klioufis@decus.com.au (02/05/91)
hi,
I am experiencing a couple of problems with DSV11''s that i
could really use some help with.
The first one seems to be a bug in the SJDRIVER from the WAN
DRIVER KIT 1.1a. The problem presents itself in the SJDRIVER when the
LAPB protocol is used. So here goes....
/*
* Set an AST on the device
*/
dsv_desc->input_flg = NO
status =SYS$QIOW(0, /* THE EVENT FLAG */
dsv_desc->channel, /* THE CHANNEL */
IO$_SETCHAR|IO$M_CTRL|IO$M_ATTNAST /* Attention ast*/
&iosb, /* I/O STATUS BLOCK */
0,0, /* NO AST COMPLETION */
attention[fd % MAXIMUM_DSV11] /* A Unsolicited event
handler */
0, /* Ignored */
psl$c_user,0,0,0);
if (!STATUS_BAD(status))
status = iosb.cc;
if (STATUS_BAD(STATUS))
{
PutError(_FATAL_, 0, "ReadDSV - %s SYS$QIOW-
IO$_SETCHAR|IO$M_CRTL|IO$M_ATTNAST SS$_%s\n",dsv_desc->devname,ssname(status));
return (-1);
}
return (0);
It looks like when the attention AST completes VMS dies...(note that
both reading and writing work OK) The WAN driver is 1.1a and VMS is 5.3. What
follows is the entry in the error log of the crash....
**************************** ENTRY 652. **********************************
ERROR SEQUENCE 2138. LOGGED ON: SID 08000000
DATE/TIME 25-JAN-1991 12:04:20.05 SYS_TYPE 01010000
SCS NODE VAX/VMS V5.3
NON FATAL BUGCHECK KA630
BADDALRQSZ, Bad memory deallocation request size or address
PROCESS NAME PAUL
PROCESS ID 0002002E
ERROR PC 81910C4
ERROR PSL 00020000
INTERRUPT PRIORITY LEVEL = 02
PREVIOUS MODE = KERNEL
CURRENT MODE = KERNEL
STACK POINTERS
KSP 7FFE775C ESP 7FFE9800 SSP 7FFECA4C USP 7FF35DA0 ISP 80419200
GENERAL REGISTERS
R0 802A1504 R1 0000150C R2 00000000 R3 803A99B0 R4 802DB530
R5 803A99B0 R6 802A1504 R7 00000000 R8 8028C310 R9 00000030
R10 00000071 R11 OCO42000 AP 7FF35DF8 FP 7FFE77E4 SP 7FFE77A0
APPROX TODR 1C610D9F
VAX/VMS SYSTEM ERROR REPORT COMPILED 4-FEB-1991 09:27
PAGE 9.
*************************** ENTRY 653. ************************************
ERROR SEQUENCE 2139. LOGGED ON: SID 08000000
DATE/TIME 25-JAN-1991 12:04:20.66 SYS_TYPE 01010000
SCS NODE: VAX/VMS 5.3
FATAL BUGCHECK KA630
INVEXCEPTN, Exception while above ASTDEL or on onterrupt stack
process name NULL
PROCESS ID 00010000
ERROR PC 8014DA32
ERROR PSL 04080009
C-BIT
N-BIT
INTERUPPT PRIORITY LEVEL = 08
PREVIOUS MODE = KERNEL
CURRENT MODE = KERNEL
INTERRUPT STACK
STACK POINTERS
KSP 7FFE7800 ESP 7FFE9800 SSP 7FFED800 USP 7FF96A88 ISP 804190FC
GENERAL REGISTERS
RO 00000008 R1 04080000 R2 802A25EB R3 801811A0 R4 80197A00
R5 802A13F0 R6 802DD6A0 R7 00000001 R8 802BC038 R9 0011A713
R10 0011A715 R11 802A25EB AP 7FF96AC8 FP 7FF96A8C SP 80419164
SYSTEM REGISTERS
POBR 80933E00
P0 PTE BASE (VIRT ADDRS)
POLR 00000485
TOTAL P0 PAGES
P1BR 8016E800
P1 PTE BASE (VIRT ADDRS)
P1LR 001FFCA2
TOTAL NON EXISTENT P1 PAGES
SBR 00FDEE00
SYSTEM PTE BASE (PHYS ADDRS)
SLR 00007F80
TOTAL PAGES "SYSTEM" VIRT MEM
PCBB 00A13020
PCB BASE (PHYS ADDRS)
SCBB 00FDA800
SCB BASE (PHYS ADDRS)
ASTLVL 00000004
NO AST'S PENDING
SISR 00000000
INTERRUPT REQUEST ACTIVE = 0.
ICCS 00000040
INTERRUPT ENABLE
Has anyone struck this problem?, if so, how was it overcome?
The other problem is with the same version of WAN driver and VMS..
We have two uVAX III's connected via a 64k circuit using decnet
over DDCMP. The default buffer size is 1498... the two machines are next
to each other, and the cct is provided by a V35 NULL MODEM.
Every few days (sometimes weeks) this line will go down, and upon investigation
the cause is that the dsv11 SJA0: AND SJA1: are offline (sho dev), ncp shows
the circuit as syncing... This failure occurs on both machines..
any assistance is welcome....
thanks and regards
Arthur Klioufis
MIS Manager
Pont Data sydney, Australia
note.
please respond via news system or forward mail to
DECUS MAIL ACCOUNT KLIOUFIS
CHEERS...
warwick@wikkit.enet.dec.com (02/12/91)
hi, > I am experiencing a couple of problems with DSV11''s that i > could really use some help with. The first one seems to be a > bug in the SJDRIVER from the WAN DRIVER KIT 1.1a. The > problem presents itself in the SJDRIVER when the LAPB > protocol is used. So here goes.... > I have recently seen the same program submitted via the UK Customer Support Centre, exhibiting the same bug. We have found a problem in the V1.1A SJDRIVER (also present in SLDRIVER and ZSDRIVER) with the processing of the IO$M_NOW modifier on IO$_READxBLK requests. There are two workarounds: a) When the startup qio is sent to the driver, request more than 7 receive buffers. This work around is not guaranteed since the bug will take place in the following circumstances: - If an io$m_now read is queued to the driver and there are currently less than 7 receive command blocks queued to the device. Obviously, predicting whether the device has 7 cmd's down on the board is a bit of a lottery - I would recommend option b. b) Don't use the io$m_now modifier. The UK customer who reported the problem went for option a), and is no longer seeing the crashes. Although I must stress this is not a guaranteed workaround - there is still a window where the problem can happen. We plan to fix this bug in the next release of VAX WAN Device Drivers. We can also supply a fix kit for you, if you contact your local customer support centre. > The other problem is with the same version of WAN driver and > VMS.. We have two uVAX III's connected via a 64k circuit > using decnet over DDCMP. The default buffer size is 1498... > the two machines are next to each other, and the cct is > provided by a V35 NULL MODEM. > Every few days (sometimes weeks) this line will go down, and > upon investigation the cause is that the dsv11 SJA0: AND > SJA1: are offline (sho dev), ncp shows the circuit as > syncing... This failure occurs on both machines.. > The most likely cause of this is that the DSV11's loadable firmware has crashed. Please contact your local customer support centre, and they will help you work with us to find the problem. Trevor Warwick -------------- Telecommunications and Networks Engineering, Digital Equipment Corporation, Reading, England. Internet: warwick@marvin.enet.dec.com warwick%marvin.enet@decwrl.dec.com PSImail: 234284400135::warwick "the opinions expressed herein do not necessarily reflect the views or opinions of Digital Equipment Corporation"