Hi,
fortunately we where able to capture the following message from Proc32:
Version: 425.L Feb 15 2001 Technical Support: +1 (613) 591-0941
Proc fault 1, ldt 100 /boot/sys/Proc32; fault e+0
cs:eip=5:10568 ss:esp=d:f7c0f84 efl=12246 ds=d es=d fs=0 gs=0
eax/0 ebx/f7c0fd0 ecx/3822 edx/14 esi/5991 edi/3820 ebp/f7c0fb4
Stack (d:f7c0f84)
00000020 00015c35 000287d0 000161e0 00018319 000028f0 00000000 00000000
00000320 0000bf23 00028920 000287d0 0f7c0fe8 00000000 00005991 0000001f
00003822 0f7c0fd0 00005836 00003820 0000000d 00000610 00003820 0000000d
000001d2 00000001 00005979 00005991 0f7c0959 0000060c 000059f4 00000000
Process Entry (addr b050)
00000000 00000001 00000000 00000001 00000000 00000000 30020207 00001e1e
0000586c 0100000d 0000b108 ffffffff 00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000501 000d0005 00007a58 00000000 00000002
00000038 00000000 0000019c 0000bdc0 00000000 00000000 00010880 00000000
00000000 00000000 00000000 ffff0001 00000000 00000000 00000000
After giving this message the system hungs except for the ISR’s. Coul’d someone explain this? How is the Process Entry to be interpreted?
TIA
Peter Stoeckigt
STN Atlas Marine Electronics GmbH
Dipl.-Ing. Peter Stoeckigt
Department MST21
Bremen, Germany
Phone: +49 (421) 457-1190
Fax: +49 421 457 2026
mailto:peter.stoeckigt@sam-electronics.de
http://www.sam-electronics.de
this is in the vc_attach code of Proc … the local side of the connect.
we have had no other reports of problems with that section of kernel code …
is this repeatable? under what conditions does it occur?
Peter St??ckigt <stoeckigt.p@stn-atlas.de> wrote:
Hi,
fortunately we where able to capture the following message from Proc32:
Version: 425.L Feb 15 2001 Technical Support: +1 (613) 591-0941
Proc fault 1, ldt 100 /boot/sys/Proc32; fault e+0
cs:eip=5:10568 ss:esp=d:f7c0f84 efl=12246 ds=d es=d fs=0 gs=0
eax/0 ebx/f7c0fd0 ecx/3822 edx/14 esi/5991 edi/3820 ebp/f7c0fb4
Stack (d:f7c0f84)
00000020 00015c35 000287d0 000161e0 00018319 000028f0 00000000 00000000
00000320 0000bf23 00028920 000287d0 0f7c0fe8 00000000 00005991 0000001f
00003822 0f7c0fd0 00005836 00003820 0000000d 00000610 00003820 0000000d
000001d2 00000001 00005979 00005991 0f7c0959 0000060c 000059f4 00000000
Process Entry (addr b050)
00000000 00000001 00000000 00000001 00000000 00000000 30020207 00001e1e
0000586c 0100000d 0000b108 ffffffff 00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000501 000d0005 00007a58 00000000 00000002
00000038 00000000 0000019c 0000bdc0 00000000 00000000 00010880 00000000
00000000 00000000 00000000 ffff0001 00000000 00000000 00000000
After giving this message the system hungs except for the ISR’s. Coul’d someone explain this? How is the Process Entry to be interpreted?
TIA
Peter Stoeckigt
STN Atlas Marine Electronics GmbH
Dipl.-Ing. Peter Stoeckigt
Department MST21
Bremen, Germany
Phone: +49 (421) 457-1190
Fax: +49 421 457 2026
mailto:> peter.stoeckigt@sam-electronics.de
http://www.sam-electronics.de
–
Randy Martin randy@qnx.com
Manager of FAE Group, North America
QNX Software Systems www.qnx.com
175 Terence Matthews Crescent, Kanata, Ontario, Canada K2M 1W8
Tel: 613-591-0931 Fax: 613-591-3579
Yes, it is repeatable. The conditions are as follows (and admittedly a little bit tricky):
We play with tree nodes connected via two lan’s. Node 1 and 2 are connected via Net.ether1000 and node 2 and 3 are connected via Net.fd. On node 1 we run +ACI-on -n2 on -n3 ls+ACI-. This leads immediately to the fault.
On the other hand running +ACI-on -n3 ls+ACI- on node 2 works perfectly.
The server for Net.fd is a combination of two processes able to transport the Net.fd messages via a CAN (Controller Area Network) bus. It is surely possible that there are errors in CAN-Driver even if they work so far. So it would help a lot to now the exact condition that leads to the Proc fault.
Peter
STN Atlas Marine Electronics GmbH
Dipl.-Ing. Peter Stoeckigt
Department MST21
Bremen, Germany
Phone: ±49 (421) 457-1190
Fax: ±49 421 457 2026
mailto:peter.stoeckigt+AEA-sam-electronics.de
http://www.sam-electronics.de
Peter St+APY-ckigt <stoeckigt.p@stn-atlas.de> wrote:
Yes, it is repeatable. The conditions are as follows (and admittedly a little bit tricky):
We play with tree nodes connected via two lan’s. Node 1 and 2 are connected via Net.ether1000 and node 2 and 3 are connected via Net.fd. On node 1 we run +ACI-on -n2 on -n3 ls+ACI-. This leads immediately to the fault.
On the other hand running +ACI-on -n3 ls+ACI- on node 2 works perfectly.
The server for Net.fd is a combination of two processes able to transport the Net.fd messages via a CAN (Controller Area Network) bus. It is surely possible that there are errors in CAN-Driver even if they work so far. So it would help a lot to now the exact condition that leads to the Proc fault.
from the info posted i have a question…
do you use messaging of any data sizes larger than 64K? there is a potential
in there for a problem with Net.fd/Proc in that condition.
please check this. make sure all messages are < 64K
when you setup your Net.fd link over CAN, if you try just a simple
sin -n3 from node2 does this succeed?
Peter
STN Atlas Marine Electronics GmbH
Dipl.-Ing. Peter Stoeckigt
Department MST21
Bremen, Germany
Phone: ±49 (421) 457-1190
Fax: ±49 421 457 2026
mailto:peter.stoeckigt+AEA-sam-electronics.de
http://www.sam-electronics.de
–
Randy Martin randy@qnx.com
Manager of FAE Group, North America
QNX Software Systems www.qnx.com
175 Terence Matthews Crescent, Kanata, Ontario, Canada K2M 1W8
Tel: 613-591-0931 Fax: 613-591-3579
Randy Martin +ADw-randy+AEA-qnx.com+AD4- wrote in message news:9lc7ri+ACQ-7q4+ACQ-1+AEA-nntp.qnx.com…
+AD4- Peter St+APY-ckigt +ADw-stoeckigt.p+AEA-stn-atlas.de+AD4- wrote:
+AD4- +AD4- Yes, it is repeatable. The conditions are as follows (and admittedly a little bit tricky):
+AD4-
+AD4- +AD4- We play with tree nodes connected via two lan’s. Node 1 and 2 are connected via Net.ether1000 and node 2 and 3 are connected via Net.fd. On node 1 we run +ACI-on -n2 on -n3 ls+ACI-. This leads immediately to the fault.
+AD4-
+AD4- +AD4- On the other hand running +ACI-on -n3 ls+ACI- on node 2 works perfectly.
+AD4-
+AD4- +AD4- The server for Net.fd is a combination of two processes able to transport the Net.fd messages via a CAN (Controller Area Network) bus. It is surely possible that there are errors in CAN-Driver even if they work so far. So it would help a lot to now the exact condition that leads to the Proc fault.
+AD4-
+AD4- from the info posted i have a question…
+AD4-
+AD4- do you use messaging of any data sizes larger than 64K? there is a potential
+AD4- in there for a problem with Net.fd/Proc in that condition.
+AD4- please check this. make sure all messages are +ADw- 64K
No, we don’t use that large messages. Our CAN bus runs at only 31250 bits/sec, so we have short messages and proxies (Isn’t message passing limited to 64KByte anyway?).
+AD4-
+AD4- when you setup your Net.fd link over CAN, if you try just a simple
+AD4- sin -n3 from node2 does this succeed?
Sure. And many more commands suporting the -n option, This is why we like it.
Also running cat // 3/etc/config/sysinit.3 works (slow, but…).
+AD4-
+AD4-
+AD4- +AD4- Peter
+AD4-
+AD4- +AD4- STN Atlas Marine Electronics GmbH
+AD4- +AD4- Dipl.-Ing. Peter Stoeckigt
+AD4- +AD4- Department MST21
+AD4- +AD4- Bremen, Germany
+AD4- +AD4- Phone: ±49 (421) 457-1190
+AD4- +AD4- Fax: ±49 421 457 2026
+AD4- +AD4- mailto:peter.stoeckigt+AEA-sam-electronics.de
+AD4- +AD4- http://www.sam-electronics.de
+AD4-
+AD4-
+AD4- –
+AD4- Randy Martin randy+AEA-qnx.com
+AD4- Manager of FAE Group, North America
+AD4- QNX Software Systems www.qnx.com
+AD4- 175 Terence Matthews Crescent, Kanata, Ontario, Canada K2M 1W8
+AD4- Tel: 613-591-0931 Fax: 613-591-3579
+AD4-
please stay tuned for more info as we investigate this. it sounds like you
have isolated this to the point where you know how to cause it, so you can
avoid it which is good.
as with any kernel modification this will take time to fix and prove… do
you have the ability to work around this problem in the shorter term?
it sounds like you do…
Peter St+APY-ckigt <stoeckigt.p@stn-atlas.de> wrote:
Randy Martin +ADw-randy+AEA-qnx.com+AD4- wrote in message news:9lc7ri+ACQ-7q4+ACQ-1+AEA-nntp.qnx.com…
+AD4- Peter St+APY-ckigt +ADw-stoeckigt.p+AEA-stn-atlas.de+AD4- wrote:
+AD4- +AD4- Yes, it is repeatable. The conditions are as follows (and admittedly a little bit tricky):
+AD4-
+AD4- +AD4- We play with tree nodes connected via two lan’s. Node 1 and 2 are connected via Net.ether1000 and node 2 and 3 are connected via Net.fd. On node 1 we run +ACI-on -n2 on -n3 ls+ACI-. This leads immediately to the fault.
+AD4-
+AD4- +AD4- On the other hand running +ACI-on -n3 ls+ACI- on node 2 works perfectly.
+AD4-
+AD4- +AD4- The server for Net.fd is a combination of two processes able to transport the Net.fd messages via a CAN (Controller Area Network) bus. It is surely possible that there are errors in CAN-Driver even if they work so far. So it would help a lot to now the exact condition that leads to the Proc fault.
+AD4-
+AD4- from the info posted i have a question…
+AD4-
+AD4- do you use messaging of any data sizes larger than 64K? there is a potential
+AD4- in there for a problem with Net.fd/Proc in that condition.
+AD4- please check this. make sure all messages are +ADw- 64K
No, we don’t use that large messages. Our CAN bus runs at only 31250 bits/sec, so we have short messages and proxies (Isn’t message passing limited to 64KByte anyway?).
+AD4-
+AD4- when you setup your Net.fd link over CAN, if you try just a simple
+AD4- sin -n3 from node2 does this succeed?
Sure. And many more commands suporting the -n option, This is why we like it.
Also running cat // 3/etc/config/sysinit.3 works (slow, but…).
+AD4-
+AD4-
+AD4- +AD4- Peter
+AD4-
+AD4- +AD4- STN Atlas Marine Electronics GmbH
+AD4- +AD4- Dipl.-Ing. Peter Stoeckigt
+AD4- +AD4- Department MST21
+AD4- +AD4- Bremen, Germany
+AD4- +AD4- Phone: ±49 (421) 457-1190
+AD4- +AD4- Fax: ±49 421 457 2026
+AD4- +AD4- mailto:peter.stoeckigt+AEA-sam-electronics.de
+AD4- +AD4- > http://www.sam-electronics.de
+AD4-
+AD4-
+AD4- –
+AD4- Randy Martin randy+AEA-qnx.com
+AD4- Manager of FAE Group, North America
+AD4- QNX Software Systems > www.qnx.com
+AD4- 175 Terence Matthews Crescent, Kanata, Ontario, Canada K2M 1W8
+AD4- Tel: 613-591-0931 Fax: 613-591-3579
+AD4-
–
Randy Martin randy@qnx.com
Manager of FAE Group, North America
QNX Software Systems www.qnx.com
175 Terence Matthews Crescent, Kanata, Ontario, Canada K2M 1W8
Tel: 613-591-0931 Fax: 613-591-3579