Compaq Wide Area Networking for Tru64 UNIX Release Notes August 2000 These Release Notes apply to the Compaq Wide Area Networking for Tru64 UNIX , Version 3.1A product. You should read these Release Notes before attempting to install the product. Operating System: Compaq Tru64 UNIX, V5.0, 5.0A, 5.1 Software Version: Compaq Wide Area Networking for Tru64 UNIX , Version 3.1A Compaq Computer Corporation PROPRIETARY INFORMATION Furnished for Field Test Purpose Only The information contained herein is furnished in confidence and is subject to the terms and conditions of a License Agreement for field testing Compaq Software. Compaq Computer Corporation Houston, Texas © 2000 Compaq Computer Corporation COMPAQ, Tru64 UNIX, the Compaq logo, and the DIGITAL logo Registered in U.S. Patent and Trademark Office. DECnet and DECnet-Plus are trademarks of Compaq Information Technologies Group, L.P. Motif, OSF/1, UNIX, and the "X" device are trademarks of The Open Group. All other product names mentioned herein may be the trademarks or registered trademarks of their respective companies. Confidential computer software. Valid license from Compaq or authorized sublicensor required for possession, use, or copy- ing. Consistent with FAR 12.211 and 12.212, Commercial Computer Software, Computer Software Documentation, and Technical Data for Commercial Items are licensed to the U.S. Government under vendor's standard commercial license. Compaq shall not be liable for technical or editorial errors or omissions contained herein. The information in this publication is subject to change without notice and is provided "AS IS" WITHOUT WARRANTY OF ANY KIND. THE ENTIRE RISK ARISING OUT OF THE USE OF THIS INFORMATION REMAINS WITH THE RECIPIENT. IN NO EVENT SHALL COMPAQ BE LIABLE FOR ANY DIRECT, CONSEQUENTIAL, INCIDENTAL, SPECIAL, PUNITIVE, OR OTHER DAMAGES WHATSOEVER (INCLUDING WITHOUT LIMITATION, DAMAGES FOR LOSS OF BUSINESS PROFITS, BUSINESS INTERRUPTION, OR LOSS OF BUSINESS INFORMATION), EVEN IF COMPAQ HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. THE FOREGOING SHALL APPLY REGARDLESS OF THE NEGLIGENCE OR OTHER FAULT OF EITHER PARTY AND REGARDLESS OF WHETHER SUCH LIABILITY SOUNDS IN CONTRACT, NEGLIGENCE, TORT, OR ANY OTHER THEORY OF LEGAL LIABILITY, AND NOTWITHSTANDING ANY FAILURE OF ESSENTIAL PURPOSE OF ANY LIMITED REMEDY. The limited warranties for Compaq products are exclusively set forth in the documentation accompanying such products. Nothing herein should be construed as constituting a further or additional warranty. This document was prepared using VAX DOCUMENT Version 2.1. Contents PREFACE vii CHAPTER 1 INSTALLATION INSTRUCTIONS 1-1 1.1 INSTALLING THE KIT 1-1 CHAPTER 2 RESOLVED PROBLEMS 2-1 2.1 API 2-1 2.1.1 Facilities Data Incorrectly Set Throughput Value 2-1 2.1.2 X25Encode() Was Not Calculating the Size of List Parameters Correctly 2-1 2.1.3 Multithreaded Processes Linked with x25lib_r.a Aborted with a Segmentation Fault 2-2 2.1.4 The NCL Command disable x25 access application app Caused x25applicationd to Exit 2-2 2.1.5 x25applicationd Not Handling Termination of Children 2-2 2.2 ISDN 2-3 2.2.1 ISDN Clear Command Exited Early 2-3 2.2.2 ISDN Call Connect Daemon Dumped Core During rlogin 2-3 2.2.3 ISDN Call Connect Daemon Dumped Core When Started from the Command Line 2-3 2.2.4 ISDN Call Connect Daemon Dumped Core During rlogin Over ISDN Line 2-3 2.2.5 ISDN Call Connect Daemon Dumped Core When Started from the Command Line 2-4 2.3 WAN DEVICE DRIVERS 2-4 2.3.1 PBXDP Card Lost Data During High Utilization 2-4 2.3.2 PBXDP Synch Card Experienced Transmit Underruns 2-4 2.3.3 PBXDP Synch Driver Hangs EV6 SMP Systems 2-4 2.4 PBXDP SYNCH CARD RUNNING AT 3/4 SPEED WHEN CLOCK SET TO INTERNAL 2-5 2.4.1 X.25 Recognized Unsupported EMULEX Cards During Reboot 2-5 2.4.2 Problems with PBXDP Synchronous Communications Card 2-5 2.4.3 PBXDP Synchronous Communications Card Driver Did Not Handle Spurious Interrupts 2-6 iii Contents 2.5 X.25 MANAGEMENT 2-6 2.5.1 Repeated Enabling/Disabling of X.25 and LLC2 Entities Could Result in Kernel Memory Faults 2-6 2.5.2 X25Decode() Occasionally Returned a Bad Status 2-6 2.5.3 Adding LLC2 Poll Timer Functionality 2-7 2.5.4 wdd_datalinks Lost LAPB Configuration at System Reboot 2-7 2.5.5 x25config Created Duplicate Lines in the Startup File 2-7 2.5.6 x25config Crashed When Modifying Configuration to Add IP Over X.25 2-7 2.6 WAN DATALINKS 2-8 2.6.1 X.25 Halted Datalink When a DM Was Received in the Running State 2-8 2.7 X.25 2-8 2.7.1 X.25 Access Not Notifying Clients of Lost Link 2-8 2.7.2 Connections Configured Over X.25 Hung Due to a Lack of Credits 2-8 2.7.3 Added Dynamic Support for Accepting the D-bit 2-8 2.7.4 AccProcessWaitingServiceQ Thread Is Preempted, Halting Data Transmission 2-9 2.8 CTF 2-9 2.8.1 CTF list Command Sometimes Resulted in CTF Display Problems 2-9 2.8.2 CTF Trace File Included Nonprinting Characters 2-9 2.8.3 CTF stop Command Deleted the Last File of a Detached Trace Session 2-10 CHAPTER 3 KNOWN PROBLEMS 3-1 3.1 INSTALLATION 3-1 3.1.1 DMS Not Supported 3-1 3.1.2 Installation Subset Interdependencies 3-1 3.2 CONFIGURATION 3-1 3.2.1 Configurator (wansetup) May Crash When Configuring Relay PVCs 3-1 3.3 SOFTWARE 3-2 3.3.1 Application Programmer Interface 3-2 3.3.1.1 Example Program call_sender.c Causes PSDN Network to Reject an X.25 Call Initiated by the Program 3-2 iv Contents 3.3.2 X.25 Management 3-2 3.3.2.1 NCL SHOW Commands May Initiate Core Dumps 3-2 3.3.2.2 Modifying the x25startup.ncl File Results in Unpredictable Behavior 3-2 3.3.3 X.25 Mail 3-3 3.3.3.1 Received Messages 3-3 3.3.3.2 Configuring X.25 Mail 3-3 3.3.3.3 Error Messages in mail.log When Sending and Receiving X.25 Mail 3-3 3.3.4 WAN Device Drivers 3-3 3.3.4.1 PCI Configuration 3-3 3.3.4.2 PBXDI Line Drivers 3-4 3.3.4.3 Creating NCL Management Entities for Users of the WAN API 3-4 3.3.4.4 DSY and DNSES Line Drivers 3-4 3.3.4.5 Modems and Modem Eliminators 3-5 3.3.4.6 Scripts That Need /vmunix to Be Active 3-6 3.3.5 Common Trace Facility 3-6 3.3.5.1 Installing CTF After X.25 3-6 3.3.5.2 CTF Crashes After Several Hours of Live Tracing 3-6 3.3.5.3 Multiple Traces of "x25 client" on Gateways Not Supported 3-7 3.3.5.4 Time Displayed Inaccurately 3-7 3.3.5.5 CTF Remote Trace Over TCP 3-7 3.3.5.6 CTF Cannot be Paused for a Long Time 3-7 3.3.5.7 Tracing on the Local Node While the Node Name is Specified and DECnet is Used as Transport 3-8 3.3.5.8 Option "notruncate" in CTF Does Not Work Without "full" Option 3-8 3.3.6 ISDN 3-8 3.3.6.1 Starting IP/ISDN When Kernel Rebuild Was Required 3-8 3.3.6.2 Delay When Configuring IP/ISDN Hosts 3-9 3.3.6.3 Lockmode 4 Not Supported for ISDN 3-9 3.3.6.4 X.25 Clear Confirm Packets Sometimes Lost in X.25 over ISDN 3-9 3.3.6.5 Lockup While Attempting X.25 Call Over ISDN 3-9 CHAPTER 4 DOCUMENTATION 4-1 4.1 CLARIFICATION OF BEHAVIOR OF X25DECODE 4-1 4.2 AVAILABLE X.25 DOCUMENTATION 4-1 4.3 LOCATION OF DOCUMENTATION 4-2 v ___________________________________________________________________ Preface ___________________________________________________________________ Intended Audience This manual is intended for all Compaq Wide Area Networking for Tru64 UNIX  (WAN) software users. Read this manual before you install, upgrade, or use Version 3.1A of Compaq Wide Area Networking for Tru64 UNIX . ___________________________________________________________________ Structure of This Book This manual contains the following chapters: o Chapter 1 contains information on installation instructions. o Chapter 2 contains information about the problems that have been found and corrected since the release of WAN Support for DIGITAL UNIX Systems, Version 3.0. o Chapter 3 contains information that you should be aware of and possible problems that may arise during installation, configuration, and operation of the product. Where appro- priate, solutions to those problems, or workarounds, are provided. o Chapter 4 details the documentation available for this version of the Compaq Wide Area Networking for Tru64 UNIX  product. ___________________________________________________________________ Conventions The following conventions are used in this book. vii Preface _______________________________________________________________ Convention_______Meaning_______________________________________ UPPERCASE and The Compaq Tru64 UNIX operating system dif- lowercase ferentiates between lowercase and uppercase characters. Literal strings that appear in text, examples, syntax descriptions, and function descriptions must be typed exactly as shown. special type Indicates a literal example of system output or user input. In text, this typeface indi- cates the exact name of a function, command, option, node, pathname, directory, or file. italic type Indicates a variable or a variable portion of a command. # A number sign is the default superuser prompt. Ctrl/x Indicates that you press the Control key while you press the key noted by x. _________Indicates_that_you_press_the_Return_key.______ viii Chapter 1 Installation Instructions This chapter provides notes about installing the Version 3.1A release of Compaq Wide Area Networking for Tru64 UNIX . NOTE Compaq Wide Area Networking for Tru64 UNIX , Version 3.1A is for Compaq Tru64 UNIX 5.0, 5.0A, and 5.1. To verify that you are running the correct version of Tru64 UNIX, enter the following command: uname -a Verify that the returned system information contains the strings V5.x and alpha. 1.1 Installing the Kit Install this kit with the setld command by logging into the root account and typing the following at the shell prompt: 1. Verify that the Version 5.0A-1 release of the Compaq DECnet- Plus for Tru64 UNIX software is installed. If it is not, install it. Note: The Compaq Wide Area Networking for Tru64 UNIX , Version 3.1A software requires that at least the subsets DNAKBIN501 and DNANETMAN501 be installed. 2. Remove the existing subsets: # setld -d `setld -i \ | grep installed \ | awk '/^CTA|^ISD|^WDA|^XXA|^ZZA/ {printf ("%s ",$1)}'` 3. Reboot the system. Installation Instructions 1-1 4. cd to the directory containing the Version 3.1A kit. 5. Load the Version 3.1A kit. # setld -l . 6. Build a new kernel using the following command: # /usr/sbin/wansetup KERNEL 7. Reboot the system. The kit location may be a tape drive, or a disk directory that contains the Version 3.1A kit. 1-2 Installation Instructions Chapter 2 Resolved Problems This chapter discusses the problems that have been resolved since the release of WAN Support for DIGITAL UNIX Systems, Version 3.0. 2.1 API 2.1.1 Facilities Data Incorrectly Set Throughput Value The function that handled the throughput request was treat- ing throughput classes of 64,000 and 48,000 the same way. The result was that when a throughput class of 64,000 bits per second was specified in the X.25 API function X25Encode() the facilities data was indicating a throughput class of 48,000. Source: CFS.37264 2.1.2 X25Encode() Was Not Calculating the Size of List Parameters Correctly X25Encode() was not calculate the size of list parame- ters correctly. As a result, when a list parameter such as X25I_ETEDELAY was put in an encoded parameter list (EPL) by X25Encode() the last argument was omitted (when the ETE delay list has 2 or more parameters). This problem has been cor- rected. Source: CFS.51473 Resolved Problems 2-1 2.1.3 Multithreaded Processes Linked with x25lib_r.a Aborted with a Segmentation Fault X25ClosePort() closed a socket and then continued process- ing using the closed socket's file descriptor to access data structures and deallocate memory internal to X.25's library. If another thread within the same process called X25OpenPort(), it was possible for both threads to use the same file descriptor at the same time. This provided an opportunity for the thread that was opening the port to access internal data structures that the thread that was closing the port was in the process of deallocating. Under this condition, the system responded with a segmentation fault. The stack trace showed the segmentation fault occurring during a call to X25TestState(). The solution was to modify X25ClosePort() to close the socket after it is done using the file descriptor. This ensures that the file de- scriptor will not be given to another thread until X.25 has completed all processing associated with the file descriptor. Source: CFS.50840 2.1.4 The NCL Command disable x25 access application app Caused x25applicationd to Exit The NCL command disable x25 access application app caused x25applicationd to exit. The system call that was reading the message from NCL was exiting with an error as the result of an interrupt. The error checking code did not attempt to recover; instead, an error was logged and the program exited. If the operator attempted to use NCL to enable the X25 Access appli- cation app, an error message was displayed indicating that no resource was available. Source: CFS.61970 2.1.5 x25applicationd Not Handling Termination of Children The function that was handling the SIGCHLD signal was occasion- ally interrupted before it completed processing. As a result, x25applicationd's child processes remained on the system in the defunct state. If allowed to continue for a period of time, X.25 stopped functioning. Source: CFS.61972 2-2 Resolved Problems 2.2 ISDN 2.2.1 ISDN Clear Command Exited Early isdnclear uses the open() system call to open the call controls maintenance channel. That call was being interrupted. The code did not include logic to handle interrupted system calls. The code has been modified to recall open() when it's return value indicates that it is returning as the result of an interrupt. Source: CFS.69775 2.2.2 ISDN Call Connect Daemon Dumped Core During rlogin When an rlogin to a host's alias was attempted on a line that was set up as IP over ISDN, the ISDN Call Connect Daemon crashed. The code that was performing the alias lookup was us- ing the wrong field from a structure causing isdnccd to crash. The code has been changed to use the correct field. Source: Laboratory test environment 2.2.3 ISDN Call Connect Daemon Dumped Core When Started from the Command Line When isdnccd was started from the command line it crashed. The code was setting up a data structure and it never returned the pointer to the caller. This caused a crash. The code has been modified to return the pointer to the caller. Source: Laboratory test environment 2.2.4 ISDN Call Connect Daemon Dumped Core During rlogin Over ISDN Line When an rlogin to a host's alias was attempted on a line that was set up as IP over ISDN, the ISDN Call Connect Daemon crashed. The code that was performing the alias lookup was us- ing the wrong field from a structure causing isdnccd to crash. The code has been changed to use the correct field. Source: Laboratory test environment Resolved Problems 2-3 2.2.5 ISDN Call Connect Daemon Dumped Core When Started from the Command Line When isdnccd was started from the command line, it crashed. The code was setting up a data structure and never returning the pointer to the caller, which resulted in the crash. The code has been modified to return the pointer to the caller. Source: Laboratory test environment 2.3 WAN Device Drivers 2.3.1 PBXDP Card Lost Data During High Utilization The interrupt service routine that is called when the card completes DMA was occasionally being called prior to DMA com- pletion. The logic assumed that the function was called only after DMA had completed and that it could reclaim at least one of the transmit buffers. If the routine was entered when DMA had not completed and if all of the buffers where in use, the entire transmit buffer chain was reclaimed resulting in lost data. The function was modified to check to see if all of the buffers are in use before reclaiming any buffers. If they are, the reclaim step is skipped. Source: CFS.68583 2.3.2 PBXDP Synch Card Experienced Transmit Underruns When the PBXDP synch card was used without the X.25 protocol, the driver occasionally reported transmit underruns. The un- derruns could be observed in the Oerrs column of the output generated by netstat -i for the pwa? device. The driver was not initializing the card to the settings that were recommended by the manufacturer. The code has been modified to initialize the card to the most efficient settings. Source: CFS.70893 2.3.3 PBXDP Synch Driver Hangs EV6 SMP Systems When the NCL disable lapb link * command was entered, the sys- tem usually hung. If it did not hang there, it hung when the enable command was entered. The code was not configuring the Local Arbitration Register correctly on the PBXDP card. The code has been modified to configure the register as described in the errata sheets issued by PLX for the 9060SD chip. Source: DS20/ES40 PBXDP qualification 2-4 Resolved Problems 2.4 PBXDP Synch Card Running at 3/4 Speed When Clock Set to Internal The wrong clock frequency was used when setting the internal clock for the PBXDP synch card. As a result, the data transfer rate was only 3/4 of the rate set in NCL. Source: CFS.61490 2.4.1 X.25 Recognized Unsupported EMULEX Cards During Reboot X.25 was configuring all of the EMULEX X-Series Sync/Async Non-Intelligent PCI Boards. This resulted in a conflict with other device drivers that use PCI cards from EMULEX. The code was modified so that only the EMULEX cards that are supported by X.25 are configured. The /etc/sysconfigtab file should have entries for Vendor_Id 0x10df with Device_ Ids 0x1133, 0x1155, and 0x1156. Device_Ids 0x1131, 0x1132, 0x1151, 0x1152, 0x1153, and 0x1154 are not supported by X.25. Source: CFS.53662 2.4.2 Problems with PBXDP Synchronous Communications Card There were a number of problems seen with the PBXDP PCI syn- chronous communications card. These have been rectified in the wdd_pwa.mod image. Testing revealed the following problems: o Interrupt on Transmit Underrun Bit On lines with full modem control, the "interrupt on transmit underrun" bit was not getting set. The driver was informed of the underrun and did not clear the condition. (This bit was already set on lines with no modem control.) Due to this condition, transmission stalled until the user of the line issued a "driver abort". o Crash During Buffer Check When performing consistency checks on buffer addresses, the driver issued a panic call (which crashed the system) if the check failed. The check was failing on large memory systems (e.g. > 2GB). o Tracing Facility Not Called The driver was not calling the Common Trace Facility (CTF) to trace transmitted and received frames. Resolved Problems 2-5 Source: AlphaServer 4100 PBXDP Option Qualification Testing CFS.56455 (for the panic call problem) 2.4.3 PBXDP Synchronous Communications Card Driver Did Not Handle Spurious Interrupts The code that handled the interrupt from the card assumed that there was at least one DMA transmit buffer that could be re- claimed when the interrupt was received. If an interrupt was received when all of the buffers were full the logic resulted in all of the buffers being reclaimed. As a result, data was lost. Source: CFS.68583 2.5 X.25 Management 2.5.1 Repeated Enabling/Disabling of X.25 and LLC2 Entities Could Result in Kernel Memory Faults The problem occurred when the enable llc2 link command was issued in NCL after disabling the LLC2 link. The code that processes the disable command, deallocated memory that was created for the link, but it did not reset a pointer. When the user attempted to enable the link, the pointer was used to access the deallocated memory resulting in a system panic. The change corrects the code so that the pointer is reset to NULL after the memory that it accesses is deallocated. Source: CFS.52803 2.5.2 X25Decode() Occasionally Returned a Bad Status When multiple X.25 calls were made within a process, the X25ReadAccept() function eventually returned an encoded pa- rameter list (EPL) without a send buffer size parameter. When the X25Decode() function was called to decode the EPL, it re- turned an X25RC_BADEPL error. The code was not initializing a local variable prior to using it. The code has been modified to initialize the local variable. Source: CFS.54754 2-6 Resolved Problems 2.5.3 Adding LLC2 Poll Timer Functionality If the LAN was disconnected, there was no mechanism for datalinks to detect and report the lost connection. The code was modified to restore logic that implements functionality that provides a poll timer, reject timer, and busy timer. Source: CFS.59445 2.5.4 wdd_datalinks Lost LAPB Configuration at System Reboot When configuring a pwa sync card into a system using wansetup the wdd_datalinks configured properly. After a reboot, wdd_ datalinks did not configure lapb. The only way to get LAPB back was to run wansetup again. The messages file contained the following line: date node vmunix: wdd_datalinks: hdlc configured The line expected would be: date node vmunix: wdd_datalinks: hdlc lapb configured The code was redesigned to work the way the Tru64 UNIX device driver manual recommends. Source: CFS.55161 2.5.5 x25config Created Duplicate Lines in the Startup File Running wansetup to modify a configuration that included X.25 over IP resulted in duplicate lines being added to the X.25 startup file (/sbin/rc3.d/X28.50x25). The duplicate lines showed up between the #X25IP_CONFIG_START and #X25IP_CONFIG_ END tags in the file. Every time that wansetup was executed, more duplicate lines were added. The duplicate lines resulted in errors being generated at X.25 startup. The code was modi- fied so that the IP information is not duplicated in the X.25 startup file every time that the configuration is modified. Source: CFS.58129 2.5.6 x25config Crashed When Modifying Configuration to Add IP Over X.25 If an existing configuration was modified to add more than one IP over X.25 circuits, the configuration program generated a segmentation fault and crashed while exiting. The code was attempting to deallocate static memory. The crash only occurred when multiple IP over X.25 circuits were added. The code that was deallocating the static memory has been removed. Resolved Problems 2-7 Source: CFS.58131 2.6 WAN Datalinks 2.6.1 X.25 Halted Datalink When a DM Was Received in the Running State If a DM was received while in the running state, X.25 responded by sending a DM and then halted the datalink. The code was changed to reinitialize in response to receiving a DM while in the running state. Source: CFS.54326 2.7 X.25 2.7.1 X.25 Access Not Notifying Clients of Lost Link The logic that notifies clients that there is no link was in- cluded in the section of code that is processed when a lost link is reestablished. The logic was moved to the section of code that detects that the link has been lost so that the client is notified when the link is lost. Source: CFS.65356 2.7.2 Connections Configured Over X.25 Hung Due to a Lack of Credits Code in X.25 Access was releasing the SMP lock prior to calling assert_wait_mesg_head(). This opened a timing window where preemptions (or another CPU on an SMP system) could insert an entry on the queue and do the wakeup, causing the job that X.25 Access was scheduling to be missed. The code has been modified to hold the lock until after the assert call. Source: CFS.67412 2.7.3 Added Dynamic Support for Accepting the D-bit D-bit processing is now configurable. To configure D-bit use the following command: sysconfig -r x25_access accept-D-bit-1. This can also be set in the /etc/sysconfigtab file. Source: QAR.62130 2-8 Resolved Problems 2.7.4 AccProcessWaitingServiceQ Thread Is Preempted, Halting Data Transmission The code that scheduled that AccProcessWaitingServiceQ was dropping the SMP lock prior to calling the assert wait. This opened a window where a preemption could insert an entry on the queue, do the wake-up, and cause the AccProcessWaitingServiceQ request to be missed. Source: CFS.67412 2.8 CTF 2.8.1 CTF list Command Sometimes Resulted in CTF Display Problems When the CTF list command was issued on a single CPU system with a large number of tracepoints, CTF would occasionally hang for 30 seconds, display a partial list of tracepoints, and then display the "Timeout waiting for SHOW_TRACEPOINT_RESPONSE" mes- sage. The CTF daemon was writing a message that was too large for the socket buffer that is used to transmit data between the daemon and the CTF sink. As a result the write() call would oc- casionally return with an EWOULDBLOCK error. After 30 seconds, the sink timed out and displayed a the portion of the message that was received and the error message indicating that the message was incomplete. The daemon was modified to handle the EWOULDBLOCK error by resending the message to ensure that the sink gets the complete message. Source: CFS.68700 2.8.2 CTF Trace File Included Nonprinting Characters When CTF was used to perform detached tracing with the de- fault file name, the file that was written to disk contained nonprintable ASCII characters. CTF responded with the message: Trace file ctf_trace.dat^C.000 created. A listing of the directory showed the trace file as ctf_ trace.dat?.0000. The correct file name is ctf_trace.dat.0000. When CTF performed the initialization of the default file name, the file name was being copied incorrectly. The code has been modified to copy the file name correctly. Source: CFS.71866 Resolved Problems 2-9 2.8.3 CTF stop Command Deleted the Last File of a Detached Trace Session When the CTF stop command was issued while performing detached tracing, the last file in the sequence was removed causing trace data to be lost. The code was not passing an argument to the function that handles the last write to the file resulting in unpredictable behavior. The code has been modified to pass the correct value so that the last file is no longer deleted. Source: CFS.74388 2-10 Resolved Problems Chapter 3 Known Problems 3.1 Installation 3.1.1 DMS Not Supported The Compaq Wide Area Networking for Tru64 UNIX  product is not supported for installation in a dataless environment (DMS). 3.1.2 Installation Subset Interdependencies If you do not intend to install all the subsets provided in the kit, refer to Appendix B of the WAN Support for DIGITAL UNIX Systems Installation Guide for details of the subset interde- pendencies. You should do this before invoking the installation script. 3.2 Configuration 3.2.1 Configurator (wansetup) May Crash When Configuring Relay PVCs After you add a Relay PVC and modify it, if you attempt to add another Relay PVC, the Configurator may crash. In addition, adding more than one Relay PVC in a single session of wansetup may corrupt the PVC list. To avoid this problem, add one Relay PVC, then save the config- uration and exit. Repeat this operation for as many Relay PVCs as you need to configure. Known Problems 3-1 3.3 Software This section is divided into the following major subsections: o Application Programming Interface: refer to Section 2.3.1 o X.25 Management: refer to Section 2.3.2 o X.25 Mail: refer to Section 2.3.3 o WAN Device Drivers: refer to Section 2.3.4 o Common Trace Facility: refer to Section 2.3.5 o ISDN: refer to Section 2.3.6 3.3.1 Application Programmer Interface 3.3.1.1 Example Program call_sender.c Causes PSDN Network to Reject an X.25 Call Initiated by the Program When the example program is used over a PSDN, the call is cleared with a C=03 D-41. There is no workaround. Source: QAR.00082 3.3.2 X.25 Management 3.3.2.1 NCL SHOW Commands May Initiate Core Dumps Issuing such NCL SHOW commands as the following may initiate NCL core dumps: ncl> show x25 access template * all To avoid this, specify the names of the attributes to be dis- played rather than specifying ALL. 3.3.2.2 Modifying the x25startup.ncl File Results in Unpredictable Behavior This file is generated by the configuration program as de- scribed in Chapter 2 of the WAN Support for DIGITAL UNIX Systems Configuration Guide. The configuration program pro- vides extensive error checking to ensure that the resultant NCL commands create a valid X.25 configuration. If you modify the resultant NCL script file or if you later use NCL commands to create new X.25 entities, you run the risk of creating an unworkable X.25 configuration. NCL can only check for the le- gality of the requested operation not the larger question of 3-2 Known Problems whether the requested operation makes sense given the current X.25 configuration. Therefore, Compaq encourages you to per- form all configuration tasks using the configuration program included with the X.25 software. Source: QAR.00079 3.3.3 X.25 Mail 3.3.3.1 Received Messages Five blank lines are prepended to messages received by the x25mail utility. There is no workaround. 3.3.3.2 Configuring X.25 Mail You must specify a template when configuring X.25 Mail. If this is not done, the following message appears in the mail.log file: /var/adm/syslog.dated/22-Jul-14:29/mail.log < Jul 22 15:45:32 pardon Master Mailer[734]: Error: MakeCall Specified template does not exist 3.3.3.3 Error Messages in mail.log When Sending and Receiving X.25 Mail When sending and receiving X.25 Mail, the following error mes- sage may appear in the mail.log file: /var/adm/syslog.dated/September 2, 1999/mail.log Apr 3 11:24:12 alfons Slave Mailer[1609]: X25ReadData Error: Operation failed - call cleared This message does not represent a failure to transfer the mail. 3.3.4 WAN Device Drivers 3.3.4.1 PCI Configuration With certain combinations of I/O adapters including PBXDPs, the console firmware on some systems (particularly AlphaServer 4100 and AlphaServer 4000 systems) will be unable to complete PCI configuration. This is due to an otherwise harmless bug in the PCI interface chip on the PBXDP. The bug results in the firmware thinking the PBXDP is making multiple requests for huge blocks of PCI I/O address space. Known Problems 3-3 It is recommended that any system using a PBXDP upgrade the firmware to the latest appropriate version of firmware that is applicable for the operating system and hardware being used. It should be noted that the version of the firmware CD is not the same as the version number of all the different console firmware images contained on the CD. 3.3.4.2 PBXDI Line Drivers Using PBXDI Cards on EISA Buses The use of the PBXDI card on EISA bus systems is not supported at this time. Reported Interface Types Due to limitations in the hardware of the PBXDI card, it is not possible to distinguish between EIA- 232-D (V.24) and V.35 cables. Both these types of cables are reported by the modem connect entity as V.35 interfaces. 3.3.4.3 Creating NCL Management Entities for Users of the WAN API The network management entities necessary to establish a configuration for use with the Direct Driver Application Programming Interface to the WAN Drivers are created when the wddsetup utility is run. This utility is either run directly when the WAN Drivers and Datalinks subsets are installed or as part of the wansetup utility when other parts of the X.25 product are installed. Programmers wishing to use the HDLC and LAPB Application Programming Interface to the WAN Drivers and Datalinks will have to manually configure the HDLC and LAPB management en- tities. The HDLC and LAPB entities are usually created when the wansetup utility is run to build an X.25 configuration. WAN Datalinks programmers should read the WAN Support for DIGITAL UNIX Systems Configuration Guide and the WAN Support for DIGITAL UNIX Systems Management Guide to become familiar with these entities. 3.3.4.4 DSY and DNSES Line Drivers DSY and DNSES Line Drivers Require Device Unit to be Enabled The DSY and DNSES line drivers will not function properly unless the device unit entity is present and in the Running state. To find out whether the device unit entity for a par- ticular device is in the correct state, enter the following NCL command: show device unit state 3-4 Known Problems If it is not in the correct state, enter the following NCL commands: ncl> create device unit name ncl> enable device unit If you enter a large number of NCL commands while the device unit is not running, some commands (for example, enable lapb link) may fail with the error message No Resource Available. NCL Command Sequences Care is required when issuing disable, set, and enable NCL command sequences on DSY Modem Connect Line entities, or on LAPB/HDLC entities that use DSY Modem Connect Line entities as the physical line. If multiple instances of these commands are issued in quick succession, for example from a script, the DSY driver will exhaust its command buffer pool and halt operation. A firmware reload is required to bring the device out of this state. Boot-Time Hanging If a line on the DSY is connected to a mo- dem that does not terminate the TestMode, RingIndicate, or SpeedSelect lines correctly, the system can hang on a reboot with the message intr: stray interrupt vector=2048, ipl=4 displayed repeatedly. The solution to this problem is to halt the machine, and set the BOOT_RESET console environment variable to ON using the following command: >>> setenv BOOT_RESET ON The machine can now be booted safely, and the problem will not show up in future reboots. 3.3.4.5 Modems and Modem Eliminators Interchange Circuits Some types of modems and modem elimina- tors may leave interchange circuits, such as Ring Indicator (CCITT 125) and Test Indicator (CCITT 142), unterminated. This condition will allow these signals to float and cause con- tinual device interrupts, with a corresponding loss of system performance. You should verify that signals are correctly terminated, and rectify this situation if they are not. Local and Remote Loopback Tests Information sheet EK-BS19D- IS-001 outlines conditions under which the adapter connector 12-27591-01 should be used. It should be noted that if this adapter connector is used then automatic local and remote loop- back tests can no longer be performed. Known Problems 3-5 Retraining (Loss of Clear To Send) If the modem drops CTS (CCITT 106) for a period greater than the value of the Maximum Enable Transmit Timer attribute of the Modem Connect Line en- tity (which has a default of 2 seconds), Modem Control will be cycled by dropping DTR (CCITT 108.2) and RTS (CCITT 105). If this situation causes problems, the value of the Maximum Enable Transmit Timer attribute should be increased. Modem Connect Management If a Modem Connect line is deleted, then the Modem Connect data ports must also be deleted before reusing the associated communications port. The data port can be deleted by disabling the Modem Connect Client. 3.3.4.6 Scripts That Need /vmunix to Be Active The following scripts depend on netstat -i and cannot be run if the active kernel is something other than /vmunix: setld -v WDABASE310 setld -d XXAIP310 cons_setup 3.3.5 Common Trace Facility 3.3.5.1 Installing CTF After X.25 If CTF is installed following an X.25 installation, then the following messages appear: To complete your installation of DEC CTF for DIGITAL UNIX, please perform the following step: -- Configure CTF by executing the command: # /usr/sbin/ctfsetup Instead of executing /usr/sbin/ctfsetup, you should execute the following command: # /usr/sbin/wansetup KERNEL This command configures all the components of X.25, including CTF. 3.3.5.2 CTF Crashes After Several Hours of Live Tracing When you use the live trace function in CTF for an extended time, CTF may experience a memory fault that causes the program to halt and generate a core file. To avoid this situation, use the detached tracing functionality described in Sections 3.5.2, 3.6.2, and 3.7.3 of the CTF for Digital UNIX Systems - CTF User's Guide. 3-6 Known Problems 3.3.5.3 Multiple Traces of "x25 client" on Gateways Not Supported There is a limitation when tracing the x25 client tracepoint on remote Gateways. The Gateway will support only one trace session at a time for this tracepoint. If there is already an x25 client trace session started on a Gateway, any attempts to create further x25 client trace sessions will appear to work, with CTF returning notification of tracepoints being traced, but no trace records will ever be received from the Gateway. 3.3.5.4 Time Displayed Inaccurately It is not possible to display the times from remote non Tru64 UNIX nodes as local times. Using the set time local command while tracing such nodes will cause timestamps to be displayed incorrectly as the remote nodes do not supply time zone infor- mation. 3.3.5.5 CTF Remote Trace Over TCP The TCP port number for CTF remote tracing has been corrected. Consequently, remote tracing over TCP between two machines, one with the corrected version of CTF and one with an earlier version, is no longer possible. Either the older version of CTF should be updated, or DECnet should be used as an alternative transport. 3.3.5.6 CTF Cannot be Paused for a Long Time An error in CTF will not allow high activity traces to be paused for an extended length of time. For instance, this error will occur if you start tracing a X.25 Protocol DTE dte-0 with the following command: CTF> start x25 proto dte dte-0 ,live ,full and pause the trace with the Pause command. If there is a lot of activity over this circuit, after approxi- mately 30 minutes the CTF buffers will fill, and CTF will exit with a access violation. To remedy this error, tracepoints should not be paused for an extended length of time. Where possible, Stop should be used in preference to Pause. Known Problems 3-7 3.3.5.7 Tracing on the Local Node While the Node Name is Specified and DECnet is Used as Transport If the user started a trace in CTF specifying the transport to be DECnet and the node to be the local node as follows: CTF> start node , transport=decnet the following error message will be displayed : Connection failed: Node Name Translation Failed Cannot process command: Connection Failed It is recommended that the user specify the node name to be 0 (zero) instead of the actual name when the tracepoint being traced is on the local node. CTF> start node 0 , transport=decnet or CTF> set default node 0 CTF> start , transport=decnet 3.3.5.8 Option "notruncate" in CTF Does Not Work Without "full" Option CTF trace output is truncated if it is wider than the screen. If you are interested in the call packet data, the use of the option full is recommended. For a more readable display, you should increase the width of the window that is running the trace. 3.3.6 ISDN 3.3.6.1 Starting IP/ISDN When Kernel Rebuild Was Required When configuring IP over ISDN, certain configurations require the kernel to be rebuilt and the system rebooted. After the reboot, you must configure IP over ISDN using wansetup ad- vanced. After this, to start ISDN correctly, you must issue the following command: # wanstartup RESTART That is, you must specify the RESTART option instead of the START option. Subsequent reboots will not require this. 3-8 Known Problems 3.3.6.2 Delay When Configuring IP/ISDN Hosts When configuring IP/ISDN, you enter a hostname that identifies a remote IP/ISDN host. If your ISDN system uses a name server, such as BIND or NIS, there will be a delay after you exit the configuration screen if the hostname you enter is not known to the name server. The delay could be quite long, but eventually the configuration program will display the next screen. This delay will also occur if you enter a local IP/ISDN hostname not configured in the name server. If you add the local and remote hostnames to the name server, this delay will not occur. 3.3.6.3 Lockmode 4 Not Supported for ISDN Lockmode=4 is not supported for ISDN in this release. It is suggested that lockmode=3 be used. This can be done by editing the /etc/sysconfigtab file. In the generic: definitions section, insert the following: lockmode=3 3.3.6.4 X.25 Clear Confirm Packets Sometimes Lost in X.25 over ISDN In dynamic call control for ISDN, when one end of an ISDN link sends a CLEAR CONFIRM packet for the last VC in use, it clears the ISDN call and shuts down LAPB. However, the CLEAR CONFIRM packet can be lost during LAPB shutdown. 3.3.6.5 Lockup While Attempting X.25 Call Over ISDN When an X.25 call is attempted over the ISDN interface and the call is unsuccessful, a timeout can cause the LAPB machine to become locked in the halted state. From that time on, no ISDN calls can be made over the X.25 interface to ISDN. If the LAPB machine is locked in the halted state, then: o in the NCL facility, the LAPB link state is shown as halted; and o an attempt to perform an X29 login will return the message Layer 2 is disconnected Known Problems 3-9 The LAPB state machine moves from the initialising state to the halted state when it times out. Normally (for synchronous devices) when the physical layer comes up the state machine will move back to the running layer. However, ISDN call control does not automatically reschedule calls, and so the physical layer never comes up and LAPB remains in the halted state. Enter the following NCL commands: ncl> disable lapb link ncl> enable lapb link and reattempt the call. 3-10 Known Problems Chapter 4 Documentation 4.1 Clarification of Behavior of X25Decode The description of X25Decode in both the man page and program- ming documentation is contradictory. The definition for the value returned for packet-size, window-size, etc, is defined in one location as a short int and in another as an int value. Both documents should consistently say a short int is returned. 4.2 Available X.25 Documentation The following documentation for the Compaq Wide Area Networking for Tru64 UNIX  product is available: WAN Support for DIGITAL UNIX Systems Introduction WAN Support for DIGITAL UNIX Systems Installation Guide WAN Support for DIGITAL UNIX Systems Configuration Guide WAN Support for DIGITAL UNIX Systems Management Guide WAN Support for DIGITAL UNIX Systems Programmer's Reference Manual WAN Support for DIGITAL UNIX Systems Problem Solving Guide WAN Support for DIGITAL UNIX Systems Security Guide WAN Support for DIGITAL UNIX Systems PAD User's Guide CTF for DIGITAL UNIX Systems CTF User's Guide WAN Support for DIGITAL UNIX Systems ISDN User's Guide- DEC 3000 Systems WAN Support for DIGITAL UNIX Systems RoamAbout Transport Server User's Guide WAN Support for DIGITAL UNIX Systems Device Drivers- Application Programmer's Guide All manuals were last published in May 1997 for the WAN Support for DIGITAL UNIX Systems, Version 3.0 product (a predecessor to the current release). Documentation 4-1 4.3 Location of Documentation Bookreader versions of the manuals are available on the Tru64 UNIX Online Documentation Library CD-ROM. Details of how to access the Bookreader documentation are provided in the Compaq Wide Area Networking for Tru64 UNIX Cover Letter. Hardcopy versions of each manual can be ordered from Compaq. See the Preface for ordering information. 4-2 Documentation