| Author |
TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 19 2017 01:31 |
|
|
Hi,
When I start TCPIP I got this:
...
TCPIP>
**** OpenVMS Alpha Operating System V8.4 - BUGCHECK ****
** Bugcheck code = 000001CC: INVEXCEPTN, Exception while above ASTDEL
DEC PW 433au with defected UTP port (BNC is working with FreeBSD).
TIA, |
|
|
| Author |
RE: TCPIP bugcheck |
malmberg
Moderator
Posts: 530
Joined: 15.04.08 |
| Posted on October 19 2017 18:34 |
|
|
I would check the sysgen parameters to see if you are short of any resources.
https://sourceforge.net/p/vms-ports/wiki/SystemTuning/
If you have an unterminated BNC ethernet connection it can generate a high interrupt rate, which might cause crashes if the driver is short of resources. Not sure though.
!
! Site Preference settings
!--------------------------
TTY_DEFCHAR2 = 143366 ! autobaud + hanguup + edit + insert + discon
TTY_TIMEOUT = 3600 * 16 ! Allow 8 hours for reconnects
TTY_ALTYPAHD = 2048 ! Large buffer for Packet Modem
!
!
min_procsectcnt = 200
min_maxprocesscnt = 150
min_gblpagfil = 600000
min_gblpages = 10060000
min_gblsections = 1400
min_channelcnt = 4096
!
min_gh_exec_code = 1536
min_gh_exec_data = 448
min_gh_res_code = 1584
!
min_npagedyn = 8300000
!
min_wsmax=250000
!
min_maxbuf = 8200
!
!
! RMS improvements
!-------------------
MIN_RMS_DFMBFSDK = 2
MIN_RMS_DFMBFSMT = 2
MIN_RMS_DFMBFSUR = 2
MIN_RMS_DFMBFREL = 2
MIN_RMS_DFMBFIDX = 9
!
!
pagefile = 0
swapfile = 0
dumpfile = 0
savedump = 1
|
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 20 2017 00:01 |
|
|
BNC is not unterminated. There is one terminator both end of the coax cable.
I think the problem is hardware related because I've got a VMS supported Digital PCI VGA card which gives the same problem if I push it in a 32 bit PCI slot instead of the 64 bit slot. The card is 32 bit ... and VMS is v7.11-H2. |
|
|
| Author |
RE: TCPIP bugcheck |
malmberg
Moderator
Posts: 530
Joined: 15.04.08 |
| Posted on October 20 2017 03:12 |
|
|
OpenVMS can be sensitive to what PCI cards are in which slots. There should be a manual for your hardware that tells you what options are supported and what slot restrictions there are.
With higher end PCI configurations, there can be multiple PCI "controllers" on the motherboard, and some PCI slots can share the same "controller". An older card on a controller can force all PCI slots on the controller to the slower bandwidth/speed.
|
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 21 2017 00:40 |
|
|
OK.
I'll test my install removing all IDE and other cards and check again booting from SCSI disk.
I don't think so it will solve the problem.
I'm waiting for a new network PHY card from eBay putting in working state UTP again.
I can accept your hardware related advices but I don't understand FreeBSD 6.2 is working perfectly on the same hw with more cards! I think FreeBSD 6.2 (2007?) is older than OpenVMS 8.4 (2010) |
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 26 2017 20:05 |
|
|
Today test:
The coaxial cable was removed from the BNC connector and network was replaced a dummy T + two pices terminator. The TCPIP was started and no bugcheck, yet.
There is something strange. Look at this:
TCPIP> sh int/full
Interface: LO0
IP_Addr: 127.0.0.1 NETWRK: 255.0.0.0 BRDCST:
MTU: 4096
Flags: UP LOOP NOARP MCAST SMPX
RECEIVE SEND
Packets 0 0
Errors 0 0
Collisions: 0
Interface: WE0
IP_Addr: 192.168.52.198 NETWRK: 255.255.255.0 BRDCST: 192.168.52.255
Ethernet_Addr: 00-00-00-00-00-00 MTU: 1500
Flags: UP BRDCST MCAST SMPX
RECEIVE SEND
Packets 0 0
Errors 0 0
Collisions: 0
TCPIP>
MAC of WE0 is invalid! (full 00). Is that correct? I don't think so. :-(
$ sh dev ewa0/full
Device EWA0:, device type DE500, is online, network device, error logging is
enabled, device is a template only.
Error count 0 Operations completed 0
Owner process "" Owner UIC [SYSTEM]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G,W
Reference count 0 Default buffer size 512
Operating characteristics: Link down.
Speed (Mbits/sec) 10
Def. MAC addr 00-00-F8-76-61-8C Current MAC addr 00-00-F8-76-61-8C
I'm waiting for a "new" UTP+BNC panel from the eBay and I hope UTP will work soon.
I'll check my DECserver 90TL+ with its BNC connector because it is so strange for me the TX and RX LEDs are always blinking on the same time on the AUI/BNC media converter.
TIA, |
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 26 2017 23:02 |
|
|
|
A little bit interesting for me that OpenVMS 7.11-H2 works with TCPIP and LAT on the same hardware. |
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 28 2017 06:54 |
|
|
Today I installed VMS and TCPIP patch.
The Bugcheck disappeared but:
TCPIP> sh arp
TCPIP> ping 192.168.52.6
PING 192.168.52.6 (192.168.52.6): 56 data bytes
ping: sendto: network is down
ping: wrote 192.168.52.6 64 chars, ret=-1
ping: sendto: network is down
ping: wrote 192.168.52.6 64 chars, ret=-1
ping: sendto: network is down
ping: wrote 192.168.52.6 64 chars, ret=-1
Cancel
----192.168.52.6 PING Statistics----
3 packets transmitted, 0 packets received, 100% packet loss
%SYSTEM-F-TIMEOUT, device timeout
TCPIP> sh route
DYNAMIC
Type Destination Gateway
AN 0.0.0.0 192.168.52.254
AH 127.0.0.1 127.0.0.1
AN 192.168.52.0/24 192.168.52.198
AH 192.168.52.198 192.168.52.198
TCPIP> sh int we0/full
Interface: WE0
IP_Addr: 192.168.52.198 NETWRK: 255.255.255.0 BRDCST: 192.168.52.255
Ethernet_Addr: 00-00-00-00-00-00 MTU: 1500
Flags: UP BRDCST MCAST SMPX
RECEIVE SEND
Packets 994 0
Errors 0 0
Collisions: 0
$ sh dev
Device Device Error Volume Free Trans Mnt
Name Status Count Label Blocks Count Cnt
ATEST$DKA0: Online 0
ATEST$DKA200: Mounted 0 ALPHASYS 138425040 373 1
ATEST$DQA0: Online 0
ATEST$DQA1: Online 0
ATEST$DQB0: Online 0
ATEST$DQB1: Online 0
ATEST$DVA0: Online 0
Device Device Error
Name Status Count
...
MPA0: Online 0
EWA0: Online 0
EWA2: Online 0
EWA3: Online 0
EWA4: Online 0
EWA5: Online 0
EWA6: Online 0
GYA0: Online 0
INA0: Offline 0
IMA0: Offline 0
IKA0: Offline 0
WSA0: Offline 0
WSA1: Online 0
PPP0: Online 0
$
Why network is down??? :-O |
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on October 30 2017 20:35 |
|
|
Today I got the new Ethernet media module/card from eBay.
Under 7.1-1H2 it is working with a 3Com small Office Connect HUB with UTP+BNC+AUI. (I just tested UTP, not BNC. BNC will be tested later.)
OpenVMS 8.4 said: LinkDown. :-( I think that DE500 driver is full off bugs. :-(
|
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on November 07 2017 06:06 |
|
|
Can I get or download v7.3 iso from somewere?
I hope it works on my Alpha with networking.
TIA, |
|
|
| Author |
RE: TCPIP bugcheck |
malmberg
Moderator
Posts: 530
Joined: 15.04.08 |
| Posted on November 08 2017 03:14 |
|
|
OpenVMS Engineering only tests with specific models/revisions of hardware that is listed on the SPD (Software Product Description) .
In many cases the HP/HPE/Compaq/DEC branded card is slightly different than the a commodity card that may have a similar part number. OpenVMS may not be able to detect that a commodity card has been substituted and so will report it as a known device.
Some of the differences may be hardware/software fixes to the specific hardware that is sold for VMS that are not in the commodity part.
An Open Source volunteer generally does not have access to get a hardware/firmware issue fixed, and needs to work around the issue in the driver, hence you can find cases were BSD/Linux works on specific hardware and VMS does not.
The mass market hardware is primarily tested for Microsoft Windows, and secondarily for Linux.
OpenVMS because of its design to use ASTs for non-blocking I/O also is able run cards at higher stress levels than the typical application. OpenVMS Engineering generally stress tests I/O hardware at the maximum speed the hardware can obtain. This stress testing often finds defects in commodity hardware. While the Vendor may fix the issue for the card resold by DEC/Compaq/HP/HPE, they may or may not decide to fix the issue in the Windows/Linux version of the cards.
A thinkwire or thinwire ethernet connection is half duplex. Only UTP can be full duplex.
You may want to ask on comp.os.vms newsgroup on recommendations on Ethernet hardware known to work with OpenVMS instead of trying different generic cards.
Also the sysgen parameter INTSTKPAGES (Spelling) may need to be increased based on what hardware you have on your system. I recall seeing references to it about TCP/IP somewhere. |
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on November 09 2017 06:54 |
|
|
OK, I understand everything.
I changed the media card for an other one. UTP is working under 7.1-1H2. It is true the speed is 10Mbps/half duplex. I don't know why. I tried to use SRM command ewa0_media FastFD for example but no 100Mbps/FD.
OpenVMS 8.4 is not able to use the same hardware. Link down was said by LANCP for both UTP and BNC media. I think Digital PWS 433 au is supported by VMS so I think this LinkDown is a bug in the driver (?). |
|
|
| Author |
RE: TCPIP bugcheck |
abrsvc
Member
Posts: 108
Joined: 12.03.10 |
| Posted on November 09 2017 10:43 |
|
|
These systems were configured with EITHER a 10mbps or 100mbps MAU. If you can not get to 100Mbps, I suspect yours has the 10Mbps MAU. Even so, I don't recall OpenVMS dropping support for these models although it might. I will research this a bit more. These are small systems that would run poorly with the latest versions of OpenVMS anyway. Unless there is a specific feature in the newer version that you require, V7.1-1h1 should work just fine.
Dan |
|
|
| Author |
RE: TCPIP bugcheck |
malmberg
Moderator
Posts: 530
Joined: 15.04.08 |
| Posted on November 09 2017 13:05 |
|
|
It very well may be a bug, but with out a support contract there is no way to get any ECO kits not supplied with the hobbyist distribution.
If that is not a specifically supported card, then it is possible that the driver is now using a feature that is not supplied by that model of card.
|
|
|
| Author |
RE: TCPIP bugcheck |
rattila
Member
Posts: 56
Location: Budapest
Joined: 11.08.17 |
| Posted on November 10 2017 05:56 |
|
|
I know the card is DE500. Chip is Tulip 21143. I think it is well supported and the machine is a well known DEC PWS 433au.
That's why I'm searching for a maybe working OpenVMS 7.3 install kit. I think 8.4 is for Itanium machines because they are newer ones. |
|
|
| Author |
RE: TCPIP bugcheck |
malmberg
Moderator
Posts: 530
Joined: 15.04.08 |
| Posted on November 10 2017 12:34 |
|
|
I would suggest posting on comp.os.vms (Free accounts from eternal-september.org) about what works and what does not work.
Someone may have some idea of what is going on.
OpenVMS 8.4 is definitely for Alpha, and has a lot of fixes in it, particularly on the CRTL.
|
|
|
