linux-kernel - Re: [PATCH V3 1/3] vfio, platform: add support for ACPI while detecting the reset driver

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] [day] [month] [year] [list]

Date:	Sun, 1 May 2016 13:55:43 -0400
From:	Sinan Kaya <okaya@...eaurora.org>
To:	Arnd Bergmann <arnd@...db.de>
Cc:	linux-arm-kernel@...ts.infradead.org, kvm@...r.kernel.org,
	timur@...eaurora.org, cov@...eaurora.org, jcm@...hat.com,
	eric.auger@...aro.org, mark.rutland@....com,
	devicetree@...r.kernel.org,
	Baptiste Reynal <b.reynal@...tualopensystems.com>,
	vikrams@...eaurora.org, marc.zyngier@....com,
	linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
	vinod.koul@...el.com, Alex Williamson <alex.williamson@...hat.com>,
	agross@...eaurora.org, Dan Carpenter <dan.carpenter@...cle.com>,
	shankerd@...eaurora.org
Subject: Re: [PATCH V3 1/3] vfio, platform: add support for ACPI while
 detecting the reset driver

On 3/29/2016 8:44 AM, Arnd Bergmann wrote:
>>  can certainly drop patch #3 and push the reset responsibility to acpi.
>> > 
>> > I never liked having a fragmented sw design across multiple drivers.
>> > 
>> > I need something for patch #1. Compatible is a DT property not ACPI.but 
>> > then, I won't have a reset driver anymore.
>> > 
>> > If we think about how vfio pci works, we pass the pci vendor and device 
>> > id to new_id file to find out which pci device needs to be pass thru.
>> > 
>> > I can go to a similar route. This time we pass the object id through 
>> > new_id and I call reset method on this object.
>> > 
>> > Let me know what you think?
> It would certainly be nice to make it work more like PCI VFIO does
> here, where you can assign any device as long as it has an IOMMU
> (and a _rst method in this case).
> 
> 	Arnd

I looked at the code today. This doesn't have to be as convoluted as PCI is. 
The VFIO platform driver already has a pointer to the actual object in
vdev->pdev.dev.

This is really as simple as calling _RST method on the passed oject at the last
step below. We don't need to create a dynamic list like PCI does. 

This is how the driver gets called. 

echo vfio-platform | tee -a /sys/bus/platform/devices/QCOM8061:00/driver_override
echo QCOM8061:00 | tee -a  /sys/bus/platform/devices/QCOM8061:00/driver/unbind
echo QCOM8061:00 |tee -a /sys/bus/platform/drivers_probe

I'll post a patch as soon as I test it.

-- 
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project