This is the mail archive of the ecos-patches@sourceware.org mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [patch] Philips D12 USB Slave Driver


Andrew Lunn wrote:

Hi Frank

On Tue, Apr 11, 2006 at 09:35:28AM -0400, Frank Pagliughi wrote:


This is a driver for the Philips PDIUSBD12 (D12) USB Slave Controller Chip. It was written for the D12 on a PC/104 (ISA) board. With some minor adjustments and testing should go with any CPU. I (tried to) removed all the board-specific stuff, which, unfortunately also meant removing DMA support.



I just had a very quick look at the code. I does seem to have a few
x386ism's in it but they are all nicely abstracted. Do you think you
can split this into two packages. A generic D12 driver and an i386
specific part. It would contain the functions to actually access the
hardware. You could put these into an inline header file so there
would be no loss of performance.


Yeah, I thought of trying something like that at the last minute, especially to get the DMA code back in there. But I'm leaving town for a few weeks, so I thought I would throw it at you to get some suggestions. I can have a look when I get back. Can you point me to any example that does this sort of thing? The PC serial ports, maybe?

I noticed there is also an extensive test program. What is required on
the host to run this test?



Oops. I messed up the patch. Didn't clear out my directory very well before creating it. The "target.c" is actually an emacs backup file of the existing "usbtarget.c" that's in the USB tests directory (packages/io/usb/slave/current/tests). Note the "~" in the file name. Sorry about that. Ignore that file.

I will submit a few minor changes to that file, though, in the near future. Probably as configuration options. There are a few things hard-wired into the test program that need to be configured for the specific chip - such as the size of control endpoint zero.

In the patch, there is, however a valid, minor, update to the host side of the existing test program that runs on Linux, "usbhost.c" The host didn't recognize the target when run on Fedora Core 4. The fix should allow it to run on FC4, plus whatever system it was originnaly designed for.

A similar driver for the Philips ISP1181 is being tested and should be delivered within a few weeks.



Sounds good.


We will need a copyright asignement in order to incorperate this into
eCos. Please could you take a look at:

http://ecos.sourceware.org/assign.html



That won't be a problem.

       Thanks
               Andrew







Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]