dragonfly.git
17 years agoUse .Dv for ioctls.
Sascha Wildner [Thu, 15 Mar 2007 21:26:05 +0000 (21:26 +0000)]
Use .Dv for ioctls.

17 years agoUse .Dv for defined values.
Sascha Wildner [Thu, 15 Mar 2007 21:17:45 +0000 (21:17 +0000)]
Use .Dv for defined values.

17 years ago* Use .Dv for ioctls.
Sascha Wildner [Thu, 15 Mar 2007 20:58:57 +0000 (20:58 +0000)]
* Use .Dv for ioctls.

* Fix two "typos":

  s/CDIOCREADTOCHEADER/CDIOREADTOCHEADER/
  s/CDIOCREADTOCENTRYS/CDIOREADTOCENTRYS/

  I think the real typo is in the header file but let's just document
  the status quo.

17 years agoInitial syslink system call documentation and overview.
Matthew Dillon [Wed, 14 Mar 2007 21:09:34 +0000 (21:09 +0000)]
Initial syslink system call documentation and overview.

17 years agoAdd a note that getopt_long_only(3) is deprecated.
Sascha Wildner [Wed, 14 Mar 2007 14:50:07 +0000 (14:50 +0000)]
Add a note that getopt_long_only(3) is deprecated.

Requested-by: joerg
17 years agoOops, really add the MLINK.
Sascha Wildner [Wed, 14 Mar 2007 10:37:42 +0000 (10:37 +0000)]
Oops, really add the MLINK.

17 years ago1) Fix typo in getopt_long's prototype.
Sascha Wildner [Wed, 14 Mar 2007 10:34:17 +0000 (10:34 +0000)]
1) Fix typo in getopt_long's prototype.

2) Document the index argument.

3) s/floride/fluoride/

4) Document getopt_long_only() and add an MLINK.

Taken-from:  NetBSD (1-3) and FreeBSD (4)
Reported-by: Slava Semushin <php-coder@altlinux.ru> (4)
17 years agoReword a couple of sentences and do some minor mdoc cleanup.
Sascha Wildner [Tue, 13 Mar 2007 10:16:56 +0000 (10:16 +0000)]
Reword a couple of sentences and do some minor mdoc cleanup.

OK'd-by: corecode

17 years agoAdd /dev/vkd0 to the list of devices to be searched when the
Sascha Wildner [Tue, 13 Mar 2007 09:42:20 +0000 (09:42 +0000)]
Add /dev/vkd0 to the list of devices to be searched when the
device name is omitted.

17 years agoAdapt libthread_xu to use the new lwp kernel functions.
Simon Schubert [Tue, 13 Mar 2007 00:19:29 +0000 (00:19 +0000)]
Adapt libthread_xu to use the new lwp kernel functions.

17 years agoDrop this strange start/endzero thing for procs and lwps.
Simon Schubert [Tue, 13 Mar 2007 00:18:59 +0000 (00:18 +0000)]
Drop this strange start/endzero thing for procs and lwps.

Instead, unconditionally bzero the new proc/lwp.  This also allows us
to remove now superfluous initializations.

17 years agoRegen.
Simon Schubert [Mon, 12 Mar 2007 23:45:04 +0000 (23:45 +0000)]
Regen.

17 years agoRework stopping of procs.
Simon Schubert [Mon, 12 Mar 2007 21:08:15 +0000 (21:08 +0000)]
Rework stopping of procs.

Before, proc_stop() would sleep until all running lwps stopped.  This
break when a stop signal is actually coming from the console and is
executed in the context of the idle thread.

Now we count all sleeping threads as stopped and also set LWP_WSTOP to
indicate so.  These threads will stop before return to userland.
Running threads (including the current one) will eventually stop when
returning to userland and will increase p_nstopped.  The last thread
stopping will then send a signal to the parent process.

Discussed-with:  Thomas E. Spanjaard <tgen@netphreax.net>

17 years ago1:1 Userland threading stage 4.8/4:
Simon Schubert [Mon, 12 Mar 2007 21:07:42 +0000 (21:07 +0000)]
1:1 Userland threading stage 4.8/4:

Add syscalls lwp_gettid() and lwp_kill().

17 years agoAdd man page for lwp_create().
Simon Schubert [Mon, 12 Mar 2007 21:06:18 +0000 (21:06 +0000)]
Add man page for lwp_create().

17 years agoAdd prototype for lwp_create().
Simon Schubert [Mon, 12 Mar 2007 21:06:00 +0000 (21:06 +0000)]
Add prototype for lwp_create().

17 years agoAdd precautions so that threads won't race to exit1() and get stuck there.
Simon Schubert [Mon, 12 Mar 2007 21:05:48 +0000 (21:05 +0000)]
Add precautions so that threads won't race to exit1() and get stuck there.

17 years agoRemove trailing whitespace.
Sascha Wildner [Mon, 12 Mar 2007 17:12:11 +0000 (17:12 +0000)]
Remove trailing whitespace.

17 years agoMake the vkernel disk (vkd) known to vinum.
Sascha Wildner [Mon, 12 Mar 2007 12:10:13 +0000 (12:10 +0000)]
Make the vkernel disk (vkd) known to vinum.

17 years ago* Add callout(9) MLINK.
Sascha Wildner [Sun, 11 Mar 2007 20:04:18 +0000 (20:04 +0000)]
* Add callout(9) MLINK.

* List all functions in the NAME section.

* Use .Fn

17 years agoConvert to mdoc.
Sascha Wildner [Sun, 11 Mar 2007 19:05:31 +0000 (19:05 +0000)]
Convert to mdoc.

Obtained from (except for bs(6)): NetBSD

17 years ago* Reduce differences to FreeBSD.
Sascha Wildner [Sun, 11 Mar 2007 11:57:27 +0000 (11:57 +0000)]
* Reduce differences to FreeBSD.

* s/fixit floppy/LiveCD/

17 years agoAdd stge(4) and vge(4).
Sascha Wildner [Sun, 11 Mar 2007 11:15:33 +0000 (11:15 +0000)]
Add stge(4) and vge(4).

17 years agoAdd missing config declaration.
Sascha Wildner [Sun, 11 Mar 2007 10:22:52 +0000 (10:22 +0000)]
Add missing config declaration.

17 years agomy(4) doesn't use miibus
Sascha Wildner [Sun, 11 Mar 2007 03:46:48 +0000 (03:46 +0000)]
my(4) doesn't use miibus

17 years agoAdd nfe(4) and vge(4).
Sascha Wildner [Sun, 11 Mar 2007 03:39:25 +0000 (03:39 +0000)]
Add nfe(4) and vge(4).

17 years agoUse correct parameter order for strcpy().
Sascha Wildner [Sat, 10 Mar 2007 09:47:29 +0000 (09:47 +0000)]
Use correct parameter order for strcpy().

17 years agoMake sure all string buffers passed from userland are terminated before
Matthew Dillon [Thu, 8 Mar 2007 16:42:34 +0000 (16:42 +0000)]
Make sure all string buffers passed from userland are terminated before
proceeding.

17 years agoAdd reference to em(4).
Sascha Wildner [Thu, 8 Mar 2007 12:27:07 +0000 (12:27 +0000)]
Add reference to em(4).

17 years agoAdd a little documentation for the hw.em.int_throttle_ceil tunable.
Sascha Wildner [Thu, 8 Mar 2007 11:20:56 +0000 (11:20 +0000)]
Add a little documentation for the hw.em.int_throttle_ceil tunable.

17 years agoAdd a better description for the hw.bge.fake_autoneg tunable.
Sascha Wildner [Thu, 8 Mar 2007 09:46:42 +0000 (09:46 +0000)]
Add a better description for the hw.bge.fake_autoneg tunable.

Submitted-by: sephe
17 years agoDon't try holding serializer in ti_rxeof(). This function is only called by
Sepherosa Ziehau [Wed, 7 Mar 2007 12:33:07 +0000 (12:33 +0000)]
Don't try holding serializer in ti_rxeof().  This function is only called by
intr routine which alrealy holds serializer.

Reported-by: "RusE39" <russiane39@gmail.com>
17 years agoUse random octet stream to create IBSS's BSSID as suggested by
Sepherosa Ziehau [Tue, 6 Mar 2007 14:51:22 +0000 (14:51 +0000)]
Use random octet stream to create IBSS's BSSID as suggested by
IEEE Std 802.11, 1999 Edition.

Obtained-from: Sam Leffler

17 years agoCorrect the unit of "powersavesleep" argument: it is in TU's, not in
Sepherosa Ziehau [Tue, 6 Mar 2007 12:22:39 +0000 (12:22 +0000)]
Correct the unit of "powersavesleep" argument: it is in TU's, not in
milliseconds.  The difference is marginal though, since 1TU~=1.024msec.

17 years agoCorrect the unit of assoc/reassoc req listen interval: it is in number
Sepherosa Ziehau [Tue, 6 Mar 2007 12:17:51 +0000 (12:17 +0000)]
Correct the unit of assoc/reassoc req listen interval: it is in number
of beacons, not in TU's.  It will probably affect how AP handles PS.

17 years agoCorrect TU <-> millisecond and TU -> ticks convertion.
Sepherosa Ziehau [Mon, 5 Mar 2007 14:17:36 +0000 (14:17 +0000)]
Correct TU <-> millisecond and TU -> ticks convertion.

17 years ago1 TU is 1024 usecs
Sepherosa Ziehau [Mon, 5 Mar 2007 12:25:05 +0000 (12:25 +0000)]
1 TU is 1024 usecs

17 years agoerr() and warn() print the strerror() message by themselves.
Sascha Wildner [Mon, 5 Mar 2007 02:41:45 +0000 (02:41 +0000)]
err() and warn() print the strerror() message by themselves.

17 years agoDocument some tunables:
Sascha Wildner [Sun, 4 Mar 2007 20:10:34 +0000 (20:10 +0000)]
Document some tunables:

hw.bge.fake_autoneg
hw.cardbus.cis_debug
hw.cardbus.debug
hw.cbb.debug
hw.em.rx_abs_int_delay
hw.em.rx_int_delay
hw.em.rxd
hw.em.tx_abs_int_delay
hw.em.tx_int_delay
hw.em.txd
hw.pccard.cis_debug
hw.pccard.debug
kern.dfldsiz
kern.dflssiz
kern.maxdsiz
kern.maxssiz
kern.maxtsiz
kern.sgrowsiz

Taken-from: FreeBSD

17 years agoRemove weird license clause which has expired.
Sascha Wildner [Sun, 4 Mar 2007 18:51:59 +0000 (18:51 +0000)]
Remove weird license clause which has expired.

17 years agoAdd wlandebug(8) tool for turning on/off debug flags in netproto/802_11
Sepherosa Ziehau [Sun, 4 Mar 2007 13:15:48 +0000 (13:15 +0000)]
Add wlandebug(8) tool for turning on/off debug flags in netproto/802_11
Thank swildner@ to help with the manpage.

Obtained-from: FreeBSD (sam@freebsd.org) w/ modification

17 years agoExpose IEEE80211_MSG_ for user space programs.
Sepherosa Ziehau [Sun, 4 Mar 2007 13:09:15 +0000 (13:09 +0000)]
Expose IEEE80211_MSG_ for user space programs.

17 years agoFix typo.
Sascha Wildner [Sun, 4 Mar 2007 12:18:05 +0000 (12:18 +0000)]
Fix typo.

17 years ago- Add additional parameter to ieee80211_mgmt_output(). Use this new
Sepherosa Ziehau [Sun, 4 Mar 2007 07:52:36 +0000 (07:52 +0000)]
- Add additional parameter to ieee80211_mgmt_output().  Use this new
  parameter instead of protocol specific mbuf flag M_LINK0 to inform
  ieee80211_mgmt_output() that the mgmt frame needs encryption.
- Revoke protocol specific mbuf flag M_LINK0

These changes save one protocol specific mbuf flag, which will be used
by TXCB mechanism for mgmt frames in the future.

17 years ago1) Fix prototype in vget(9).
Sascha Wildner [Sun, 4 Mar 2007 02:45:08 +0000 (02:45 +0000)]
1) Fix prototype in vget(9).

2) Add vhold/vdrop manual page.

3) Some mdoc cleanup.

[2] and some of [3] taken from: FreeBSD

17 years agoBring back in some previously commented out references to ifmedia(4).
Sascha Wildner [Sun, 4 Mar 2007 02:32:59 +0000 (02:32 +0000)]
Bring back in some previously commented out references to ifmedia(4).

17 years agoFix typo.
Sascha Wildner [Sat, 3 Mar 2007 23:19:25 +0000 (23:19 +0000)]
Fix typo.

17 years agoUse .Er for errno value and fix typo.
Sascha Wildner [Sat, 3 Mar 2007 23:16:54 +0000 (23:16 +0000)]
Use .Er for errno value and fix typo.

17 years agoFix prototype.
Sascha Wildner [Thu, 1 Mar 2007 16:18:11 +0000 (16:18 +0000)]
Fix prototype.

17 years agoNo . needed here.
Sascha Wildner [Thu, 1 Mar 2007 12:57:15 +0000 (12:57 +0000)]
No . needed here.

17 years agoSort by section.
Sascha Wildner [Thu, 1 Mar 2007 12:37:19 +0000 (12:37 +0000)]
Sort by section.

17 years agoRegen.
Simon Schubert [Thu, 1 Mar 2007 01:48:51 +0000 (01:48 +0000)]
Regen.

17 years ago1:1 Userland threading stage 4.7/4:
Simon Schubert [Thu, 1 Mar 2007 01:46:53 +0000 (01:46 +0000)]
1:1 Userland threading stage 4.7/4:

Add a new system call lwp_create() which spawns a new lwp with a
given thread function address and given stack pointer.  Rework
and add some associated functions to realize this goal.

In-collaboration-with: Thomas E. Spanjaard <tgen@netphreax.net>

17 years agoAdd sc_newstate_arg to ural_softc, so ieee80211_new_state() 'arg' can be
Sepherosa Ziehau [Wed, 28 Feb 2007 13:04:26 +0000 (13:04 +0000)]
Add sc_newstate_arg to ural_softc, so ieee80211_new_state() 'arg' can be
passed to ieee80211_newstate().  This makes 802_11 internal state machine
run correctly for certain state changes, e.g. RUN->AUTH

17 years agoRemove trailing space.
Sascha Wildner [Tue, 27 Feb 2007 17:33:32 +0000 (17:33 +0000)]
Remove trailing space.

17 years agoValidate msg_controllen, as per RFC2292.
Sascha Wildner [Tue, 27 Feb 2007 12:05:26 +0000 (12:05 +0000)]
Validate msg_controllen, as per RFC2292.

Submitted-by: Hasso Tepper <hasso@estpak.ee>
17 years ago1:1 Userland threading stage 4.6/4:
Simon Schubert [Mon, 26 Feb 2007 21:41:08 +0000 (21:41 +0000)]
1:1 Userland threading stage 4.6/4:

Factor out a new function `lwp_fork()' from fork1().  Rearrange various
fork helpers to operate exclusively on procs or lwps.  Collect some MI
code from MD sources into lwp_fork().

Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
17 years agoSync zoneinfo database with tzdata2007c from elsie.
Sascha Wildner [Mon, 26 Feb 2007 14:54:00 +0000 (14:54 +0000)]
Sync zoneinfo database with tzdata2007c from elsie.

europe:         8.7  -> 8.8
northamerica:   8.11 -> 8.12
southamerica:   8.8  -> 8.9
zone.tab:       8.8  -> 8.9

From Paul Eggert's comments:

* Turkey switches at 01:00 standard time, not at 01:00 UTC.  (Thanks to
  Amar Devegowda and Steffen Thorsen for this.)

* Pulaski County, Indiana, is switching from CST/CDT to EST/EDT next
  month (which means a 2-hour change on March 11).  This requires a
  new Zone, America/Indiana/Winamac.  (Thanks to Gwillim Law for the
  heads-up.)

* A slight correction to Easter Island Mean Time (1890-1932).

* Update comments for Canada.  (Thanks to Chris Walton.)

17 years agoAdd forgotten file.
Simon Schubert [Mon, 26 Feb 2007 13:14:04 +0000 (13:14 +0000)]
Add forgotten file.

17 years agoDiscard deauth/disassoc that are not for us. Reception of these frames
Sepherosa Ziehau [Mon, 26 Feb 2007 12:28:23 +0000 (12:28 +0000)]
Discard deauth/disassoc that are not for us.  Reception of these frames
can only happen when promisc mode is enabled.

Obtained-from: FreeBSD-p4/sam_wifi (sam@freebsd.org)

17 years agoFix suspend behavior by really stopping the process and setting its signal.
Simon Schubert [Mon, 26 Feb 2007 03:55:22 +0000 (03:55 +0000)]
Fix suspend behavior by really stopping the process and setting its signal.

17 years agoGet rid of struct user/UAREA.
Simon Schubert [Sun, 25 Feb 2007 23:17:13 +0000 (23:17 +0000)]
Get rid of struct user/UAREA.

Merge procsig with sigacts and replace usage of procsig with
sigacts, like it used to be in 4.4BSD.

Put signal-related inline functions in sys/signal2.h.

Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
17 years agoDrop a.out coredump support from gcore.
Simon Schubert [Sun, 25 Feb 2007 23:07:08 +0000 (23:07 +0000)]
Drop a.out coredump support from gcore.

17 years agoAdd missing .El
Sascha Wildner [Sun, 25 Feb 2007 19:12:28 +0000 (19:12 +0000)]
Add missing .El

17 years agoFix comments.
Sascha Wildner [Sun, 25 Feb 2007 19:09:15 +0000 (19:09 +0000)]
Fix comments.

17 years ago.ar -> .Ar
Sascha Wildner [Sun, 25 Feb 2007 19:08:40 +0000 (19:08 +0000)]
.ar -> .Ar

17 years agoRemove empty line (silence manlint).
Sascha Wildner [Sun, 25 Feb 2007 18:50:48 +0000 (18:50 +0000)]
Remove empty line (silence manlint).

17 years agoAdd missing .Nm
Sascha Wildner [Sun, 25 Feb 2007 18:35:05 +0000 (18:35 +0000)]
Add missing .Nm

17 years agoRegen
Simon Schubert [Sun, 25 Feb 2007 14:08:08 +0000 (14:08 +0000)]
Regen

17 years ago1:1 Userland threading stage 4.5/4:
Simon Schubert [Sun, 25 Feb 2007 14:07:13 +0000 (14:07 +0000)]
1:1 Userland threading stage 4.5/4:

Add a new syscall, extexit(), which follows approximately the semantics
discussed about two years ago.  This function will be used by userland
threading libraries to exit just a single lwp of a process.

17 years agoAdd an ifmedia(4) reference to the manpages of all drivers that support
Sascha Wildner [Sun, 25 Feb 2007 12:03:07 +0000 (12:03 +0000)]
Add an ifmedia(4) reference to the manpages of all drivers that support
the ifmedia interface.

17 years agoFix references.
Sascha Wildner [Sun, 25 Feb 2007 01:09:09 +0000 (01:09 +0000)]
Fix references.

17 years ago1:1 Userland threading stage 4.4/4:
Simon Schubert [Sat, 24 Feb 2007 14:25:07 +0000 (14:25 +0000)]
1:1 Userland threading stage 4.4/4:

Implement killlwps() and flesh out lwp_exit().  Lwps which have set the
LWP_WEXIT flag will terminate themselves by calling lwp_exit() from
userret().

Reap lwps in a per-CPU taskqueue.  NOTE: Even the last lwp of an exiting
proc will be reaped by the task queue, effectively leaving a BARE proc
to reap in wait4().

In-collaboration-with: Thomas E. Spanjaard <tgen@netphreax.net>

17 years ago1:1 Userland threading stage 4.3/4:
Simon Schubert [Sat, 24 Feb 2007 14:24:06 +0000 (14:24 +0000)]
1:1 Userland threading stage 4.3/4:

Add service functions for the handling of exiting/dead lwps.
Rework the exiting code to use these functions.  This moves MI code
back to MI land.

17 years agoRemove superfluous conditional.
Simon Schubert [Sat, 24 Feb 2007 14:23:18 +0000 (14:23 +0000)]
Remove superfluous conditional.

sig is always == 0 at this place, and even if a signal arrived in between,
we after all were woken up by the timeout and the signal didn't break
our sleep.

17 years agoFix wording.
Sascha Wildner [Sat, 24 Feb 2007 13:00:32 +0000 (13:00 +0000)]
Fix wording.

17 years agoFix config declaration.
Sascha Wildner [Sat, 24 Feb 2007 09:07:09 +0000 (09:07 +0000)]
Fix config declaration.

17 years ago* Sync the list of devices in the polling(4) manpage and the polling
Sascha Wildner [Sat, 24 Feb 2007 08:27:38 +0000 (08:27 +0000)]
* Sync the list of devices in the polling(4) manpage and the polling
  section in LINT with the current state of affairs.

* Add references to polling(4) to the manpages of the drivers that
  support it.

Reported-by: Rick De Bella (on IRC)
17 years agoFix taskqueues to truely work on SMP systems.
Simon Schubert [Sat, 24 Feb 2007 04:36:21 +0000 (04:36 +0000)]
Fix taskqueues to truely work on SMP systems.

The taskqueue_task_td threads were created with kthread_create(), which
defaults to creating the thread on the current CPU.  Achieve correct
operation by using lwkt_create(), which allows specification of the
CPU the thread should run on.

Add a comment about the suggested usage of task structures.

17 years agoUnbreak wrong offset calculations by using the correct functions.
Simon Schubert [Sat, 24 Feb 2007 04:36:10 +0000 (04:36 +0000)]
Unbreak wrong offset calculations by using the correct functions.

17 years agoDon't print error messages on AE_TYPE from AcpiGetCurrentResources().
YONETANI Tomokazu [Fri, 23 Feb 2007 01:18:00 +0000 (01:18 +0000)]
Don't print error messages on AE_TYPE from AcpiGetCurrentResources().
Resource Manager has been changed to return AE_TYPE for any non-device
objects earlier in validation instead of going through functions and
return AE_NOT_FOUND.  Since acpi_parse_resources() is only interested
in device objects, don't complain on AE_TYPE either.

Noticed-by: many people
17 years agoAdd a 'NATA' kernel config file to go with the previous commit by Simon
Thomas E. Spanjaard [Thu, 22 Feb 2007 22:13:17 +0000 (22:13 +0000)]
Add a 'NATA' kernel config file to go with the previous commit by Simon
'corecode' Schubert.

17 years agoEnable building/installing of multiple kernels in nrelease.
Simon Schubert [Thu, 22 Feb 2007 21:02:49 +0000 (21:02 +0000)]
Enable building/installing of multiple kernels in nrelease.

Directly make use of this feature and build a NATA kernel.

17 years ago1:1 Userland threading stage 4.2/4:
Simon Schubert [Thu, 22 Feb 2007 15:50:50 +0000 (15:50 +0000)]
1:1 Userland threading stage 4.2/4:

Make signal system fully lwp-aware by splitting ksignal() in appropriate
functions.  Introduce lwpsignal(), which now contains the logic of
ksignal(), but can be used to deliver a signal to a specific lwp.

Convert consumers of ksignal() to use lwpsignal() when they actually
generate a thread-specific signal.

Fully implement proc_stop() and proc_unstop().

Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
17 years agoMake setrunnable() behave more like it is intended.
Simon Schubert [Thu, 22 Feb 2007 15:49:08 +0000 (15:49 +0000)]
Make setrunnable() behave more like it is intended.

Last changes made setrunnable() change lwp_stat, which is definitely
wrong.  Rework the code to match the original intent.

Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
17 years agoFix stopping of the current process, which was broken by the last commit.
Simon Schubert [Thu, 22 Feb 2007 15:48:55 +0000 (15:48 +0000)]
Fix stopping of the current process, which was broken by the last commit.

If the current process is still far inside the kernel and is going to
tsleep(), CURSIG()/issignal() in tsleep() have to see the pending signal;
holding the process in SSTOP is not enough.

However, if this is happening to a process which already received a stopping
signal and thus already is in SSTOP, but which wakes up for some reason, and
then posts a SA_STOP signal to itself, we may not ignore this, in spite of
already being in SSTOP.

All this process stopping business inside CURSIG()/issignal() makes all of
this very fragile, but we will have to live with it for now.

The bug can be reproduced by doing
% cat
^Z
% ps lx|grep '[c]at'

The correct output would be something like:
 4156 stop    p2  DL     0:00.00 cat

The broken behavior shows with:
17417 ttybg2  p2  TL     0:00.00 cat

Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
17 years agoFix symbol resolution for objects loaded by ELF dependencies.
Simon Schubert [Thu, 22 Feb 2007 13:15:55 +0000 (13:15 +0000)]
Fix symbol resolution for objects loaded by ELF dependencies.

We previously used the dagmembers of the specified object to resolve
a symbol.  However, dagmembers is only set for objects loaded by dlsym().
The correct behavior, implemented in this patch is to instead search
the needed objects.

This should fix KDE/SSL.

Taken-from: FreeBSD
Identified-by: Kimura Fuyuki <fuyuki@hadaly.org>
17 years ago- Add noise floor to stats
Sepherosa Ziehau [Thu, 22 Feb 2007 05:17:10 +0000 (05:17 +0000)]
- Add noise floor to stats
- Include current tx rate in stats
- Remove stub radar support
- Update ath_hal(4) to 0.9.20.3:
  o  Add support for newer parts that do not require separate keycache
     entries for tx+rx mic keys
  o  No more ds_vdata in tx/rx descriptors
  o  Split h/w tx/rx descriptor from s/w status
  o  As part of the descriptor split change the rate control module API
     so the ath_buf is passed in to the module so it can fetch both
     descriptor and status information as needed
- Replace replace h/w descriptor struct's with proper mask+shift defs for
  ath(4) sample TX rate control algorithm.
- Due to the license change in ah_osdep.[ch] (now dual-BSD/GPL), move them
  out of sys/contrib/dev/ath/dragonfly and put them into sys/dev/ath/ath
- Nuke the local patch for ah_osdep.c, ah_osdep.c can be modified directly.

Obtained-from: FreeBSD (sam@freebsd.org)

17 years agoDon't loop beyond maximum cpu ID (ncpus - 1); this probably hasn't done
YONETANI Tomokazu [Thu, 22 Feb 2007 04:02:50 +0000 (04:02 +0000)]
Don't loop beyond maximum cpu ID (ncpus - 1); this probably hasn't done
any harm because the corresponding bit in smp_active_mask is zero anyway.

17 years ago1:1 Userland threading stage 4.1/4:
Simon Schubert [Wed, 21 Feb 2007 15:47:02 +0000 (15:47 +0000)]
1:1 Userland threading stage 4.1/4:

Make ksignal() aware of multiple lwps.  Streamline logic.
Introduce find_lwp_for_signal(), which returns the best
lwp to deliver a signal to.

17 years ago1:1 Userland threading stage 2.20/4:
Simon Schubert [Wed, 21 Feb 2007 15:46:48 +0000 (15:46 +0000)]
1:1 Userland threading stage 2.20/4:

Unify access to pending threads with a new function, lwp_sigpend(), which
returns pending signals for the lwp, which includes both lwp-specific
signals and signals pending on the process.  The new function lwp_delsig()
is used to remove a certain signal from the pending set of both process and
lwp.

Rework the places which access the pending signal list to either use those
two functions or, where not possibly, to work on both lwp and proc signal
lists.

17 years ago1:1 Userland threading stage 2.19/4:
Simon Schubert [Wed, 21 Feb 2007 15:45:37 +0000 (15:45 +0000)]
1:1 Userland threading stage 2.19/4:

Move p_sig and p_code to struct lwp and add comment for them.
Make elf coredumps and checkpointing more lwp-aware.
Pass the signal responsible for the coredump down to the service routine.

17 years agoReplace magic number with constant.
Simon Schubert [Wed, 21 Feb 2007 02:19:39 +0000 (02:19 +0000)]
Replace magic number with constant.

Spotted-by: Jason Smethers
17 years agoAdd NATA devices to LINT, and hook the natacontrol binary into the build,
Thomas E. Spanjaard [Tue, 20 Feb 2007 15:58:27 +0000 (15:58 +0000)]
Add NATA devices to LINT, and hook the natacontrol binary into the build,
as it does no harm even when not using NATA.

Prodded-by: Justin C. Sherrill <justin@shiningsilence.com>
Sascha Wildner <saw@online.de>

17 years agoFix typo.
Sascha Wildner [Tue, 20 Feb 2007 15:52:43 +0000 (15:52 +0000)]
Fix typo.

17 years agoDon't assume mgtq will always be empty, drain mgtq explicitly.
Sepherosa Ziehau [Tue, 20 Feb 2007 14:24:21 +0000 (14:24 +0000)]
Don't assume mgtq will always be empty, drain mgtq explicitly.

Tested-by: Johannes Hofmann <Johannes.Hofmann@gmx.de> (iwi(4))
17 years agoFix typos.
Sascha Wildner [Mon, 19 Feb 2007 11:10:12 +0000 (11:10 +0000)]
Fix typos.

17 years agoDo some syncing with reality.
Sascha Wildner [Mon, 19 Feb 2007 09:34:33 +0000 (09:34 +0000)]
Do some syncing with reality.

17 years agoSilence compiler warning.
Simon Schubert [Mon, 19 Feb 2007 01:14:54 +0000 (01:14 +0000)]
Silence compiler warning.

17 years agoRemove outdated comment.
Simon Schubert [Mon, 19 Feb 2007 01:14:40 +0000 (01:14 +0000)]
Remove outdated comment.