Dariush Debian Diary

Sat Nov 10 13:38:45 CET 2007 Realtek's r8168 driver, this one works reliably with:

04:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 01)

in contrast to lenny's r8169 driver which oops'es every ~few hours with:

------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:721!
invalid opcode: 0000 [#1]
SMP
Modules linked in: vznetdev vzethdev vzrst vzcpt vzdquota vzmon vzdev ipt_ttl xt_tcpmss ipt_TCPMSS ipt_tos ext2 mbcache xt_length xt_CLASSIFY iptable_mangle cls_u32 sch_sfq sch_htb ipt_MASQUERADE ipt_REDIRECT iptable_nat ip_nat xt_multiport ipt_REJECT xt_limit xt_tcpudp xt_state ip_conntrack nfnetlink iptable_filter ip_tables x_tables ata_piix ip_set softdog tg3 evdev parport_pc pcspkr parport rtc xfs dm_mirror dm_log dm_snapshot dm_mod raid10 raid456 xor raid1 raid0 multipath linear md_mod ide_disk ahci libata sd_mod 8139too jmicron ehci_hcd aic7xxx uhci_hcd e100 8139cp generic scsi_transport_spi mii r8169 ide_core scsi_mod usbcore thermal processor fan
CPU:    3, VCPU: -1.3
EIP:    0060:[<c0220b2d>]    Not tainted VLI
EFLAGS: 00010202   (2.6.18.bsd40l-vz #1)
EIP is at pskb_expand_head+0x20/0x115
eax: 00000001   ebx: f58a2600   ecx: ffffffe2   edx: f71b6900
esi: f71b6800   edi: 00000000   ebp: 00000002   esp: f7a45e24
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 0, veid: 0, ti=f7a44000 task=f7a9e640 task.ti=f7a44000)
Stack: f6ea9d8c f7a5991c f58a2600 f78172e0 f7817000 00000002 c022120a 00000020
       02000000 0000003a f78172e0 f7817000 f58a2600 f8a28674 f71b6910 f6f6e000
       f7a59918 f687b600 0000003c f6f6e3c0 f8a7c000 00000000 dd66a540 f687b600
 Call Trace:
 [<c022120a>] skb_pad+0xa0/0x10b
 [<f8a28674>] rtl8169_start_xmit+0x222/0x370 [r8169]
 [<c0224f6f>] dev_hard_start_xmit+0x1a7/0x1fd
 [<c0232d25>] __qdisc_run+0xf5/0x1d1
 [<c0226974>] dev_queue_xmit+0x142/0x24a
 [<c025e717>] arp_xmit+0x4e/0x50
 [<c025fb66>] arp_solicit+0x14e/0x158
 [<c022b9b9>] neigh_timer_handler+0x247/0x2aa
 [<c022b772>] neigh_timer_handler+0x0/0x2aa
 [<c0124820>] run_timer_softirq+0x10e/0x177
 [<c012067e>] __do_softirq+0x84/0x109
 [<c0120739>] do_softirq+0x36/0x3a
 [<c010344b>] apic_timer_interrupt+0x1f/0x24
 [<c0101842>] mwait_idle+0x25/0x38
 [<c0101807>] cpu_idle+0x5e/0x74
Code: 00 00 5e e9 f7 08 f4 ff 5b 5e c3 55 57 89 d7 56 53 89 c3 83 ec 08 8b 90 b0 00 00 00 8b b0 a4 00 00 00 8b 80 a0 00 00 00 48 74 0b <0f> 0b 66 b8 d1 02 b8 a3 77 2a c0 29 f2 8d 44 17 1f 8b 54 24 1c
EIP: [<c0220b2d>] pskb_expand_head+0x20/0x115 SS:ESP 0068:f7a45e24
Kernel panic - not syncing: Fatal exception in interrupt
.....
....
.....
Unbalanced enable for IRQ 169
BUG: warning at kernel/irq/manage.c:118/enable_irq()
 [<c01485a9>] enable_irq+0x54/0x87
 [<f8a9d8ba>] probe_hwif+0x556/0x627 [ide_core]
 [<c020c4b8>] __driver_attach+0x0/0x5d
 [<f8a9e1b5>] probe_hwif_init_with_fixup+0xb/0x63 [ide_core]
 [<f8a9ff89>] ide_setup_pci_device+0x3c/0x6d [ide_core]
 [<f8a2411e>] generic_init_one+0x92/0xa0 [generic]
 [<c01bd75c>] pci_device_probe+0x36/0x57
 [<c020c419>] driver_probe_device+0x42/0x8b
 [<c020c4f0>] __driver_attach+0x38/0x5d
 [<c020bf3a>] bus_for_each_dev+0x33/0x55
 [<c020c383>] driver_attach+0x11/0x13
 [<c020c4b8>] __driver_attach+0x0/0x5d
 [<c020bc53>] bus_add_driver+0x64/0xfd
 [<c01bd898>] __pci_register_driver+0x47/0x63
 [<c013c175>] sys_init_module+0x16c9/0x184c
 [<c016c454>] cp_new_stat64+0xfd/0x10f
 [<f8a3cdf8>] e100_probe+0x2ea/0x5f7 [e100]
 [<f8a3c7b0>] e100_get_wol+0xc/0x2d [e100]
 [<f8a3c6c0>] e100_get_drvinfo+0x1b/0x4c [e100]
 [<f8a3c7d8>] e100_set_wol+0x7/0x41 [e100]
 [<f8a3b000>] e100_setup_ucode+0x0/0xcc [e100]
 [<f8a3c494>] e100_tx_clean+0x3a/0xb8 [e100]
 [<f8a3c800>] e100_set_wol+0x2f/0x41 [e100]
 [<c010297f>] syscall_call+0x7/0xb
------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:721!
invalid opcode: 0000 [#1]
SMP
Modules linked in: vznetdev vzethdev vzrst vzcpt vzdquota vzmon vzdev ipt_ttl xt_tcpmss ipt_TCPMSS ipt_tos ext2 mbcache xt_length xt_CLASSIFY iptable_mangle cls_u32 sch_sfq sch_htb ipt_MASQUERADE ipt_REDIRECT iptable_nat ip_nat xt_multiport ipt_REJECT xt_limit xt_tcpudp xt_state ip_conntrack nfnetlink iptable_filter ip_tables x_tables ata_piix ip_set softdog tg3 rtc parport_pc evdev parport pcspkr xfs dm_mirror dm_log dm_snapshot dm_mod raid10 raid456 xor raid1 raid0 multipath linear md_mod ide_disk sd_mod ahci libata jmicron 8139too ehci_hcd e100 aic7xxx scsi_transport_spi scsi_mod uhci_hcd generic ide_core 8139cp r8169 mii usbcore thermal processor fan
CPU:    0, VCPU: -1.0
EIP:    0060:[<c0220b2d>]    Not tainted VLI
EFLAGS: 00010202   (2.6.18.bsd40l-vz #1)
EIP is at pskb_expand_head+0x20/0x115
eax: 00000001   ebx: da11a980   ecx: ffffffe2   edx: eca99f00
esi: eca99e00   edi: 00000000   ebp: 00000002   esp: c030fe48
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 0, veid: 0, ti=c030e000 task=c02c26c0 task.ti=c030e000)
Stack: f74a7d8c f6ebeadc da11a980 f78fc2e0 f78fc000 00000002 c022120a 00000020
       02000000 0000003a f78fc2e0 f78fc000 da11a980 f8a47674 eca99f10 f6f3f000
       f6ebead8 f6c97600 0000003d f6f3f3d0 f8a2a000 00000000 f0f7cc80 f6c97600
 Call Trace:
 [<c022120a>] skb_pad+0xa0/0x10b
 [<f8a47674>] rtl8169_start_xmit+0x222/0x370 [r8169]
 [<c0224f6f>] dev_hard_start_xmit+0x1a7/0x1fd
 [<c0232d25>] __qdisc_run+0xf5/0x1d1
 [<c0226974>] dev_queue_xmit+0x142/0x24a
 [<c025e717>] arp_xmit+0x4e/0x50
 [<c025fb66>] arp_solicit+0x14e/0x158
 [<c022b9b9>] neigh_timer_handler+0x247/0x2aa
 [<c022b772>] neigh_timer_handler+0x0/0x2aa
 [<c0124820>] run_timer_softirq+0x10e/0x177
 [<c012067e>] __do_softirq+0x84/0x109
 [<c0120739>] do_softirq+0x36/0x3a
 [<c010344b>] apic_timer_interrupt+0x1f/0x24
 [<c0101842>] mwait_idle+0x25/0x38
 [<c0101807>] cpu_idle+0x5e/0x74
 [<c0314723>] start_kernel+0x39c/0x3a3
Code: 00 00 5e e9 f7 08 f4 ff 5b 5e c3 55 57 89 d7 56 53 89 c3 83 ec 08 8b 90 b0 00 00 00 8b b0 a4 00 00 00 8b 80 a0 00 00 00 48 74 0b <0f> 0b 66 b8 d1 02 b8 a3 77 2a c0 29 f2 8d 44 17 1f 8b 54 24 1c
EIP: [<c0220b2d>] pskb_expand_head+0x20/0x115 SS:ESP 0068:c030fe48
Kernel panic - not syncing: Fatal exception in interrupt
------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:721!
invalid opcode: 0000 [#1]
SMP
Modules linked in: vznetdev vzethdev vzrst vzcpt vzdquota vzmon vzdev ipt_ttl xt_tcpmss ipt_TCPMSS ipt_tos ext2 mbcache xt_length xt_CLASSIFY iptable_mangle cls_u32 sch_sfq sch_htb ipt_MASQUERADE ipt_REDIRECT iptable_nat ip_nat xt_multiport ipt_REJECT xt_limit xt_tcpudp xt_state ip_conntrack nfnetlink iptable_filter ip_tables x_tables ata_piix ip_set softdog tg3 parport_pc parport rtc pcspkr evdev xfs dm_mirror dm_log dm_snapshot dm_mod raid10 raid456 xor raid1 raid0 multipath linear md_mod ide_disk ahci libata jmicron sd_mod aic7xxx generic scsi_transport_spi ehci_hcd uhci_hcd e100 8139too 8139cp scsi_mod mii r8169 ide_core usbcore thermal processor fan
CPU:    1, VCPU: -1.1
EIP:    0060:[<c0220add>]    Not tainted VLI
EFLAGS: 00010202   (2.6.18.bsd40m-vz #1)
EIP is at pskb_expand_head+0x20/0x115
eax: 00000001   ebx: d2f5a500   ecx: ffffffe2   edx: d9ad1b00
esi: d9ad1a00   edi: 00000000   ebp: 00000002   esp: f7a03e24
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 0, veid: 0, ti=f7a02000 task=f7abf220 task.ti=f7a02000)
Stack: f6e73d8c f7d5bc9c d2f5a500 f78c02e0 f78c0000 00000002 c02211ba 00000020
       02000000 0000003a f78c02e0 f78c0000 d2f5a500 f8a285cc d9ad1b10 f79c7000
       f7d5bc98 f6ee3a00 00000030 f79c7300 f8a3c000 00000000 f6f26680 f6ee3a00
 Call Trace:
 [<c02211ba>] skb_pad+0xa0/0x10b
 [<f8a285cc>] rtl8169_start_xmit+0x222/0x370 [r8169]
 [<c0224f1f>] dev_hard_start_xmit+0x1a7/0x1fd
 [<c0232cd5>] __qdisc_run+0xf5/0x1d1
 [<c0226924>] dev_queue_xmit+0x142/0x24a
 [<c025e6be>] arp_xmit+0x4e/0x50
 [<c025fb0d>] arp_solicit+0x14e/0x158
 [<c022b969>] neigh_timer_handler+0x247/0x2aa
 [<c022b722>] neigh_timer_handler+0x0/0x2aa
 [<c0124820>] run_timer_softirq+0x10e/0x177
 [<c012067e>] __do_softirq+0x84/0x109
 [<c0120739>] do_softirq+0x36/0x3a
 [<c010344b>] apic_timer_interrupt+0x1f/0x24
 [<c0101842>] mwait_idle+0x25/0x38
 [<c0101807>] cpu_idle+0x5e/0x74
Code: 00 00 5e e9 47 09 f4 ff 5b 5e c3 55 57 89 d7 56 53 89 c3 83 ec 08 8b 90 b0 00 00 00 8b b0 a4 00 00 00 8b 80 a0 00 00 00 48 74 0b <0f> 0b 66 b8 d1 02 b8 72 77 2a c0 29 f2 8d 44 17 1f 8b 54 24 1c
EIP: [<c0220add>] pskb_expand_head+0x20/0x115 SS:ESP 0068:f7a03e24
Kernel panic - not syncing: Fatal exception in interrupt

Thu Nov 1 13:38:45 CET 2007 Backports for r8169 support, JMicron PATA support. Various fixes.. OpenVZ 028stab047.1 plus compilation fix

time bzip2 -dc /tmp/ecos.tar.bz2 | ../lzma -x -b64 -s25 > ecos.tar.lzma

real    23m31.643s
user    19m17.570s
sys     0m18.510s

eyck@dev:/tmp/lzma-0.04/tst$ du -sh *
22M     ecos.tar.lzma
52K     lzma-0.04.tar.bz2
64K     lzma-0.04.tar.gz
52K     lzma-0.04.tar.lzma
eyck@dev:/tmp/lzma-0.04/tst$ du -sh /tmp/ecos.tar.bz2
35M     /tmp/ecos.tar.bz2

seems interesting, should be able to beat rzip on smaller files.

WAR (What A Rack!) - Rack management/accounting interface.

Tue May 24 18:08:27 CEST 2005

bsd29b
1. security fix - vserver barrier fixed.

Fri Jan 21 7:55:24 CET 2005

2.4.29 brakes my X (radeon), thus BSD is only at 2.4.28 with security fixes. I'm testing 2.4.29 with few lines backed out, this should work. The patch is ~10 lines long.

Fri Aug 6 12:17:08 CEST 2004 (6 Aug 2004):

1. Stable: 18f, upgrades openwall to ow3, removes fixes for i387 (included in ow3). This is still based on 2.4.25.
2. New stable is getting ready: 20a. Basically this is 21n with ocfs and e2compr riped out.
3. New dev: 21n is coming..., this will come with drbd 0.7.x, which works with xfs..

(4 Jul 2004):

1. Stable: 18e, includes fixed vserver (vs1.28) and fixed openswan (1.0.6), I'm going to update stable branch to 2.4.26 base, this means 20a release is coming soon, will be based on 21m with ext2 compression ripped off. The rest of 21m seem to be solid enough to move to dead branch.
2. Dev: 21m, includes fixed vserver (vs1.28) and fixed openswan (1.0.6),

Two releases (7 Apr 2004):

1. Stable: 18a,
2. Development: 19a, start of new development tree, at the moment it contains device mapper for lvm2

   UPDATE: Development is now at 21XXX, based off of new(huh) 2.4.26, currently at revision 21k 21k.html

Links: http://www.rshk.co.uk/projects/cpucap.php

NAME
       Net::Lite::FTP - Perl FTP client

SYNOPSIS
              use Net::Lite::FTP;
	      my $tlsftp=Net::Lite::FTP->new();
	      $tlsftp->open("ftp.tls.pl","21");
	      $tlsftp->user("user");
	      $tlsftp->pass("password");
	      $tlsftp->cwd("pub"); 
	      my $files=$tlsftp->nlst("*.exe");
	      foreach $f (@files) {
	      	$tlsftp->get($f); 
	      };

DESCRIPTION
	Very simple FTP client with support for TLS
See more ...


	

Network Working Group                                        M. Horowitz
Request for Comments: 2228                              Cygnus Solutions
Updates: 959                                                     S. Lunt
Category: Standards Track                                       Bellcore
                                                            October 1997

                        FTP Security Extensions

Status of this Memo

   This document specifies an Internet standards track protocol for the
   Internet community, and requests discussion and suggestions for
   improvements.  Please refer to the current edition of the "Internet
   Official Protocol Standards" (STD 1) for the standardization state
   and status of this protocol.  Distribution of this memo is unlimited.

Copyright Notice

   Copyright (C) The Internet Society (1997).  All Rights Reserved.

Abstract

   This document defines extensions to the FTP specification STD 9, RFC
   959, "FILE TRANSFER PROTOCOL (FTP)" (October 1985).  These extensions
   provide strong authentication, integrity, and confidentiality on both
   the control and data channels with the introduction of new optional
   commands, replies, and file transfer encodings.

   The following new optional commands are introduced in this
   specification:

      AUTH (Authentication/Security Mechanism),
      ADAT (Authentication/Security Data),
      PROT (Data Channel Protection Level),
      PBSZ (Protection Buffer Size),
      CCC (Clear Command Channel),
      MIC (Integrity Protected Command),
      CONF (Confidentiality Protected Command), and
      ENC (Privacy Protected Command).

   A new class of reply types (6yz) is also introduced for protected
   replies.

   None of the above commands are required to be implemented, but
   interdependencies exist.  These dependencies are documented with the
   commands.

   Note that this specification is compatible with STD 9, RFC 959.



Horowitz & Lunt             Standards Track                     [Page 1]
See more ...


	

Fri Jan 7 11:05:47 CET 2005 Todays release (25c), https://ghost.anime.pl/~eyck/Projects/bsd/25c/, based on 2.4.28 introduces openswan 2.3.x (2.3.0), this is the first major backwards-incompatible release in BSD family, I don't know yet if it'll trickle down to stable branch.

Also, together with patchset I prepare 'default' kernel, which configuration is based on debian kernels. This time around the configuration contains proper blue support ( TTY was missing ), without it you couldn't play with bluetooth phones ("rfcomm bind /dev/rfcomm0 MAC CHANNEL" command was failing)

Next-gen FTP, fixing some problems with FTP, namely:

1. performance problems with small files
2. uptight firewall admin-friendly (use one long-lived TCP session for all comunication)

Generally FTP is very old and very evolved protocol, so all/most known problems already have been solved ( TLS/SSL FTP encryption, then you hit problem with FTP conntrack not working, then you solve it by temporarily decrypting control session etc..).

Solving those problems with XML is possible and straight-forward.

1. buttons
   • power - instat restore/suspend
   • keyboard
2. touchscreen - it's still flaky
3. apps: mplayer, rdesktop, mail, ...

Wed Jun 2 20:37:54 CEST 2004 21k is now ready/in testing https://ghost.anime.pl/~eyck/Projects/bsd/21k/, contains

• FRESH: it now compiles with gcc 3.4.0
• FRESH: it now compiles with gcc 3.3.3
• FRESH: contains gcc 3.4.0 compilation fixes from Rock Linux, plus few personal tweaks (drbd, honeynet logging etc...) (btw, drbd is still 0.6.1, while current is 0.6.12, I'm not using drbd in production so it may take me some more time before I get to updating this)
• UPDATE: ocfs updated to 0.1.9
• jfs deadlock fixed: http://linux.bkbits.net:8080/linux-2.4/cset@1.1359.20.3
• rfc2385 - md5 protection on long-running tcp sessions
• Bunch of fixes from http://www.csd.uu.se/~mikpe/linux/patches
• UPDATE: openswan updated to 1.0.4

Plus new stuff from earlier non-released bsd21XXXs:

• UPDATE: squashfs 2.0 ( GREAT STUFF! )
• FRESH: esfq 2.0
• FRESH: ext2 compression ( chattr +c ) (this is not for production, will be removed in stable branch)

This time around I'll probably release debian kernel packages compiles with gcc : 2.95, 3.3.3, 3.4.0 with and without space-conserving optimisations (-Os)

NOTE: -Os brakes stuff, for example, usbcore no longer works

	unresolved symbol memcmp
	

with 3.4.0 I still have those problems:

/lib/modules/2.4.26-bsd21k/kernel/fs/ocfs/ocfs.o: /lib/modules/2.4.26-bsd21k/kernel/fs/ocfs/ocfs.o: unresolved symbol ocfs_version
/lib/modules/2.4.26-bsd21k/kernel/net/khttpd/khttpd.o: /lib/modules/2.4.26-bsd21k/kernel/net/khttpd/khttpd.o: unresolved symbol tcp_v4_md5_lookup

To be investigated...

/dev/random ( needed for crypto, supposedly is safe to put inside vservers, because it can't be used to brake random pool, ie - you can update pool, but not in a way that could harm entrophy or sth..) /dev/tty ( used by lots of perly stuff and more ) /dev/null ( some stuff needs it, especially sh scripts )

Wed Apr 14 21:42:00 CEST 2004 Todays release (21c), https://ghost.anime.pl/~eyck/Projects/bsd/21c/, is based on recently released 2.4.26 contains

• Tiny patch fixing this:

   
   In file included from gamma_drv.c:34:
   drmP.h:180: redefinition of `__cmpxchg'
   /home/eyck/bsd/patches/21b/work/linux-2.4.26/include/asm/system.h:245: `__cmpxchg' previously defined here
   make[4]: *** [gamma_drv.o] Error 1
   make[4]: Leaving directory `/home/eyck/bsd/patches/21b/work/linux-2.4.26/drivers/char/drm
  

• vserver routing fixing patch from Herbert of Vserver fame
• floppy-fallback patch from Herbert of Vserver fame
• drbd returns to tree, relativly old version..

Also, together with patchset I prepare 'default' kernel, which configuration is based on debian kernels. This time around I added strangely missing pcmcia-support, and enabled advanced routing ( multipath ... )..

Stable release (18a) http://eyck.forumakad.pl/Projects/bsd/18a/, is just 17j renamed, there were no issues found with 17j.

http://anduin.net/jailctl/

Todays release (17i), http://eyck.forumakad.pl/Projects/bsd/17i/ contains

• lufs 0.9.7

Todays release (17h), http://eyck.forumakad.pl/Projects/bsd/17h/ contains

• updated mount -o ro,bind patch from Herbert of Vserver fame (0.4 pre-release)
• reemoved old framebuffer logos
• mppe/mppc module downgraded to avoid 'PFC bug'

Warnings:

1. netconsole is not yet final, don't use it on production machines
2. Same goes for mount -o ro,noatime,bind option

Todays release (17f), https://ghost.anime.pl/~eyck/Projects/bsd/17f/ contains

• updated mount -o ro,bind patch from Herbert of Vserver fame
• updated netconsole patch
• re-added mga_vid driver for matrox and radeon cards
• re-added old OCFS (Oracle Cluster Filesystem)
• re-added old framebuffer logos

Warnings:

1. 17f includes Jan Dubiec's newest mppe/mppc module. Unfortunatelly it contains fix for 'PFC bug', which, unless you're affected by 'PFC bug', triggers 'PFC bug'. 17g will be available with older (0.98) version of this module
2. netconsole patch should not be trusted. As Herbert Poezl says:

   11:51 < Bertl> eyck: it seems that some parts of the netpoll api ahve sneaked in for the following cards:
   11:52 < Bertl> net/8139cp.c, net/b44.c, e1000/e1000, net/gt64240eth.c, net/mv64340, net/tg3.c
   

   thus, you should be cautious with using this feature on those cards.

Release (17e), https://ghost.anime.pl/~eyck/Projects/bsd/17e/ contains

• XFS ACL - xfs enter mainline kernel, but unfortunatelly stripped-down :(, this brings back some of expected functionality
• XFS DMAPI
• epoll - /dev/epoll
• squashfs - add this line deb http://eyck.forumakad.pl/woody/squashfs/ ./ to your /etc/apt/sources.list to get some userspace squashfs-tools
• mount -o ro,bind patch( from vserver's Herbert )
• netconsole logging patch ( from vserver's Herbert )[BROKEN!]