Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756531AbZKRJDt (ORCPT ); Wed, 18 Nov 2009 04:03:49 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752485AbZKRJDr (ORCPT ); Wed, 18 Nov 2009 04:03:47 -0500 Received: from mail-px0-f180.google.com ([209.85.216.180]:54490 "EHLO mail-px0-f180.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755450AbZKRJDn convert rfc822-to-8bit (ORCPT ); Wed, 18 Nov 2009 04:03:43 -0500 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=H0RP40HHqiDoRjzyzeuuRha+TML5+n9oYUYcfUOhqRI02gAlYtzRKXNF9g4mzqZG2m bIun04165WWczV1Ck0oZc8OU8tSX3euIHt2S7/0Yyt7qpm/jHZzP5NxOoyhU9SNwvuJv FEihUnFfngc0K5QNT6wlcphtLN7myZFiggq0U= MIME-Version: 1.0 In-Reply-To: <4B038BB7.2050902@qualcomm.com> References: <4B035081.8040900@qualcomm.com> <4B038BB7.2050902@qualcomm.com> Date: Wed, 18 Nov 2009 20:03:49 +1100 Message-ID: Subject: Re: [PATCH] Documentation, clarify tuntap IPX example. From: andrew hendry To: Max Krasnyansky Cc: "netdev@vger.kernel.org" , linux-kernel Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 6692 Lines: 152 Hi Max, Thanks. TUNSETLINK looks like exactly what I need. I'll have to have a closer look at kernel x.25 next, i don't think it likes the tunx25 going away with routes still too it. deb32:~# cat /proc/net/x25/route Address Digits Device 000000000000000 0 @�� @� [ 662.425648] BUG: unable to handle kernel NULL pointer dereference at 00000017 [ 662.426291] IP: [] dup_fd+0x21f/0x2fa [ 662.427849] *pde = 00000000 [ 662.428028] Oops: 0002 [#1] SMP [ 662.428028] last sysfs file: /sys/class/net/lo/operstate [ 662.428028] Modules linked in: x25 tun ipv6 loop ac button parport_pc parport snd_pcm snd_timer serio_raw psmouse snd soundcore snd_page_alloc pcspkr i2c_piix4 i2c_core evdev ext3 jbd mbcache ide_cd_mod cdrom ide_gd_mod ata_generic ata_piix libata scsi_mod ide_pci_generic piix floppy pcnet32 mii ide_core thermal fan thermal_sys [last unloaded: tun] [ 662.428028] [ 662.428028] Pid: 12252, comm: net.agent Not tainted (2.6.31.6 #1) VirtualBox [ 662.428028] EIP: 0060:[] EFLAGS: 00010286 CPU: 0 [ 662.428028] EIP is at dup_fd+0x21f/0x2fa [ 662.428028] EAX: 00000017 EBX: ce5262a0 ECX: 00000220 EDX: ffffffff [ 662.428028] ESI: 00000088 EDI: ce41e400 EBP: 00000100 ESP: c2849eec [ 662.428028] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0069 [ 662.428028] Process net.agent (pid: 12252, ti=c2848000 task=ce9d51c0 task.ti=c2848000) [ 662.428028] Stack: [ 662.428028] 00000202 00000020 00000008 cf466cc0 cf466cc4 cf4667c0 ce41e400 00000078 [ 662.428028] <0> ce5262c0 cf466d00 cf4667c8 ce02bc00 ce18db60 00000000 00000000 ce9d51c0 [ 662.428028] <0> c102d03b 080a8120 cf649668 c2849fb4 bfe05a6c 01200011 ce18dca4 fffffff4 [ 662.428028] Call Trace: [ 662.428028] [] ? copy_process+0x469/0xf96 [ 662.428028] [] ? do_fork+0x115/0x282 [ 662.428028] [] ? vfs_llseek+0x30/0x34 [ 662.428028] [] ? sys_clone+0x21/0x27 [ 662.428028] [] ? syscall_call+0x7/0xb [ 662.428028] Code: 4c 24 08 8b 70 08 f3 a5 8b 4c 24 04 83 e1 03 74 02 f3 a4 31 c9 31 f6 89 6c 24 1c eb 29 8b 7c 24 18 8b 14 0f 85 d2 74 09 8d 42 18 <3e> ff 42 18 eb 06 8b 43 0c 0f b3 30 8b 44 24 2c 46 89 14 08 83 [ 662.428028] EIP: [] dup_fd+0x21f/0x2fa SS:ESP 0069:c2849eec [ 662.428028] CR2: 0000000000000017 [ 662.452941] ---[ end trace aab1610cb3ca8b89 ]--- Message from syslogd@deb32 at Nov 19 06:58:56 ... kernel:[ 904.324109] unregister_netdevice: waiting for xotdtun to become free. Usage count = -3 Andrew On Wed, Nov 18, 2009 at 4:52 PM, Max Krasnyansky wrote: > andrew hendry wrote: >> Hi Max, >> >> I started looking through the documentation because im looking to use >> tun for X.25 over TCP. The IPX example sounded similar but i couldn't >> get it to work as I was expecting. Looking for a hint in the right >> direction. > Ah, X25 is different. > >> There is an old non-mainline patch called x25tap (clone of obsolete >> ethertap). I'm hoping to replace x25tap with tun. >> >> The old x25tap creates a device like: >> x25tap0 Link encap:generic X.25 HWaddr >> UP BROADCAST RUNNING NOARP MULTICAST MTU:1024 Metric:1 >> RX packets:3040785 errors:0 dropped:0 overruns:0 frame:0 >> TX packets:2895530 errors:0 dropped:0 overruns:0 carrier:0 >> collisions:0 txqueuelen:10 >> RX bytes:138787339 (132.3 Mb) TX bytes:91395983 (87.1 Mb) >> When the x25tap is loaded it has dev->type = ARPHRD_X25, after which >> x.25 routing and AF_X25 sockets can be created. Along with an XoT >> userspace program which handles some basic headers and TCP port 1998 >> it works well. >> >> If using tun, how does something get registered to handle protocol AF_X25? >> > > I've never tried this personally so ... :). But I just glanced over net/ax25 > and drivers/wan/ax25 and it looks TUNSETLINK might work for you. > In other words you want to allocate TUN device (not TAP) and then call > TUNSETLINK with ARPHRD_X25. Seems like you'll get exactly what you need. > > Max > > >> Regards, >> Andrew. >> >> On Wed, Nov 18, 2009 at 12:40 PM, Max Krasnyansky wrote: >>> On 11/17/2009 05:30 PM, andrew hendry wrote: >>>> Can the TUNSETIFF ioctl change a tap's protocol to IPX as the >>>> documentation suggests? >>>> I think tun.c would need IFF_IPX_TAP added for it to work as described? >>>> Otherwise tap can only be ptp or ethernet, and there is no way to >>>> route or use AF_IPX. >>> TAP is an Ethernet device. No special handling is required for IPX or for >>> that matter any other protocol. Example seems fine too. >>> >>> Max >>> >>>> Signed-off-by: Andrew Hendry >>>> >>>> --- a/Documentation/networking/tuntap.txt 2009-11-11 >>>> 14:03:22.676167648 +1100 >>>> +++ b/Documentation/networking/tuntap.txt 2009-11-18 >>>> 11:34:18.106647029 +1100 >>>> @@ -127,12 +127,14 @@ Ethernet device, which instead of receiv >>>> media, receives them from user space program and instead of sending >>>> packets via physical media sends them to the user space program. >>>> >>>> -Let's say that you configured IPX on the tap0, then whenever >>>> -the kernel sends an IPX packet to tap0, it is passed to the application >>>> -(VTun for example). The application encrypts, compresses and sends it to >>>> -the other side over TCP or UDP. The application on the other side >>>> decompresses >>>> -and decrypts the data received and writes the packet to the TAP device, >>>> -the kernel handles the packet like it came from real physical device. >>>> +Let's say for the purpose of example, IPX support was added to tuntap. >>>> +Then whenever the kernel routes an IPX packet to tap0, it is passed to >>>> the >>>> +application reading the file descriptor from /dev/net/tun (VTun for >>>> example). >>>> +The application encrypts, compresses and sends it to the other side over >>>> TCP >>>> +or UDP. The application on the other side decompresses and decrypts the >>>> data >>>> +received and writes the packet to the TAP device, the remote kernel >>>> handles >>>> +the packet like it came from real physical device. The IPX applications >>>> are >>>> +able to communicate as if there was a real IPX network. >>>> >>>> 4. What is the difference between TUN driver and TAP driver? >>>> TUN works with IP frames. TAP works with Ethernet frames. >>> > > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/