For virtual interfaces created using the ifcfg-ethN:X scripts the
ifconfig and netstat -i report statistics of the last real adapter
listed in /proc/dev/net.

URL: https://bugzilla.redhat.com/143981

The display of "netstat -o" (on|off|keepalive|unkn-X) is not accurate.
I think that "unkn-4" should be "probe0" that is working "probe0 timer
(= ICSK_TIME_PROBE0 status = TCP Window Size is Zero)" in kernel.

URL: https://bugzilla.redhat.com/466845

A bunch of these funcs are only used by interface.c, so drop them from
interface.h and mark them static.  Also mark a bunch of buffers as const
when they aren't written to.

Yay for the ABI breakage.

Signed-off-by: Catalin Patulea <cat@vv.carleton.ca>

After coverity tool scan, some warnings about signed/unsigned conflicts in
sscanf format string were found. Not a serious problem, but better to be
correct.

RHBZ#1032563

Patch from TAKADA Yoshihito
https://bugzilla.redhat.com/show_bug.cgi?id=1021109

They are needed to be able to install binaries into
/usr/bin & /usr/sbin instead of /bin & /sbin.

1) Fall-back to numeric address if getnameinfo fails.
   Reverse lookup is not mandatory, therefore its fail
   is not an error. Just return numeric address in that case.
   This makes netstat/route show IPv6 address instead of
   [UNKNOWN] in case of DNS problems.

2) Pass length of 'name' buffer into function.
   'name' is a pointer and therefore sizeof(name)
   returns size of pointer and not size of the buffer.
   see http://stackoverflow.com/questions/14298710/c-pointers-and-arrays-sizeof-operator
   The sizeof() usage was added with commit 604785ad,
   so I checked all the other changes in that commit
   and they seem to be OK.

3) remove unused 's' variable

Thanks to Elias Probst for mentioning the problem.

Signed-off-by: Lehner Florian <dev@der-flo.net>

Signed-off-by: Lehner Florian <dev@der-flo.net>

[PATCH] replace strcpy() with safe_strncpy()

Signed-off-by: Lehner Florian <dev@der-flo.net>

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

Currently, we pass around sockaddr to various locations where they
might case things to IPv4 or IPv6 addresses.  In the latter case,
this struct is not large enough to hold an IPv6 address, so we end
up smashing the stack.

For example:
	ifconfig sit1 inet6 pointopoint ::1.2.3.4
This calls ap->input() with an IPv6 address but a sockaddr struct.
Internally, we cast this to sockaddr_in6 and then write too much.

Rather than fixing this one cast, change the code to use the new
sockaddr_storage struct.  Then point sockaddr_in and sockaddr_in6
to that location.  This way we can use "sa" everywhere and then
use "sin" and "sin6" automatically (since they point to the same
storage).

URL: http://sourceforge.net/p/net-tools/bugs/3/Signed-off-by: Mike Frysinger <vapier@gentoo.org>

An IPv4 address is 32bits.  Using unsigned long to represent that results
in misbehavior on 64bit hosts as sizeof(ulong) == 8.  Use in_addr_t in its
place to represent IPv4 addresses.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

Signed-off-by: Mike Frysinger <vapier@gentoo.org>