Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:8 as permitted sender) client-ip=2620:137:e000::3:8;
Feedback-ID: ie8e14432:Fastmail
Date:   Tue, 17 Oct 2023 00:56:57 +0900
From:   Takashi Sakamoto <o-takashi@sakamocchi.jp>
To:     Jiri Slaby <jirislaby@kernel.org>
Cc:     linux-kernel@vger.kernel.org
Subject: Re: [GIT PULL] firewire updates for 6.5-rc1
Message-ID: <20231016155657.GA7904@workstation.local>
Mail-Followup-To: Takashi Sakamoto <o-takashi@sakamocchi.jp>,
        Jiri Slaby <jirislaby@kernel.org>, linux-kernel@vger.kernel.org
References: <20230704121847.GA26576@workstation.local>
 <f3d8c72b-cd30-490c-833f-5807bd74511f@kernel.org>
 <20230926140922.GA6538@workstation.local>
 <a12593c9-9ced-4bb8-b7a5-8247b08d0e0f@kernel.org>
 <0ed4012a-83a7-4849-92c4-87a86e1bbb84@app.fastmail.com>
 <9e335ce7-a4ac-45a0-afdf-9afca8a300d5@kernel.org>
 <20231004002407.GA48535@workstation.local>
 <13363c8e-22ec-48de-a471-91535ad1249f@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <13363c8e-22ec-48de-a471-91535ad1249f@kernel.org>
Precedence: bulk

Hi Jiri,

I purchased 1394 OHCI hardware quite similar to the issued hardware
(VIA VT6307 + Asmedia ASM1083), then experienced that it causes the
issue in AMD chipset, but it does not in Intel chipset.

I use below machines:

* AMD Ryzen 5 2400G in Gigabyte AX370-Gaming 5 rev.1.0 (Bios version F51h) [1]
* Intel Core i3-2120T in Intel DH77DF (Bios version 0112) [2]

As long as seeing filed issues, the reporters use AMD chipset.

AMD Ryzen 5 3400G, MSI B450-A Pro: https://bugzilla.suse.com/show_bug.cgi?id=1215436
AMD Ryzen 7 3700X, MSI X570-A Pro: https://bugzilla.redhat.com/show_bug.cgi?id=2240973

Furthermore, I experienced that:

* guest machine can cause host reboot in Ryzen machine even if the 1394
  OHCI hardware is passed-through to the guest by vfio-pci (and
  firewire-ohci is loaded in the guest machine side)
* execution of lspci(8) in the host can cause the reboot when the hardware
  is associated to vfio-pci  (thus firewire-ohci is not loaded in the case).
  This issue is in Ubuntu 23.04 (v6.2 kernel).

At present, I suspect either or both iommu function and kernel
implementation.

For further investigation, I would like you to give some instructions
about the way to use OpenSUSE distribution. I want to boot the distribution
in 'low graphics mode' which debian/ubuntu distribution has. In the mode,
I expect to boot it without loading amdgpu module which depends on iommuv2
module.

Anyway, I would like you to check reporters' chipset when the similar
issues are reported. If they encounter the issue still in Intel chipset,
please inform it to me, since it could deny my current presumption.


[1] https://www.gigabyte.com/Motherboard/GA-AX370-Gaming-5-rev-10#kf
[2] https://www.intel.com/content/www/us/en/products/sku/59503/intel-desktop-board-dh77df/specifications.html

Thanks

Takashi Sakamoto