mace:transmit timeout - resetting

mace:transmit timeout - resetting

[Patrick Berge]

 I have am receiving "mace:transmit timeout - resetting" sporadically.
It occurs while running VNC ( http://www.uk.research.att.com/vnc) .
Ping, telnet and even IP Masquerading are working fine on all machines.
I have one Power Mac with PPCLinux (R5-Kernel 2.2.6-15apmac) , an Intel
box (Red Hat 6.0 - Kernel-2.2.5-15) which is the gateway  and an AMD
box (RedHat 5.1- Kernel 2.0.34).    I only see this error on PPC Linux
console when transmitting to the Intel box  with either Linux or Windows
95 (Sorry it came with PC and used it to see if it was just Red Hat 6.0
causing my problem).  I am betting I need to increase the timeout in
the  /usr/src/linux/drivers/net/mace.c file? Although I don not
understand why, since the connection to the AMD box is fine.  I swapped
the ethernet cards (except the built-in) /cables and hub ports and
always get the  same problem.  I even disabled IP Masquerading think
that was the problem ... but I still had the problem.

 I am not sure but I think the line I need to change is in
/usr/src/linux/drivers/net/mace.c file shown below? Not sure what to
change it to?

 #define TX_TIMEOUT HZ /* 1 second */

Any ideas about my problem and/or the meaning of the HZ would be
helpful.

Patrick Berge


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: mace:transmit timeout - resetting

[Paul Mackerras]

Patrick Berge <pberge@stny.rr.com> wrote:

>  I have am receiving "mace:transmit timeout - resetting" sporadically.

But do you see any problems with the network apart from the message?

The reason for the message is that there are some bugs in the mace
chip which cause it to lock up very occasionally.  When the driver
detects this condition, it prints this message and tries to reset the
chip.  If your network keeps working, then it means that the driver
successfully got the mace chip going again.

> Any ideas about my problem and/or the meaning of the HZ would be
> helpful.

HZ is the number of timer interrupts per second.

Paul.

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: mace:transmit timeout - resetting

[Patrick Berge]

Paul Mackerras wrote:

>
> The reason for the message is that there are some bugs in the mace
> chip which cause it to lock up very occasionally.  When the driver
> detects this condition, it prints this message and tries to reset the
> chip.  If your network keeps working, then it means that the driver
> successfully got the mace chip going again.
>
> > Any ideas about my problem and/or the meaning of the HZ would be
> > helpful.
>
> HZ is the number of timer interrupts per second.
>
> Paul.

Oh No... yea it still works but is very very slow so bad I can not even use
VNC !   Well this is miserable!  Thank you very much for the quick
response.  Looks like I will be buying a PCI ethernet card for my PowerMac
7200.  I just got it donated to me so guess I can't complain too much.
Does Apple know about the problem since mace is used with the built-in
ethernet?  I am sure this is asking too much but I will try ...will Apple
replace it?

I would still like to get a software fix for my problem? If possible?  Will
increasing the timeout value improve the problem?  It would if the software
detection of the timeout is bogus sometimes.  If the software detection of
this problem is always detecting the hardware hang then increasing the
timeout will just make it slower.  Any ideas ?

Also what company made the mace chip ... I am hardware designer myself and
I am very frustrated to see companies ship buggy hardware!  I work on
hardware for UNIX machines that must be 24x7 and  blazingly fast.  It is a
shame PC customers put up with buggy hardware/software!  (That is why
Windows/MacOS are still around.)  I guess rebooting your machine because of
lock ups or the blue screen death is acceptable... yea but the UNIX
customers do pay lots of bucks.  Sorry I had to vent my frustrations.

Patrick Berge


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: mace:transmit timeout - resetting

[Patrick Berge]

Paul Mackerras wrote:

> Patrick Berge <pberge@stny.rr.com> wrote:
>
> >  I have am receiving "mace:transmit timeout - resetting" sporadically.
>
> But do you see any problems with the network apart from the message?
>
> The reason for the message is that there are some bugs in the mace
> chip which cause it to lock up very occasionally.  When the driver
> detects this condition, it prints this message and tries to reset the
> chip.  If your network keeps working, then it means that the driver
> successfully got the mace chip going again.
>
> > Any ideas about my problem and/or the meaning of the HZ would be
> > helpful.
>
> HZ is the number of timer interrupts per second.
>
> Paul.

Thanks for the info.  For the record I have a PowerPC 7200/120 and was
wondering if any other extended ethernet users have this same problem.  The
network works but is incredibly slow do to the resetting every second.   I
really don't think the network ever was completely hung...  just very very
slow ..like 10,100,1000 times slower than all the other non-mace
machines.   I talked to Apple (they just told me the only known problem is
with certain 7200/90 machines (see
http://til.info.apple.com/techinfo.nsf/artnum/n18944) .   I guess this is
the issue Keith mentioned in a previous post.  I ask them to review there
database and search for any ethernet issues for 7200/120 and they claimed
there are no known ethernet problems.

I guess my question to you, Paul, is what do I need to do to the mace.c
file to turn off the resetting of the chip.  I am just guessing that the
reset condition you looked for does not hang the mace chip in the
7200/120.   If so then removing the reset should fix my problem.    I am
again assuming the the mace chip in my 7200/120 hardware does not hang.
If I remove the resetting and it hangs then well I just have to but a PCI
ethernet card.



[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]