Jun Kuriyama wrote: >At Thu, 24 Jun 2004 6:21:30 -0400, ><[EMAIL PROTECTED]> wrote: > >>If you would kindly download net-snmp-5.1.2.pre2.tar.gz from >>http://www.net-snmp.org/ >>and compare that with the patched version of 5.1.1 that you maintain, >>then we can discuss how to incorporate change into the sources. >> >>It would be a wonderful thing to fully support FreeBSD 4 and 5 >>in the 5.1.2 release. >> >>We are very close to releasing 5.1.2, so time is of the essence. > > >Thanks! I'll pass patches against 5.1.2pre2 to you. That's my >laziness, I should submit my modifications in timely fashion... > >
Please, instead of sending patches to me, send them to [EMAIL PROTECTED] That way they will receive the proper review. The best way is to submit each patch separately in the SourceForge repository, so that the discussion for a patch can be kept in one place. Adding the patch is easy; goto http://www.net-snmp.org/, click Patches. [Make sure to click "Check box to Upload an attach a file" when you submit the patch, and for updated patches] I've attached the patch that Willem sent me this morning. I'm cherry-picking them in my comments below. Ah Life! Here are a few comments that I have to start the clock: 1. Forget the tcpTable/udpTable patches. They do not apply cleanly to net-snmp-5.1.1, and they duplicate changes that have already been applied to 5.1.2.pre2, which does work on FreeBSD 4.x/5.x. 2. The agent/agent_trap.c patch is interesting. V5-1-patches already declares sinkport as an integer, which is proper for the "atoi()" call. However, it should cast that value to (u_short) when calling the createXXsession calls. The main branch should also use int for sinkport, and cast the result to u_short. 3. The agent/mibgroup/mibII/ip.c, function int ip_handler, line 892 return NULL is bad, as you've noticed. the value (int) NULL is assumed to be 0, [not universally true!] How about replacing the return with "continue", as other errored results do in this function ? 4. The patch for hr_print.c is not good. I believe the change that is is 5.1.2.pre2 makes the function more robust, but that belief hasn't been tested on a system with many printers %^) 5. Question: when you apply the patch to mibII/ipv6.c, does this fix the "OID not increasing" bug 735912 ? 6. There is an attached patch that removes the CPP macro from mibII/var_route.c. This source includes util_funcs.h, which defines the CPP macro "satosin" differently. Does this patch fix problems with the addresses seen in FreeBSD ? mibII/var_route.c and mibII/ipAddr.c both define a CPP macro named "satosin", like this : #define satosin(sa) ((struct sockaddr_in *)(sa)) which differs from the CPP macro defined in util_funcs.h : #define satosin(x) ((struct sockaddr_in *) &(x)) mibII/var_route.c includes util_funcs.h, and there is an attached patch that removes the CPP macro from mibII/var_route.c. In both sources, sa is declared as struct sockaddr *sa, and satosin is used to copy a sin_addr member to a struct in_addr. 7. Regarding the snmplib/scapi.c patch, does this match the version of OpenSSL you are using ?
net-snmp.64bit.diff
Description: Binary data