Re: [Outreachy kernel] Need help building a driver: mt7621-pci

[Deepak R Varma <mh12gx2825@gmail.com>]

On Tue, Oct 20, 2020 at 12:32:22PM +0100, Matthew Wilcox wrote:
> On Tue, Oct 20, 2020 at 09:08:36AM +0530, Deepak R Varma wrote:
> > On Mon, Oct 19, 2020 at 08:18:03PM +0100, Matthew Wilcox wrote:
> > > It's useful for drivers to build on all architectures by depending
> > > on COMPILE_TEST.  You will probably have to provide a lot of dummy
> > > interfaces; the only requirement is that the code build, not work,
> > > since it almost certainly needs to access hardware that only exists on
> > > the Mediatek platform.
> > 
> > Thank you. Tried a couple of times with different selections however,
> > not able to get across this error:
> > 
> > make[4]: *** No rule to make target
> > 'drivers/staging/mt7621-pci/pci-mt7621.o'.  Stop.
> > make[3]: *** [scripts/Makefile.build:475: __build] Error 2
> > make[2]: *** [scripts/Makefile.build:500: drivers/staging] Error 2
> > make[1]: *** [Makefile:1777: drivers] Error 2
> > make: *** [Makefile:335: __build_one_by_one] Error 2
> > 
> > It looks like a generic error message and lot of possibilities while I
> > try to explore it myself. Any quick suggestion on what is wrong will
> > be time saving.
> 
> You need to go to the Kconfig file and change:
> 
> -       depends on RALINK
> +       depends on RALINK || COMPILE_TEST
> 
> Then you'll need to add dummy definitions of various things to make it
> build.  Commit 171a9bae68c72f2d1260c3825203760856e6793b shows me doing
> horrible things to the Octeon driver to make it build on any platform.
> It's a mixture of adding structs that I copied from elsewhere and
> making random #defines.  There's no hard-and-fast rule about how to make
> things build for COMPILE_TEST, you're just trying to make things build
> so that changes across the tree (eg to common APIs) get picked up by
> the build bots.

This is really very useful information. I will definitely attempt this
once I am done with the current patch work on a different driver.
Thank you very much Matthew.

Deepak.