Fixes for IPv6 #84

BitBlit88 · 2022-03-20T18:05:08Z

This repo fixes issues with IPv6.
There were errors like:

IPv6 is not supported under this link-layer type

if the interface has only an IPv6 address but no IPv4 address.

Updated copyright notes Fixed issue with uninitialized ip6 pointer in case of packet type DLT_LINUX_SLL and DLT_RAW Removal of whitespaces

Updated copyright notes

Bump version to 8.0 (due to major IPv6 feature) Update copyright notes

Fixed handling of v6 pcap filter rule recreation in case of one time sequences

…pull request jvinet#1 from primeos/fix-ipv6-only

… which in case set will remove the generated filter strings. That reduces the allocated memory, but in case of one-time door sequences consumed requires all filters to be recreated in generate_pcap_filters(). Also take care of proper free() handling.

… fetched from jvinet#68 created by catbref

TDFKAOlli · 2022-03-20T18:22:05Z

@BitBlit88 , not sure if that works. My version derived a bit due to different work on memory safety issues (#74) which were merged here in this repo and what I did on my fork. I had questions about the last changes but they where merged before those where clarifed. @dimkr used lint or another code checker to find the memory issues in the code and I always dreamed of making the core code unit testable so to be able to really test all situations and thereby find all leaks.
I think if you want to take stuff from my fork you need to cherry pick.

dimkr · 2022-03-20T18:48:45Z

@TDFKAOlli I used Valgrind and -fsanitize=address. I can help with conflict resolution and code review, if needed.

TDFKAOlli · 2022-03-22T21:05:48Z

@dimkr Thanks. I checked through the conflicts and resolved them, creating a new pull request on my master. In the end the changes on this master seems logic and I couldn't see issues anymore (and I don't know which ones I have seen in the past 😄 ).

@BitBlit88 So the conflicts are at least resolved. I have not tested the result yet, but will try to do so with the branch created. Not sure if you can do additional commits here for your "any" feature.

@jvinet I have adapted the copyright notes in my branch just to mention the contributers. That might not been right, its your code base, so you might want to change this back to what you had before.

EDIT: Damn, seems somehow some bugs where fetched. The pcap-filter "tcp-psh" is wrong it has to be "tcp-push". That was right on my branch, now it is broken in the new branch... 😢

…variables of new features to match overall naming style. Fix tcp-psh filter to tcp-push.

dimkr · 2022-03-23T19:00:36Z

src/knockd.c

+		if (tmp->ifa_addr && tmp->ifa_addr->sa_family == AF_INET) {
+			struct sockaddr_in *addr_in = (struct sockaddr_in *)tmp->ifa_addr;
+			if (strcmp(iface,tmp->ifa_name) == 0 ) {
+				strncpy(buf, inet_ntoa(addr_in->sin_addr), bufsize-1);


If you just return inet_ntop(tmp->ifa_addr->sa_family, buf, tmp->ifa_addr, bufsize), you don't need to copy a string

Can't find any usage of get_ip() anymore. Seems to be a leftover from an old implementation. I'll remove.

dimkr · 2022-03-23T19:02:00Z

src/knockd.c

-		free(door->stop_command);
-		if(door->one_time_sequences_fd) {
+		if(door->start_command) {
+			free(door->start_command);


free(NULL) is safe, you can easily reduce the diff

Yes, you are right.

dimkr · 2022-03-23T19:02:41Z

src/knockd.c

@@ -795,6 +797,10 @@ int parseconfig(char *configfile)
 			fprintf(stderr, "error: section '%s' has an empty knock sequence\n", door->name);
 			return(1);
 		}
+		if(door->start_command == NULL && door->start_command6 == NULL) {


What if !start_command[0]? Is that OK?

Right, should also check for strlen() as also done in process_attempt().

If you only want to check if a string is empty, x[0] is more efficient than strlen(x) == 0

I prefer readabiliy over efficiency, especially as this is just done once at startup when reading in the commands.

TDFKAOlli · 2022-03-23T20:33:12Z

So, now cleaned up a bit and also built for my OpenWrt router. Seems all fine.
tcp-psh vs. tcp-push seemed to be my commit from 2019, I guess based on tcpdump information. I have tested with tcp-push and that seems to compile fine.

@dimkr Thanks, will also have a look at your comments.

… all network interfaces of the device

TDFKAOlli · 2022-03-23T22:33:05Z

@dimkr Should be fine now.
@BitBlit88 I have added the "any" interface support as described in the issue discussion. It doesn't seem to work on my OpenWrt. I can configure the 'any' interface, but the packets are not captured. Have to check what is the problem there, because that would be useful for me too. 😏

TDFKAOlli · 2022-03-29T19:12:09Z

Fixed the problem that 'any' doesn't capture on OpenWRT. That was my fault in adding the check for 'any'. 😄 Now there is another problem with setting the any interface... I do get each packet twice. Apparently from the same IP source and dest address as the logs state. That breaks the attempts as duplicates are currently not allowed. Have to check where that is coming from.

…tput.

TDFKAOlli added 12 commits December 31, 2018 15:25

Merged changes from jvinet#53 into master

d50b116

Updated copyright notes Fixed issue with uninitialized ip6 pointer in case of packet type DLT_LINUX_SLL and DLT_RAW Removal of whitespaces

Merged fix from jvinet#52 by Marius Hoch to master

e9b9cf5

Updated copyright notes

Fix tcp-psh filter setting (tcp-push => tcp-psh) in pcap filter

91bb0b2

Bump version to 8.0 (due to major IPv6 feature) Update copyright notes

Fixed some mem-leaks introduced with ipv6

7c00e40

Fixed handling of v6 pcap filter rule recreation in case of one time sequences

Remove declaration non-existing function runCommand()

4c3c5c5

Add patch for ipv6 only interfaces from jvinet/knock jvinet#58 (from …

2ac7d47

…pull request jvinet#1 from primeos/fix-ipv6-only

Add documentation of new flag.

92a7ae3

Fix incorrect TCP ACK/URG flag NOT_SET testing logic in flags_match()…

1f77a3f

… fetched from jvinet#68 created by catbref

Bump version to 0.8.1

faee547

Fixed use-after-free based on jvinet#74 and jvinet#75.

39c3693

bump to version 0.8.2

58a2ed4

Merge branch 'master' into resolve_conflict_from_upstream

4f0ad22

TDFKAOlli added 3 commits March 22, 2022 23:26

Fix build problems due to variable name changes on upstream. Renamed …

04bd747

…variables of new features to match overall naming style. Fix tcp-psh filter to tcp-push.

Correct the copyright notes

bd0b842

Further corrections on copyright notes

c07bc93

dimkr reviewed Mar 23, 2022

View reviewed changes

Additional small corrections

abbdc06

TDFKAOlli added 2 commits March 23, 2022 23:03

Address review comments of pull request jvinet/knockd#84

a7b6411

Add support for using 'any' for the configured interface to listen on…

d4c85aa

… all network interfaces of the device

Fix issue in 'any' interface handling

f3e7e7d

Added fix for VLAN Ethernet as proposed here: jvinet#89. Fixed log ou…

735379d

…tput.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixes for IPv6 #84

Fixes for IPv6 #84

BitBlit88 commented Mar 20, 2022

TDFKAOlli commented Mar 20, 2022 •

edited

Loading

dimkr commented Mar 20, 2022

TDFKAOlli commented Mar 22, 2022 •

edited

Loading

dimkr Mar 23, 2022

TDFKAOlli Mar 23, 2022

dimkr Mar 23, 2022

TDFKAOlli Mar 23, 2022

dimkr Mar 23, 2022

TDFKAOlli Mar 23, 2022

dimkr Mar 24, 2022 •

edited

Loading

TDFKAOlli Mar 24, 2022

TDFKAOlli commented Mar 23, 2022

TDFKAOlli commented Mar 23, 2022

TDFKAOlli commented Mar 29, 2022 •

edited

Loading

Fixes for IPv6 #84

Are you sure you want to change the base?

Fixes for IPv6 #84

Conversation

BitBlit88 commented Mar 20, 2022

TDFKAOlli commented Mar 20, 2022 • edited Loading

dimkr commented Mar 20, 2022

TDFKAOlli commented Mar 22, 2022 • edited Loading

dimkr Mar 23, 2022

Choose a reason for hiding this comment

TDFKAOlli Mar 23, 2022

Choose a reason for hiding this comment

dimkr Mar 23, 2022

Choose a reason for hiding this comment

TDFKAOlli Mar 23, 2022

Choose a reason for hiding this comment

dimkr Mar 23, 2022

Choose a reason for hiding this comment

TDFKAOlli Mar 23, 2022

Choose a reason for hiding this comment

dimkr Mar 24, 2022 • edited Loading

Choose a reason for hiding this comment

TDFKAOlli Mar 24, 2022

Choose a reason for hiding this comment

TDFKAOlli commented Mar 23, 2022

TDFKAOlli commented Mar 23, 2022

TDFKAOlli commented Mar 29, 2022 • edited Loading

TDFKAOlli commented Mar 20, 2022 •

edited

Loading

TDFKAOlli commented Mar 22, 2022 •

edited

Loading

dimkr Mar 24, 2022 •

edited

Loading

TDFKAOlli commented Mar 29, 2022 •

edited

Loading