[PATCH RFC] dwc2: Don't assume URB transfer_buffer are dword-aligned
Mauro Carvalho Chehab
mchehab at s-opensource.com
Wed Mar 29 11:49:09 PDT 2017
Em Wed, 29 Mar 2017 11:57:22 +0200
Greg Kroah-Hartman <gregkh at linuxfoundation.org> escreveu:
> On Tue, Mar 28, 2017 at 06:48:02AM -0300, Mauro Carvalho Chehab wrote:
> > Em Fri, 17 Mar 2017 10:24:15 +0900
> > Greg Kroah-Hartman <gregkh at linuxfoundation.org> escreveu:
> > > On Thu, Mar 16, 2017 at 09:08:40PM -0300, Mauro Carvalho Chehab wrote:
> > > > The dwc2 hardware doesn't like to do DMA transfers without
> > > > aligning data in DWORD. The driver also assumes that, even
> > > > when there's no DMA, at dwc2_read_packet().
> > > >
> > > > That cause buffer overflows, preventing some drivers to work.
> > >
> > > Why aren't the drivers being fixed? This is a well-known (hopefully)
> > > restriction on USB data buffers, can't the uvc_driver be fixed?
> > I talked to Laurent about on IRC. He told that he is willing to consider
> > that option, if the USB API explicitly states that all buffers must be
> > dword-aligned (and/or buffer sizes).
> > IMHO, he has a point: if this is a restriction of for usb transfer
> > buffers, it should be documented somewhere.
> > Yet, a quick check with:
> > $ git grep -i dword Documentation/usb/
> > $ git grep -i align Documentation/usb/
> > Didn't hit anything related to it.
> Hm, most of the USB documentation is in the kerneldoc in the USB code
> itself, and is built that way. I'm pretty sure this is documented
> somewhere, but I can't seem to find it at the moment either :(
> Care to write a patch for that? :)
Sure. Btw, I noticed that not all USB documents were converted
yet to ReST, so I took the time to convert the core documents to ReST
I kept the driver-specific documentation at Documentation/usb.
The final result is at:
I'll be sending the documentation patches in a few.
More information about the linux-rpi-kernel