| 1 |
Hello all, |
| 2 |
|
| 3 |
well finally the 2.4.20-wolk4.4s kernel has been released! It can be |
| 4 |
obtained here: http://sourceforge.net/projects/wolk. It was released |
| 5 |
last night, but if you had downloaded the patch between then and now |
| 6 |
then you should probably grab it again, as there was a problem with |
| 7 |
the new v0.99Za 3c59x (3Com) driver which has been rectified. |
| 8 |
|
| 9 |
The ebuild in Portage still installs 4.3s. The sources can be patched |
| 10 |
with the linux-2.4.20-wolk4.3s-to-4.4s patch without a hitch. |
| 11 |
|
| 12 |
If you want bootsplash, supermount or win4lin support then look here: |
| 13 |
http://wolk.sf.net/Workstation-Edition. Please note that if you had |
| 14 |
already applied any of the "workstation" patches to a 4.3 tree then |
| 15 |
there is *no* need to remove them, or refresh the sources before |
| 16 |
applying the 4.4 patch. |
| 17 |
|
| 18 |
The md5sums for the 4.4 patch and ChangeLog follow. Enjoy! |
| 19 |
|
| 20 |
--Kerin Francis Millar (kerframil) |
| 21 |
|
| 22 |
6fbff6c3c6c0e298eb203d6c2af36772 *linux-2.4.20-wolk4.3s-to-4.4s.patch.bz2 |
| 23 |
48ee6b8c23aa9d137ae5cc02c0b9882d *linux-2.4.20-wolk4.3s-to-4.4s.patch.gz |
| 24 |
|
| 25 |
Changelog from v4.3s -> v4.4 |
| 26 |
- ----------------------------- |
| 27 |
o added: High Performance Packet Classification (nf-hipac) v0.8 rev2 |
| 28 |
o added: boost the copy-user asm. |
| 29 |
o added: merged IO-Stall fixes from 2.4.22-pre3/4/5/6/7 |
| 30 |
o added: q->full, defaults to off and keeps the elvtune changes. |
| 31 |
So to turn on the q->full low latency fixes, you need to: |
| 32 |
"elvtune -b 1 /dev/xxxx" . Note that for lvm and md, you need |
| 33 |
to elvtune each underlying device. Running it on an lvm/md |
| 34 |
device doesn't do anything. |
| 35 |
o added: low latency / normal / max throughput - I/O elevator selection. |
| 36 |
o fixed: CAN-2003-0461: /proc/tty/driver/serial reveals the exact |
| 37 |
character counts for serial links. This could be used by a |
| 38 |
local attacker to infer password lengths and inter-keystroke |
| 39 |
timings during password entry. |
| 40 |
o fixed: CAN-2003-0462: Paul Starzetz discovered a file read race |
| 41 |
condition existing in the execve() system call, which could |
| 42 |
cause a local crash. |
| 43 |
o fixed: CAN-2003-0476: The execve system call in Linux 2.4.x records |
| 44 |
the file descriptor of the executable process in the file |
| 45 |
table of the calling process, allowing local users to gain |
| 46 |
read access to restricted file descriptors. |
| 47 |
o fixed: CAN-2003-0501: The /proc filesystem in Linux allows local users |
| 48 |
to obtain sensitive information by opening various entries in |
| 49 |
/proc/self before executing a setuid program. This causes the |
| 50 |
program to fail to change the ownership and permissions of |
| 51 |
already opened entries. |
| 52 |
o fixed: CAN-2003-0550: The STP protocol is known to have no security, |
| 53 |
which could allow attackers to alter the bridge topology. |
| 54 |
STP is now turned off by default. |
| 55 |
o fixed: CAN-2003-0551: STP input processing was lax in its length |
| 56 |
checking, which could lead to a denial of service. |
| 57 |
o fixed: CAN-2003-0552: Jerry Kreuscher discovered that the Forwarding |
| 58 |
table could be spoofed by sending forged packets with bogus |
| 59 |
source addresses the same as the local host. |
| 60 |
o fixed: reduced the number of requests during seeks (the latency times |
| 61 |
increased slightly during seeks with pre5/pre6). |
| 62 |
o fixed: reserved some spare request for reads. This is been measured |
| 63 |
to avoid some waiting for reads and it's beneficial in the |
| 64 |
common case |
| 65 |
o fixed: copy-namespace |
| 66 |
o fixed: RMAP: refill free list in batches, in fixup_freespace |
| 67 |
o fixed: RMAP: only wake up kswapd when low on free+clean pages |
| 68 |
o fixed: RMAP: in __alloc_pages_limit, first do rmqueue (to use per-cpu |
| 69 |
freelist) and fall back to direct_reclaim when needed |
| 70 |
o fixed: RMAP: make sure the first stage of page allocation doesn't |
| 71 |
upset the zone balancing |
| 72 |
o fixed: RANDOM: another SMP deadlock. I really wonder how many (read: |
| 73 |
tons * 2^99) deadlocks we have in the Linux kernel! |
| 74 |
I am _shouting_ out for a lock free kernel!! ;) |
| 75 |
o updated: Oracle Cluster FileSystem (OCFS) v1.0.9-pre Jul 17 2003 |
| 76 |
o updated: AIO: allow aio on blkdevices too |
| 77 |
o updated: smp-timers: merged an anti deadlock fix from lcm, 2.5 probably |
| 78 |
needs it too. In short the theory that mod_timer is the only |
| 79 |
thing that can run in parallel was wrong, add_timer and |
| 80 |
del_timer/del_timer_sync can too. Having already fixed |
| 81 |
mod_timer in a backwards compatible way before merging the |
| 82 |
smp-timers in -aa, made it easy to fix those further windows |
| 83 |
too. |
| 84 |
o updated: ksoftirqd: merged a fix from Philip Craig to be sure to make |
| 85 |
the anti-DoS logic effective. He wrote and verified the code. |
| 86 |
It makes perfect sense so it's applied. Normal usages |
| 87 |
shouldn't notice the difference, especially with the |
| 88 |
max-loop logic. |
| 89 |
o updated: Intel E100 driver v2.3.18-k1 + bugfixes from .22-BK |
| 90 |
o updated: Intel E1000 driver v5.1.11-k1 + bugfixes from .22-BK |
| 91 |
o updated: Broadcom BCM5700 driver v6.2.11 |
| 92 |
o updated: Broadcom Tigon3 v1.6 |
| 93 |
o updated: SysKonnect SK-98xx driver v6.12 |
| 94 |
o updated: HP CISS Driver v2.4.47 |
| 95 |
o updated: Compaq SMART2 Driver v2.4.25 |
| 96 |
o updated: raw vary-io 21, including more SCSI driver support |
| 97 |
o updated: Super FreeS/WAN v1.99.8 Final |
| 98 |
o updated: Qlogic QLA 2x00 v6 FC SCSI support v6.05.60 |
| 99 |
o updated: 3com 3c59x v0.99Za (vanilla ones is over 2 years old) |
| 100 |
o removed: LUFS: braindamaged, non-working. Get rid of that shit. |
| 101 |
o changed: DRM v4.3 is now seperate from DRM 4.0/4.1/4.2 |
Archives