On Sun, Dec 20, 2015 at 7:28 PM, Sudip Mukherjee <sudipm.mukherjee@xxxxxxxxx> wrote: > On Sun, Dec 20, 2015 at 04:43:53PM +0000, One Thousand Gnomes wrote: >> > v3: Alan commented on v2, few things like NULL check, name of the gpio chip, >> > moving some more code into the gpio code. He also commented on it being >> > a separate module, but since this will not be needed by someone who is not >> > using Exar chip, and even those who are using, some of them may not need it. >> > This is optional for only those who wants to use the gpio capability of that >> > chip. So kept it as a separate module. Waiting for your comments on that. >> >> That doesn't work because you reference the methods in it so it will >> always be dragged in. You would have to make the exar driver an MFD that >> provided a serial and a gpio binding to fix that up I think, or instead >> of having the serial driver call into your gpio driver (thus forcing the >> module into memory) you would have the serial driver create a platform >> device or similar that the GPIO device bound to. > > So then should I rewrite it as MFD Once we split 8250_mid.c (which is actually supports real MFD on one of newer Intel SoCs) from 8250_pci.c, but it is a comprehensive driver that uses 8250_core/port.c as a library. > or should it be like the way you > suggested in the other mail to have it as bool in the Kconfig and combine > the file with 8250_pci.c when the symbol is selected? As far as I can see, you did other way around: use your module as a lib. In that case you have to compile 8250_pci as 8250_pci.o and your 8250_exar.o, though you can't use the same name for resulting module and one of its objects AFAIR. -- With Best Regards, Andy Shevchenko -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
