v0.1: Initial release September 27/99

user space: pppoed
kernel patches against: 2.0.38,2.2.12,2.3.18

-------------------------------------------------------
V0.2: October 16/99

user space: patch agains pppd-2.3.10 to get it working with kernel 2.3.21
kernel patches against: 2.0.38,2.2.12,2.3.21
extra kernel patch: 2.3.21 from Paul Fulghum 


pppoed changes: Contribution from Irving Reid <ireid@sympatico.ca>
	- A Makefile for pppoed 
	- A fix for pppoed to get the -I option working with non-eth0
	  interfaces
kernel changes:
	-testing with 2.3.21. Required a couple of fixes and a patch
	for both pppd and kernel ppp code. [The 2.3.18 patch in v0.1
	will not work even with the kernel and pppd patches]. The ppp 
	patch is from Paul Fulghum <paulkf@microgate.com> and will 
	probably never be needed for post 2.3.21 kernels and 
	post 2.3.10 pppd.

        - a quick test with 2.0.38 failed. Analysis shows its
        quite a bit of work which requires using the ppp-2.3.8
        kernel patches. I'll make this a very low priority unless
	someone gives a very very good reason.

-------------------------------------------------------
V0.3: October 23/99

Only user space pppoed extensions.

- Extended the PADR to echo the AC Cookies. 
It seems that Deutsche Telecom (DTAG), Germany has implemented it
in their AC (and so we have to oblige to inter-operate;->).
Maximum credit goes to Oliver Bleutgen<o.bleutgen@ndh.net> for debugging 
and testing this. Without him this would have been impossible. It brings
new meaning to the term "remote debugging" and a long with it an 
appreciation that discovery lives in user space. And yes, Oliver is
one cool coder!

-Added the -L to select a different lineid/pppox device (other than the default
pppox0). Josef Drexler <jdrexler@julian.uwo.ca> requested this. He is 
the only person I know who creates two sessions to different suppliers.
This is one of the reasons PPPOE was created.

- Added the -F option to select additional ppp options file, beyond
the default /etc/ppp/options. This is obviosuly required for the -L
option but was requested for by Martin Hebert <martin@simpli.net>
because he uses PPTP to VPN.

-Added the -R option to restart pppoed, should the other side die.
This is a popular request by many people.
Wrote the persist-HOWTO to accompany this feature.

-Josef Drexler wrote the Multi-pppoed HOWTO to get multiple sessions going.
-Greg Stark <gsstark@mit.edu> contrinuted a tcpdump patch. Should probably
go into ANK's tcpdump. Included here for now.

----------------------------------------------

V0.4: Nov 27 1999

kernel changes: patch against 2.3.29 and 2.2.13 tested

pppoed changes:

+ Thanks to SUSE we now have dedicated major/minor numbers. Major
144 belongs to pppox.
I strongly recommend people to migrate to this new version.

SUSE is packaging this in 6.3.

This version is a fork from 0.3 since it uses the new major minor numbers.
i.e it is not compatible with version 0.3 because it will be looking
for a different device identifier. So delete your /dev/pppox*
devices and use the new pppoed.c instead.

You can now build the driver as a module 

- Andi Kleen <andi@suse.de> Modularization, various cleanups, port to the 
  new ioctl and endian fixes. Good stuff.
- Woody Suwalski <woody@rebel.com> fix for packed structs needed by 
  the ARM.
- jhs -P option to select a different pppd.  
- Marc Boucher <marc@mbsi.ca> sigchild and misc cleanup, typos etc
- jhs AC and service name reflection requested by Chae Yoonki <tulbo@myself.com>
  needed for the telcos in Korea. Also added overriding of those two options
  using -A and -E options. 

-----------------------------------

V0.41: Dec 09 1999

Kernel: patch against 2.3.31 and 2.2.13 tested

pppoed changes:
- jhs: PADI service name overriding
- Marc Boucher <marc@mbsi.ca> fix the patch screwup on 2.3 kernel

----------------------------------------------

V0.42: Dec 19 1999

Kernel: patches against: 2.2.13 and 2.3.33 tested

pppoed changes:
- Marc Boucher <marc@mbsi.ca> 
Some major cleanups.
support for daemonization, syslog, runtime debugging, ARM fixes, etc.
Well tested on the netwinder.

----------------------------------------------

V0.43: Jan 19 2000

General:

* Could people Please Please test this especially users in Europe and Asia
(who seem to have these weird requirements). There has been a lot
code added/rearranged/removed from pppoed; i have been testing this at
Sympatico for sometime. It works fine.

*new mirror site at:ftp://ftp.netwinder.org/pub/mirrors/pppoe/

* Marc Boucher <marc@mbsi.ca>
  *contribution of mssfwclamp module 
	+ works with both 2.2 and 2.3 kernels
	+ gets rid of requirement to change MTUs on masqueraded machines
* RPMing contrib from Ralph Siemsen <ralphs@netwinder.org>
* Redhat init files from Simon Piette <spiette@generation.net>

*This is the beggining of a major overhaul of the architecture of the
user space pppoed. 
- I will not accept architectural changes from now on -- only bug fixes.

Kernel:
- patches against: 2.2.14 and 2.3.33 tested
- got rid of the bh context check (it was scaring some folks)

pppoed changes:
- Perry Wintner <perryw@3no.com>, Service Name Tag miscopy fix
- Jeff Hildebrand <hildebrj@airwire.com>: '-C' option to allow restart
  using ppp peer files.
- kalle@datenhighway.com : FAQ error fix
-jhs: 
  * Some major cleanup and re-achitecturing.
  * Errors generated by AC not ignored

----------------------------------------------

V0.44: Feb 02  2000

This is a hurried release; please test

Kernel:
Forgot to include the patch against 2.2.14 stated in release notes
for 0.43 ;->

pppoed changes:

- Bug fixes by Marc Boucher. 

----------------------------------------------

V0.45: Feb 12  2000

The restructuring of the user space code continues ...
[There is a lot of code tuning needed; however the code works
just fine. I've been testing it for a week now on a RH6.1 P100 machine]

kernel changes:

patch against 2.2.14 and 2.3.33

** I have not been updating the 2.3 kernel patches;
This is in anticipation of a code merge between Michal Ostrowski
and myself.  We hope to merge completely with version 0.5. 
 
The 2.3 patch included should patch cleanly against newer 2.3
kernels.

- removed totaly semaphore dependency
- allow pppd to send control packets from user space during setup.

pppoed changes:

- More cleanup
- Found out the reason behind the pppd dying in some systems 
It happened to me so i had to fix it ;-> Read the FAQ question 3.
- I am including diald config from Andreas Fincke 
<andreas.fincke@netcologne.de> for dial on demand.
Thanks Andre.
- Added the -G option to enable users to invoke pppoed from pppd.
Mostly because i dislike hacks such as diald and would hope people
would use the built in dial on demand features of pppd.
** I have not tested the details. I have cut and pasted from the pppd
man pages; If someone does get this working correctly please fix the
document in doc/demand-dialling and send me a patch.

----------------------------------------------

V0.46: Feb 26  2000

- Contribs from Amy Fong<afong@furryterror.org>
  + Configure script to ease install
  + Man page
  + Debian packaging

kernel changes:
None. 

pppoed changes:

- More cleanup; getting there ....
- updated diald config by Andreas Fincke <andreas.fincke@netcologne.de>
  Fixing earlier problems.

----------------------------------------------

V0.47: March 19  2000

General:
- Added reference to the Korean translation of the docs by
  Chae yoonki<tulbo@myself.com>
  at http://soback,kornet.net/~eezen/adsl/

kernel changes:
None.  

pppoed changes:

- More cleanup; getting there ....
- Added the hostuniq option 

----------------------------------------------

V0.48: October 08  2000

General:
- Under a lot of pressure, from a lot of people, i have created
an install script. You dont need to  patch or compile the kernel
anymore. Just type "./Install_pppoe" at the top directory.

For people who know what they are doing, the kernel patch and 
instructions are still there.

There is now absolutely no good reason to use  a user-space 
solution.
Regardless of how "efficiently" it is written it will have a
relative impact on your performance.
Technically, a user-space solution adds two more context switches
and two extra copies _per packet_
You do the math.

kernel changes:
-Abstracted the module out.  

pppoed changes:
- More cleanup

