WOLk-4.4s kernel release imminent

Hi everyone,

Users of the WOLK kernel may be interested to know that a 4.4s release
is due at some point today. The ChangeLog from the 4.3s release can be
seen below. Note that it fixes several vulnerabilities for which
advisories have been recently published (see the CAN-xxxx-xxxx entries
below).

No doubt, an incremental patch will be released so I dare say that any
existing sources can be updated with something like:

  cd /usr/src/linux
  bzcat /location/of/linux-2.4.20-wolk4.3s-to-wolk4.4s.patch.bz2 | patch -p1

--Kerin Millar (kerframil)

<mcp> Changelog from v4.3s -> v4.4s
<mcp> -----------------------------
<mcp> o   added:    High Performance Packet Classification (nf-hipac) v0.8 rev2
<mcp> o   added:    boost the copy-user asm.
<mcp> o   added:    merged IO-Stall fixes from 2.4.22-pre3/4/5/6/7
<mcp> o   fixed:    CAN-2003-0461: /proc/tty/driver/serial reveals the exact
<mcp>                 character counts for serial links. This could be used by a
<mcp>                 local attacker to infer password lengths and inter-keystroke
<mcp>                 timings during password entry.
<mcp> o   fixed:    CAN-2003-0462: Paul Starzetz discovered a file read race
<mcp>                 condition existing in the execve() system call, which could
<mcp>                 cause a local crash.
<mcp> o   fixed:    CAN-2003-0476: The execve system call in Linux 2.4.x records
<mcp>                 the file descriptor of the executable process in the file
<mcp>                 table of the calling process, allowing local users to gain
<mcp>                 read access to restricted file descriptors.
<mcp> o   fixed:    CAN-2003-0501: The /proc filesystem in Linux allows local users
<mcp>                 to obtain sensitive information by opening various entries in
<mcp>                 /proc/self before executing a setuid program. This causes the
<mcp>                 program to fail to change the ownership and permissions of
<mcp>                 already opened entries.
<mcp> o   fixed:    CAN-2003-0550: The STP protocol is known to have no security,
<mcp>                 which could allow attackers to alter the bridge topology.
<mcp>                 STP is now turned off by default.
<mcp> o   fixed:    CAN-2003-0551: STP input processing was lax in its length
<mcp>                 checking, which could lead to a denial of service.
<mcp> o   fixed:    CAN-2003-0552: Jerry Kreuscher discovered that the Forwarding
<mcp>                 table could be spoofed by sending forged packets with bogus
<mcp>                 source addresses the same as the local host.
<mcp> o   fixed:    reduced the number of requests during seeks (the latency times
<mcp>                 increased slightly during seeks with pre5/pre6).
<mcp> o   fixed:    reserved some spare request for reads. This is been measured
<mcp>                 to avoid some waiting for reads and it's beneficial in the
<mcp>                 common case
<mcp> o   fixed:    copy-namespace
<mcp> o   fixed:    RMAP: refill free list in batches, in fixup_freespace
<mcp> o   fixed:    RMAP: only wake up kswapd when low on free+clean pages
<mcp> o   fixed:    RMAP: in __alloc_pages_limit, first do rmqueue (to use per-cpu
<mcp>                       freelist) and fall back to direct_reclaim when needed
<mcp> o   fixed:    RMAP: make sure the first stage of page allocation doesn't
<mcp>                       upset the zone balancing
<mcp> o   updated:  Oracle Cluster FileSystem (OCFS) v1.0.9-pre Jul 17 2003
<mcp> o   updated:  AIO: allow aio on blkdevices too
<mcp> o   updated:  smp-timers: merged an anti deadlock fix from lcm, 2.5 probably
<mcp>                 needs it too. In short the theory that mod_timer is the only
<mcp>                 thing that can run in parallel was wrong, add_timer and
<mcp>                 del_timer/del_timer_sync can too. Having already fixed
<mcp>                 mod_timer in a backwards compatible way before merging the
<mcp>                 smp-timers in -aa, made it easy to fix those further windows
<mcp>                 too.
<mcp> o   updated:  ksoftirqd: merged a fix from Philip Craig to be sure to make
<mcp>                 the anti-DoS logic effective. He wrote and verified the code.
<mcp>                 It makes perfect sense so it's applied. Normal usages
<mcp>                 shouldn't notice the difference, especially with the
<mcp>                 max-loop logic.
<mcp> o   updated:  Intel E100 driver v2.3.13 + bugfixes from .22-BK
<mcp> o   updated:  Intel E1000 driver v5.1.11 + bugfixes from .22-BK
<mcp> o   updated:  Broadcom BCM5700 driver v6.2.11
<mcp> o   updated:  Broadcom Tigon3 v1.6
<mcp> o   updated:  SysKonnect SK-98xx driver v6.12
<mcp> o   updated:  HP CISS Driver v2.4.47
<mcp> o   updated:  Compaq SMART2 Driver v2.4.25
<mcp> o   updated:  raw vary-io 21, including more SCSI driver support
<mcp> o   updated:  Super FreeS/WAN v1.99.8 Final
<mcp> o   updated:  Qlogic QLA 2x00 v6 FC SCSI support v6.05.60
<mcp> o   removed:  LUFS: braindamaged, non-working. Get rid of that shit.

--
[email protected] mailing list