Re: about syslogd and printk()

Mike Jagdis (mike@roan.co.uk)
Tue, 21 Jul 1998 10:26:58 +0100 (GMT/BST)


On Mon, 20 Jul 1998, Andi Kleen wrote:

> Interesting. The only difference for the reading path between
> (AF_PACKET, SOCK_RAW) and (AF_PACKET, SOCK_DGRAM) is:
>
> if (sk->type != SOCK_DGRAM)
> skb_push(skb, skb->data - skb->mac.raw);
>
> Does an ordinary tcpdump work on your slip line?
>
> For me AF_PACKET,SOCK_DGRAM works fine on a ippp interface.

Actually it's SOCK_DGRAM and SOCK_PACKET not SOCK_RAW. Using SOCK_DGRAM
works on ippp0, ppp0, lo, eth0, plip0 - but not sl0. Using SOCK_PACKET
works on all. In all cases the socket was bound to the interface.
I can send diald patches and the test code if it helps. My tcpdump
doesn't use SOCK_DGRAM unfortunately (it's on my upgrade list...)

There appears to be another problem with ippp, possibly related
to masquerading, where outgoing packets get source addresses of
zero (tcpdump on the remote end shows packets from 0.0.0.0 and
the local /proc/net/rt_cache shows a source of 0x00000000). As far
as I can tell, if a masqueraded packet opens the link we get a
good source address (in this case the packet is received by diald
on its slip link and forwarded to the actual device once it comes
up). If the link is already up I seem to get the zero address
problem - ifconfig, ip link and ipchains show the expected addresses
but rt_cache and what goes out is wrong.

There appear to be zero source addr checks in tcp_v4_rebuild_header.
These are not printing anything. Nor do the dynaddr checks print
anything. Perhaps there is a rebuild header missing somewhere? Or
perhaps the ISDN code is not doing something it should be when the
link comes up? (N.B. I'm not sure if locally originated packets
show the same problem. I believe they do at times).

This is with kernel 2.1.107 and the latest ISDN code from ftp.suse.com
(isdn-9806280742.tar.gz). 2.0.35 with the corresponding 2.0 version
of the ISDN code works fine (with a small hack to make ippp interfaces
use IFF_SOFTHEADERS like ppp).

Mike

-- 
.----------------------------------------------------------------------.
|  Mike Jagdis                  |  Internet:  mailto:mike@roan.co.uk   |
|  Roan Technology Ltd.         |                                      |
|  54A Peach Street, Wokingham  |  Telephone:  +44 118 989 0403        |
|  RG40 1XG, ENGLAND            |  Fax:        +44 118 989 1195        |
`----------------------------------------------------------------------'

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html