Vincent, I'm glad to hear you had the driver working so well. I look forward to seeing it.
Thanks, Rob On Tue, Aug 18, 2009 at 5:01 AM, Vincent Del Medico<[email protected]> wrote: > Rob, > > I didn't release the code, as this is not really part of openmoko. > To make this driver, I used a part of Atheros's driver, merged with a part > of openmoko's one, resulting in a "arch-independant" driver for 1.x versions > of this card. > > This worked well, and was stable enough for applications like video > streaming, so I stayed with this solution. > > I don't have access to the code yet ( I recently changed of company ) , I > have to ask a person to send it back to me. AFAIK, he is on holiday > actually, I'll be able to give it to you in about a week I think. > > Regards, > Vincent. > > 2009/8/18 Robert Emanuele <[email protected]> >> >> Hi Vincent, >> >> I would be happy to give your code a try. We are not using DMA with a >> real MMC driver. I'd appreciate a look at it. Have you released this >> code back to the openmoko project? Is this the final solution you >> went with or did you end up finding a card with the 2.0 firmware? >> >> Thanks, >> >> Rob >> >> On Mon, Aug 17, 2009 at 10:22 PM, Vincent Del >> Medico<[email protected]> wrote: >> > Hi rob, >> > >> > Openmoko's version of AR6k driver is for cards with 2.x firmware. >> > You seem to have a 1.x version ( BMI Get Target Info: Exit (ver: >> > 0x1302270f >> > type: 0x1) ), which is not compatible with 2.x. >> > >> > I faced the same problem as you before, except I was playing with the >> > OMAP >> > processor ( TI ). >> > >> > I can give you the job I did, but it only works in some conditions : >> > - your MMC/SD/SDIO driver must NOT use DMA transfers. The 32bit ARM core >> > only supports transfers of data multiple of 4 bytes, but the AR6k driver >> > sometimes sends data of 1, 2 or 3 bytes, resulting in DMA misaligned >> > errors. >> > I've not been able to solve this problem, it could be nice to >> > investigate. >> > - the driver works with a real MMC/SD/SDIO driver ( not SPI )... >> > nevertheless, it maybe in SPI, but didn't tested. >> > >> > Regards, >> > Vincent. >> > 2009/8/18 Robert Emanuele <[email protected]> >> >> >> >> Greetings, >> >> >> >> I've been toying with your port of the ar6000 driver with my atmel >> >> at91 processor. >> >> >> >> It tries to initialize the card but fails.... >> >> >> >> ar6000_available >> >> BMI Get Target Info: Exit (ver: 0x1302270f type: 0x1) >> >> Block Size Set: 128 (target address:0x8000066C) >> >> ar6000_init() Got WMI @ 0xc3a37d60. >> >> Timeout waiting for recv message >> >> Target Not Available!! >> >> ar6000_avail: register_netdev failed >> >> +ar6000_destroy >> >> ar6000_cleanup(): WMI not ready 0xc3a322c0 0xc3a37d60 >> >> ar6000_cleanup(): Shut down WMI >> >> eth0: link up (100/Full) >> >> Attempting to reset target on instance destroy.... >> >> -ar6000_destroy >> >> >> >> In an earlier post I saw that this firmware version may be too old. >> >> I've tried cards from 3 different vendors and the newest version I can >> >> come up with is 0x17 (which is on the way). Does anyone know the >> >> differences between 1.3 and the version that works with this driver? >> >> How do the CE guys get this to work? Or, does anyone know of a vendor >> >> that sells an ar6000 with the latest firmware in an SD Card package? >> >> >> >> Thanks, >> >> >> >> Rob >> >> >> > >> > > >
