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: <15a39981-37ca-ed2d-3baf-c4cb74f8c343@linux.ibm.com>
Date:   Thu, 2 Dec 2021 09:33:43 +0100
From:   Harald Freudenberger <freude@...ux.ibm.com>
To:     Thomas Huth <thuth@...hat.com>, linux-s390@...r.kernel.org,
        Tony Krowiak <akrowiak@...ux.ibm.com>,
        Halil Pasic <pasic@...ux.ibm.com>,
        Jason Herne <jjherne@...ux.ibm.com>
Cc:     linux-kernel@...r.kernel.org, Heiko Carstens <hca@...ux.ibm.com>,
        Vasily Gorbik <gor@...ux.ibm.com>
Subject: Re: [RFC PATCH] s390: vfio-ap: Register the vfio_ap module for the
 "ap" parent bus

On 02.12.21 08:13, Thomas Huth wrote:
> On 01/12/2021 18.10, Harald Freudenberger wrote:
>> On 01.12.21 15:11, Thomas Huth wrote:
>>> The crypto devices that we can use with the vfio_ap module are sitting
>>> on the "ap" bus, not on the "vfio_ap" bus that the module defines
>>> itself. With this change, the vfio_ap module now gets automatically
>>> loaded if a supported crypto adapter is available in the host.
>>>
>>> Signed-off-by: Thomas Huth <thuth@...hat.com>
>>> ---
>>>   Note: Marked as "RFC" since I'm not 100% sure about it ...
>>>         please review carefully!
>>>
>>>   drivers/s390/crypto/vfio_ap_drv.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/s390/crypto/vfio_ap_drv.c b/drivers/s390/crypto/vfio_ap_drv.c
>>> index 4d2556bc7fe5..5580e40608a4 100644
>>> --- a/drivers/s390/crypto/vfio_ap_drv.c
>>> +++ b/drivers/s390/crypto/vfio_ap_drv.c
>>> @@ -39,7 +39,7 @@ static struct ap_device_id ap_queue_ids[] = {
>>>       { /* end of sibling */ },
>>>   };
>>>   -MODULE_DEVICE_TABLE(vfio_ap, ap_queue_ids);
>>> +MODULE_DEVICE_TABLE(ap, ap_queue_ids);
>>>     /**
>>>    * vfio_ap_queue_dev_probe:
>> Hello Thomas, interesting.
>> Did you test this ? I mean did you build a kernel and have it run on a s390 with crypto cards available ?
>
> Yes, I've tested it. Without the patch, the vfio_ap module does not get loaded automatically if a crypto card is available. With the patch applied, the vfio_ap module correctly gets loaded automatically on my system (similar to the vfio_ccw module).
>
>> My strong feeling is that this will make the AP bus code stumble as the code silently assumes there are exact
>> two types of ap devices attached to the ap bus: ap cards and ap queues.
>
> This is only about getting the module loaded automatically once such a device is available ... AFAIK it does not grab any of the devices automatically, so there shouldn't be any problems?
>
>  Thomas
>
Yes, of course for the automatic module load works this way. But you understand that now
the vfio devices are childs of the ap bus and thus are siblings of the ap queue and ap card
devices. As I wrote the ap bus code is not prepared to deal with a 3th type of devices
dangling on the ap bus. So you should test what happens when there are real vfio ap devices
in use together with 'regular' ap card and queue devices.

However, I am still not sure if it is preferable to have the vfio ap module loaded automatically. The majority
of customers will never use vfio ap devices - this is specific to kvm hosts only. I think this has to be
decided by Tony and maybe some kvm architect. If there is an agreement, I will try to rework the
ap code to be able to deal with foreign devices attached to the ap bus.

So thanks for your investigations. Let's wait for Tony and see how we proceed.

Harald
Harald