Received-SPF: pass (google.com: domain of linux-kernel+bounces-123612-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Date: Thu, 28 Mar 2024 14:42:01 -0600
From: Alex Williamson <alex.williamson@redhat.com>
To: Bjorn Helgaas <helgaas@kernel.org>
Cc: linux1394-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org,
 linux-pci@vger.kernel.org, edmund.raile@proton.me, Takashi Sakamoto
 <o-takashi@sakamocchi.jp>
Subject: Re: [PATCH v2] PCI: Mark LSI FW643 to avoid bus reset
Message-ID: <20240328144201.510f6d5e.alex.williamson@redhat.com>
In-Reply-To: <20240327150119.GA1502858@bhelgaas>
References: <20240326131858.GA140624@workstation.local>
	<20240327150119.GA1502858@bhelgaas>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Wed, 27 Mar 2024 10:01:19 -0500
Bjorn Helgaas <helgaas@kernel.org> wrote:

> On Tue, Mar 26, 2024 at 10:18:58PM +0900, Takashi Sakamoto wrote:
> > On Mon, Mar 25, 2024 at 09:41:49AM -0500, Bjorn Helgaas wrote:  
> > > So even without this patch, you are able to pass the FW643 to a VM
> > > with VFIO, and you don't see any issues caused by VFIO resetting the
> > > device?  
> >  
> > Absolutely yes, at least in my VM, for recent years to maintain Linux
> > FireWire subsystem and ALSA firewire stack.  
> 
> So there must be something different between your system and Edmund's.
> Maybe we can refine the quirk so it avoids the SBR on Edmund's system
> but not yours.
> 
> Can you both collect the output of "sudo lspci -vvv" so we can try to
> figure out the difference?  Also a complete dmesg log would be helpful
> and would contain DMI information that we might need if this is
> firmware dependent.

The original patch proposed for this gave me the impression that this
was a device used on various old Mac systems, not likely applicable to
a general purpose plug-in card.  Given the expanded use case, I'd
suggest reverting the patch.

I think we need significantly more exhaustive testing on the afflicted
system to understand whether this is an issue with the endpoint, the
root port, the BIOS, etc.

In the meantime, or maybe as a permanent solution, Edmund can make use
of the reset_method interface in pci-syfs to restrict the available
reset methods for the device rather than risk removing a reset
mechanism identified as working by other users.  My 2 cents.  Thanks,

Alex