Re: [KBUILD] allow any PHONY in if_changed_dep

[Milton Miller <miltonm@bga.com>]

On Jul 1, 2006, at 4:39 AM, Sam Ravnborg wrote:

> On Sat, Jul 01, 2006 at 04:16:56AM -0500, Milton Miller wrote:
> > While all the if_changed family filter $(PHONY) from the list of newer
> > files, if_changed_dep was only filtering FORCE from the list of all
> > dependents.  This resulted in forced recompile every time.
>
> I see where you are heading with this. But on the principle of minimal
> suprise this is not good. Why does it not include file.o in the link
> when I mark it phony?

This is your suprise argument, right?  (Not my current problem).

Actually this is only changing weither file.o gets considered in
causing the build command to be executed.

> The root cause is buried in the powerpc/boot/Makefile restructuring.
> I do not see what is wrong with current approch if you do a simple
> s/cmd/if_changed/g and then assign targets += <relevant targets>

The issue is I need headers copied for some of the files.  The current
approach is any file that might need a copied header gets a dependency
on all of the copied .h files.  I was planning on coping some more
headers used in different files.  if_changed_dep

The issue is specifically COPYFILES.  The .o can not be built until
its dependencys are met, but it never exists itself.  The actual .h
files a given source will be detected by if_changed_dep and fixdep.

> You need to post a full diff of the Makefile before I can give better
> feedback.

Sure.  I'm reposting the series, I'll cc you and copy the makefile
patch to linux-kernel, its 3/5.    That patch is independent of 1 and 2 
while 4 and 5 require them.  The full series is available on the 
linuxppc-dev mailing list.

milton

-
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/