Snort 2.9.4.1 Pkg 2.5.4 – Fix for SO rules version mismatch and failed startup
-
Aint that nice…............ :(
-
Aint that nice…............ :(
Yeah, but I guess from the point of view of the Snort VRT, this is a "carrot" to entice folks to buy subscriptions instead of using the free registered user rules. To stay current with the rules, they drive you to the subscription model. Can't really fault them for that.
There is not much that can be done on the pfSense side except to fall back to the older 2.9.4.0 binary. There are down sides to that as well.
-
Its understandable, but very frustrating from the enduser perspective…. :(
-
I found the bug that is causing the whitelist parsing failure in 2.9.4.1 and the error messages saying "…non IP() parameter found...".
A misplaced call to clear memory in the Spoink plug-in is the culprit. The buffer is being initialized AFTER being filled with data instead of BEFORE being filled with data to parse. The function in the Spoink plug-in then calls a Snort API to parse the IP address data. Because it inadvertently zeroes out the buffer prior to the Snort API call, then Snort returns a parsing error because there is nothing to parse. The end result is the whitelist does not get populated, and thus WAN IPs get blocked.
Here is the errant code snippet:
while((ret = s2c_parse_line(cad, wfile)) != 0) { memset(cad, 0, WLMAX); if (ret == 1) { ipw = malloc(sizeof(struct ipwlist)); if (ipw == NULL) { ErrorMessage("Could not allocate memory"); continue; } if (sfip_pton(cad, &ipw->waddr) != SFIP_SUCCESS) { ErrorMessage("Non ip(%s) parameter passed with white list, skipping...", cad); free(ipw); continue; } // else //printf("IP(%s) parsed succesfuly", cad);
Notice the memset() call immediately after the while() statement. That is zeroing out the buffer containing the IP address to be parsed. The memset() function call should be BEFORE the while() statement instead of AFTER.
Bill
-
Have you sent this to Ermal?
-
Have you sent this to Ermal?
Yes. I have not received a reply yet, but he and I are in vastly different time zones and he may not have seen the note yet.
-
Allright mate!! You doing a hell of a job for the rest of the community!
-
bmeeks…very impressed with your debugging abilities :-) As a guy who coded a long time back, it's refreshing to see someone who can get under the hood and ID issues so quickly.
Cheers,
Dennis. -
bmeeks…very impressed with your debugging abilities :-) As a guy who coded a long time back, it's refreshing to see someone who can get under the hood and ID issues so quickly.
Cheers,
Dennis.Thanks, but I must admit I stared at that code for like an hour and did not see the bug. I knew it had to be there somewhere, and finally I noticed the misplaced memset() call. After that it was like … Doh!! ... why didn't I see that first thing ?? :D
Bill
-
Please let us know when it will be fixed.
So we can pull new package.
Thanks a lot for all efforts -
Please let us know when it will be fixed.
So we can pull new package.
Thanks a lot for all effortsI have submitted a Pull Request via GitHub that contains a fix for the whitelist parsing issue. The pfSense developers have to accept my patch into the packages repository and then compile it into the new binary for Snort. Changes to the binary are a bit more involved to publish than changes to the GUI code.
Bill
-
Any news on this ?
-
waiting as well for this.
-
Any news on this ?
My Pull Request for the Spoink patch was accepted, but so far it has not been incorporated into a new build of the binary as far as I can tell. I don't know what the process is nor the timeline for the binary side. On the GUI side, once a Pull Request is accepted by the Core Team it is immediately available for download. I know the binaries have to be built, but I don't know if that is automated (I think it is) or a human has to intervene.
Bill
-
Just uninstall , then install package.
It looks like is NOT rebuilt yet , I see my WAN blocked . :(
(Snort 2.9.4.1 Pkg 2.5.4 , Emergingthreats rules only )
Thanks -
I reported that to Bmeeks some time ago since I saw my WAN blocked as well. It must be the implementation of Snort into PFsense that is causing this behaviour…
-
I found my WAN blocked yesterday. I removed the block, but it came right back. Restarted the service and it has been running fine since. Never saw this before upgrading to Snort 2.9.4.1.
If it matters, I do have the paid VRT rules. (Well worth $2.50/month. I think it's a good value and money worth spent)
-
I found my WAN blocked yesterday. I removed the block, but it came right back. Restarted the service and it has been running fine since. Never saw this before upgrading to Snort 2.9.4.1.
If it matters, I do have the paid VRT rules. (Well worth $2.50/month. I think it's a good value and money worth spent)
Until my latest bug fix is incorporated into the binary build of Snort on pfSense, you will see your WAN IP (and any other normally whitelisted IPs) get blocked. The blocking of offenders in Snort on pfSense is done with an optional output plugin.
Snort, natively, has no "blocking" capability. The Snort team leaves that to others. There are two popular methods in use: Snortsam and Spoink. The pfSense folks chose Spoink. This works as an optional output plugin compiled into the Snort binary. The Snort source code is patched during the pfSense package build process to incorporate the Spoink output plugin. This plugin receives each Alert from Snort as it is on the way to the log files. It compares the IP addresses in the Alert (SRC, DST or BOTH according to how you configure blocking) to the list of Whitelist IPs. If the offending IP is NOT in the whitelist, then an API call is made into the pfSense packet filter code to insert a blocking rule for that IP. The IP whitelist is just a text file in the same directory as the Snort configuration files. That file is created by the GUI code and then read at Snort startup by the Spoink plugin patched into Snort.
The bug that got introduced in 2.9.4.1 is in the Spoink plugin patch. During startup, when it reads the Whitelist file and stores the addresses in there into the in-memory table of whitelist IP addresses, it zeroes out the data it reads from the file just prior to parsing it! So it sees an "empty" whitelist file and thus blocks ALL alerting IP addresses. The intent of the zero-out call was to initialize the buffer with zeros prior to reading in the whitelist, but the memory clearing call was typed in the wrong spot such that it clears the buffer immediately after it was just filled with the file's data. I submitted a fix for this bug, but it has not made its way into the compiled binary package yet. Until it gets fixed, this bug will keep causing people issues with their WAN IP and other normally whitelisted IPs getting blocked.
Bill
-
ERMAL WE NEED YOU URGENTLY!!
-
Bill, that explanation was maybe my favourite post ever here. While I make no claims on code prowess, I really appreciate the under-hood explanation of what's going on. I used to try the variety of work-arounds that are normally offered up after debugging a package. It's a lot more time efficient however to watch posts like yours, and enter back into debugging/testing contribution phase once it looks like things "should" work. Again thanks to all for their efforts.
Cheers,
Dennis.