X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C7B348.DDCDD64A@onstor-exch02.onstor.net>; Wed, 20 Jun 2007 06:39:54 -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: eee protocol and mgmt-bus driver - checkin 1
Date: Wed, 20 Jun 2007 06:39:54 -0800
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E031D889C@onstor-exch02.onstor.net>
In-Reply-To: <BB375AF679D4A34E9CA8DFA650E2B04E043607B7@onstor-exch02.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: eee protocol and mgmt-bus driver - checkin 1
Thread-Index: AcevkdcRlPL+nbdVTsaaFkfNz5sYdwCVsTLdAABbIeAAAIH2kAACVh3gAAHf14AAUrQKIA==
From: "Richard Beck" <IMCEAEX-_O=ONSTOR_OU=FIRST+20ADMINISTRATIVE+20GROUP_CN=RECIPIENTS_CN=RICHARD+2EBECK@onstor.com>
To: "Maxim Kozlovsky" <maxim.kozlovsky@onstor.com>
Cc: "Andy Sharp" <andy.sharp@onstor.com>

For the first submittal I have removed the dependency on nfx-types.h,
but the issue must be faced. If we are to have the neteee protocol
dealing with the agile_hdr struct as with the BSD version, then the
files will have to include nfx-types.h. =20

Changes to the linux/kernel/Makefile for copying neteee header files to
/usr/include/neteee have been removed from the changelist in favor of
the standard method.=20

Next set of changes shortly.

--rb

-----Original Message-----
From: Maxim Kozlovsky=20
Sent: Monday, June 18, 2007 4:05 PM
To: Richard Beck
Cc: Larry Scheer
Subject: RE: eee protocol and mgmt-bus driver - checkin 1

You should not have to include nfx-types.h in the kernel. Why there is a
dependency on this file?

1 - That would be p4 reopen.=20

2 - If you don't know how to do it, don't do it. We will figure it out
later.

-----Original Message-----
From: Richard Beck=20
Sent: Monday, June 18, 2007 4:01 PM
To: Maxim Kozlovsky
Cc: Larry Scheer
Subject: RE: eee protocol and mgmt-bus driver - checkin 1

Max & Larry,=20

Unfortunately while dealing with multiple changelists in perforce, I had
to revert a couple of the files to put them in the right numbered list.
I didn't realize that would be over-writing my files, so I will fix them
and get back with you when it's corrected.
=20
Larry -=20

1. what's the right way to move a file from one change list to another
(without doing a 'revert'!)?

2. could you look at the changes I made to linux/kernel/Makefile for
copying the include file(s) over to /usr/include/neteee?  There is
probably a more standard way to do this.  Also, where are we going to
put files like nfx-types.h?

--Rich


(Other comments that can wait until the 'revert' damage is fixed....)
1. I have updated the comment in net/neteee/Kconfig to be more accurate.
2. linux/kernel/linux-mips-2.6/net/Kconfig has been included in the
changelist.
3. I can include nfx-types.h, but wasn't sure if that really should go
in /usr/include/neteee.  The change list does compile, but you're right,
it depends on this file which we need to find a home for...

--rb

-----Original Message-----
From: Maxim Kozlovsky=20
Sent: Monday, June 18, 2007 2:05 PM
To: Maxim Kozlovsky; Richard Beck
Cc: Larry Scheer; Andy Sharp
Subject: RE: eee protocol and mgmt-bus driver - checkin 1

Net/neteee/Kconfig:

	The comment is wrong.

-----Original Message-----
From: Maxim Kozlovsky=20
Sent: Monday, June 18, 2007 1:59 PM
To: Richard Beck
Cc: Larry Scheer; Andy Sharp
Subject: RE: eee protocol and mgmt-bus driver - checkin 1

Bobcat-config - no change

linux/kernel/linux-mips-2.6/net/Kconfig - modified but is not part of
the change

I don't think the changes to linux/kernel/Makefile are right, there
should be some common method to get the kernel includes in the right
place. Nevermind, we'll let Larry figure this out.

Eee.h -=20

4: This will not compile, there is no nfx-types.h in this change.

Where is the change that enables CONFIG_NETEEE and CONFIG_MGMT_BUS?


-----Original Message-----
From: Richard Beck=20
Sent: Monday, June 18, 2007 1:43 PM
To: Maxim Kozlovsky
Cc: Larry Scheer
Subject: eee protocol and mgmt-bus driver - checkin 1

Max,

Checkin #1 is in:
p4 client:  richardb-linux
changelist: 24262

Changes for items 2 & 3 will be checked in later today.
--Rich

-----Original Message-----
From: Maxim Kozlovsky
Sent: Fri 6/15/2007 2:12 PM
To: Richard Beck; Tim Gardner; Andy Sharp
Subject: Next steps for eee protocol and mgmt-bus driver.
=20
Hello Richard,

Here is how I want us to proceed for the rest of your stay with us. I
don't expect that you (or any other person) will be able to have
complete debugged implementation of EEE protocol and the management bus
driver in a week. I would like to get checked in as much as possible of
code that we all agree works and passes some tests. On the other hand, I
don't really need the code that is not known to work since it will
probably be easier for us to start over than to do the debugging for
you. To accomplish that, I'd like to do the checkins and reviews in
steps. I will be always available to do the review in the regular
working hours and will do the review as fast as possible so you are not
slowed down by multiple reviews. I hope Andy will help as well.

Here is the list of the steps together with the exit criteria:

1)	Check in the kernel configuration changes with empty .c and .h
files. After these changes, the kernel make in the kernel directory
should compile the neteee and mgmt bus driver .c files. The
corresponding .h files should be installed in the appropriate place in
/usr/include/neteee as part of the build process.
2)	Check in the changes that add AF_EEE/PF_EEE protocol family.
After these changes, we should be able to create sockets with
socket(PF_EEE, SOCK_DGRAM, 0).
3)	Check in the changes that add the sockaddr_eee. After these
changes the following tests should pass:
a)	Bind a socket to a specified address
b)	Negative test for binding a socket with the same address as
another socket
c)	Negative test for binding a socket to invalid address
d)	Bind a socket to ADDR_ANY - should pickup the next available
address in reserved range
e)	Test to make sure the address is available again after bound
socket is closed
4)	Check in the changes to support connect call for AF_EEE sockets.
Should pass positive tests and negative tests for invalid address.
5)	Check in the code to support send and receive on the loopback
interfaces. Should pass simple client/server tests for SSC applications
only.
6)	Check in the code that registers the mgmtbus driver as the
network driver. Put just minimal amount of the code to make it
recognizable to the linux kernel, the input/output routines should do
nothing and just drop the packets. After that we expect to do ifconfig
on this device and assign ip and eee addresses to it.
7)	Check in the code for sending and receiving the packets through
the mgmt bus driver. The simple client/server tests between the
applications on txrx, fp, fc and ssc should pass.
8)	Check in the code to support fragmentation. The tests for
sending large packets between ssc and embedded cards should pass.
9)	Check in the code for IP forwarding. We will not be able to test
this part without a lot of other system running.

Max



