linux-kernel - Re: MFD: core assumes that all children are platform devices

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]

Date:	Tue, 29 Nov 2011 10:57:46 -0800
From:	Guenter Roeck <guenter.roeck@...csson.com>
To:	Jean-François Dagenais <dagenaisj@...atest.com>
CC:	Samuel Ortiz <sameo@...ux.intel.com>,
	open list <linux-kernel@...r.kernel.org>
Subject: Re: MFD: core assumes that all children are platform devices

On Mon, Nov 28, 2011 at 11:23:00AM -0500, Jean-François Dagenais wrote:
> Hi,
> 
> I have a pci driver that registers with UIO for it's operations. As a consequence, the pci device
> instance has a child device of class uio.
> 
> My driver also declares a ds1wm instance that has it's register interface at an offset in BAR0
> of the pci device, as an MFD cell.
> 
> When I call mfd_remove_devices, MFD proceeds to enumerate ALL the parent device's chilren
> and assumes that they are MFD cells, and thus platform_device, which is not true in my case.
> (...uio is a child of the parent pci device)
> 
> I had (luckily or unluckily) not seen signs of this broken assumption on certain setups I have
> used, but in my current setup, this page-faults every time now.
> 
> This is a major thing and I have not found the assumption documented anywhere.
> 
> I could first declare a new child device on my pci device and then declare it as the parent to
> the mfd cells...
> 
I had the same problem, with a Multi-function USB device. Took me a while to figure out that
mfd_remove_devices() removed the USB child devices when I used the USB device as MFD parent device.

My solution was to do what you suggested - my MFD probe function now creates a platform device
to be used as MFD parent device. Works nicely, it doesn't require much additional code,
and I think it is cleaner than other possible solutions (at least the ones I came up with).
We'll see how it flies with the MFD and USB maintainers once I submit the patch ;).

> Or, is there a way for the mfd-core, as it's doing the "for each child device", to recognize
> non-MFD-cell children and skip them?
> 
Looking at your proposed patch, I personally prefer my solution. Of course it would be nice
if it was documented that MFD parent devices MUST be dedicated (platform) devices and must not
have any non-MFD child devices. This would be a simple documentation patch and avoid making
assumptions on MFD child device removal.

Guenter 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/