29 linux-x5uw kernel: handlers:
> Jul 15 19:17:29 linux-x5uw kernel: [<ffffffffc0604030>] amd_gpio_irq_handler
> [pinctrl_amd]
> Jul 15 19:17:29 linux-x5uw kernel: Disabling IRQ #7

what kernel is that on? what distro, etc? havn't seen that before.

[edisonalvaringo (edisonalvaringo-linux-kernel-bugs)]

Opensuse 42.2.
linux-4.11.0-2.g057f66f from Opensuse tumbleweed repos.

[edisonalvaringo (edisonalvaringo-linux-kernel-bugs)]

booting with irqpoll it make thing much worse and spams dmesg with "lost hpet rtc xxxx interrupts"

[edisonalvaringo (edisonalvaringo-linux-kernel-bugs)]

I don't know if it is related but I'm having trouble resuming from hibernation sometimes the usb ports will stop working .

I get this message also

[ 877.231395] ------------[ cut here ]------------
[ 877.231395] WARNING: CPU: 0 PID: 4060 at ../arch/x86/kernel/cpu/mcheck/mce_amd.c:191 mce_amd_feature_init+0x27d/0x2c0
[ 877.231396] Modules linked in: ccm dm_crypt dm_mod loop ppdev parport_pc parport vmw_vsock_vmci_transport vsock vmw_vmci nf_log_ipv6 xt_pkttype nf_log_ipv4 nf_log_common xt_LOG xt_limit af_packet ip6t_REJECT nf_reject_ipv6 xt_tcpudp nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_raw ipt_REJECT nf_reject_ipv4 iptable_raw xt_CT iptable_filter ip6table_mangle nf_conntrack_netbios_ns nf_conntrack_broadcast nf_conntrack_ipv4 nf_defrag_ipv4 ip_tables xt_conntrack nf_conntrack libcrc32c ip6table_filter ip6_tables x_tables it87 hwmon_vid joydev fuse snd_hda_codec_realtek snd_hda_codec_hdmi snd_hda_codec_generic snd_hda_intel snd_hda_codec edac_mce_amd snd_hda_core snd_hwdep snd_pcm snd_timer kvm i2c_designware_platform snd irqbypass crct10dif_pclmul crc32_pclmul crc32c_intel ath9k ath9k_common ath9k_hw ath mac80211
[ 877.231405] cfg80211 ghash_clmulni_intel rfkill r8169 mii pcbc soundcore ccp pinctrl_amd shpchp aesni_intel i2c_piix4 aes_x86_64 acpi_cpufreq tpm_infineon tpm_tis tpm_tis_core tpm gpio_amdpt gpio_generic pcspkr crypto_simd glue_helper cryptd button i2c_designware_core wmi arc4 ppp_mppe hid_generic usbhid bcache amdgpu i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm xhci_pci xhci_hcd sr_mod drm cdrom usbcore msr sg ppp_generic slhc efivarfs
[ 877.231412] CPU: 0 PID: 4060 Comm: systemd-sleep Tainted: G W 4.12.0-2.g2399a91-default #1
[ 877.231413] Hardware name: Gigabyte Technology Co., Ltd. AB350-Gaming/AB350-Gaming-CF, BIOS F5 06/19/2017
[ 877.231413] task: ffff995671398000 task.stack: ffffbc5a81a0c000
[ 877.231414] RIP: 0010:mce_amd_feature_init+0x27d/0x2c0
[ 877.231414] RSP: 0018:ffffbc5a81a0fd08 EFLAGS: 00010096
[ 877.231415] RAX: 0000000000000028 RBX: 0000000000000005 RCX: ffffffff98e5abe8
[ 877.231415] RDX: ffffffff98e5abe8 RSI: 0000000000000096 RDI: 0000000000000002
[ 877.231415] RBP: 00000000c000205d R08: 0000000000000000 R09: 0000000000000028
[ 877.231416] R10: ffffbc5a81a0fc80 R11: 0000000000000000 R12: 0000000000000006
[ 877.231416] R13: 0000000000000000 R14: 0000000000000000 R15: ffffbc5a81a0fd24
[ 877.231417] FS: 00007f96b749e740(0000) GS:ffff9956d6600000(0000) knlGS:0000000000000000
[ 877.231417] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 877.231417] CR2: 00007f420b571926 CR3: 0000000213bc5000 CR4: 00000000003406f0
[ 877.231418] Call Trace:
[ 877.231419] mce_syscore_resume+0x1e/0x30
[ 877.231420] syscore_resume+0x4b/0x1a0
[ 877.231421] suspend_devices_and_enter+0x488/0x7e0
[ 877.231422] pm_suspend+0x31a/0x390
[ 877.231423] state_store+0x42/0x90
[ 877.231424] kernfs_fop_write+0xfa/0x180
[ 877.231424] __vfs_write+0x23/0x140
[ 877.231426] ? __getnstimeofday64+0x3b/0xd0
[ 877.231427] ? getnstimeofday64+0xa/0x20
[ 877.231427] ? __audit_syscall_entry+0xab/0x100
[ 877.231428] ? security_file_permission+0x36/0xb0
[ 877.231429] v...

[edisonalvaringo (edisonalvaringo-linux-kernel-bugs)]

Should I open a new bug? the irq problem is similar on pinctrl_amd, if compile a kernel without pinctrl there are no strange messages.

[ldkxingzhe (ldkxingzhe-linux-kernel-bugs)]

I got the same problem. Arch, kernel: 4.13, CPU: Ryzen 7 1700, motherboard: B350 Plus.

[rootexpression (rootexpression-linux-kernel-bugs)]

https://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response

Perhaps this is related to the hardware bug from that first batch of processors.

Just don't want to see you wasting your time fixing something with software that you can't fix with software.

[ldkxingzhe (ldkxingzhe-linux-kernel-bugs)]

maybe. but my win10 is fine.

[rootexpression (rootexpression-linux-kernel-bugs)]

(In reply to ldk from comment #22)
> maybe. but my win10 is fine.

Uh huh. Please read up on the hardware bug.

"AMD engineers found the problem to be very complex and characterize it as a performance marginality problem exclusive to certain workloads on Linux."

[LCID Fire (lcid-fire)]

I encountered a similar or same problem on Gigabyte AB350 with Ubuntu 16.04 running `4.13.0-21-generic #24~16.04.1-Ubuntu SMP Mon Dec 18 19:39:31 UTC 2017`. It even shows up on 18.04 Alpha.
Only thing that makes it halfway work is setting `acpi=off`.

Isn't this supposed to be fixed by now?

[Kai-Heng Feng (kaihengfeng)]

> On 25 Dec 2017, at 4:44 AM, LCID Fire <email address hidden> wrote:
>
> I encountered a similar or same problem on Gigabyte AB350 with Ubuntu 16.04 running `4.13.0-21-generic #24~16.04.1-Ubuntu SMP Mon Dec 18 19:39:31 UTC 2017`. It even shows up on 18.04 Alpha.
> Only thing that makes it halfway work is setting `acpi=off`.
>
> Isn't this supposed to be fixed by now?

Please file a new bug with log attached.

>
> --
> You received this bug notification because you are a member of AMD Team,
> which is subscribed to the bug report.
> https://bugs.launchpad.net/bugs/1671360
>
> Title:
> System doesn't boot properly on Gigabyte AM4 motherboards (AMD Ryzen)
>
> Status in Linux:
> Unknown
> Status in linux package in Ubuntu:
> Fix Released
> Status in linux source package in Zesty:
> Fix Released
> Status in linux source package in Artful:
> Fix Released
>
> Bug description:
> [Impact]
> Gigabyte AM4 boards users cannot boot Ubuntu successfully.
> Commit linux-gpio/fixes babdc22b0ccf4ef5a3075ce6e4afc26b7a279faf "pinctrl/amd: Use regular interrupt instead of chained" can fix the issue.
>
> [Test Case]
> All Gigabyte AM4 boards can reproduce the issue.
> With the patch, the issue is resolved, per comment #170.
>
> [Regression Potential]
> Regression Potential is low. It limits to rather new AMD platform which has pinctrl-amd.
> As the commit log says, use chained interrupt is not a good idea. Use regular interrupt is the correct way.
>
> I also test the patch on an AMD laptop, where its touchpad depends on
> pinctrl-amd. No regression found.
>
> Original bug report:
> I'm trying to run ubuntu on Ryzen 1700x with Gigabyte GA-AB350-gaming-3 motherboard,
> and it has a load of problems, starting with not being able to boot normally.
>
> During normal boot, on 16.10 as well as 17.04 beta:
> system doesn't boot normally, hangs with a lot of "unexpected irq trap at vector 07"
> messages displayed.
>
> Following advice from various places, I've tried:disable cpu freq
> governor and cpu handling in acpi settings
>
> 1. add "acpi=off" to boot params
>
> That helps, allowing me to boot into recovery mode, though it leaves
> me with system seeing only one core, is really slow and still only
> boots in recovery mode.
>
> 2. Compile own kernel using 4.11.rc1 and disabling cpu freq governor
> and cpu handling in acpi settings. Boot with "quiet loglevel=3"
> option.
>
> That gets me even further - system sees all cores now. Still only recovery mode though,
> but its enough to get info for this bug report.
>
> Some observed problems:
>
> 1. dmesg reports *a lot* of messages like this all the time:
>
> [ 163.362068] ->handle_irq(): ffffffff87a7e090,
> [ 163.362081] bad_chained_irq+0x0/0x40
> [ 163.362089] ->handle_irq(): ffffffff87a7e090,
> [ 163.362090] amd_gpio_irq_handler+0x0/0x200
> [ 163.362090] ->irq_data.chip(): ffffffff88587e20,
> [ 163.362090] ioapic_ir_chip+0x0/0x120
> [ 163.362090] ->action(): ffffffff884601c0
> [ 163.362091] IRQ_NOPROBE set
> [ 163.362099] ->handle_irq(): ffffffff87a7e090,
> [ 163.362099] amd_gpio_irq_handler+0x0/0x200
> [ 163.362100] ->irq_data....

[jpujades (jpujades-linux-kernel-bugs)]

Similar problem with Acer TravelMate B117-M (SSD disk).
Upgraded the BIOS system from 1.07 to 1.13 and 1.15 (the latest)
Lubuntu 16.04 LTS HWE 64 bit

System works fine with 4.10.0-42
System shows the messages at start-up when using 4.13.0-26 or 4.13.0-31 kernels.

Curiously, if I turn off the system, connect/disconnect the battery (battery reset hole at the bottom of the laptop) and turn on the system, NO errors.

At next boot, errors reappear.

However, the system seems to work without problems, despite the start-up IRQ errors. I have to test more. I don't know if I will have performance problems or other kind of troubles.

[jpujades (jpujades-linux-kernel-bugs)]

It seems to be related with https://bugzilla.kernel.org/show_bug.cgi?id=194945#c73

I can see also bad interrupts for chv-gpio

No more IRQ errors with kernel 4.14.15-041415-generic

http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.14.15/

[jpujades (jpujades-linux-kernel-bugs)]

"System shows the messages at start-up when using 4.13.0-26 or 4.13.0-31 kernels." I'm sorry, the messages were:

unexpected irq trap at vector 73

Tested on HP Convertible x360 11-ab0XX with same results. No more messages about IRQ errors with kernel 4.14.15-041415-generic

[rootexpression (rootexpression-linux-kernel-bugs)]

(In reply to Josep Pujadas-Jubany from comment #24)
> Similar problem with Acer TravelMate B117-M (SSD disk).
> Upgraded the BIOS system from 1.07 to 1.13 and 1.15 (the latest)
> Lubuntu 16.04 LTS HWE 64 bit
>
> System works fine with 4.10.0-42
> System shows the messages at start-up when using 4.13.0-26 or 4.13.0-31
> kernels.
>
> Curiously, if I turn off the system, connect/disconnect the battery (battery
> reset hole at the bottom of the laptop) and turn on the system, NO errors.
>
> At next boot, errors reappear.
>
> However, the system seems to work without problems, despite the start-up IRQ
> errors. I have to test more. I don't know if I will have performance
> problems or other kind of troubles.

This bug report here is for AMD Ryzen cpus. You'll need to find or make a bug report for your laptop. The bug discussed here is specific to only AMD Ryzen CPUs.