[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <56FBDA0D.4030507@ti.com>
Date: Wed, 30 Mar 2016 16:52:13 +0300
From: Grygorii Strashko <grygorii.strashko@...com>
To: Felipe Balbi <felipe.balbi@...ux.intel.com>,
"Thang Q. Nguyen" <tqnguyen@....com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
<linux-usb@...r.kernel.org>, <linux-omap@...r.kernel.org>,
<linux-kernel@...r.kernel.org>, <linux-arm@...ts.infradead.org>,
Arnd Bergmann <arnd@...db.de>,
"Karicheri, Muralidharan" <m-karicheri2@...com>,
Peter Ujfalusi <peter.ujfalusi@...com>
CC: Phong Vo <pvo@....com>, Loc Ho <lho@....com>, <patches@....com>,
Santosh Shilimkar <ssantosh@...nel.org>,
"Ben Dooks (embedded platforms)" <ben-linux@...ff.org>
Subject: Re: [PATCH v3 2/2] usb:dwc3: pass arch data to xhci-hcd child
On 03/30/2016 04:10 PM, Felipe Balbi wrote:
> "Thang Q. Nguyen" <tqnguyen@....com> writes:
>
>> [ text/plain ]
>> From: "Thang Q. Nguyen" <tqnguyen@....com>
>>
>> The xhci-hcd child node needs to inherit archdata attribute to use
>> dma_ops functions and attributes. This patch enables the USB DWC3
>> driver to pass archdata attributes to its xhci-hcd child node.
>>
>> Changes from v2:
>> - None
>>
>> Changes from v1:
>> - None
>
> changes should be between tearline and diffstat.
>
uh. This become a real problem :(, especially with LPAE enabled.
DMA properties need to be inherited not only here, but also in
usb_add_gadget_udc_release(). And probably in other places
where devices are created manually - the worst case : device is created
manually but doesn't belong to any bus.
And DMA configuration must include dma_pfn_offset also!
And how about iommu staff?
FYI. Solution used for PCI
c49b8fc of/pci: Add of_pci_dma_configure() to update DMA configuration
Rejected: introduce dma_init_dev_from_parent() or smth. like this
http://permalink.gmane.org/gmane.linux.ports.arm.kernel/378317
https://lkml.org/lkml/2014/11/4/519
--
regards,
-grygorii
Powered by blists - more mailing lists