kientzle [Sat, 11 Dec 2004 07:11:42 +0000 (07:11 +0000)]
Be more careful about assembling/disassembling
device numbers. In particular, this should fix
a bug where archiving a device node with a very
large minor number would sometimes overflow and
corrupt the major number.
marcel [Sat, 11 Dec 2004 06:15:12 +0000 (06:15 +0000)]
Use primitive types to avoid creating an artificial header dependency:
o s/u_long/unsigned long/
o s/uint32_t/unsigned int/g
o s/uint64_t/unsigned long/g
sam [Fri, 10 Dec 2004 16:35:37 +0000 (16:35 +0000)]
fix wep key seting: ENETRESET is the code to return to signal the driver
should push software state to the hardware (was ERESTART which caused the
system call to be retried)
trhodes [Fri, 10 Dec 2004 15:24:40 +0000 (15:24 +0000)]
According to the information on:
http://www.opengroup.org/onlinepubs/009695399/functions/swab.html
the prototype for swab() should be in <unistd.h> and not in <string.h>.
Move it, and update to match SUS. Leave the prototype in string.h for
now, for backwards compat.
PR: 74751
Submitted by: Craig Rodrigues <rodrigc@crodrigues.org>
Discussed with: das
harti [Fri, 10 Dec 2004 10:13:40 +0000 (10:13 +0000)]
Remove a useless list where just all command lines are stuffed onto, never
used and just freed at the end. The idea might have been to be able
to free all the strings, but what's the point to free just before exiting?
glebius [Fri, 10 Dec 2004 07:47:17 +0000 (07:47 +0000)]
Revert last change.
Andre:
First lets get major new features into the kernel in a clean and nice way,
and then start optimizing. In this case we don't have any obfusication that
makes later profiling and/or optimizing difficult in any way.
ps [Fri, 10 Dec 2004 03:29:02 +0000 (03:29 +0000)]
In nfs_rename(), skip the otw rename operation if the fsync (to
either src or dst) fails. This closes a potential data loss case
(where the fsync failed with ENOSPC, for example).
Submitted by: Mohan Srinivasan mohans at yahoo-inc dot com
Obtained from: Yahoo!
ps [Fri, 10 Dec 2004 03:27:12 +0000 (03:27 +0000)]
Store a hint in the nfsnode to detect sequential access of the file.
Kick off a readahead only when sequential access is detected. This
eliminates wasteful readaheads in random file access.
Submitted by: Mohan Srinivasan mohans at yahoo-inc dot com
Obtained from: Yahoo!
csjp [Fri, 10 Dec 2004 02:17:18 +0000 (02:17 +0000)]
This commit adds a shared locking mechanism very similar to the
mechanism used by pfil. This shared locking mechanism will remove
a nasty lock order reversal which occurs when ucred based rules
are used which results in hard locks while mpsafenet=1.
So this removes the debug.mpsafenet=0 requirement when using
ucred based rules with IPFW.
It should be noted that this locking mechanism does not guarantee
fairness between read and write locks, and that it will favor
firewall chain readers over writers. This seemed acceptable since
write operations to firewall chains protected by this lock tend to
be less frequent than reads.
rsm [Thu, 9 Dec 2004 22:27:11 +0000 (22:27 +0000)]
Add Ethernet part of Intel EtherExpress PRO/100 LAN/Modem card. This is a
rebadged Xircom REM56 RealPort card. Short MFC timeout to beat the 4.11
code freeze.
PR: 53027
Submitted by: John Merryweather Cooper <coop9211 at uidaho dot edu>
Approved by: imp (mentor)
MFC after: 2 days
mckusick [Thu, 9 Dec 2004 21:24:00 +0000 (21:24 +0000)]
Fixes a bug that caused UFS2 filesystems bigger than 2TB to
prematurely report that they were full and/or to panic the kernel
with the message ``ffs_clusteralloc: allocated out of group''.
Submitted by: Henry Whincup <henry@jot.to>
MFC after: 1 week
emax [Thu, 9 Dec 2004 18:57:12 +0000 (18:57 +0000)]
Add workaround for {not so|too} smart phones that cannot handle SDP attribute
ID ranges that consist of exactly one attribute ID. libsdp(3) will check start
and end of the attribute ID range and if they are the same the range will be
collapsed to one atribute ID.
The problem was observed on Audiovox SMT5600 and Palm Treo 650.
mlaier [Thu, 9 Dec 2004 11:47:30 +0000 (11:47 +0000)]
Start the protocol timeouts only after all domains have been initialized
completely. For some reason (that I am still curious about) we started to no
longer manage to finish the initialization before the timeouts run the first
time leading to panics when using uninitialized mutex etc.
The root of this problem is that we currently first link a domain to the
domains list and only later initialize the domain's protocols. This should
be reworked in the future, but with the current API it is not possible in
all situations. We settle with this lazy fix for now.
harti [Thu, 9 Dec 2004 10:00:55 +0000 (10:00 +0000)]
If a path element directory was never opened it is not on the list of
open directories so there is no need to remove it from there. This fixes
a core dump introduced by removing the run-time check from Lst_Remove.
sam [Wed, 8 Dec 2004 19:18:07 +0000 (19:18 +0000)]
Overhaul to cleanup some of the tangled logic that's grown over the years.
o break per-address family support out into separate files
o modularize per-address family and functional operations using
a registration mechanism; this permits configuration according
to which files you include (but beware that order of the files
is important to insure backwards compatibility)
o many cleanups to eliminate incestuous behaviour, global variables,
and poor coding practices (still much more to fix)
The original motivation of this work was to support dynamic addition
of functionality based on the interface so we can eliminate the various
little control programs and so that vendors can distribute ifconfig
plugins that support their in-kernel code. That work is still to be
completed.
o Update 802.11 support for all the new net80211 functionality; some
of these operations (e.g. list *) may be better suited in a different
program
kbyanc [Wed, 8 Dec 2004 19:03:55 +0000 (19:03 +0000)]
If the parent process has the trap bit set (i.e. a debugger had single
stepped the process to the system call), we need to clear the trap flag
from the new frame unless the debugger had set PF_FORK on the parent.
Otherwise, the child will receive a (likely unexpected) SIGTRAP when it
executes the first instruction after returning to userland.
sam [Wed, 8 Dec 2004 18:06:09 +0000 (18:06 +0000)]
Version 0.9.14.12:
[Changes listed only since last public release 0.9.12.14; for changes
prior to that consult the CVS logs at http://madwifi.sourceforge.net]
o reorg directory structure to have a single set of public binary builds
shared by all systems
o support for new parts (all shipping pci/cardbus parts to this date work)
o new capabilities for identifying various chip features
o set/get tx power cap for supporting 802.11h information element
o revised api for set/get tx queue properties
o support for updating CTS in frames when doing packet bursting
o support for querying which tx queues have pending interrupts
sam [Wed, 8 Dec 2004 17:26:47 +0000 (17:26 +0000)]
Update 802.11 support; too much new functionality to fully describe
here but it includes completed 802.11g, WPA, 802.11i, 802.1x, WME/WMM,
AP-side power-save, crypto plugin framework, authenticator plugin framework,
and access control plugin frameowrk.
harti [Wed, 8 Dec 2004 16:22:01 +0000 (16:22 +0000)]
Get rid of the sequential access feature of the lists. This was used
only in a couple of places and all of them except for one were easily
converted to use Lst_First/Lst_Succ. The one place is compatibility
mode in job.c where the it was used to advance to the next command on
each invocation of JobStart. For this case add a pointer to the node to
hold the currently executed command.
harti [Wed, 8 Dec 2004 12:59:27 +0000 (12:59 +0000)]
Constify the arguments to the list compare function. This temporarily
requires to make a copy of the filename in ReadMakefile and to duplicate
two small functions in suff.c. This hopefully will go away when everything
is constified.
Submitted by: Max Okumoto <okumoto@ucsd.edu> (partly)
rwatson [Wed, 8 Dec 2004 11:46:44 +0000 (11:46 +0000)]
Add a new sysctl/tunable to mac_portacl:
security.mac.portacl.autoport_exempt
This sysctl exempts to bind port '0' as long as IP_PORTRANGELOW hasn't
been set on the socket. This is quite useful as it allows applications
to use automatic binding without adding overly broad rules for the
binding of port 0. This sysctl defaults to enabled.
This is a slight variation on the patch submitted by the contributor.
MFC after: 2 weeks
Submitted by: Michal Mertl <mime at traveller dot cz>
harti [Wed, 8 Dec 2004 08:11:59 +0000 (08:11 +0000)]
Consify the arguments to str_concat. Remove the STR_DOFREE flag for that
purpose and explicitely free the input string in the one place that was
calling str_concat with that flag.
sam [Wed, 8 Dec 2004 05:45:59 +0000 (05:45 +0000)]
Cleanup link state change notification:
o add new if_link_state_change routine that deals with link state changes
o change mii to use if_link_state_change