[NEW] suricata 1.4.7

[jul_bsd@…]

Open Source IDS / IPS / NSM engine

• compile/run
• devel subport

[jul_bsd@…]

• port lint --nitpick
• livecheck
• /tab/spacex4/

[neverpanic (Clemens Lang)]

• --prefix=${prefix} is automatically passed configure in configure.pre_args by MacPorts. You don't have to pass it again.
• --sysconfdir defaults to PREFIX/etc, so passing --sysconfdir=${prefix}/etc isn't necessary.
• The same applies to --localstatedir.
• revision 0 is the default and can be omitted.
• Don't set configure.compiler directly unless absolutely necessary. Instead, use compiler.blacklist to blacklist the compilers that will fail to compile suricata and let MacPorts choose a fallback. Even better would be fixing the build with clang.
• Conflicts should be specified both ways, so the suricata subport should conflicts ${name}-devel
• Instead of listing a variant that does nothing, consider commenting it.
• I'm not sure hardcoding the startupitem to the en1 interface is a good solution. I don't have a better idea either, though.
• When adding optional features with dependencies in variants you need to make sure the configure script doesn't autodetect the presence of the dependency and links against libraries when the dependency isn't specified. For example, suricata must not link against libgeoip when installed without the geoip variant even when the libgeoip port is installed. To do this, pass --disable-geoip in the default configure.args and replace it with --enable-geoip in the geoip variant.

[neverpanic (Clemens Lang)]

Also, configure picks up python as $prefix/bin/python

checking for python... /opt/local/bin/python

which is a symlink that can be adjusted by users and might cause the package to build different depending on what a user has chosen as python version. You should choose a version of python you want to use and set configure.python to avoid that to make the build reproducible.

[neverpanic (Clemens Lang)]

Also, here's a patch to fix the build problem with clang due to the system headers defining strlcat(3) and strlcpy(3) as macros:

configure.ac

@@ -155 +155 @@
     # Checks for library functions.
     AC_FUNC_MALLOC
     AC_FUNC_REALLOC
-    AC_CHECK_FUNCS([gettimeofday memset strcasecmp strchr strdup strerror strncasecmp strtol strtoul memchr])
+    AC_CHECK_FUNCS([gettimeofday memset strcasecmp strchr strdup strerror strncasecmp strtol strtoul memchr strlcat strlcpy])
 
     # Add large file support
     AC_SYS_LARGEFILE

src/suricata-common.h

@@ -304 +304 @@
 #include "util-optimize.h"
 #include "util-path.h"
 
+#ifndef strlcat
+/* strlcat is a macro on some systems; if it is, we probably don't need this
+ * declaration and the macro expansion will break the declaration. */
 size_t strlcat(char *, const char *src, size_t siz);
+#endif
+#ifndef strlcpy
+/* strlcpy is a macro on some systems; if it is, we probably don't need this
+ * declaration and the macro expansion will break the declaration. */
 size_t strlcpy(char *dst, const char *src, size_t siz);
+#endif
 
 #endif /* __SURICATA_COMMON_H__ */

Please submit this patch upstream as well.

[jul_bsd@…]

• update to 2.0 and subport suricata1, disabled suricata-devel for now, review patch for 2.0
• 2.0 is NOT compiling, see main.log
• add test command but not sure if it is executed
• review livecheck for 2.0

[jul_bsd@…]

• suricata 2.0.1 + subport suricata1 => both install/run
• configure.blacklist need to be reviewed but seems to be a long one. works ok w gcc4.9
• universal variant in work in progress but failing if enable build.args (be it in universal or not)

  util-affinity.h:70:5: error: unknown type name 'cpu_set_t'
  

• test.run
• switch tasks to LaunchDaemons
• newsyslog configuration to rotate logs
• fix livecheck

[jul_bsd@…]

• bump to 2,0,3, -devel 2,1beta1 and keep old 1,4,7
• Merge some patch from duplicate submission #44227 (#44691 "port search: return Trac ticket for submission" would mostly avoid that...)
• the 3 versions are up to destroot and 2,0,3+universal
• add variant nss

[jul_bsd@…]

sadly, running test mode 2,1/devel and also 2,0,3 gives

$ sudo suricata -c /opt/local/etc/suricata/suricata.yaml -T
30/8/2014 -- 20:07:45 - <Info> - Running suricata under test mode
30/8/2014 -- 20:07:45 - <Notice> - This is Suricata version 2.1beta1 RELEASE
/usr/share/file/magic/animation, 712: Warning: Printf format `l' is not valid for type `belong' in description `%ld x'
/usr/share/file/magic/animation, 713: Warning: Printf format `l' is not valid for type `belong' in description `%ld'
/usr/share/file/magic/animation, 720: Warning: Printf format `l' is not valid for type `belong' in description `%ld x'
/usr/share/file/magic/animation, 721: Warning: Printf format `l' is not valid for type `belong' in description `%ld'
/usr/share/file/magic/animation, 785: Warning: Current entry does not yet have a description for adding a MIME type
/usr/share/file/magic/animation, 787: Warning: Current entry does not yet have a description for adding a MIME 
[...]
/usr/share/file/magic/xwindows, 34: Warning: Printf format `h' is not valid for type `leshort' in description `version %hd'
/usr/share/file/magic/xwindows, 35: Warning: Printf format `h' is not valid for type `leshort' in description `.%hd'
30/8/2014 -- 20:07:45 - <Error> - [ERRCODE: SC_ERR_MAGIC_LOAD(197)] - magic_load failed: File 5.19 supports only version 12 magic files. `/usr/share/file/magic.mgc' is version 7

[Schamschula (Marius Schamschula)]

Cc Me!

[jul_bsd@…]

• update to 2.0.4, destroot ok
• v1 destroot nok

[Schamschula (Marius Schamschula)]

Update from 2.0.4 to 2.0.7

[Schamschula (Marius Schamschula)]

Updated suricata to version 2.0.7. A number of patches have been included upstream, but a new patch is needed to build output-json.c.

[Schamschula (Marius Schamschula)]

Update post-destroot

[Schamschula (Marius Schamschula)]

Copy a few more rules and files and a config file in post-destroot.

[jul_bsd@…]

I merged your changes for 2.1.7 but

• not sure you want rules files directly in etc, especially as from a security point of view, you will not use default set but other like EmergingThreats (that you can download w oinkmaster) and it includes them
• the jansson header was already added in common.h but an update was needed
• strlcat/copy patch included by upstream and not needed anymore
• destroot/install is fine

but at configuration check, still have

# suricata -c /opt/local/etc/suricata/suricata.yaml -T 2>&1 |grep -v Warning
10/3/2015 -- 19:37:19 - <Info> - Running suricata under test mode
10/3/2015 -- 19:37:19 - <Notice> - This is Suricata version 2.0.7 RELEASE
10/3/2015 -- 19:37:19 - <Error> - [ERRCODE: SC_ERR_MAGIC_LOAD(197)] - magic_load failed: File 5.22 supports only version 12 magic files. `/usr/share/file/magic.mgc' is version 7

which I referenced here ​https://redmine.openinfosecfoundation.org/issues/1268

did you met this issue? solved it?

Thanks

[Schamschula (Marius Schamschula)]

Replying to jul_bsd@…:

I merged your changes for 2.1.7 but

• not sure you want rules files directly in etc, especially as from a security point of view, you will not use default set but other like EmergingThreats (that you can download w oinkmaster) and it includes them
• the jansson header was already added in common.h but an update was needed
• strlcat/copy patch included by upstream and not needed anymore
• destroot/install is fine

but at configuration check, still have

# suricata -c /opt/local/etc/suricata/suricata.yaml -T 2>&1 |grep -v Warning
10/3/2015 -- 19:37:19 - <Info> - Running suricata under test mode
10/3/2015 -- 19:37:19 - <Notice> - This is Suricata version 2.0.7 RELEASE
10/3/2015 -- 19:37:19 - <Error> - [ERRCODE: SC_ERR_MAGIC_LOAD(197)] - magic_load failed: File 5.22 supports only version 12 magic files. `/usr/share/file/magic.mgc' is version 7

which I referenced here ​https://redmine.openinfosecfoundation.org/issues/1268

did you met this issue? solved it?

Thanks

As you noted, the changes here are not enough to get things working.

1) There currently is no oinkmaster port, so I downloaded it and put the perl script into ~/bin

2) I manually edited a few more things:

Make sure that logging to file is enabled in /opt/local/etc/suricata/suricata.yaml

 - file:
     enabled: yes
     filename: /opt/local/var/log/suricata/suricata.log

also use MacPort's libmagic:

#magic-file: /opt/local/share/misc/magic
magic-file: /usr/share/file/magic

to

magic-file: /opt/local/share/misc/magic
#magic-file: /usr/share/file/magic

[mkae (Marko Käning)]

What's the status of this? Stable is by now at 2.0.8...

[mkae (Marko Käning)]

Cc Me!

[Schamschula (Marius Schamschula)]

Replying to mk@…:

What's the status of this? Stable is by now at 2.0.8 and devel at 2.1beta4.

As this port has not been officially added, I don't run livecheck on it. I just updated my local Portfile to version 2.0.8. It built w/o any problems. My notes from comment:14 still apply.

[mkae (Marko Käning)]

There is no port oinkmaster up to now and you copied its perl code into your ~/bin!?

But ​org.macports.oinkmaster-suricata.plist seems to assume it being installed below the MacPorts prefix. Obviously I am missing something here.

What about the SC_ERR_MAGIC_LOAD error? Is that fixed?

[Schamschula (Marius Schamschula)]

Replying to mk@…:

There is no port oinkmaster up to now and you copied its perl code into your ~/bin!?

oinkmaster is just a perl script. I put it in ~/bin to keep it out of the /usr and the /opt trees. Needless to say, once there is a port it will be in /opt/local/bin

But ​org.macports.oinkmaster-suricata.plist seems to assume it being installed below the MacPorts prefix. Obviously I am missing something here.

There is not such a thing (yet). At least it didn't get installed on my machines.

What about the SC_ERR_MAGIC_LOAD error? Is that fixed?

I don't see that issue with 2.0.7 or 2.0.8. I've got File 5.23. I just ran the 'test' mode and saw no errors.

[mkae (Marko Käning)]

So, it is only missing port oinkmaster which prevents publishing this port? (Ping.)

[Schamschula (Marius Schamschula)]

Updated to suricata 2.0.9:

• stream: retransmission not detected (2.0.x)
• Segmentation Fault at detect-engine-content-inspection.c:438
• defrag: evasion issue
• last_ack update issue leading to stream gaps (2.0.x)
• 2.0.x backport: Leading whitespace in flowbits variable names
• http_host payload validation erroring on uppercase PCRE metacharacters
• 2.0.x backport: Add HUP coverage to output json-log
• 2.0.x: address var parsing issue
• stream_size <= and >= modifiers function as < and > (equality is not functional) (2.0.x)
• Update bundled libhtp to 0.5.18

[Schamschula (Marius Schamschula)]

Replying to mk@…:

So, it is only missing port oinkmaster which prevents publishing this port? (Ping.)

Created oinkmaster port: #48977

[jul_bsd@…]

Hello

Sorry for the delayed answer. not much time for those in past months. oinkmaster port was part of the past submissions (#42859). never committed though.

Need to merge mschamschula@ work. also with brew formula (​https://github.com/Homebrew/homebrew/blob/master/Library/Formula/suricata.rb). Magic issue seems fixed with a config change

[Schamschula (Marius Schamschula)]

Updated to suricata 2.0.10.

[Schamschula (Marius Schamschula)]

Updated to suricata 2.0.11.

[Schamschula (Marius Schamschula)]

Updated suricata-devel to the current "stable" branch, which is currently at version 3.0.

Fixed livecheck.

[Schamschula (Marius Schamschula)]

Updated suricata-devel to version 3.0.1.

[Schamschula (Marius Schamschula)]

I've stopped posting updates here. See

​https://github.com/Schamschula/macports/tree/master/net/suricata

for the current version.