X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C81775.9359635F@onstor-exch02.onstor.net>; Thu, 25 Oct 2007 18:11:53 -0800
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Content-class: urn:content-classes:message
Subject: RE: Re-review of 26028
Date: Thu, 25 Oct 2007 18:11:53 -0800
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E056C916D@onstor-exch02.onstor.net>
In-Reply-To: <20071025180832.0fb0520e@ripper.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Re-review of 26028
Thread-Index: AcgXbLtDWqjuU5vfRCiaoM4BhAtnZgAAEqfw
From: "Larry Scheer" <larry.scheer@onstor.com>
To: "Andy Sharp" <andy.sharp@onstor.com>

Looks like I need to spend some time with you going over my changes and
what needs to be done, explaining how it works and why I need to do
things the way I did.

The linux/kernel Makefile needs work I don't think failing is the
correct thing to do if you invoke make in the linux/kernel directory.
Depending upon a hierarchical make is just wrong. If we spend some time
together I think we can quickly come up with something we both like, for
this Makefile.

Right now we can't build a cougar kernel until you get more of your
stuff checked in. No sense rushing this. We will just have to put up
with the nightly build failing for a while.


-----Original Message-----
From: Andy Sharp=20
Sent: Thursday, October 25, 2007 6:09 PM
To: Larry Scheer
Subject: Re: Re-review of 26028

On Thu, 25 Oct 2007 14:12:27 -0700 "Larry Scheer"
<larry.scheer@onstor.com> wrote:

> Andy,
>    I was not satisfied with the top level Makefile changes for the
> kernel build. I decided to change the rules.
> Please review these changes.
>=20
> Also the kernel build would fail if the ROOTFS didn't exist so I
> added a rule to make it.
>=20
> Change 26028 by larrys@larrys on 2007/10/24 16:49:35 *pending*
>=20
>         Change top level Linux kernel targets for specific products to
>         use an implicit rule and to call make kernel recursively
> passing in
>         PROD to the kernel target.=20
>         Added a default setting for TFTPDIR in case it is not passed
> in or set in the environment.
>         Reviewed by: AndyS
>=20
> Affected files ...
>=20
> ... //depot/dev/Makefile#2 edit
> ... //depot/dev/linux/kernel/Makefile#7 edit

Makefile

     line 47: is that a trailing space?  yes, it is!  this just
     needlessly re-invokes make.  i don't understand why any kernel
     target in the top level makefile should proceed into linux/rootfs
     directory.  probably shouldn't.

     kernel: kernel-bl # kernel-cg

     kernel-%:
     	make -C linux/kernel $@ PROD=3D$*  (<-- $*? is that even
     	right?)

     these targets in this directory don't invoke kernel-install,
     so there are no rootfs dependencies.

linux/kernel/Makefile

     line 37

     no need for this stuff here, it just makes my life more difficult.
     this is caught in higher level makefiles.

     line 68

     the rootfs target is bogus: the make should die not knowing how to
     make rootfs, or, it should invoke make in the rootfs directory,
     but doing a mkdir could be very bad right here.  i don't like
     the mkdir $(rootfs)/boot either for the same reasons plus it's
     procedural code that should be avoided in a makefile.  add /boot
     to the end of the kernel-install dependency line like thus:

     kernel-install: $(KERNEL_FILES) $(ROOTFS)/boot

     and remove the test/mkdir line for the boot directory.  if rootfs
     has been properly made, it is there, otherwise, make should fail.


