On Tuesday 21 November 2006 6:10 am, Rodolfo Giometti wrote: > Hello, > > here my proposal for PXA27x USB device support. This code is strongly > derived from linux-hh version but with some fixes by me. This driver > as been tested on a PXA270 custom board.
Do you know if the new PXA 3xx chips chips use this UDC, or if the new-owners-of-PXA have actually fixed some of the problems which have made it so hard to get a good driver working? It'd be nice if you could compare this driver to the others which have circulated, and describe its status. What tests does it pass, what gadget drivers does it work with, how robustly, and so on. How do the HH.org folk view this version? > I know this driver is quite far from perfection but, in my opinion, it > could be a good starting point to add this support into the kernel. Linux could really use a good starting point for PXA 27x USB peripheral support; that hardware was fighting back all too successfully! :( So it's good to see you take this up. The endpoint autoconfig stuff would need to be split completely out from this driver ... could you do that? That change would involve updating all the other users of that autoconfig code. Such changes must not depend on being used only with a pxa27x USB controller. A brief skim of this makes me think you're doing more than just generalizing things. It looks like you're changing the model of an endpoint in the API ... making it work less well for non-PXA hardware, where endpoints and altsettings are managed easily in software, rather than buggily in hardware. This stack works on too much other hardware to accept such changes. You need to split out that work, so that it can be reviewed on its own; and possibly, to develop a better solution. (I sort of liked the idea of a hook into the UDC that gets called by the endpoint autoconfig logic, to pick the best endpoint or configure the FIFO memory appropriately. Your patch does that. But the ep_alloc gadget op doesn't use the usb_endpoint_config thing, which is surprising, and it wasn't clear to me how best to handle the more usual case where config/interface/altsetting values are completely irrelevant to the hardware.) I notice you still haven't managed to get the PXA to properly handle CDC-Ethernet. Those chip errata seem to act like they're never going to get fixed! The downside of that is that it seems like your patch likely doesn't handle altsettings well. The update to gadget_chips.h could be submitted and merged separately. > I use the driver from my platform specific file into > "arch/arm/mach-pxa/" by using the following code: Could you try to follow more closely the model used in the latest pxa2xx_udc code (for pxa210/25x/26x)? The old model was a bit too funky/strange. ;) - Dave ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
