snmpstrucfiller(3) snmpkit functions for taking making snmp requests based upon a structure.


#include <snmpkit.h>

SNMPSTRUCTFILLER *new_snmpstructfiller(SNMPSESSION *session);
void delete_snmpstructfiller(SNMPSTRUCTFILLER *doomed);
void snmpstructfiller_append(SNMPSTRUCTFILLER *sf, const char *oidstr,Tags tag, unsigned int offset);
void snmpstructfiller_remove(SNMPSTRUCTFILLER *sf,const char *oidstr);
void *snmpstructfiller_get(SNMPSTRUCTFILLER *sf,void *tobefilled);
void *snmpstructfiller_get_next(SNMPSTRUCTFILLER *sf, void *tobefilled);


A snmpstructfiller is an opaque data structure used fill C structures with data from the snmp requests.

The SNMP protocol is designed in a way where there is significant overhead in process of encoding and exchanging a packet. Therefore it is very inefficient and time consuming to exchange many variables in individual request packets. The way around this is to bulk up a bunch of snmp requests into one packet. Since you are most frequently fetching information which is related the snmp structure filler is designed to associate the snmp objects and data types with the offsets into the structure. That way when you do a get or a get_next you will get back a completely filled in structure.

The new_snmpstructfiller() function creates new structfiller. You will need one of these for every different set of objects that you want to fetch. It starts out not having any requests associated with it. To add requests to this structure use the snmpstructfiller_append() function. To actually do the request, use the snmpstructfiller_get() or the snmpstructfiller_get_next() functions.

The delete_snmpstructfiller() function is the opposite of the new function. It frees all the resources used by the structfiller structure.

The snmpstructfiller_append() function appends a new snmp object to the structure filler. It requires the oidstring. This snmp library doesn't bother attempting to read MIBS. You must specify the snmp object in dotted decimal notation. For example, system.sysDesc is "". This means that you must look up the objects ID's before hand. This sounds like a lot of work but I have found that when writing an application it is really not that much of a problem. The tag is the type of data that the object is supposed to have as defined in the MIB and cooresponds directly to the space used in the structure you are filling. The tag can be one of the following:

INT_TAG long
IPADDR_TAG unsigned long
You get the IPADDR back in binary format. You must translate it byte for byte to get an IPADDR in the format that is usually expected.
This could concievably be an unsigned long but enough implementations of MIBS are fouled up in such that even though the mib specifies a counter, the device returns an integer. That it is best to store this number in an integer rather than an unsigned..

The final parameter is offset into the structure. For example if you have a structure and a variable like: struct info { char *descr; char *contact; } foo;

Then the offset of descr would be (char*)&(foo.descr)-(char*)&foo and the offset of contact would be (char*)&(*)&foo I know that this is a very bad way of doing things and I intend to change it in the future. Probably what I will do is change this to a function pointer which when called takes the pointer to the structure that it is supposed to insert the data in as well as the value of the data and it will insert it into the data into the structure.


Ben Woodard <[email protected]>


The library can possibly throw different kinds of C++ exceptions that won't be caught by the glue code and therefore it can cause your program to crash inexplicably.

There is no support for SNMP sets.

There is no support for getting back an OID.

There is no support for