]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoRemove a list that used to hold all the GNodes just to be able
Hartmut Brandt [Fri, 10 Dec 2004 10:32:10 +0000 (10:32 +0000)]
Remove a list that used to hold all the GNodes just to be able
to free them just before exiting.

19 years agoRemove a useless list where just all command lines are stuffed onto, never
Hartmut Brandt [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?

19 years agoIn certain cases ip_output() can free our route, so check
Gleb Smirnoff [Fri, 10 Dec 2004 07:51:14 +0000 (07:51 +0000)]
In certain cases ip_output() can free our route, so check
for its presence before RTFREE().

Noticed by: ru

19 years agoRevert last change.
Gleb Smirnoff [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.

Requested by: csjp, sam

19 years agoClean up from '#if __FreeBSD_version'.
Roman Kurakin [Fri, 10 Dec 2004 05:54:45 +0000 (05:54 +0000)]
Clean up from '#if __FreeBSD_version'.

19 years agoRemove unneeded code from the zero-copy receive path.
Alan Cox [Fri, 10 Dec 2004 04:49:13 +0000 (04:49 +0000)]
Remove unneeded code from the zero-copy receive path.

Discussed with: gallatin@
Tested by: ken@

19 years agoAdd pthread_atfork().
Greg Lehey [Fri, 10 Dec 2004 03:36:45 +0000 (03:36 +0000)]
Add pthread_atfork().

PR:   bin/68841
Submitted by: Dan Nelson <dnelson@allantgroup.com>

19 years agoIn nfs_rename(), skip the otw rename operation if the fsync (to
Paul Saab [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!

19 years agoStore a hint in the nfsnode to detect sequential access of the file.
Paul Saab [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!

19 years agoUpdate the IPFW man page to reflect reality. mpsafenet=0 is no longer
Christian S.J. Peron [Fri, 10 Dec 2004 02:38:21 +0000 (02:38 +0000)]
Update the IPFW man page to reflect reality. mpsafenet=0 is no longer
required when using ucred based rules.

Pointed out by: seanc (thanks!)
MFC after: 1 month

19 years agoThis commit adds a shared locking mechanism very similar to the
Christian S.J. Peron [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.

Reviewed by: andre, rwatson
Tested by: myself, seanc
Silence on: ipfw@
MFC after: 1 month

19 years agoFix compilation and correct mapping from struct ifnet to
Sam Leffler [Fri, 10 Dec 2004 00:59:27 +0000 (00:59 +0000)]
Fix compilation and correct mapping from struct ifnet to
struct ieee80211com after net80211 import.

Submitted by: Tor Egge

19 years agoPartially backout 1.34. These ioctls and function cx_modem_status () are
Roman Kurakin [Thu, 9 Dec 2004 23:43:03 +0000 (23:43 +0000)]
Partially backout 1.34. These ioctls and function cx_modem_status () are
used by sync part of driver, so put them back.

19 years agoAdd Intel EtherExpress PRO/100 LAN/Modem card to manpage.
Scott Mitchell [Thu, 9 Dec 2004 23:13:18 +0000 (23:13 +0000)]
Add Intel EtherExpress PRO/100 LAN/Modem card to manpage.

MFC after: 2 days

19 years agoAdd Ethernet part of Intel EtherExpress PRO/100 LAN/Modem card. This is a
Scott Mitchell [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

19 years agoComplete the repo-copy of aac_ioctl.h from sys/dev/aac/to sys/sys.
Scott Long [Thu, 9 Dec 2004 22:20:25 +0000 (22:20 +0000)]
Complete the repo-copy of aac_ioctl.h from sys/dev/aac/to sys/sys.

19 years agoFixes a bug that caused UFS2 filesystems bigger than 2TB to
Kirk 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

19 years agoAdd workaround for {not so|too} smart phones that cannot handle SDP attribute
Maksim Yevmenkin [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.

MFC after: 3 days

19 years agoConnect acpi_ibm to build.
Takanori Watanabe [Thu, 9 Dec 2004 16:47:07 +0000 (16:47 +0000)]
Connect acpi_ibm to build.

Submitted by: rushani, sumikawa

19 years agoCheck that DUMMYNET_LOADED before seeking dummynet m_tag.
Gleb Smirnoff [Thu, 9 Dec 2004 16:41:47 +0000 (16:41 +0000)]
Check that DUMMYNET_LOADED before seeking dummynet m_tag.

Reviewed by: andre
MFC after: 1 week

19 years agoUpdate the description of SCSI_DELAY and annotate IDE_DELAY as a
Giorgos Keramidas [Thu, 9 Dec 2004 15:52:51 +0000 (15:52 +0000)]
Update the description of SCSI_DELAY and annotate IDE_DELAY as a
PC98-only option.

Submitted by: Joel Dahl (joel at automatvapen dot se)
PR: docs/74862
MFC after: 3 days

19 years agoNobody actually checked the return codes from Lst_Append and Lst_Insert
Hartmut Brandt [Thu, 9 Dec 2004 15:31:32 +0000 (15:31 +0000)]
Nobody actually checked the return codes from Lst_Append and Lst_Insert
so don't return anything.

19 years agoFix typos.
Christian Brueffer [Thu, 9 Dec 2004 13:57:12 +0000 (13:57 +0000)]
Fix typos.

19 years agoAdd IBM Laptop extra device driver.
Takanori Watanabe [Thu, 9 Dec 2004 13:54:29 +0000 (13:54 +0000)]
Add IBM Laptop extra device driver.
This depends on ACPI and RTC registers.

Reviewed by: njl

19 years agoFollow the language currently used in the manual page.
Tom Rhodes [Thu, 9 Dec 2004 13:48:33 +0000 (13:48 +0000)]
Follow the language currently used in the manual page.

Submitted by: Michal Mertl <mime@traveller.cz>

19 years agoURL of the data sheet has changed.
Christian Brueffer [Thu, 9 Dec 2004 13:27:44 +0000 (13:27 +0000)]
URL of the data sheet has changed.

Obtained from: OpenBSD

19 years agoVarious language tweaks.
Christian Brueffer [Thu, 9 Dec 2004 13:11:52 +0000 (13:11 +0000)]
Various language tweaks.

Obtained from: OpenBSD

19 years ago- Turn off 'fast' mode by default and increase maximum memory to consume
Pawel Jakub Dawidek [Thu, 9 Dec 2004 12:26:47 +0000 (12:26 +0000)]
- Turn off 'fast' mode by default and increase maximum memory to consume
  when this mode is used.
- Manual page update.

19 years agoStart the protocol timeouts only after all domains have been initialized
Max Laier [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.

Tested by: gnn, ru, myself

19 years agoIf a path element directory was never opened it is not on the list of
Hartmut Brandt [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.

19 years agostyle the last change
Gleb Smirnoff [Thu, 9 Dec 2004 09:52:58 +0000 (09:52 +0000)]
style the last change

19 years agoMFinet4:
Gleb Smirnoff [Thu, 9 Dec 2004 09:48:47 +0000 (09:48 +0000)]
MFinet4:

  - Make route cacheing optional, configurable via IFF_LINK0 flag.
  - Turn it off by default.

Reminded by: suz

19 years agoUse ng_callout() and ng_uncallout() instead of home-grown
Gleb Smirnoff [Thu, 9 Dec 2004 07:49:02 +0000 (07:49 +0000)]
Use ng_callout() and ng_uncallout() instead of home-grown
implementation.

Tested by: Savchuk Taras
Reviewed by: archie
Approved by: julian (mentor)

19 years agoCompensate for off by one bugs in disk firmware for 48BIT addressing cutover.
Søren Schmidt [Thu, 9 Dec 2004 07:31:06 +0000 (07:31 +0000)]
Compensate for off by one bugs in disk firmware for 48BIT addressing cutover.

19 years agoMore fixing of multiple addresses in the same prefix. This time do not try
Max Laier [Thu, 9 Dec 2004 00:12:41 +0000 (00:12 +0000)]
More fixing of multiple addresses in the same prefix. This time do not try
to arp resolve "secondary" local addresses.

Found and submitted by: ru
With additions from: OpenBSD (rev. 1.47)
Reviewed by: ru

19 years agofixup printf arguments for 64-bit machines
Sam Leffler [Wed, 8 Dec 2004 22:34:07 +0000 (22:34 +0000)]
fixup printf arguments for 64-bit machines

19 years agoDiff-reduction before merging if_axe to RELENG_4.
Scott Mitchell [Wed, 8 Dec 2004 22:02:58 +0000 (22:02 +0000)]
Diff-reduction before merging if_axe to RELENG_4.

Approved by: imp (mentor)
MFC after: 1 week

19 years agoOverhaul to cleanup some of the tangled logic that's grown over the years.
Sam Leffler [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

19 years agoIf the parent process has the trap bit set (i.e. a debugger had single
Kelly Yancey [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.

Reviewed by: bde
MFC after: 3 days

19 years agoOooops I forgot to commit that.
Olivier Houchard [Wed, 8 Dec 2004 18:41:42 +0000 (18:41 +0000)]
Oooops I forgot to commit that.
Use fp_rnd_t, not fp_rnd.

Reported by: Jia-Shiun Li (jiashiun at gmail dot com)

19 years agoclear old files now in public
Sam Leffler [Wed, 8 Dec 2004 18:27:44 +0000 (18:27 +0000)]
clear old files now in public

19 years agoupdate for new ath hal
Sam Leffler [Wed, 8 Dec 2004 18:20:53 +0000 (18:20 +0000)]
update for new ath hal

19 years agoresolve merge conflict
Sam Leffler [Wed, 8 Dec 2004 18:18:39 +0000 (18:18 +0000)]
resolve merge conflict

19 years agoWordsmith previous commit and bump document date.
Tom Rhodes [Wed, 8 Dec 2004 18:11:38 +0000 (18:11 +0000)]
Wordsmith previous commit and bump document date.

Poked by: simon

19 years agoThis commit was generated by cvs2svn to compensate for changes in r138583,
Sam Leffler [Wed, 8 Dec 2004 18:06:09 +0000 (18:06 +0000)]
This commit was generated by cvs2svn to compensate for changes in r138583,
which included commits to RCS files with non-trunk default branches.

19 years agoVersion 0.9.14.12:
Sam Leffler [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

19 years agoUpdate the ICH6 support so all 4 SATA channels can be used.
Søren Schmidt [Wed, 8 Dec 2004 18:00:46 +0000 (18:00 +0000)]
Update the ICH6 support so all 4 SATA channels can be used.
We still use "normal" mode, AHCI mode is in the works still.

HW donated by: Sentex
HW donated by: Yahoo!

19 years agoNow that circular lists are gone remove stuff for them. Simplify
Hartmut Brandt [Wed, 8 Dec 2004 17:48:15 +0000 (17:48 +0000)]
Now that circular lists are gone remove stuff for them. Simplify
somewhat so that we can remove a local variable.

19 years agodocument m_append
Sam Leffler [Wed, 8 Dec 2004 17:47:49 +0000 (17:47 +0000)]
document m_append

19 years agoUpdates for revised ath and net80211 support; the new man pages are
Sam Leffler [Wed, 8 Dec 2004 17:47:32 +0000 (17:47 +0000)]
Updates for revised ath and net80211 support; the new man pages are
placeholders for real material.

19 years agoLoadable modules that run test vectors for net80211 crypto plugins.
Sam Leffler [Wed, 8 Dec 2004 17:45:53 +0000 (17:45 +0000)]
Loadable modules that run test vectors for net80211 crypto plugins.

19 years agoNo caller checks the return code from Lst_Remove, so don't return one.
Hartmut Brandt [Wed, 8 Dec 2004 17:43:43 +0000 (17:43 +0000)]
No caller checks the return code from Lst_Remove, so don't return one.
Simplify the algorithm now that circular lists are gone.

19 years agoo new tools for futzing with 80211 and ath debug masks symbolically
Sam Leffler [Wed, 8 Dec 2004 17:42:53 +0000 (17:42 +0000)]
o new tools for futzing with 80211 and ath debug masks symbolically
o update stats programs to reflect driver and net80211 changes

19 years agoadd ath rate control module(s)
Sam Leffler [Wed, 8 Dec 2004 17:40:55 +0000 (17:40 +0000)]
add ath rate control module(s)

19 years agoUpdate for ath and net80211 changes.
Sam Leffler [Wed, 8 Dec 2004 17:39:51 +0000 (17:39 +0000)]
Update for ath and net80211 changes.

19 years agoUpdate/new modules for net80211 and ath changes.
Sam Leffler [Wed, 8 Dec 2004 17:38:38 +0000 (17:38 +0000)]
Update/new modules for net80211 and ath changes.

19 years agoUpdate for net80211 changes.
Sam Leffler [Wed, 8 Dec 2004 17:36:51 +0000 (17:36 +0000)]
Update for net80211 changes.

19 years agoo Update for net80211 changes
Sam Leffler [Wed, 8 Dec 2004 17:36:28 +0000 (17:36 +0000)]
o Update for net80211 changes
o Import some minor fixes from netbsd

19 years agoUpdate with last year of work.
Sam Leffler [Wed, 8 Dec 2004 17:34:36 +0000 (17:34 +0000)]
Update with last year of work.

19 years agoTransmit rate control modules for the ath driver.
Sam Leffler [Wed, 8 Dec 2004 17:32:02 +0000 (17:32 +0000)]
Transmit rate control modules for the ath driver.

19 years agoUpdate 802.11 support; too much new functionality to fully describe
Sam Leffler [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.

19 years agoDon't check the return code from Lst_Remove. There is no way
Hartmut Brandt [Wed, 8 Dec 2004 16:50:14 +0000 (16:50 +0000)]
Don't check the return code from Lst_Remove. There is no way
that the list's first element is not on the list.

19 years agoRemove return value from Lst_Concat. None of the callers ever checked
Hartmut Brandt [Wed, 8 Dec 2004 16:47:19 +0000 (16:47 +0000)]
Remove return value from Lst_Concat. None of the callers ever checked
it. Remove stuff that was needed for circular lists.

19 years agoDon't free the second list in Lst_Concat for LST_CONCLINK; free it
Hartmut Brandt [Wed, 8 Dec 2004 16:28:53 +0000 (16:28 +0000)]
Don't free the second list in Lst_Concat for LST_CONCLINK; free it
in the caller instead.

19 years agoGet rid of the sequential access feature of the lists. This was used
Hartmut Brandt [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.

19 years agoDocument recently added security.mac.portacl.autoport_exempt tunable.
Tom Rhodes [Wed, 8 Dec 2004 15:58:38 +0000 (15:58 +0000)]
Document recently added security.mac.portacl.autoport_exempt tunable.

Outraced the speed of doc request: trhodes

19 years ago- Use ng_callout() instead of timeout()
Gleb Smirnoff [Wed, 8 Dec 2004 13:27:54 +0000 (13:27 +0000)]
- Use ng_callout() instead of timeout()
- remove spl(9) calls

Tested by: Ilya Pizik
Approved by: julian (mentor)

19 years agoConstify the arguments to the list compare function. This temporarily
Hartmut Brandt [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)

19 years agoMake sure we don't pass garbage to the kernel.
Poul-Henning Kamp [Wed, 8 Dec 2004 11:54:50 +0000 (11:54 +0000)]
Make sure we don't pass garbage to the kernel.

The filesystem is named "ufs" not "ffs".

19 years agoFix snapshot creation.
Poul-Henning Kamp [Wed, 8 Dec 2004 11:54:06 +0000 (11:54 +0000)]
Fix snapshot creation.

19 years agoAdd a new sysctl/tunable to mac_portacl:
Robert Watson [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>

19 years agoAdd support for the ITE IT8212F controller.
Søren Schmidt [Wed, 8 Dec 2004 11:17:38 +0000 (11:17 +0000)]
Add support for the ITE IT8212F controller.

HW donated by: Yahoo!

19 years agoReset timeout when we are back from interrupt.
Søren Schmidt [Wed, 8 Dec 2004 11:16:33 +0000 (11:16 +0000)]
Reset timeout when we are back from interrupt.

19 years agoAdd first shot on support for the new Promise SATAII chips.
Søren Schmidt [Wed, 8 Dec 2004 10:02:41 +0000 (10:02 +0000)]
Add first shot on support for the new Promise SATAII chips.

HW donated by: pil.dk

19 years agoCorrect logical error, result was that retries wasn't always made but
Søren Schmidt [Wed, 8 Dec 2004 09:19:03 +0000 (09:19 +0000)]
Correct logical error, result was that retries wasn't always made but
failure reported instead.

19 years agoConstify some calls of Buf_AddBytes.
Hartmut Brandt [Wed, 8 Dec 2004 08:18:13 +0000 (08:18 +0000)]
Constify some calls of Buf_AddBytes.

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoConsify the arguments to str_concat. Remove the STR_DOFREE flag for that
Hartmut Brandt [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.

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoProperly handle vnode_if.h dependencies, and don't forget to clean all files.
Ruslan Ermilov [Wed, 8 Dec 2004 07:23:12 +0000 (07:23 +0000)]
Properly handle vnode_if.h dependencies, and don't forget to clean all files.

19 years agoDon't obtain the HCDP address directly from the bootinfo structure.
Marcel Moolenaar [Wed, 8 Dec 2004 05:46:54 +0000 (05:46 +0000)]
Don't obtain the HCDP address directly from the bootinfo structure.
Use a function to keep the details at arms length from uart(4).

19 years agoCleanup link state change notification:
Sam Leffler [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

19 years agoadd m_append utility function to be used in forthcoming changes
Sam Leffler [Wed, 8 Dec 2004 05:42:02 +0000 (05:42 +0000)]
add m_append utility function to be used in forthcoming changes

19 years agoDon't require a device to be marked up when issuing BIOCSETIF.
Sam Leffler [Wed, 8 Dec 2004 05:40:02 +0000 (05:40 +0000)]
Don't require a device to be marked up when issuing BIOCSETIF.

19 years agoTidy up the zero-copy receive path: Remove an unneeded argument to
Alan Cox [Wed, 8 Dec 2004 05:25:08 +0000 (05:25 +0000)]
Tidy up the zero-copy receive path: Remove an unneeded argument to
uiomoveco() and userspaceco().

19 years agoWith the removal of kern/uipc_jumbo.c and sys/jumbo.h,
Alan Cox [Wed, 8 Dec 2004 05:01:47 +0000 (05:01 +0000)]
With the removal of kern/uipc_jumbo.c and sys/jumbo.h,
vm_object_allocate_wait() is not used.  Remove it.

19 years agoFix comments to match last commit, and minor reformatting...
Warner Losh [Wed, 8 Dec 2004 04:35:19 +0000 (04:35 +0000)]
Fix comments to match last commit, and minor reformatting...

19 years agoIt turns out that a lot of newer systems have 'base peripherals' on
Warner Losh [Wed, 8 Dec 2004 04:16:18 +0000 (04:16 +0000)]
It turns out that a lot of newer systems have 'base peripherals' on
the PCI bus.  We presently have no drivers for these devices, so they
are powered down.  This is undesirable behavior since it breaks the
system when the base peripherals go away suddenly in the middle of
boot.

# if we ever get generic drivers for memory and/or base peripherals, then
# we can remove the tests here.

19 years agoRemove some (no longer vaild) notes about PFIL_HOOKS.
Max Laier [Wed, 8 Dec 2004 04:08:33 +0000 (04:08 +0000)]
Remove some (no longer vaild) notes about PFIL_HOOKS.

19 years agoUnbreak KLDs that use VFS by accounting for changes in the generation
Brian Feldman [Wed, 8 Dec 2004 03:54:32 +0000 (03:54 +0000)]
Unbreak KLDs that use VFS by accounting for changes in the generation
process for the vnode header files.

19 years agoAdd the devclass_get_count(9) function and man page. It gets a count of
Nate Lawson [Wed, 8 Dec 2004 02:39:56 +0000 (02:39 +0000)]
Add the devclass_get_count(9) function and man page.  It gets a count of
the number of devices in a devclass and is a subset of
devclass_get_devices(9).

Reviewed by: imp, dfr

19 years agoAlmost nine years ago, when support for 1TB files was introduced in
Alan Cox [Tue, 7 Dec 2004 22:05:38 +0000 (22:05 +0000)]
Almost nine years ago, when support for 1TB files was introduced in
revision 1.55, the address parameter to vnode_pager_addr() was changed
from an unsigned 32-bit quantity to a signed 64-bit quantity.  However,
an out-of-range check on the address was not updated.  Consequently,
memory-mapped I/O on files greater than 2GB could cause a kernel panic.
Since the address is now a signed 64-bit quantity, the problem resolution
is simply to remove a cast.

Reviewed by: bde@ and tegge@
PR: 73010
MFC after: 1 week

19 years agoFix for a Lock Order Reversal in the nfs_flush() path, between the
Paul Saab [Tue, 7 Dec 2004 21:16:32 +0000 (21:16 +0000)]
Fix for a Lock Order Reversal in the nfs_flush() path, between the
vnode interlock and the proc lock.

Reported by: marcel
Submitted by: Mohan Srinivasan mohans at yahoo-inc dot com

19 years agoAvoid more than two pending IPI interrupt vectors per local APIC
Stephan Uphoff [Tue, 7 Dec 2004 20:15:01 +0000 (20:15 +0000)]
Avoid more than two pending IPI interrupt vectors per local APIC
as this may cause deadlocks.

This should fix kern/72123.

Discussed with: jhb
Tested by: Nik Azim Azam, Andy Farkas, Flack Man, Aykut KARA
           Izzet BESKARDES, Jens Binnewies, Karl Keusgen
Approved by:    sam (mentor)

19 years agoPropagate TDF_NEEDRESCHED to replacement thread in sched_switch().
Stephan Uphoff [Tue, 7 Dec 2004 18:17:24 +0000 (18:17 +0000)]
Propagate TDF_NEEDRESCHED to replacement thread in sched_switch().

Reviewed by:    julian, jhb (in October)
Approved by:    sam (mentor)
MFC after:      4 weeks

19 years agoInclude <sys/signalvar.h> for trapsignal().
Olivier Houchard [Tue, 7 Dec 2004 17:39:42 +0000 (17:39 +0000)]
Include <sys/signalvar.h> for trapsignal().

19 years agoNEC PC-98 machines do not have and cannot have an EISA bus. They have
Warner Losh [Tue, 7 Dec 2004 15:36:19 +0000 (15:36 +0000)]
NEC PC-98 machines do not have and cannot have an EISA bus.  They have
only C-Bus and PCI busses.  Therefore, don't create an eisa0 node on
the legacy bus that can never attach.

PC-98 info verified by: nyan-san

19 years agoFirst save from editor, *then* commit.
Poul-Henning Kamp [Tue, 7 Dec 2004 15:25:36 +0000 (15:25 +0000)]
First save from editor, *then* commit.

19 years agoFix exports.
Poul-Henning Kamp [Tue, 7 Dec 2004 15:13:35 +0000 (15:13 +0000)]
Fix exports.

19 years agoFix nfs exports (for now). The real fix is to teach mountd about
Poul-Henning Kamp [Tue, 7 Dec 2004 15:09:30 +0000 (15:09 +0000)]
Fix nfs exports (for now).  The real fix is to teach mountd about
nmount.

19 years agoFix builds with a read-only directory and a make upgrade. This is done
Hartmut Brandt [Tue, 7 Dec 2004 14:41:16 +0000 (14:41 +0000)]
Fix builds with a read-only directory and a make upgrade. This is done
by forcing the creation of an object directory for the make regression
tests. Let make handle the tracking of the dependency and installation
of test_shell script.

Submitted by: ru

19 years agoDon't clobber mnt_stat.f_mntonname
Poul-Henning Kamp [Tue, 7 Dec 2004 14:26:39 +0000 (14:26 +0000)]
Don't clobber mnt_stat.f_mntonname