EST SBC8260 and Canary CTX-1170

EST SBC8260 and Canary CTX-1170

[Brian Ford]

I am having the same problem that was referenced in:
http://lists.linuxppc.org/listarcs/linuxppc-embedded/200010/msg00167.html

Basically, I can't generate any traffic on the fast Ethernet as seen
by a sniffer.  I could not find any responses to that message, so I will
pose those questions again.

Does Linux support the CTX-1170 MII?  The arch/ppc/8260_io/fcc_enet.c
source code talks about the QS6612.  Will I need to extended fcc_enet.c to
support the CTX-1170?

I know that the hardware can work, because the VxWorks boot rom loads
Linux successfully via this interface and MII.

Thanks for any pointers on where to start.

--
Brian Ford
Software Engineer
Vital Visual Simulation Systems
FlightSafety International
Phone: 314-551-8460
Fax:   314-551-8444


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

Brian Ford wrote:


> Does Linux support the CTX-1170 MII?  The arch/ppc/8260_io/fcc_enet.c
> source code talks about the QS6612.  Will I need to extended fcc_enet.c to
> support the CTX-1170?

That driver either "supports" any PHY, or none at all, depending upon
your perspective.  None of the command/status MII interface is in
place to do anything.  None of those functions are called in the
driver.  The QS6612 comment is left over from the 860T FEC driver I
copied to get the framework for the FCC driver.

The Linux 8260 FCC driver works with the supplied EST hardware and the
CTX-1170 I have.  I know it doesn't work on some others, and I don't
know if they use the same Ethernet PHY or the deal of the week.

>From looking at the drivers, it appears they are a little out of
date.  I have a newer one (that I was certain I checked in) that will
interrogate the PHY.  In any case, this driver should work fine on
any half duplex Ethernet, and you should see transmit packets in any
case.

> I know that the hardware can work, because the VxWorks boot rom loads
> Linux successfully via this interface and MII.

Then the driver must not be configuring the I/O lines correctly for
some unknown reason, or you are running full duplex without the FCC
initialized to do that.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Brian Ford]

On Mon, 27 Nov 2000, Dan Malek wrote:

> That driver either "supports" any PHY, or none at all, depending upon
> your perspective.  None of the command/status MII interface is in
> place to do anything.  None of those functions are called in the
> driver.  The QS6612 comment is left over from the 860T FEC driver I
> copied to get the framework for the FCC driver.
>
So does this mean that a "complete" implementation would have a
"command/status MII interface" or that a "complete" implementation does
not need one?

> From looking at the drivers, it appears they are a little out of
> date.  I have a newer one (that I was certain I checked in) that will
> interrogate the PHY.  In any case, this driver should work fine on
> any half duplex Ethernet, and you should see transmit packets in any
> case.
>
That seems to be my problem.  It is negotiating a full duplex link.

> > I know that the hardware can work, because the VxWorks boot rom loads
> > Linux successfully via this interface and MII.
>
> Then the driver must not be configuring the I/O lines correctly for
> some unknown reason, or you are running full duplex without the FCC
> initialized to do that.
>
Can you help point me to where I need to write code that would initialize
the FCC to do full duplex, or to have it auto-negotiate to half
duplex?  It is fairly problematic for me to force the other end to half
duplex and, ultimately, I would really like to run full.

Thank you for any information that would allow me to write such code.

--
Brian Ford
Software Engineer
Vital Visual Simulation Systems
FlightSafety International
Phone: 314-551-8460
Fax:   314-551-8444


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

Brian Ford wrote:

> So does this mean that a "complete" implementation would have a
> "command/status MII interface" or that a "complete" implementation does
> not need one?

The complete implementation should have a real command/status MII
interface.


> Can you help point me to where I need to write code that would initialize
> the FCC to do full duplex,

I just checked in the newer driver that will properly do this on
the EST board with the National PHY in the CTX-1170 gizmo.  If for
some reason this doesn't work, just modify mii_discover_phy_poll() to
set 'phy_duplex' to '1'.

I checked this into linuxppc_2_5 BitKeeper on FSM Labs.  My 2.3/2.4
kernel is broken and I need to clone a new tree before I can make
changes.  That will take a little while......


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Brian Ford]

On Tue, 28 Nov 2000, Dan Malek wrote:

> I just checked in the newer driver that will properly do this on
> the EST board with the National PHY in the CTX-1170 gizmo.  If for
> some reason this doesn't work, just modify mii_discover_phy_poll() to
> set 'phy_duplex' to '1'.
>
Thank you very much.

> I checked this into linuxppc_2_5 BitKeeper on FSM Labs.  My 2.3/2.4
> kernel is broken and I need to clone a new tree before I can make
> changes.  That will take a little while......
>

Ok.  Do you know if there is anonymous bk access to the linuxppc_2_5
tree?  If so, on what port?  http://www.fsmlabs.com/linuxppcbk.html
doesn't mention it and I would like to play with it if possible.

As an off topic side note, wouldn't it make sense for there to be a common
MII library for all linux network drivers to share?  Is this not possible,
or has no one volunteered yet?

--
Brian Ford
Software Engineer
Vital Visual Simulation Systems
FlightSafety International
Phone: 314-551-8460
Fax:   314-551-8444


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

Brian Ford wrote:

> Ok.  Do you know if there is anonymous bk access to the linuxppc_2_5
> tree?  If so, on what port?

I think so, 5005 comes to mind.  Search the linuxppc-dev archives
for some references.

> As an off topic side note, wouldn't it make sense for there to be a common
> MII library for all linux network drivers to share?  Is this not possible,
> or has no one volunteered yet?

I don't ever remember such a discussion.  If it made sense, you would
see it in the generic network drivers, but it's not there.  Logically,
it makes sense, but you will notice the implementations are quite
different because of the hardware.  Even the 860 and 8260 are different.
The 860 has an interrupt driven piece of hardware, while the 8260 is
single threaded software bit shifting.  The support functions are quite
different because of this (which is why I am not using any of the 860
functions in the 8260 driver).  If you have time, take a look at it.
Just remember simplicity wins....I'm sure you can force some kind of
"common functions" if that is your goal, but if it's a complicated
spaghetti mess that only you understand it doesn't help :-).


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Brian Ford]

On Tue, 28 Nov 2000, Dan Malek wrote:

> I just checked in the newer driver that will properly do this on
> the EST board with the National PHY in the CTX-1170 gizmo.  If for
> some reason this doesn't work, just modify mii_discover_phy_poll() to
> set 'phy_duplex' to '1'.
>

I cracked open my CTX-1170 gizmo and it has a Lucent LU3X51FT-JE80 PHY in
it.  I manually set the phy_duplex to 1 and the phy_speed to 100, still
no dice.

Do you know what parts I need to modify in order to get the Lucent PHY
working too?  I am just looking for a few hints on what needs to be
done.  I'll be glad to try and do the "dirty" coding work.

Thanks.

--
Brian Ford
Software Engineer
Vital Visual Simulation Systems
FlightSafety International
Phone: 314-551-8460
Fax:   314-551-8444


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[diekema_jon]

> I cracked open my CTX-1170 gizmo and it has a Lucent LU3X51FT-JE80 PHY in
> it.  I manually set the phy_duplex to 1 and the phy_speed to 100, still
> no dice.

We have several of the CTX-1170 MII, and we have at least two
different PHY's in them.  One of the types is the Lucent
LU3X51FT-JE80.  I was unable to find a data sheet for the LU3X51FT
from http://www.lucent.com/micro, however I was able to find one on
the LU3X54FT.  The LU3X54-FT is a quad version of the LU3X51-FT.

The cep->phy_type variable in arch/ppc/8260_io/fec_enet.c gets derived
from the PHY identifer registers 1 / 2.  When I run with FEC2 active,
I get a type of 0x00437412.  I would expect this to look something
like 0x0180????.

#
# MPC8260 Communication Options
#
# CONFIG_SCC_ENET is not set
CONFIG_FEC_ENET=y
# CONFIG_FCC1_ENET is not set
CONFIG_FCC2_ENET=y
# CONFIG_FCC3_ENET is not set

MPC8260 CPU/CPM/BUS: 200/133/33 Mhz

fec: Phy @ 0x0, type 0x00437412


When Dan Malek runs with the National Semi PHY DP83840A, how fast
is the MPC8260 running?  I think that Dan is running at 166Mhz,
if so we might have a timing problem on I2C bus.


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Jerry Van Baren]

[oops, forgot to send it to the list.  Sorry for the duplicate Jon,
Brian, Dan]

...and the other Canary dongle had a chip in it labeled:

Enable
(c) 1998
5v Simple
3001AD1E2MJ1C
3440903 9934P

I have not tried to track down the manufacturer.

The primary point is that PHYs have different chip sets, even ones from
a single company and labeled with the same part number (maybe Canary is
the only company to do this, but somehow I doubt that :-).

The secondary point is that you should stick to standard MII registers
if at all possible.  According to 802.3 (1998) Section 22.2.4,
registers 0-1 are "Basic", 2-10 are "Extended", 11-14 are reserved, 15
("Extended Status") is "Basic", and 16-31 are vendor specific.

gvb


At 09:47 AM 11/30/00 -0500, diekema_jon wrote:

> > I cracked open my CTX-1170 gizmo and it has a Lucent LU3X51FT-JE80
> PHY in
> > it.  I manually set the phy_duplex to 1 and the phy_speed to 100, still
> > no dice.
>
>We have several of the CTX-1170 MII, and we have at least two
>different PHY's in them.  One of the types is the Lucent
>LU3X51FT-JE80.  I was unable to find a data sheet for the LU3X51FT
>from http://www.lucent.com/micro, however I was able to find one on
>the LU3X54FT.  The LU3X54-FT is a quad version of the LU3X51-FT.
>
>The cep->phy_type variable in arch/ppc/8260_io/fec_enet.c gets derived
>from the PHY identifer registers 1 / 2.  When I run with FEC2 active,
>I get a type of 0x00437412.  I would expect this to look something
>like 0x0180????.
>
>#
># MPC8260 Communication Options
>#
># CONFIG_SCC_ENET is not set
>CONFIG_FEC_ENET=y
># CONFIG_FCC1_ENET is not set
>CONFIG_FCC2_ENET=y
># CONFIG_FCC3_ENET is not set
>
>MPC8260 CPU/CPM/BUS: 200/133/33 Mhz
>
>fec: Phy @ 0x0, type 0x00437412
>
>
>When Dan Malek runs with the National Semi PHY DP83840A, how fast
>is the MPC8260 running?  I think that Dan is running at 166Mhz,
>if so we might have a timing problem on I2C bus.
>
>


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

diekema_jon wrote:

> ....  When I run with FEC2 active,
> I get a type of 0x00437412.  I would expect this to look something
> like 0x0180????.

Probably an MII timing problem.


> When Dan Malek runs with the National Semi PHY DP83840A, how fast
> is the MPC8260 running?

I run a variety of clock speed combinations depending upon
performance data customers request.

> ....  I think that Dan is running at 166Mhz,
> if so we might have a timing problem on I2C bus.

You mean MII control/status lines, I assume.  I have had lots of
trouble with this and am constantly changing the MII function in the
driver.  I have used about six different PHYs on a variety of boards,
none of them will respond the same.  Some seem very sensitive to the
clock duty cycle, and I may just have to write some function that
carefully provides the clock using the timebase registers and interrupts
disabled.  I don't want to do this because of the interrupt latencies
it could cause.  You can tune this function to work with one or two
PHYs, and when you try something else it just doesn't work.


	-- Dan

--

	I like MMUs because I don't have a real life.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: EST SBC8260 and Canary CTX-1170

[Gessner, Matt]

This is a Lucent LU3X31FT IIRC.
I actually wrote some code to initialize the PHY
to put it in full duplex mode.

> -----Original Message-----
> From: Jerry Van Baren [mailto:vanbaren_gerald@si.com]
> Sent: Thursday, November 30, 2000 11:04 AM
> To: linuxppc-embedded@lists.linuxppc.org
> Subject: Re: EST SBC8260 and Canary CTX-1170
>
>
>
> [oops, forgot to send it to the list.  Sorry for the duplicate Jon,
> Brian, Dan]
>
> ...and the other Canary dongle had a chip in it labeled:
>
> Enable
> (c) 1998
> 5v Simple
> 3001AD1E2MJ1C
> 3440903 9934P
>
> I have not tried to track down the manufacturer.
>
> The primary point is that PHYs have different chip sets, even
> ones from
> a single company and labeled with the same part number (maybe
> Canary is
> the only company to do this, but somehow I doubt that :-).
>
> The secondary point is that you should stick to standard MII registers
> if at all possible.  According to 802.3 (1998) Section 22.2.4,
> registers 0-1 are "Basic", 2-10 are "Extended", 11-14 are reserved, 15
> ("Extended Status") is "Basic", and 16-31 are vendor specific.
>
> gvb
>
>
> At 09:47 AM 11/30/00 -0500, diekema_jon wrote:
>
> > > I cracked open my CTX-1170 gizmo and it has a Lucent LU3X51FT-JE80
> > PHY in
> > > it.  I manually set the phy_duplex to 1 and the phy_speed
> to 100, still
> > > no dice.
> >
> >We have several of the CTX-1170 MII, and we have at least two
> >different PHY's in them.  One of the types is the Lucent
> >LU3X51FT-JE80.  I was unable to find a data sheet for the LU3X51FT
> >from http://www.lucent.com/micro, however I was able to find one on
> >the LU3X54FT.  The LU3X54-FT is a quad version of the LU3X51-FT.
> >
> >The cep->phy_type variable in arch/ppc/8260_io/fec_enet.c
> gets derived
> >from the PHY identifer registers 1 / 2.  When I run with FEC2 active,
> >I get a type of 0x00437412.  I would expect this to look something
> >like 0x0180????.
> >
> >#
> ># MPC8260 Communication Options
> >#
> ># CONFIG_SCC_ENET is not set
> >CONFIG_FEC_ENET=y
> ># CONFIG_FCC1_ENET is not set
> >CONFIG_FCC2_ENET=y
> ># CONFIG_FCC3_ENET is not set
> >
> >MPC8260 CPU/CPM/BUS: 200/133/33 Mhz
> >
> >fec: Phy @ 0x0, type 0x00437412
> >
> >
> >When Dan Malek runs with the National Semi PHY DP83840A, how fast
> >is the MPC8260 running?  I think that Dan is running at 166Mhz,
> >if so we might have a timing problem on I2C bus.
> >
> >
>
>

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

EST SBC8260 and Canary CTX-1170

[Oliver Brown]

I am having the same problem that was referenced in

http://lists.linuxppc.org/listarcs/linuxppc-embedded/200010/msg00167.htm
l

I can't generate any traffic on the fast ethernet port. There were many
suggestions in the above thread, but I was not sure what the final
solution was. Here what I'm using
-2.4.1 kernel
-Canary CTX-1170 with Lucent LU3X51FT Phy
-133 CPU and CPM clocks
-66 Bus clock

Ppcboot works fine on this setup, so I know that the hardware is
working.

Oliver Brown
Advent Networks

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

Oliver Brown wrote:

> I can't generate any traffic on the fast ethernet port.

The CTX-1170 worked at one time on the EST board.  Unfortunately,
I had to make little changes for just about every different 8260
board I have seen.  Further, my EST boards have all failed in some
way or another so I can't even test this anymore.

Even when it "worked", I had trouble with some 10/100 switches.

> Ppcboot works fine on this setup, so I know that the hardware is
> working.

Is this driver available from sourceforge, or is it something you have
locally?  I would like to look at it and I may see some differences.

Most of the changes surround the MDIO interface and the software
"clocking" of the bits on those lines.  The PHYs seem sensitive to
timing I just can't get right with the software.  I was able to get
one, or a small subset, working at the same time, but not all of them
at once.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[David Schleef]

On Thu, Mar 08, 2001 at 07:36:05PM -0500, Dan Malek wrote:
>
> Oliver Brown wrote:
>
> > I can't generate any traffic on the fast ethernet port.
>
> The CTX-1170 worked at one time on the EST board.  Unfortunately,
> I had to make little changes for just about every different 8260
> board I have seen.  Further, my EST boards have all failed in some
> way or another so I can't even test this anymore.
>
> Even when it "worked", I had trouble with some 10/100 switches.

I've noticed this as well.  In my case, the problem was
because the PHY was advertising and connecting full-duplex
and the CPM was doing half-duplex -- yet another problem due
to inadequate PHY communication.

Also, there is a problem with the 8260 CPM generating spurious
TX underrun errors and late collisions -- the driver does things
according to the spec, but there is an errata about this problem.
I'm about halfway through a fix for this.  The problem only
seems to occur on 10 Mbit hubs.




dave...


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

David Schleef wrote:

> .... but there is an errata about this problem.
> I'm about halfway through a fix for this.

That should have been checked in long ago....maybe I messed up
and only put it into one of the kernel trees.  I'll check into it.

My plan with the MII control/status was to utilize the new
configurable PHY support that several people contributed to
the MPC8xx FEC driver.  Two problems occurred.  One is that it
just took too long to develop, and the second is it relied on
the async/interrupt capability of the FEC.  It will get there
for the 8260...


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Dan Malek]

David Schleef wrote:

> I can confirm that the driver in the 2.4 tree does not have
> the workaround.

OK.  Sorry.  Is this from kernel.org or FSM Labs?


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[David Schleef]

On Thu, Mar 08, 2001 at 11:23:21PM -0500, Dan Malek wrote:
>
> David Schleef wrote:
>
> > .... but there is an errata about this problem.
> > I'm about halfway through a fix for this.
>
> That should have been checked in long ago....maybe I messed up
> and only put it into one of the kernel trees.  I'll check into it.


I can confirm that the driver in the 2.4 tree does not have
the workaround.




dave...


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[David Schleef]

On Thu, Mar 08, 2001 at 11:54:52PM -0500, Dan Malek wrote:
> David Schleef wrote:
>
> > I can confirm that the driver in the 2.4 tree does not have
> > the workaround.
>
> OK.  Sorry.  Is this from kernel.org or FSM Labs?


Last night's BK patch to 2.4.3-pre3.  Silly of me not to
mention.




dave...


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: EST SBC8260 and Canary CTX-1170

[Oliver Brown]

The fcc driver that was working for ppcboot was in version 0.8.2,
however the 0.8.3 will have the fixes for the EST SBC8260.

The 2.4.1 kernel I am using came from kernel.org. I didn't realize that
there were updates in Bitkeeper tree. I'll try the latest Bitkeeper
kernel.

Regards,
Oliver

-----Original Message-----
From: Dan Malek [mailto:dan@mvista.com]
Sent: Thursday, March 08, 2001 6:36 PM
To: Oliver Brown
Cc: linuxppc-embedded@lists.linuxppc.org
Subject: Re: EST SBC8260 and Canary CTX-1170


Oliver Brown wrote:

> I can't generate any traffic on the fast ethernet port.

The CTX-1170 worked at one time on the EST board.  Unfortunately,
I had to make little changes for just about every different 8260
board I have seen.  Further, my EST boards have all failed in some
way or another so I can't even test this anymore.

Even when it "worked", I had trouble with some 10/100 switches.

> Ppcboot works fine on this setup, so I know that the hardware is
> working.

Is this driver available from sourceforge, or is it something you have
locally?  I would like to look at it and I may see some differences.

Most of the changes surround the MDIO interface and the software
"clocking" of the bits on those lines.  The PHYs seem sensitive to
timing I just can't get right with the software.  I was able to get
one, or a small subset, working at the same time, but not all of them
at once.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: EST SBC8260 and Canary CTX-1170

[Wolfgang Denk]

In message <CD5CA14300A9D54A8D13D58CD3C87B422DD72B@advcorp-exc01.advent.com>
Oliver Brown wrote:
>
> The fcc driver that was working for ppcboot was in version 0.8.2,
> however the 0.8.3 will have the fixes for the EST SBC8260.

s/will have/has/

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
Every time history repeats itself the price goes up.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/