iftop-users

iftop-users@lists.beasts.org

195 discussions

Portability to GNU/kFreeBSD and OpenSolaris
by Mats Erik Andersson 23 Oct '11

23 Oct '11

Clearly no-one listened to my portability suggestions in [1]. Let me therefore point out changes mandatory to 1.0pre2, should you desire to support the OpenSolaris variant known as NexentaCore. Two changes are also relevant for GNU/kFreeBSD. First off, renew the four files {,config/}config.{guess,sub} The present versions are six years old. They are incapable of working with GNU/kFreeBSD. These are necessary changes (variations present also in [1]): * The correct conditionals in "addrs_ioctl.c" is defined __GLIBC__ & ! defined __linux__ in order to catch GNU/kFreeBSD and to avoid NexentaCore. This MUST be changed, whether or not you decide to ignore OpenSolaris. * OpenSolaris does _not_ contain any "<sys/sysctl.h>" Exclude it"! * OpenSolaris _must_ test HAVE_SYS_SOCKIO_H and then include "<sys/sockio.h>". * You can not declare sigalrm() as non-static, and then define sigalrm() as static. That leads to compiler error, unless you ignore sanity checking. * OpenSolaris can not rely on "<curses.h>", since the 32-bit calls are only present in "<ncurses.h>". I demonstrated in [1] a functional technique to let the configuration script adapt to whatever Curses compatible library happens to be available in five different operating systems. Implement that method, or use the shortcut shown below. Make your choice, I have given the options. Mats Erik Andersson [1] http://lists.beasts.org/pipermail/iftop-users/2011-August/000344.html diff -Naur iftop-1.0pre2.orig/addrs_ioctl.c iftop-1.0pre2/addrs_ioctl.c --- iftop-1.0pre2.orig/addrs_ioctl.c 2011-10-02 23:11:30.000000000 +0200 +++ iftop-1.0pre2/addrs_ioctl.c 2011-10-23 23:33:41.777439591 +0200 @@ -19,12 +19,18 @@ #include <netinet/in.h> #if defined __FreeBSD__ || defined __OpenBSD__ || defined __APPLE__ \ - || ( defined __GNUC__ && ! defined __linux__ ) + || ( defined __GLIBC__ && ! defined __linux__ ) #include <sys/param.h> +# if !__sun__ #include <sys/sysctl.h> +# endif #include <net/if_dl.h> #endif +#if HAVE_SYS_SOCKIO_H +# include <sys/sockio.h> +#endif + #ifdef USE_GETIFADDRS #include <ifaddrs.h> #endif @@ -81,7 +87,7 @@ } #else #if defined __FreeBSD__ || defined __OpenBSD__ || defined __APPLE__ \ - || ( defined __GNUC__ && ! defined __linux__ ) + || ( defined __GLIBC__ && ! defined __linux__ ) { int sysctlparam[6] = {CTL_NET, PF_ROUTE, 0, 0, NET_RT_IFLIST, 0}; size_t needed = 0; diff -Naur iftop-1.0pre2.orig/dlcommon.c iftop-1.0pre2/dlcommon.c --- iftop-1.0pre2.orig/dlcommon.c 2003-06-07 00:42:34.000000000 +0200 +++ iftop-1.0pre2/dlcommon.c 2011-10-23 23:19:48.803230766 +0200 @@ -462,7 +462,7 @@ err("dlphysaddrack: short response ctl.len: %d", ctl.len); } -static void +void sigalrm() { (void) err("sigalrm: TIMEOUT"); diff -Naur iftop-1.0pre2.orig/options.c iftop-1.0pre2/options.c --- iftop-1.0pre2.orig/options.c 2011-10-04 22:39:42.000000000 +0200 +++ iftop-1.0pre2/options.c 2011-10-23 23:07:20.794488198 +0200 @@ -19,6 +19,10 @@ #include <arpa/inet.h> #include <net/if.h> +#if HAVE_SYS_SOCKIO_H +# include <sys/sockio.h> +#endif + #include "iftop.h" #include "options.h" #include "cfgfile.h" diff -Naur iftop-1.0pre2.orig/ui.c iftop-1.0pre2/ui.c --- iftop-1.0pre2.orig/ui.c 2011-10-03 22:30:03.000000000 +0200 +++ iftop-1.0pre2/ui.c 2011-10-23 23:21:11.527269307 +0200 @@ -6,7 +6,11 @@ #include <sys/types.h> #include <ctype.h> -#include <curses.h> +#ifdef __sun__ +# include <ncurses.h> +#else +# include <curses.h> +#endif #include <errno.h> #include <string.h> #include <math.h>

1 0

Monitoring all traffic
by joedson marques 21 Oct '11

21 Oct '11

Hello Guys, I have the scenario following: One freeBSD that works as router, so, most of the traffic that pass through an interface is not connection, but just packages that are being routed. I wanna monitoring all traffic passing through an interface, no just the traffic generated by connections. Is possible make this monitoring without change the source code of the iftop? If not, which part of source code should I change? Thanks!

3 7

iftop 1.0pre2
by Paul Warren 04 Oct '11

04 Oct '11

After an unreasonably long delay, I've finally got round to releasing iftop-1.0pre2. Hopefully this fixes compilation issues on various platforms, including FreeBSD and Mac OS X. It's also got: * reverse IPv6 using all methods except ares * faster switching of graph scale * fixed support for DLT_NULL * fix for segfault on interfaces with no MAC address * improved interface autodetect * a stab at implementing support for radiotap interfaces (DLT 127). Specific items I'm looking for feedback on: * Does it compile? If you could drop me an off-list email with platforms that you've tested on, that would be useful. * Does anyone still use the ares resolver implementation? * Does the interface autodetection work any better or worse than 1.0pre1 or earlier? (i.e. do you get the interface you expected when run with no options) I hope I've dealt with all the patches I've received since 1.0pre1. If you've sent me a patch and I've not acknowledged it, please let me know. Barring any serious issues, I hope to push this to 1.0 soon. It's getting close to iftop's 10th anniversary, so it's probably about time :-) Paul

1 0

Compile failure off HEAD
by Garrett Cooper 21 Aug '11

21 Aug '11

Hi, Trying to determine whether or not the prerelease version has the same SI units bug as 0.17, I checked out the source from CVS HEAD and ran into this build bug on FreeBSD: $ make all make all-recursive Making all in config gcc -DHAVE_CONFIG_H -I. -g -O2 -MT addr_hash.o -MD -MP -MF .deps/addr_hash.Tpo -c -o addr_hash.o addr_hash.c addr_hash.c: In function 'hash': addr_hash.c:46: error: 'struct in6_addr' has no member named 's6_addr32' addr_hash.c:54: error: 'struct in6_addr' has no member named 's6_addr32' *** Error code 1 Stop in /scratch/iftop. *** Error code 1 Stop in /scratch/iftop. *** Error code 1 Stop in /scratch/iftop. $ uname -a FreeBSD bayonetta.local 9.0-BETA1 FreeBSD 9.0-BETA1 #0 r224946M: Thu Aug 18 10:02:09 PDT 2011 gcooper@bayonetta.local:/usr/obj/usr/src/sys/BAYONETTA amd64 Looks like some linuxisms have crept into the source. Thanks, -Garrett

2 1

can Applications be displayed in the iftop scroll?
by nick rundy 11 Jul '11

11 Jul '11

Is there any way to display the Application that is causing an internet connection to be established? For example, I often run the following command: sudo iftop -N -P -bis there a command or some way to get the Application responsible for the connection to be listed in the scroll?

3 2

patch for FreeBSD tunnel interface
by Scott Bertilson 05 Jun '11

05 Jun '11

It also adds support for the OpenBSD DLT_LOOP. Based on handling of these DLT types from tcpdump.c. @@ -432,10 +432,15 @@ static void handle_raw_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { handle_ip_packet((struct ip*)packet, -1); } +static void handle_null_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) +{ + handle_ip_packet((struct ip*)(packet+4), -1); +} + #ifdef DLT_PFLOG static void handle_pflog_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { register u_int length = pkthdr->len; u_int hdrlen; @@ -680,13 +685,21 @@ #ifdef DLT_PFLOG else if (dlt == DLT_PFLOG) { packet_handler = handle_pflog_packet; } #endif - else if(dlt == DLT_RAW || dlt == DLT_NULL) { + else if(dlt == DLT_RAW) { packet_handler = handle_raw_packet; } + else if(dlt == DLT_NULL) { + packet_handler = handle_null_packet; + } +#ifdef DLT_LOOP + else if(dlt == DLT_LOOP) { + packet_handler = handle_null_packet; + } +#endif else if(dlt == DLT_IEEE802) { packet_handler = handle_tokenring_packet; } else if(dlt == DLT_PPP) { packet_handler = handle_ppp_packet;

1 0

alternative, hopefully more portable patch for s6_addr32
by Scott Bertilson 05 Jun '11

05 Jun '11

This patch doesn't require extra fiddling on either Linux (Gentoo) or FreeBSD 6.1. Since both have the 8 bit array for V6 addresses, neither requires any additional OS-specific hacks. I think this might be more generally portable. This is a diff against the CVS archive as of today. Index: addr_hash.c =================================================================== RCS file: /home/pdw/vcvs/repos/iftop/addr_hash.c,v retrieving revision 1.6 diff -u -r1.6 addr_hash.c --- addr_hash.c 27 Nov 2010 11:06:12 -0000 1.6 +++ addr_hash.c 5 Jun 2011 16:57:30 -0000 @@ -43,7 +43,7 @@ addr_pair* ap = (addr_pair*)key; if (ap->af == AF_INET6) { - uint32_t* addr6 = ap->src6.s6_addr32; + uint32_t* addr6 = (uint32_t*)ap->src6.s6_addr; hash = ( hash_uint32(addr6[0]) + hash_uint32(addr6[1]) @@ -51,7 +51,7 @@ + hash_uint32(addr6[3]) + ap->src_port) % 0xFF; - addr6 = ap->dst6.s6_addr32; + addr6 = (uint32_t*)ap->dst6.s6_addr; hash = ( hash + hash_uint32(addr6[0]) + hash_uint32(addr6[1]) + hash_uint32(addr6[2]) Index: iftop.c =================================================================== RCS file: /home/pdw/vcvs/repos/iftop/iftop.c,v retrieving revision 1.56 diff -u -r1.56 iftop.c --- iftop.c 27 Nov 2010 11:06:12 -0000 1.56 +++ iftop.c 5 Jun 2011 16:57:30 -0000 @@ -328,11 +328,11 @@ /* First reduce the participating addresses using the netfilter prefix. * We need scratch pads to do this. */ - for (j=0; j < 4; ++j) { - scribdst.s6_addr32[j] = ip6tr->ip6_dst.s6_addr32[j] - & options.netfilter6mask.s6_addr32[j]; - scribsrc.s6_addr32[j] = ip6tr->ip6_src.s6_addr32[j] - & options.netfilter6mask.s6_addr32[j]; + for (j=0; j < 16; ++j) { + scribdst.s6_addr[j] = ip6tr->ip6_dst.s6_addr[j] + & options.netfilter6mask.s6_addr[j]; + scribsrc.s6_addr[j] = ip6tr->ip6_src.s6_addr[j] + & options.netfilter6mask.s6_addr[j]; } /* Now look for any hits. */ Index: ns_hash.c =================================================================== RCS file: /home/pdw/vcvs/repos/iftop/ns_hash.c,v retrieving revision 1.4 diff -u -r1.4 ns_hash.c --- ns_hash.c 27 Nov 2010 11:06:12 -0000 1.4 +++ ns_hash.c 5 Jun 2011 16:57:30 -0000 @@ -29,7 +29,7 @@ int ns_hash_hash(void* key) { int hash; - uint32_t* addr6 = ((struct in6_addr *) key)->s6_addr32; + uint32_t* addr6 = (uint32_t*)((struct in6_addr *) key)->s6_addr; hash = ( hash_uint32(addr6[0]) + hash_uint32(addr6[1]) Index: options.c =================================================================== RCS file: /home/pdw/vcvs/repos/iftop/options.c,v retrieving revision 1.46 diff -u -r1.46 options.c --- options.c 27 Nov 2010 11:06:12 -0000 1.46 +++ options.c 5 Jun 2011 16:57:30 -0000 @@ -538,16 +538,16 @@ } else { int bl, rem; - const uint32_t mm = 0xffffffff; - uint32_t part = mm; + const uint8_t mm = 0xff; + uint8_t part = mm; - bl = n / 32; - rem = n % 32; - part <<= 32 - rem; + bl = n / 8; + rem = n % 8; + part <<= 8 - rem; for (j=0; j < bl; ++j) - options.netfilter6mask.s6_addr32[j] = htonl(mm); + options.netfilter6mask.s6_addr[j] = mm; if (rem > 0) - options.netfilter6mask.s6_addr32[bl] = htonl(part); + options.netfilter6mask.s6_addr[bl] = part; options.netfilter6 = 1; } } @@ -560,8 +560,8 @@ } } /* Prepare any comparison by masking the provided filtered net. */ - for (j=0; j < 4; ++j) - options.netfilter6net.s6_addr32[j] &= options.netfilter6mask.s6_addr32[j]; + for (j=0; j < 16; ++j) + options.netfilter6net.s6_addr[j] &= options.netfilter6mask.s6_addr[j]; return 1; }

1 0

New release?
by Eric Smith 31 May '11

31 May '11

Hi, Would it be possible to release an updated tarball soon with all of the fixes? It would make packaging iftop for my distro much easier. Thanks, Eric

1 0

Re: 1.0pre1
by Doug Chalmers 03 Apr '11

03 Apr '11

> The default make target includes something that shouldn't be there > (iftop-dump). If you remove references to iftop-dump from > Makefile.am it should work. > > I need to release a pre2 to fix all the obvious brokenness in pre1, > but keep failing to find the time. > > Paul Alright, thanks for the reply, and for your work on iftop. Doug

1 0

1.0pre1
by Doug Chalmers 01 Apr '11

01 Apr '11

I'm trying to put together a slackbuild for 1.0pre1. I was getting errors with net/bpf.h, but was able to get around that with some sed magic. What I'm having problems with now is with having to use "make iftop", instead of just "make". Just curious why this is so I can maybe get this slackbuild working properly. Thanks, Doug -- ** This message was composed on a 100% Microsoft free computer **

2 1

← Newer
1
...
4
5
6
7
8
9
10
...
20
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

iftop-users