linux-kernel - Re: [RFC PATCH] s390: vfio-ap: Register the vfio

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20211216122552.0ee1b998.pasic@linux.ibm.com>
Date:   Thu, 16 Dec 2021 12:25:52 +0100
From:   Halil Pasic <pasic@...ux.ibm.com>
To:     Cornelia Huck <cohuck@...hat.com>
Cc:     Thomas Huth <thuth@...hat.com>,
        Tony Krowiak <akrowiak@...ux.ibm.com>,
        Harald Freudenberger <freude@...ux.ibm.com>,
        linux-s390@...r.kernel.org, Jason Herne <jjherne@...ux.ibm.com>,
        linux-kernel@...r.kernel.org, Heiko Carstens <hca@...ux.ibm.com>,
        Vasily Gorbik <gor@...ux.ibm.com>,
        Christian Borntraeger <borntraeger@...ibm.com>,
        Halil Pasic <pasic@...ux.ibm.com>
Subject: Re: [RFC PATCH] s390: vfio-ap: Register the vfio_ap module for the
 "ap" parent bus

On Thu, 16 Dec 2021 11:39:13 +0100
Cornelia Huck <cohuck@...hat.com> wrote:

> > Also what does vfio-pci do? As far as I can tell vfio-pci does not
> > participate in module auto loading just because there are pci devices.
> > The have some smart override I don't quite understand:
> > https://patchwork.ozlabs.org/project/linux-pci/patch/20210826103912.128972-11-yishaih@nvidia.com/
> > Before, I don't think they had a MODULE_DEVICE_TABLE:
> > https://elixir.bootlin.com/linux/v5.8.18/source/drivers/vfio/pci/vfio_pci.c  
> 
> I don't think it makes sense to look at pci for inspiration here; they
> have a myriad of different device types, while ap only has a few (and
> probably not that many different ones on a given system), and css only
> has one that really matters.

I really don't understand how they having a myraid of different device
types is relevant here. Are they not taking care of that by using
PCI_ANY_ID anyway? Can you please explain?

Please have another look at the commit message of 
cc6711b0bf36 ("PCI / VFIO: Add 'override_only' support for VFIO PCI sub system").
They generate alias(es) that catch every PCI device (via PCI_ANY_ID), but
in a way that precludes auto-loading and enables a generic 'override'
algorithm.

Regards,
Halil