Hi Andreas, On Mon, 21 Jul 2003, Major A wrote:
> > Ok, here we go. Attached the EZ-USB firmware release providing own > > descriptors, EP0 standard requests and bulk sink/source support. Ready for > > AN213x and FX - but not (yet) FX2. Successfully tested with ehci-hcd > > and usbtest - as shown in the other mail it passes all current tests. > > Sorry I can't test it right now, but a quick check has revealed the > following: > > - It won't renumerate on the FX revision A, because that chip has a > known bug (I discovered it, in fact). To make that chip disconnect, > you have to set DISCOE, not just DISCON. There is no way for the > firmware to determine what revision the chip is. Well, AFAICS there's nothing mentioned even in the latest TRM. However, I saw some errata there at the stuff that came bundled with the development board. To me it wasn't quite clear what's going on and more important, in my case I have full control over the hardware so I know there is no Rev A silicon in the field. As you discovered it, would you mind to explain? My understanding was the DISCON pin would float pretty long at H (maybe due to some similar soft-pullup like they have on the ports) so it is required to actively drive it L. However, I have never used the simple setup with the 1k5 resistor on USBD+ directly connected to the pin - I've always added a transistor there like they did on the development board. So I have never seen any issue with disconnect not working. Anyway, to work around, is it sufficient to just set both DISCON and DISCOE in USBCS? My impression from some of their sample code I've seen was there might also some timing and/or repeating requirements? > - Are you sure the "mov int2clr, a" really works? On the FX and FX2 > that I've tried, it never did, although I used the sample code > provided by Cypress. I had to do it the old-fashioned way (EXIF, > then USBIRQ). I cannot tell for the FX2 but I've never seen any problem with INT2CLR or INT4CLR. I'm running the FX with up to 100000 interrupts per second, mostly USB and GPIF, so I'm pretty sure I would have noticed. I'm sure you have it enabled in the USBBAV and INT4SETUP registers respectively? And you still need the EXIF first with INTxCLR as well, it's only a replacement for the USBIRQ or IN/OUT07IRQ. > - If you need help getting the FX2 working, please ask. I've got > firmware working that uses pretty much all the features of it > (including 16-bit GPIF with AUTOIN/OUT), I can share that knowledge Well, I've added some support for FX2 just from reading the specs. I've no resources available to work on real support for this chip at the moment, sorry. I think I have most stuff in place according to the TRM. What is missing is mostly proper endpoint reset, HS configuration and the appropriate descriptors you've mentioned below - and a lot of testing. Patches welcome of course ;-) > (it took me quite a while to acquire, because I was used to the DMA > way of doing transfers as in the FX). One thing I don't use is iso > transfers, though. Yes it's really a shame they've killed the DMA engine in the FX2. I've complained to them and they said it was because the 8051 couldn't keep with HS bandwidth anyway. This is right of course but there are other benefits f.e. when combining GPIF/FIFO transfer with XMEM buffering, which are lost without DMA. In fact this was the main reason not to use the FX2 in some project recently. > - I'm sure you know, but the one obvious thing missing that is > required for FX2 support is the extra descriptors for hi-speed USB, > and that the descriptors also have to depend on whether hi-speed or > full-speed is used. Right. Adding the descriptors shouldn't be no big deal anyway. Thanks for your comments! Martin ------------------------------------------------------- This SF.net email is sponsored by: VM Ware With VMware you can run multiple operating systems on a single machine. WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the same time. Free trial click here: http://www.vmware.com/wl/offer/345/0 _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
