- Libsmi - lists.ibr.cs.tu-bs.de

bug in libsmi-0.2.17
by Damian Gilmurray 16 Sep '01

16 Sep '01

Hi, The following code and MIB results in a SIGSEGV from libsmi if the IMPORTed MIBs are unavailable. The offending code is check.c:933 which attempts to dereference object->typePtr which is NULL. I don't know the code well enough to produce a sensible patch. Regards, Damian.

2 1

problem with mib parsing
by mayagarwal＠hss.hns.com 03 Sep '01

03 Sep '01

Hi! I have a problem parsing a mib that uses the UNITS definition for an object. eg. mem-avail OBJECT-TYPE SYNTAX Integer32 MAX-ACCESS read-only STATUS current UNITS MB DESCRIPTION "Physical memory available" ::= { memory-usage 1 } This node is parsed fine if I remove the line "UNITS MB". Is there any way to make this work? The documentation for the tool generating this MIB says that it is ASN.1 compliant. Thanks and regards, Mayank. -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

3 2

Some clarifications about smilib errors reporting
by Deshmukh, Abhay 31 Aug '01

31 Aug '01

Hello, While working with various mib files and their processing with smilib, I came across following points which I felt that I should mention. (# 1) I have two files : 1) A-MIB 2) B-MIB A-MIB contains 5 modules: 1) A-MIB 2) A1-MIB 3) A2-MIB 4) A3-MIB 5) A4-MIB B-MIB contains one module: 1) B-MIB B-MIB imports two identifiers id1 and id2 from module A-MIB(i.e first module in the file), which are defined in that module. while loading these two modules (using smiLoadModule()) A-MIB and B-MIB in the following order : First -> B-MIB Second -> A-MIB (These two files are present in the working directory, i.e where libsmi searches for the presence of the modules) While loading the first file (B-MIB) , I get following error messages from the compiler : identifier `id1' cannot be imported from module `A4-MIB' identifier `id2' cannot be imported from module `A4-MIB' ... I think that while loading the module B-MIB from the file B-MIB, libsmi loads the A-MIB from which B-MIB imports some identifiers. And the access pointer (or module pointer whatever ) in the file A-MIB has reached the last module (A4-MIB). Instead of searching the identifiers in the first module in that file, they are searched in the last module (where they are not present) of that file. (#2) Whenever following error message is obtained while compiling a faulty module by smilib :(File 'TEST'containing module XXX) "C:\TestDir\SNMP\\XXX: unable to determine SMI version" I observed that, the file from which A-MIB is being read is locked or some handle to that file which was obtained by smilib is not released ( Even after releasing the smilib resources i.e doing smiExit()). Unless the calling application is closed, the file remains locked by some process and can not be deleted. Can this be the case ? Any solution ? At the same time whenever this error is obtained, the 'path' name is indicated by 'NULL'. I felt that it would be more helpful if the full path is reported in the 'path' variable. (Althoug it is properly reported in the 'message' parameter of the errorhandler function). Some other errors also report 'path' as null for severe errors found in the module being processed.For example : "failed to locate MIB module `SNA-NAU-MIB'". Is it possible to get the complete file/module name 'path'(in the path parameter) for such errors? (#3) Few mib files contain following kind of syntax: " IBM-8271-EtherStreamer-Switch-MIB {iso org(3) dod(6) internet(1) private(4) enterprises(1) 2} DEFINITIONS ::= BEGIN -- IBM 8271 EtherStreamer Switch MIB Release 1.1 ... " smilib does not accept this. A parse error is obtained. The {....} construct appearing before DEFINITIONS clause seems to be causing the problem. Should the mib files be changed or this is a valid construct ? Thanks, Abhay -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

2 1

Obtaining sources from CVS
by Deshmukh, Abhay 30 Aug '01

30 Aug '01

Hello, I tried getting the latest sources from the CVS repository. I'm getting a time out. (see below) : cvs [login aborted]: connect to anoncvs.ibr.cs.tu-bs.de:2401 failed: Connection timed out What can be the reason for this ? Thanks, Abhay -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

2 1

Windows versions
by David T. Perkins 28 Aug '01

28 Aug '01

HI, I've built in the past windows versions using the cygwin tools, and they work great. However, the licensing on the tools appears to me to be oppressive. That is, you can create programs for personal use, but to distribute the executables to others requires a license, and you have to negotiate the license, and the fee doesn't seem reasonable to me (several US$1000s). So, has anyone looked at Microsoft's version of "Unix tools" for windows? The URL is http://www.microsoft.com/windows2000/interix/default.asp If so, would appreciate finding out the following: 1) how good are they (as good/complete as cygwin) 2) cost to get the software (it looks like either a) free, b) US$100, but could not determine) 3) is there a license fee for distributing the executables? 4) can you link static? (I couldn't figure out how to get the cygwin stuff to link statically. Does any one have a solution. (It took the flags with no error, but the result still reference the DLL files.) Regards, /david t. perkins -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

2 1

libsmi0.2.17 - Win32 compilation problems
by Deshmukh, Abhay 27 Aug '01

27 Aug '01

Hello All, I'm extremely thankful to Frank and his team for coming up with an improved version of libsmi, in which most of the memory leaks have been removed. However, while compiling the latest release of smilib (0.2.17) (which I obtained from ftp://ftp.ibr.cs.tu-bs.de/pub/local/libsmi/ ) on Win32 platform, I noticed the following things : 1) data.c contains #include <sys/wait.h> , which is not present in VC++6.0.The same file (line 3538-3560, caching module) contains calls like fork(), execv() , which are causing trouble during compilation on Win32. 2) parse-smi.tab.c and parse-sming.tab.c not present. 3) dump-stools.c not present. 4) References to initScli() from sumidump.c and .h files could not be resolved while linking. Unknown identifier _initScli . Changes done in order to compile the library : a) I resolved the first problem just by commenting the #include sys/wait.h and the code from line 3538 to 3560 in data.c b) Copied parse-smi.c to parse-smi.tab.c and parse-sming.c to parse-sming.tab.c. This worked. c) dump-stools.c from earlier version was used. d) References to initScli() were removed from smidump.c and smidump.h. After taking these measures, I could compile the smi.lib. I used this library in place of the earlier smi.lib (version 0.2.16), for my applications. I found that the memory leaks have reduced greatly. But at the same time, the library has become a bit unstable in the sense that it now crashes for certain mib files( e.g : for a combination of following modules, I got an unhandled exception in smiFreeData() : RFC1213-MIB, Modem-MIB, SNMPv2-MIB) and also for a large number of input mib files (for which version 0.2.16 worked fine) Has anybody worked with this library on Win32 and any remedy ? Thanks, Abhay -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

3 2

RE: [libsmi] libsmi release 0.2.17
by Mark Kaplun 27 Aug '01

27 Aug '01

Hi Frank, It seems that the code implementing the caching (in data.c) is Unix oriented and do not compile under windows. After removing thie code everything compiled successfully. Mark Kaplun Tel: +1-972-3-7370751 Mobile: +1-972-55-636751 e-mail: mark(a)xor-t.com -----Original Message----- From: Frank Strauss [mailto:strauss@ibr.cs.tu-bs.de] Sent: Friday, August 24, 2001 12:51 PM To: libsmi Mailinglist Subject: [libsmi] libsmi release 0.2.17 Hi! After a long silent period there's now a new libsmi release on the FTP server. It... o fixes a lot of memory leaks (though not yet all), o updates and adds some MIB checks, o fixes and updates some output drivers, o adds a simple MIB file caching scheme (experimental!), o includes the (expired!) Internet-Draft on the currently implemented XML output format of smidump and the according DTD, o introduces a scheme to allow a single application to use multiple sets of MIB data. Furthermore, CVS users will recognize that a number of files are gone and an autogen.sh script has been added to ease the build process from the CVS repository. See the ChangeLog file for more details. As usual the current release can be found at http://www.ibr.cs.tu-bs.de/projects/libsmi/ ftp://ftp.ibr.cs.tu-bs.de/pub/local/libsmi/ Enjoy, -frank -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information. -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

3 6

RE: [libsmi] libsmi release 0.2.17
by Mark Kaplun 27 Aug '01

27 Aug '01

I guess that the important missing information is that I'm working under the microsoft visual tools. At this enviroment there is no sys/wait.h include file, and the functions fork and waitpid do not exist. I guess that the other user had worked with a more Unix or GNU like compilation enviroments. I had no such problems with the 0.2.16 version. Mark -----Original Message----- From: Frank Strauss [mailto:strauss@ibr.cs.tu-bs.de] Sent: Monday, August 27, 2001 1:00 PM To: Mark Kaplun Cc: libsmi(a)ibr.cs.tu-bs.de Subject: Re: [libsmi] libsmi release 0.2.17 Hi! Mark> It seems that the code implementing the caching (in data.c) is Unix Mark> oriented and do not compile under windows. Mark> After removing thie code everything compiled successfully. Could you provide more detailed information why it failed, maybe a make output dump? AFAIK, at least one other W32 user did not have these problems. -frank -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

1 0

f:/yigal/dmh-libsmi/libsmi-0.2.17/lib/data.c
by Yigal Hochberg 27 Aug '01

27 Aug '01

Frank, This is the code I had to compile out for ms compiler/environment. -- - Yigal file: f:/yigal/dmh-libsmi/libsmi-0.2.17/lib/data.c:3570 loadModule() .................... if (!path && smiHandle->cache && smiHandle->cacheProg) { /* Not found in the path; now try to fetch & cache the module. */ path = smiMalloc(strlen(smiHandle->cache) + strlen(modulename) + 2); sprintf(path, "%s%c%s", smiHandle->cache, DIR_SEPARATOR, modulename); if (access(path, R_OK)) { cmd = smiMalloc(strlen(smiHandle->cacheProg) + strlen(modulename) + 2); sprintf(cmd, "%s %s", smiHandle->cacheProg, modulename); #if !defined(_MSC_VER) pid = fork(); if (pid != -1) { if (!pid) { argv[0] = "sh"; argv[1] = "-c"; argv[2] = cmd; argv[3] = 0; execv("/bin/sh", argv); exit(127); } waitpid(pid, &status, 0); } #endif /* _MSC_VER */ smiFree(cmd); if (access(path, R_OK)) { smiFree(path); path = NULL; } } } .................... -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

2 1

libsmi release 0.2.17
by Frank Strauss 24 Aug '01

24 Aug '01

Hi! After a long silent period there's now a new libsmi release on the FTP server. It... o fixes a lot of memory leaks (though not yet all), o updates and adds some MIB checks, o fixes and updates some output drivers, o adds a simple MIB file caching scheme (experimental!), o includes the (expired!) Internet-Draft on the currently implemented XML output format of smidump and the according DTD, o introduces a scheme to allow a single application to use multiple sets of MIB data. Furthermore, CVS users will recognize that a number of files are gone and an autogen.sh script has been added to ease the build process from the CVS repository. See the ChangeLog file for more details. As usual the current release can be found at http://www.ibr.cs.tu-bs.de/projects/libsmi/ ftp://ftp.ibr.cs.tu-bs.de/pub/local/libsmi/ Enjoy, -frank -- !! This message is brought to you via the `libsmi' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>. !! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.

1 0