Hello!
While building the iproute package in Debian I get warnings from
package helpers like this:
dpkg-shlibdeps: warning: package could avoid a useless dependency if debian/iproute/sbin/tc debian/iproute/usr/bin/lnstat debian/iproute/bin/ip debian/iproute/bin/ss debian/iproute/sbin/bridge debian/iproute/sbin/rtmon were not linked against libresolv.so.2 (they use none of the library's symbols)
The -lresolv in ./Makefile seems to come from pre-historic times (before
iproute2 git history, possibly from libc5/pre-glibc days).
I couldn't find out if/why there was any reason for linking to libresolv.
Does anyone know if there are any valid reasons for keeping it still?
If not, I'd be happy to see it go.... while at it I also removed includes
of <resolv.h> which I also couldn't find any reason for, but this is
just an added bonus of the patch (and there are probably more unneeded
includes that could be dropped in the same sources).
Signed-off-by: Andreas Henriksson <andreas@fatal.se>
That is -- write the code, that sends diag request in new format. It's
mostly copied from tcp-diag code. Plus, sock-diag differentiates sockets
by families, thus we have to send two requests sequentially.
If we fail to submit new sock-diag request, try to fall-back and submit
the legacy tcp-diag one.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
The existing function inet_show_netlink sends tcp-diag request and
then receives back the response and prints it on the screen.
The sock-diag and legacy tcp-diag have different request types, but
report sockets in the same format. In order to support both it's
convenient to split the code into sending and receiving parts.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
The sock-diag is capable to diag udp sockets as well. Prepare the
ss code for this by first renaming soon-to-be-generic tcp-s names
into inet-s.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
linux-3.6-rc1 supports SK_MEMINFO_BACKLOG with commit d594e987c6f54
(sock_diag: add SK_MEMINFO_BACKLOG)
ss command can display it if provided by the kernel.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Vijay Subramanian <subramanian.vijay@gmail.com>
INET_DIAG_SKMEMINFO is used to monitor socket memory information
which contains more information than INET_DIAG_MEMINFO.
-m option is retained for old kernel that don't surpport INET_DIAG_SKMEMINFO.
Signed-off-by: Shan Wei <davidshan@tencent.com>
LIBNETLINK will be defined in the main Makefile, so
both ../lib/libnetlink.a ../lib/libutil.a will be
automatically appended during linking. Otherwise
../lib/libnetlink.a ../lib/libutil.a will appear
twice during linking.
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
A new option -p is added to the arpd command that accepts
a time indicating the number of seconds
to wait between kernel arp table polling attempts.
The minimum value is .1 (100ms).
If not specified, polling defaults to 30 seconds.
Patch by Erik Hugne <erik.hugne@ericsson.com> with
modifications
ss -x currently fails if CONFIG_INET_UDP_DIAG=n or old kernels
Also close file descriptors while we are at it.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Pavel Emelyanov <xemul@parallels.com>
Get the same info as from /proc file plus the peer inode.
Applies on top of new sock diag patch and udp diag patch.
Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Both rtnl_talk and rtnl_dump had a callback for handling portions
of netlink message that do not match the correct pid or seq.
But this callback was never used by any part of iproute2 so remove
it.
Support ECNSEEN reporting in ss command.
ESTAB 0 0 10.170.73.123:4900
10.170.73.125:51001 uid:501 ino:385994 sk:f31e5f00
mem:(r0,w0,f0,t0) ts sack ecn ecnseen bic wscale:8,8 rto:210
rtt:18.75/15 ato:40 cwnd:10 send 69.9Mbps rcv_space:32768
"ecn" means TCP session negociated ECN capability (TCP layer) at setup
time
"ecnseen" at least one frame with ECT(0) or ECT(1) or ECN (IP layer) was
received from peer.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
This patch emulates 'netstat -ul' behavior, showing 'closed'
(state 07) UDP sockets when ss is called with '-ul' options.
Although dirty, this seems like the least invasive way to fix
it and shouldn't really break anything.
Signed-off-by: Petr Šabata <contyk@redhat.com>
Fixes following error. We currently provide garbage data to kernel, that
can abort the validation process or produce unexpected results.
$ ss -a autobound
State Recv-Q Send-Q Local Address:Port Peer Address:Port
TCPDIAG answers: Invalid argument
After patch:
$ misc/ss -a autobound
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 128 *:44624 *:*
ESTAB 0 0 192.168.1.21:47141 74.125.79.109:imaps
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Le lundi 23 août 2010 à 10:33 -0700, Stephen Hemminger a écrit :
> I think this breaks the wraparound detection code in this command.
>
>
OK lets fix the bug only, before adding 64bit counters capabilities.
Thanks
[PATCH] iproute2: add 64bit arches support to ifstat
ifstat assumes IFLA_STATS fields are "unsigned long", but they are
__u32. This fix is needed to let ifstat run on 64bit arches.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
In recent kernels, net_device_stats is not exposed and the code
shoulf have used rt_link_statistics. Also, fix use of sprintf
with user supplied value.
> On closer inspection, it appears that ss -p does a quadratic scan. It
> rescans every entry in /proc/*/fd/* repeatedly (once per listening
> port? per process? I don't remember what I figured out.)
>
> I humbly suggest that this is not a good idea.
Yep, this is junk. Please give this patch a try:
ss: Avoid quadradic complexity with '-p'
Scan the process list of open sockets once, and store in a hash
table to be used by subsequent find_user() calls.
Reported-by: Steve Fink <sphink@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Peter Palfrader said in http://bugs.debian.org/545008 that
"--help output, if explicitly requested, should go to stdout, not stderr."
which this patch fixes.
Additionally, the exit code was adjusted to success if help was
explicitly requested.
(Syntax error still outputs to stderr and has the same exit code.)
Signed-off-by: Andreas Henriksson <andreas@fatal.se>
A bunch of misc utils basically reimplement the daemon() function (the
whole fork/close/chdir/etc...). Rather than do that, use daemon() as
that will work under nommu Linux systems that lack fork().
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
When load the tcp_diag module, and use "ss -f inet" command to show the socket
information of inet domain. But this command also shows the information of inet6
domain, but not only inet domain. and "ss -f inet6", "ss -4", "ss -6" have the
same problem.
Signed-off-by: Li Yewang <lyw@cn.fujitsu.com>
Signed-off-by: Stephen Hemminger <stephen.hemminger@vyatta.com>
when using the "ss -f inet -t" command to show the tcp socket information of
inet domain, we get the error message: "ss: no socket tables to show with such filter."
This patch can fix this bug.
Signed-off-by: Li Yewang <lyw@cn.fujitsu.com>
Signed-off-by: Stephen Hemminger <stephen.hemminger@vyatta.com>
The current kernel generates 71 possible header fields, but
MAX_FIELDS in lnstat is only 64. This leads to referencing outside
of the array. To fix, increase size of array and chop off parsing
at MAX_FIELDS - 1.
Fix a bug of generic_proc_open(), so environment variables
(e.g. PROC_NET_SNMP, PROC_NET_RTACCT) can be used to specify procfile.
Signed-off-by: Yu Zhiguo <yuzg@cn.fujitsu.com>
The function get_hz() returns the psched hz value which is wrong
for anything other than tc usage. Should be user hz instead, but
kernel is broken (patch sent) and this code doesn't get hit on
current systems (netlink is used first).
In ss.c, generic_proc_open(), for which the net_*_open functions are just
convenient wrappers, uses fopen, so errors are signalled by a NULL return
value. Some checks were expecting negative values instead, fix them.
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de>
Signed-off-by: Stephen Hemminger <stephen.hemminger@vyatta.com>
