On 05/15/2012 11:39 PM, Greg KH wrote: > On Tue, May 15, 2012 at 10:36:32AM +0800, edwin_rong@xxxxxxxxxxxxxx wrote: >> From: edwin_rong <edwin_rong@xxxxxxxxxxxxxx> >> >> Recently we find that many warm-hearted people are helping improving the >> coding style and something else of driver rts5139 & rts_pstor, which will >> be replaced with one new refactored uniform Realtek cardreader driver in >> the future, update the TODO file to inform people not to do modifications >> to both of the drivers any more to avoid wasting their time and energy. > Is there anything that you can point people to today for this "new > driver"? Given that there is no other driver at this point in time, I > don't blame people for fixing up this one to work properly. So until we > see a new driver, I don't feel comfortable asking people to stop working > here, do you? > > thanks, > > greg k-h Hi Greg, No, it's not polite to stop people helping you do some fixing work. I totally agree your point, but my original purpose is to avoid people wasting their time and energy :-) About the "new driver", we make significant modifications and re-design it with Object-Oriented theory, which is adopted in kernel everywhere. We divide the driver into several portions: . "card component", which is used to implement all kinds of card protocols in h/w-independent way; ."card manager component" which manipulate all the cardreader supported cards; ."card reader component" which is responsible for concrete operations of IC; ."scsi component", whose work is to accept SCSI commands from SCSI subsystem and forward them to "card component" for further process, and ."transport component" which hidden the specfic physical interface from the compoents mentioned above. As you have seen that there's 2 drivers "rts5139" and "rts_pstor" under staging folder in Linux kernel code, the former drives USB-based card reader, while the later supports PCIe-based card reader and both of them are driver-based driver, which implement SD/MS/xD card protocol, respectively, to support these series of cards. In fact, It doesn't need to do so, since they are standard protocols, if we implement them in a hardware-independent way, it can be reused, which decreases code redundance. Also, we have to consider expansion issue of the driver, since our company will design out new card reader ICs. Actually, the "new driver" could, to some degree, be treated as an subsystem which supports all Realtek cardreader ICs. Thanks & BRs Edwin _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/devel
