sytem reboots when initializing Edirol FA-101 firewire audio interface

Thu Feb 20 02:57:27 CET 2020

Hi,

On Wed, Feb 19, 2020 at 06:19:56AM -0800, Steve Morris wrote:
> I'm running Arch Linux x86_64
> 
> My system consistently reboots when I power on my FA-101 when running
> kernels 5.5.1-4. Downgrading to 5.4.15 allows everything to work
> properly.
> 
> Here's the outpu of:
> journalctl  | grep -E 'Reboot|firewire|fw|bebob|alsa|jack'
> 
> This is a good init under 5.4.15:
> -- Reboot --
> powered on interface while running 5.4.15, initialized properly
> Feb 18 18:35:37 hostname kernel: firewire_ohci 0000:05:00.0: enabling device (0080 -> 0083)
> Feb 18 18:35:37 hostname kernel: firewire_ohci 0000:05:00.0: added OHCI v1.10 device as card 0, 4 IR + 8 IT contexts, quirks 0x11
> Feb 18 18:35:37 hostname kernel: firewire_core 0000:05:00.0: created device fw0: GUID 0011223333666677, S400
> Feb 18 18:35:37 hostname systemd-udevd[541]: controlC0: Process '/usr/bin/alsactl restore 0' failed with exit code 99.
> Feb 18 18:35:37 hostname systemd-udevd[541]: controlC1: Process '/usr/bin/alsactl restore 1' failed with exit code 99.
> Feb 18 18:35:38 hostname kernel: amdgpu: [powerplay] smu driver if version = 0x00000033, smu fw if version = 0x00000035, smu fw version = 0x002a3200 (42.50.0)
> Feb 18 18:36:08 hostname kernel: firewire_ohci 0000:05:00.0: isochronous cycle inconsistent
> Feb 18 18:36:08 hostname kernel: firewire_core 0000:05:00.0: created device fw1: GUID 0040ab0000c20bc1, S400
> Feb 18 18:36:08 hostname kernel: firewire_core 0000:05:00.0: phy config: new root=ffc1, gap_count=5
> Feb 18 18:36:10 hostname kernel: firewire_core 0000:05:00.0: BM lock failed (timeout), making local node (ffc0) root
> Feb 18 18:36:10 hostname kernel: firewire_core 0000:05:00.0: phy config: new root=ffc0, gap_count=5
> Feb 18 18:36:12 hostname systemd-udevd[1532]: controlC2: Process '/usr/bin/alsactl restore 2' failed with exit code 99.
> Feb 18 18:36:15 hostname kernel: snd-bebob fw1.0: transaction failed: no ack
> Feb 18 18:36:15 hostname kernel: snd-bebob fw1.0: fail to get an input for MSU in plug 7: -5
> Feb 18 18:36:15 hostname kernel: snd-bebob fw1.0: transaction failed: no ack
> Feb 18 18:36:15 hostname kernel: snd-bebob fw1.0: fail to get an input for MSU in plug 7: -5
> 
> This is a bad init under 5.5.4:
> -- Reboot --
> Powered on interface when machine was running.
> Feb 18 18:13:37 hostname kernel: firewire_ohci 0000:05:00.0: enabling device (0080 -> 0083)
> Feb 18 18:13:37 hostname kernel: firewire_ohci 0000:05:00.0: added OHCI v1.10 device as card 0, 4 IR + 8 IT contexts, quirks 0x11
> Feb 18 18:13:37 hostname kernel: firewire_core 0000:05:00.0: created device fw0: GUID 0011223333666677, S400
> Feb 18 18:13:37 hostname kernel: audit: type=1130 audit(1582078417.360:3): pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=ufw comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
> Feb 18 18:13:37 hostname audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=ufw comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
> Feb 18 18:13:37 hostname systemd-udevd[539]: controlC0: Process '/usr/bin/alsactl restore 0' failed with exit code 99.
> Feb 18 18:13:37 hostname systemd-udevd[530]: controlC1: Process '/usr/bin/alsactl restore 1' failed with exit code 99.
> Feb 18 18:13:38 hostname kernel: amdgpu: [powerplay] smu driver if version = 0x00000033, smu fw if version = 0x00000035, smu fw version = 0x002a3200 (42.50.0)
> Feb 18 18:19:45 hostname kernel: firewire_core 0000:05:00.0: phy config: new root=ffc1, gap_count=5
> Feb 18 18:19:48 hostname kernel: firewire_core 0000:05:00.0: phy config: new root=ffc1, gap_count=5
> Feb 18 18:19:48 hostname kernel: firewire_core 0000:05:00.0: created device fw1: GUID 0040ab0000c20bc1, S400
> Feb 18 18:19:52 hostname systemd-udevd[1682]: controlC2: Process '/usr/bin/alsactl restore 2' failed with exit code 99.
> Feb 18 18:20:08 hostname kernel: snd-bebob fw1.0: transaction failed: no ack
> Feb 18 18:20:08 hostname kernel: snd-bebob fw1.0: fail to get an input for MSU in plug 7: -5
> Feb 18 18:20:08 hostname kernel: snd-bebob fw1.0: transaction failed: no ack
> Feb 18 18:20:08 hostname kernel: snd-bebob fw1.0: fail to get an input for MSU in plug 7: -5
> Feb 18 18:20:12 hostname kernel: firewire_core 0000:05:00.0: phy config: new root=ffc1, gap_count=5
> Feb 18 18:21:24 hostname kernel: firewire_ohci 0000:05:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0001 address=0xd5080000 flags=0x0000]
> Followed by 125 more O_PAGE_FAULT and then the reboot,
> 
> I'll be happy to provide additional information if needed.

Hm. I think your system rushed into too complicated issue, At least, three
software stacks are related to your issue:
 - ALSA bebob driver(snd-bebob) and helper module(snd-firewire-lib)
 - Linux firewire core(firewire-core) and 1394 OHCI controller driver (firewire-ohci)
 - Driver stack for AMD IOMMU

>From the log, ALSA bebob driver takes your unit to generate bus reset on
IEEE 1394 bus. We can see 1394 OHCI controller handles the bus reset and
the corresponding driver takes Linux firewire core to process the event.
Your unit's generation of bus reset is responsible for ALSA bebob driver,
and this situation is quite similar to what I reported in this patch[1].

However, the system reboot itself is not responsible for ALSA bebob
driver since it is just an application of kernel API of Linux firewire
application and sound subsystem (ALSA). I don't know exactly yet but
something happened between 1394 OHCI controller driver and AMD IOMMU.
It's probably that the cause is in more platform-specific domain.

I'm sorry not to help you. For developers information, I C.C.ed this
issue to iommu at lists.linux-foundation.org and
linux1394-devel at lists.sourceforge.net. If you can get some responses
from the others, please follow their instruction. But in advance it's
better for you to open information about your kernel configuration
and hardware, especially for the version of AMD AGESA for your CPU,
like:

 * AMD Ryzen 5 2400G with Radeon Vega Graphics (family: 0x17, model: 0x11, stepping: 0x0)
 * Gigabyte Technology Co., Ltd. AX370-Gaming 5/AX370-Gaming 5
 * BIOS F42b 08/01/2019 (=AGESA 1.0.0.3 ABB, perhaps)
 * kernel configuration can be retrieved from Ubuntu package repository[2]

The above is a sample from my development environment. I think it would
be possible to regenerate the issue if you give your kernel
configuration, however I'm on vacation till the beginning of March. So
I cannot practice it now, sorry.

[1] d3eabe939aee ("ALSA: bebob: expand sleep just after breaking connections for protocol version 1")
https://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git/commit/sound/firewire/bebob?id=d3eabe939aee
[2] https://packages.ubuntu.com/eoan/kernel/linux-image-5.3.0-40-generic

Regards

Takashi Sakamoto