2006/07/02 : 2.4.32-wt6
  * fixed a small inode leak on ext3
  * updated epoll-lt to 0.23 (lockless)
  * updated squashfs to 3.0. It now lacks support for 1.x format but is
    still compatible with 2.x.

2006/07/01 : 2.4.32-wt5
  * updated to 2.4.33-rc2 which permitted to remove many pending patches
  * removed 2.6-binfmt_elf-zero-filesize-segments as those executables are
    allowed and *do* exist
  * replaced PaX with just the ASLR portion of PaX which is far less intrusive.
  * returned to standard strict overcommit, with stats
  * added the lowmem 3G address space extension
  * another fix for ip_set to allow removal of all bindings
  * disabled annoying TCP debug messages

2006/02/28 : 2.4.32-wt4
  * fixed build of lm_sensors
  * merged strict-overcommit together with PAX

2006/02/26 : 2.4.32-wt3
  * updated to 2.4.33-pre2 which fixes a few vulnerabilities
  * build fix for PPC64
  * rediffed PAX
  * ramfs-limits now working fine
  * added config_ip_stealth to hide closed ports
  * updated nsawdt to 0.2
  * merged small fixes into libata
  * bumped forcedeth to 0.42
  * some ethtool fixes for e1000
  * sk98lin updated to 8.31.2.3
  * added a config option to make the boot logo optional
  * updated i2c+lm_sensors to 2.10.0
  * JFFS2 updates from uClinux
  * two ip_set fixes for a panic and unremovable bindings
  * the bonding driver now reports its link status via ethtool

2005/12/06 : 2.4.32-wt2
  * removed unused patches
  * updated kernel to 2.4.32-git-20051205
  * build fix for IPv4 when brnf is disabled
  * build fix for ipt_route when IPT_NAT is disabled
  * added the vmwarefb framebuffer driver from DirectFB
  * made the gcc size optimizations compatible with 3.3/3.4

2005/12/04 : 2.4.32-wt1
  * fixed long-standing epoll() bug leading to lost events
  * added /proc/boot/{bootdev,opmode} settable from cmdline
  * added poll/epoll optimization for jiffies computation
  * implemented ramfs limits
  * kernel can now be built without any keyboard driver at all
  * tcp_timewait timeout lowered from 60 to 15 seconds
  * many updates :
    - kernel 2.4.32 + various fixes, pax-20051112
    - patch-o-matic-ng to 20051202 + nf_log + tcp_window_tracking
    - layer7 classifier v1.5, netconsole-0.03, openswan-1.0.10
    - ebtables-brnf-11-2, loop-aes-v3.1b, lm-sensors-2.9.2
    - libata-2.4.32 (large 2.6 backport), squashfs v2.2
    - sk98lin driver v8.28.1.3

2005/06/02 : 2.4.31-wt1
  * updated kernel to 2.4.31
  * updated pax to 20050416
  * updated virtual servers to 2.4.30-vs1.2.10
  * updated wireless extensions to we18
  * rediffed netdev-random, loop-aes

2005/05/15 : 2.4.30-wt3
  * updated kernel to 2.4.31-pre2 which also fixes CAN-2005-1263
  * updated i2c+lm_sensors to CVS-20050512

2005/04/15 : 2.4.30-wt2
  * merged L7-filter 0.9.2 from http://l7-filter.sf.net/
  * added dondiff file from Randy Dunlap
  * fixed wrong locks in usb-bluetooth
  * added new device IDs for upcoming NVidia chipsets

2005/04/14 : 2.4.30-wt1
  * updated to 2.4.30 + 2 bk fixes
  * updated PaX to 20050307 (build fix)
  * acl+ea updated to 0.8.73
  * ebtables updated to v9
  * loop-aes updated to 3.0c.
  * i2c+lm_sensors updated to 2.9.1
  * libata1 updated to 2.4.30-pre3.
  * sk98lin driver updated to 8.16

2005/03/20 : 2.4.29-wt3
  * updated to 2.4.30-rc1 (fixes a security issue on PPP)
  * sk98 driver updated to 8.15

2005/03/06 : 2.4.29-wt2
  * updated to 2.4.30-pre2 + cset-20050304 + a few related fixes
  * updated PaX to 20050305 to fix a critical bug
  * fixed 2 bugs in netdev-random-drivers
  * rediffed lowlat
  * rediffed evfs to to remove dependency on locks_verify_area()
  * updated patch-o-matic-ng to 20050305 + tcp_window_tracking updates from 2.6
  * made openswan compilable by a cross-compiler
  * rediffed pspa
  * updated i2c + lmsensors to CVS-20050305
  * rediffed libata
  * removed megaraid patch now that it's merged in mainline
  * rediffed aic7xxx

2005/02/07 : 2.4.29-wt1
  * i2c-amd756 did not build because it was not in the exported-objs. Fixed.
  * disabled ipv6:nflog since it's broken (bad arguments passed to dump_packet)
  * backported skb_header_pointer() from 2.6 because it's needed by some of
    the patch-o-matic-ng modules

2005/02/06 : 2.4.29-wt1-pre2
  * updated all patches to match 2.4.29 and patch-o-matic-ng-20050205
  * worked around a long standing bug on sparc which shows up only when vhz-j64
    is applied
  * updated bonding to 2.6.2 (fixes one bug, allows zero-copy sending)
  * updated squashfs to 2.1, loop-aes to 3.0b, lm_sensors to 2.9.0
  * updated wireless extensions to v17, and allowed the use of wildcards when
    renaming an interface
  * SATA support for ICH5/6/7, sound fixes, acpi updates
  * automatic scsi-add/remove when plugging/unplugging usb-storage
  * faster signal delivery by avoiding systematic use of a debug register

2005/02/05 : 2.4.29-wt1-pre1
  * re-organized the build tools to be more consistent with 2.4-hf tree and
    ease maintenance.

2004/09/12 : 2.4.27-wt6
  * updated to 2.4.28-pre3
  * updated loop-aes to 2.2b
  * added the ct_stat patch from pom-ng separately
  * added config-date
  * fixed compilation on non-preempt compatible targets
  * fixed hang on alpha caused by preempt which mangled task_struct
  * occasional NFS hang with SMP+preempt seems to be fixed

2004/08/21 : 2.4.27-wt5 -> never released
  * fixed compilation on sparc due to vhz
  * added missing sparc/sparc64 syscall for tux
  * exported missing memchr() on sparc/sparc64

2004/08/19 : 2.4.27-wt4
  * fixed a wrong offset patch when apply preempt and pax
  * fix for an xfs oops
  * misc network updates & fixes, hfsc and pcnet32, NFS over TCP not
    experimental anymore, few SATA fixes
  * added a new script : 'fix' to easily unlink, copy, edit and diff a file
    in one operation
  * removed the chown-sgid patch which claimed to conform to posix while we
    already do. Linux 2.4 and 2.6 do not behave the same way when chowning
    non-executable files, but both are posix-compliant.

2004/08/18 : 2.4.27-wt3
  * upgraded base kernel to 2.4.28-pre1 + cset-20040816-2001 ;
  * upgraded bme to 0.05 ;
  * finally found and fixed the merge error causing Tux to oops when stopping ;
  * changed the workaround for the gcc-2.95 bug affecting fs/proc/array.c when
    -march=i686 was used either with SMP+PREEMPT or none of them because it
    didn't always work ;
  * fixed a random build error because of a missing 'compile.h' file during
    parallel builds, which was caused by ikconfig ;
  * added the NSAWDT driver which had been forgotten in 2.4.27-wt2.
  * cleaned up the CONTENTS file.
  * enhanced the 'apply' script to report fuzz, reverted and failed patches
    with maximum information while not being too much verbose.

2004/08/16 : 2.4.27-wt2