Re:[patch, rfc] kbuild: implement checksrc without building Cources...

[Oleg Verych <olecom@flower.upol.cz>]

On Tue, Oct 24, 2006 at 01:45:08PM -0700, Randy Dunlap wrote:
> On Tue, 24 Oct 2006 19:43:40 +0000 (UTC) Oleg Verych wrote:
> > This doesn't work, use ifs instead. Updated.
> > I have no idea what to do with generated sources and headers.
> > One may be: check target `if_changed' to be %.c or %.h and let it be
> > built.
> 
> Hi Oleg,
> 
> Yes, it works for me, with the exception of host-generated
> files, as you mentioned.  I ran into those with:
> IKCONFIG (the one that you mentioned), ATM_FORE200E firmware,
> IEEE 1394 OUI database (which I sent a patch for -- it should
> not be generated when the config option is not enabled),
> RAID456 tables, VIDEO_LOGO files, and CRC32 table.

I'm glad, that semi-working thing was helpful. But it's ugly hack.

Idea is to substitute objects (*.o) with sparse output, thus targets
like
,--
|%.o : %.c
+--
will be updated and new check will pass them (force check may be applied).
Also having results in files (even in so messy called *.o) is good for
collecting and sorting errors. Headers will be generated as needed.

Finally short statistics maybe printed in the end of the check:
[(stat -c %s *.o > 0 | wc -l) / find --name *.o]
  ^ with error file size > 0  / overall error files

But implementing all this in non-hack way isn't easy. Maybe Sam will
came back and will do it magically quickly. Until that, i'll try to
fight with current makefiles myself.

> Thanks for your time and effort.  Maybe Sam will have some ideas.

I've found "sparse" and read some philosophy from README. Nice.
Thanks you also !
____
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/