]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoDo not remove logging sockets. This fixes an issue where logging
csjp [Thu, 14 Apr 2005 03:56:06 +0000 (03:56 +0000)]
Do not remove logging sockets. This fixes an issue where logging
sockets placed into prisons from the host environment get clobbered
by the prison's instance of cleanvar. (assuming /etc/rc is run in
the prison).

Discussed with: pjd, green, cperciva
MFC after: 1 week

19 years agoThe divide by zero panic must have been due to a bogus
mjacob [Thu, 14 Apr 2005 03:52:50 +0000 (03:52 +0000)]
The divide by zero panic must have been due to a bogus
period value. I suppose the BT adapter driver should be
fixed, but more importantly we should protect against
dividing by zero.

PR: kern/75603
MFC after: 1 week

19 years agoAllow user processes to completely empty out their LDT, now that user
peter [Thu, 14 Apr 2005 03:16:58 +0000 (03:16 +0000)]
Allow user processes to completely empty out their LDT, now that user
processes run from segment selectors that live in the GDT.  Doing this
used to be equivalent to committing suicide, but now this is a NOP.

19 years agoAdapt the libpthread patch for using i386_set_gsbase() to libthr.
peter [Thu, 14 Apr 2005 00:44:07 +0000 (00:44 +0000)]
Adapt the libpthread patch for using i386_set_gsbase() to libthr.

19 years agoUse the i386_set_gsbase() syscall if it is implemented in the kernel.
peter [Thu, 14 Apr 2005 00:13:20 +0000 (00:13 +0000)]
Use the i386_set_gsbase() syscall if it is implemented in the kernel.

This is a little hairy here because the allocation and usage of this
functionality is split into two places in libpthread.

19 years agoAttempt to use i386_set_gsbase(), and gracefully fall back to LDT methods
peter [Thu, 14 Apr 2005 00:04:50 +0000 (00:04 +0000)]
Attempt to use i386_set_gsbase(), and gracefully fall back to LDT methods
if the direct access methods are not implemented.

19 years agoAttempt i386_set_gsbase() before using the user_ldt code. Unimplemented
peter [Thu, 14 Apr 2005 00:02:37 +0000 (00:02 +0000)]
Attempt i386_set_gsbase() before using the user_ldt code.  Unimplemented
sysarch() calls return EINVAL, not SIGSYS.. so we can trivially adapt.

19 years agoAdd stubs for the %fs/%gs base management calls.
peter [Thu, 14 Apr 2005 00:01:35 +0000 (00:01 +0000)]
Add stubs for the %fs/%gs base management calls.

19 years ago - Remove a debugging printf that slipped in.
jeff [Wed, 13 Apr 2005 23:36:28 +0000 (23:36 +0000)]
 - Remove a debugging printf that slipped in.

Spotted by: Peter Wemm

19 years agoChange the segment limits to 4GB, we set the user accessible bit on all
peter [Wed, 13 Apr 2005 22:57:17 +0000 (22:57 +0000)]
Change the segment limits to 4GB, we set the user accessible bit on all
of the kernel address space already.  Intel recommend this anyway, because
using a non-4GB limit adds an additional clock cycle to address generation.
We were able to install 4GB segments into the LDT, so any limits we imposed
on %cs and %ds were academic anyway.  More importantly, this allows us to
make a page in the kernel readable to user applications, for holding things
like the signal trampoline and other fun things.

Move the user %cs/%ds segments from the LDT to the GDT.  There was no good
reason for them to be there anyway.  The old LDT entries are still there
but we can now relax the restriction that prevented users from emptying
the default LDT entries.

Putting user and kernel %cs and %ds together allows us to access the fast
sysenter/sysexit/syscall/sysret instructions.  syscall/sysret in particular
require that the user/kernel segments be laid out this way.  Reserve a slot
specifically for NDIS while here.

Create two user controllable slots in the GDT that are context switched
with the (kernel) thread.  This allows user applications to set two
user privilige selectors to arbitary values.  Create
i386_set_fsbase(void *base) and friends. (get/set, fs/gs).  For i386,
%gs is used by tls and the thread libraries and this means that user
processes no longer have to have the cost of having a custom LDT, and
we will no longer to do a ldt switch when activating a kthread/ithread in
the usual case any more.

In other words, we can now set the base address for %fs and %gs to arbitary
addresses without the pain of messing with ldt segments.

19 years agosort and expand the prunelist in a more or less generally sensibly
phk [Wed, 13 Apr 2005 21:22:20 +0000 (21:22 +0000)]
sort and expand the prunelist in a more or less generally sensibly
order.  Put some strategic comments in about how much storage is
necessary and a longer explanation on the top.

It is now pretty trivial to put nanoBSD on as little as a 64MB CF card.

Sponsored by: Soekris Engineering.

19 years agoIntroduce NANOBSD_PRUNE which can be used to zap out bits we don't
phk [Wed, 13 Apr 2005 21:20:16 +0000 (21:20 +0000)]
Introduce NANOBSD_PRUNE which can be used to zap out bits we don't
want to carry forward.

19 years agoUse df -i
phk [Wed, 13 Apr 2005 21:19:28 +0000 (21:19 +0000)]
Use df -i

19 years agoFix mss byte order, only affects synproxy code path.
glebius [Wed, 13 Apr 2005 21:05:55 +0000 (21:05 +0000)]
Fix mss byte order, only affects synproxy code path.

Submitted by: John L. Scarfone via OpenBSD
Reviewed by: mlaier
Obtained from: OpenBSD, rev. 1.483
MFC after: 2 days

19 years agoAdd a few regression tests for printf(1).
stefanf [Wed, 13 Apr 2005 20:08:17 +0000 (20:08 +0000)]
Add a few regression tests for printf(1).

19 years agoAssign 0.0 to the variable passed to getfloating() if the argument is missing.
stefanf [Wed, 13 Apr 2005 19:54:03 +0000 (19:54 +0000)]
Assign 0.0 to the variable passed to getfloating() if the argument is missing.

MFC after: 1 week

19 years agoFix compile error :-(.
imp [Wed, 13 Apr 2005 19:10:27 +0000 (19:10 +0000)]
Fix compile error :-(.

19 years agoFix an evil bug that appeared in September 2003. VM86 bios calls use two
peter [Wed, 13 Apr 2005 18:13:40 +0000 (18:13 +0000)]
Fix an evil bug that appeared in September 2003.  VM86 bios calls use two
of the __pcb_spare longs.  Except that fields were changed and one of the
spare values was used and the __pcb_spare field was reduced from two to one
long.  Now VM86 bios calls can trash the first 4 bytes of the next page
following the kernel stack/pcb.  This Is Bad(TM).  This bug has been
present in 5.2-release and onwards, and is still in RELENG_5.

Instead of tempting fate and trying to use "spare" fields, explicitly
reserve them.

19 years agoTest for NULL before use.
mdodd [Wed, 13 Apr 2005 17:39:43 +0000 (17:39 +0000)]
Test for NULL before use.

Submitted by: sam (Coverity)

19 years agoAdd KASSERT() to warn against NULL deref.
mdodd [Wed, 13 Apr 2005 17:36:18 +0000 (17:36 +0000)]
Add KASSERT() to warn against NULL deref.

Submitted by: sam (Coverity)

19 years agoUse pci_find_bsf() to retrieve the PCI device associated with
bms [Wed, 13 Apr 2005 17:34:38 +0000 (17:34 +0000)]
Use pci_find_bsf() to retrieve the PCI device associated with
a bus/device/function tuple.
This change enables pciconf(8) to work with CardBus devices.

Reviewed by: imp

19 years agoThe maximum allowable alloc is 16K not (16K-1).
julian [Wed, 13 Apr 2005 16:39:22 +0000 (16:39 +0000)]
The maximum allowable alloc is 16K not (16K-1).
This whole section is actually overly restrictive and
another patch is in the works.

19 years agoSince cbb implements the pcib_ interface, it must also implement the
imp [Wed, 13 Apr 2005 16:35:15 +0000 (16:35 +0000)]
Since cbb implements the pcib_ interface, it must also implement the
pcib_route_interrupt interface.  Since there's only one interrupt pin
in the CardBus form factor, everybody gets to share it.  Implement
cbb_route_interrupt to return the interrupt we have.

Suggested by: bms

19 years agoDefault to a interrupt router that returns an invalid interrupt.
imp [Wed, 13 Apr 2005 16:30:30 +0000 (16:30 +0000)]
Default to a interrupt router that returns an invalid interrupt.
Otherwise, busses that implement the pcib interface that forget to
implement pcib_route_interrupt would return EIO, which the caller
interprets as 'use interrupt 6'.  This is likely the cause of much of
the grief that we had when I enabled power modes for the cardbus
bridge, since the card needed to reroute the interrupt to it and it
was getting 6 which was d by the pccbb sanity checks.

19 years agopmap_update() is gone.
cognet [Wed, 13 Apr 2005 16:02:03 +0000 (16:02 +0000)]
pmap_update() is gone.

19 years agoNever hardcode /sys into these Makefiles. The proper way to spell it is $S.
imp [Wed, 13 Apr 2005 14:49:57 +0000 (14:49 +0000)]
Never hardcode /sys into these Makefiles.  The proper way to spell it is $S.

Also, move the -I stuff to the centralized kern.pre.mk.  However, it
might be better to add these flags to files.conf.  This is a short
term fix to fix the broken builds on my machine (I don't have a valid
/sys link).

19 years agoNG_MKRESPONSE() macro includes sizeof struct ng_mesg when doing allocation.
glebius [Wed, 13 Apr 2005 14:03:28 +0000 (14:03 +0000)]
NG_MKRESPONSE() macro includes sizeof struct ng_mesg when doing allocation.

PR: kern/79806
Submitted by: Wojciech A. Koszek

19 years agoAccording to the comment in struct tty, t_modem is optional; hence we should
avatar [Wed, 13 Apr 2005 13:56:17 +0000 (13:56 +0000)]
According to the comment in struct tty, t_modem is optional; hence we should
guard against NULL t_modem entry. Otherwise, driver doesn't have t_modem
callback implemented(such like sys/dev/usb/ucycom.c) would panic when
someone opens the driver's associated tty device.

Reviewed by: phk, sam (mentor)

19 years agoMerge two slice_type_name() functions.
nyan [Wed, 13 Apr 2005 13:42:38 +0000 (13:42 +0000)]
Merge two slice_type_name() functions.

19 years ago- Remove ifdef PC98.
nyan [Wed, 13 Apr 2005 13:26:48 +0000 (13:26 +0000)]
- Remove ifdef PC98.
- Reduce diffs from i386.

19 years agoRemove a meaningless include.
nyan [Wed, 13 Apr 2005 13:18:32 +0000 (13:18 +0000)]
Remove a meaningless include.

19 years agoMove pc98 specific parts to the pc98 specific file.
nyan [Wed, 13 Apr 2005 13:12:12 +0000 (13:12 +0000)]
Move pc98 specific parts to the pc98 specific file.

19 years agoRemove ifdef PC98.
nyan [Wed, 13 Apr 2005 13:02:58 +0000 (13:02 +0000)]
Remove ifdef PC98.

19 years agoSplit VarFind() into a series of functions tailored for the different sets
harti [Wed, 13 Apr 2005 11:40:44 +0000 (11:40 +0000)]
Split VarFind() into a series of functions tailored for the different sets
of flags originally passed to VarFind(). This eliminates the code by
removing a bunch of tests.

Patch: 7.173

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

19 years ago - Change all filesystems and vfs_cache to relock the dvp once the child is
jeff [Wed, 13 Apr 2005 10:59:09 +0000 (10:59 +0000)]
 - Change all filesystems and vfs_cache to relock the dvp once the child is
   locked in the ISDOTDOT case.  Se vfs_lookup.c r1.79 for details.

Sponsored by: Isilon Systems, Inc.

19 years ago - Change vop_lookup_post assertions to reflect recent vfs_lookup changes.
jeff [Wed, 13 Apr 2005 10:57:53 +0000 (10:57 +0000)]
 - Change vop_lookup_post assertions to reflect recent vfs_lookup changes.

Sponsored by: Isilon Systems, Inc.

19 years ago - Further simplify lookup; Force all filesystems to relock in the DOTDOT
jeff [Wed, 13 Apr 2005 10:57:13 +0000 (10:57 +0000)]
 - Further simplify lookup; Force all filesystems to relock in the DOTDOT
   case.  There are bugs in some which didn't unlock in the ISDOTDOT case
   to begin with that need to be addressed seperately.  This simplifies
   things anyway.
 - Fix relookup() to prevent it from vrele()'ing the dvp while the vp
   is locked.  Catch up to other lookup changes.

Sponsored by: Isilon Systems, Inc.
Reported by: Peter Wemm

19 years agoAdd NO_DICT handle. Saves almost 3.5 MB installed.
phk [Wed, 13 Apr 2005 10:02:58 +0000 (10:02 +0000)]
Add NO_DICT handle.  Saves almost 3.5 MB installed.

19 years agoAdd #defines for control fields and address bits.
mdodd [Wed, 13 Apr 2005 08:14:14 +0000 (08:14 +0000)]
Add #defines for control fields and address bits.

19 years agoSupport for the GTCO Digipad.
mdodd [Wed, 13 Apr 2005 07:25:45 +0000 (07:25 +0000)]
Support for the GTCO Digipad.

19 years agoAdd a ata_setmode method so we dont panic on setmode.
sos [Wed, 13 Apr 2005 07:14:17 +0000 (07:14 +0000)]
Add a ata_setmode method so we dont panic on setmode.
Note that the mode is only set on the device, we (mostly) have
no knowledge on how to set mode on the controller if at all possible.

19 years agoAdd the -S flag which produces the -s output without
mdodd [Wed, 13 Apr 2005 07:11:54 +0000 (07:11 +0000)]
Add the -S flag which produces the -s output without
mixing field separators.

19 years agoShow descriptions for type CTLTYPE_NODES.
mdodd [Wed, 13 Apr 2005 06:59:07 +0000 (06:59 +0000)]
Show descriptions for type CTLTYPE_NODES.

19 years agoZero buffers. Seatbelt against sysctl(3) returning non-NUL terminated
mdodd [Wed, 13 Apr 2005 06:56:31 +0000 (06:56 +0000)]
Zero buffers.  Seatbelt against sysctl(3) returning non-NUL terminated
string data.

19 years agoMake it clear that the statement following the conditional is a NOP.
mdodd [Wed, 13 Apr 2005 06:42:43 +0000 (06:42 +0000)]
Make it clear that the statement following the conditional is a NOP.

19 years ago Data is not information;
phk [Wed, 13 Apr 2005 06:11:49 +0000 (06:11 +0000)]
Data is not information;
Information is not knowledge;
Knowledge is not wisdom
-- Gary Flake

19 years agoFollow i386's suit and include AGP support in the generic kernel.
anholt [Wed, 13 Apr 2005 06:00:07 +0000 (06:00 +0000)]
Follow i386's suit and include AGP support in the generic kernel.

19 years agoAdd support for more than two datasets. Currently limited to 7 though
mdodd [Wed, 13 Apr 2005 05:50:56 +0000 (05:50 +0000)]
Add support for more than two datasets.  Currently limited to 7 though
the limit is only the number of meaningful graph symbols available.

Statistical comparison is performed between the first dataset and
any further datasets.

No objection by:  phk

19 years agoDefine additional commands.
mdodd [Wed, 13 Apr 2005 05:09:49 +0000 (05:09 +0000)]
Define additional commands.

Obtained from:  Linux

19 years agoWhitespace cleanup.
mdodd [Wed, 13 Apr 2005 05:06:57 +0000 (05:06 +0000)]
Whitespace cleanup.

19 years agoRemove unnecessary dpt_free().
mdodd [Wed, 13 Apr 2005 05:03:14 +0000 (05:03 +0000)]
Remove unnecessary dpt_free().

19 years ago o Code cleanup, eliminate private thread id map, directly
davidxu [Wed, 13 Apr 2005 04:57:38 +0000 (04:57 +0000)]
 o Code cleanup, eliminate private thread id map, directly
   use lwpid as thread id.
 o Export tls pointer.

19 years agoImplement SOUND_MIXER_INFO ioctl in compat layer.
mdodd [Wed, 13 Apr 2005 04:33:06 +0000 (04:33 +0000)]
Implement SOUND_MIXER_INFO ioctl in compat layer.

19 years agoAdd support for O_NOFOLLOW and O_DIRECT to Linux fcntl() F_GETFL/F_SETFL.
mdodd [Wed, 13 Apr 2005 04:31:43 +0000 (04:31 +0000)]
Add support for O_NOFOLLOW and O_DIRECT to Linux fcntl() F_GETFL/F_SETFL.

19 years agoInvert conditional and use continue to reduce nesting.
mdodd [Wed, 13 Apr 2005 03:36:24 +0000 (03:36 +0000)]
Invert conditional and use continue to reduce nesting.

19 years agoAdd ISACFGATTR_HINTS flag to allow detection of a device that was created
mdodd [Wed, 13 Apr 2005 03:26:24 +0000 (03:26 +0000)]
Add ISACFGATTR_HINTS flag to allow detection of a device that was created
as a result of the hints mechanism.

19 years agoAdd -S option which allows to change the pathname of the privileged
hrs [Wed, 13 Apr 2005 03:19:41 +0000 (03:19 +0000)]
Add -S option which allows to change the pathname of the privileged
socket /var/run/logpriv.

Reviewed by: glebius and kensmith
MFC after: 2 days

19 years agoBuild cpufreq on ia64. The upcoming Montecito processor supports the
marcel [Wed, 13 Apr 2005 02:20:17 +0000 (02:20 +0000)]
Build cpufreq on ia64. The upcoming Montecito processor supports the
Enhanced SpeedStep (that is, a follow-up of it called Foxton). Until
we actually have support for that, we build to catch regressions in
the framework.

Triggered by: njl

19 years agoSave argv[0] and use it in usage message.
mdodd [Wed, 13 Apr 2005 01:44:59 +0000 (01:44 +0000)]
Save argv[0] and use it in usage message.

19 years agoInvert conditional and use continue to reduce nesting.
mdodd [Wed, 13 Apr 2005 01:32:06 +0000 (01:32 +0000)]
Invert conditional and use continue to reduce nesting.

19 years agoBump __FreeBSD_version for LOCAL_CREDS, LOCAL_CONNWAIT.
mdodd [Wed, 13 Apr 2005 00:45:49 +0000 (00:45 +0000)]
Bump __FreeBSD_version for LOCAL_CREDS, LOCAL_CONNWAIT.

Requested by:    marcus

19 years agoProvide a sysctl (net.link.tap.user_open) to allow unpriviliged
mdodd [Wed, 13 Apr 2005 00:30:19 +0000 (00:30 +0000)]
Provide a sysctl (net.link.tap.user_open) to allow unpriviliged
acces to tap(4) device nodes based on file system permission.

Duplicate the 'debug.if_tap_debug' sysctl under the
'net.link.tap' hierarchy.

19 years agoImplement unix(4) socket options LOCAL_CREDS and LOCAL_CONNWAIT.
mdodd [Wed, 13 Apr 2005 00:01:46 +0000 (00:01 +0000)]
Implement unix(4) socket options LOCAL_CREDS and LOCAL_CONNWAIT.

- Add unp_addsockcred() (for LOCAL_CREDS).
- Add an argument to unp_connect2() to differentiate between
  PRU_CONNECT and PRU_CONNECT2. (for LOCAL_CONNWAIT)

Obtained from:  NetBSD (with some changes)

19 years agoConsistently style function declarations in kern_malloc.c.
rwatson [Tue, 12 Apr 2005 23:54:34 +0000 (23:54 +0000)]
Consistently style function declarations in kern_malloc.c.

MFC after: 3 days

19 years agoIndent structure members.
mdodd [Tue, 12 Apr 2005 23:49:57 +0000 (23:49 +0000)]
Indent structure members.

19 years agoFill traceme and events fields.
davidxu [Tue, 12 Apr 2005 23:33:08 +0000 (23:33 +0000)]
Fill traceme and events fields.

19 years agodon't use C keyword register as CPP macro argument name
imp [Tue, 12 Apr 2005 23:32:26 +0000 (23:32 +0000)]
don't use C keyword register as CPP macro argument name

19 years agoDocument the SUN_LEN macro.
mdodd [Tue, 12 Apr 2005 23:24:40 +0000 (23:24 +0000)]
Document the SUN_LEN macro.

Obtained from:  NetBSD

19 years agoUse PCPU_LAZY_INC() for cnt.v_{intr,trap,syscalls} rather than atomic
jhb [Tue, 12 Apr 2005 23:18:54 +0000 (23:18 +0000)]
Use PCPU_LAZY_INC() for cnt.v_{intr,trap,syscalls} rather than atomic
operations in some places and simple non-per CPU math in others.

19 years agoTidy vcnt() by moving a duplicated line above #ifdef and removing a useless
jhb [Tue, 12 Apr 2005 23:15:28 +0000 (23:15 +0000)]
Tidy vcnt() by moving a duplicated line above #ifdef and removing a useless
variable.

19 years agoUse NULL rather than 0 in a couple of places.
jhb [Tue, 12 Apr 2005 23:12:46 +0000 (23:12 +0000)]
Use NULL rather than 0 in a couple of places.

19 years agoThe memory operands to fldcw and ldmxcsr are inputs, not outputs.
jhb [Tue, 12 Apr 2005 23:12:00 +0000 (23:12 +0000)]
The memory operands to fldcw and ldmxcsr are inputs, not outputs.

19 years agoWe have an asm version of bcmp(), so we could use it as well.
cognet [Tue, 12 Apr 2005 22:46:09 +0000 (22:46 +0000)]
We have an asm version of bcmp(), so we could use it as well.

19 years agoThe latest release of the FreeBSD driver (twa) for
vkashyap [Tue, 12 Apr 2005 22:07:11 +0000 (22:07 +0000)]
The latest release of the FreeBSD driver (twa) for
3ware's 9xxx series controllers.  This corresponds to
the 9.2 release (for FreeBSD 5.2.1) on the 3ware website.

Highlights of this release are:

1. The driver has been re-architected to use a "Common Layer"
    (all tw_cl* files), which is a consolidation of all OS-independent
    parts of the driver.  The FreeBSD OS specific portions of the
    driver go into an "OS Layer" (all tw_osl* files).
    This re-architecture is to achieve better maintainability, consistency
    of behavior across OS's, and better portability to new OS's (drivers
    for new OS's can be written by just adding an OS Layer that's specific
    to the OS, by complying to a "Common Layer Programming Interface" API.

2. The driver takes advantage of multiple processors.

3. The driver has a new firmware image bundled, the new features of which
   include Online Capacity Expansion and multi-lun support, among others.
   More details about 3ware's 9.2 release can be found here:
   http://www.3ware.com/download/Escalade9000Series/9.2/9.2_Release_Notes_Web.pdf

Since the Common Layer is used across OS's, the FreeBSD specific include
path for header files (/sys/dev/twa) is not part of the #include pre-processor
directive in any of the source files.  For being able to integrate twa into
the kernel despite this, Makefile.<arch> has been changed to add the include
path to CFLAGS.

Reviewed by: scottl

19 years agoReplace spl protection in rtcin() and writertc() with spinlocks
mdodd [Tue, 12 Apr 2005 20:49:31 +0000 (20:49 +0000)]
Replace spl protection in rtcin() and writertc() with spinlocks
using the existing clock_lock mutex.

19 years ago- Consistently protect against NULL dereference.
mdodd [Tue, 12 Apr 2005 20:14:56 +0000 (20:14 +0000)]
- Consistently protect against NULL dereference.
- Simplify conditional logic to make code easier to read.

19 years agoAdd myself as a new ports committer.
jylefort [Tue, 12 Apr 2005 20:06:43 +0000 (20:06 +0000)]
Add myself as a new ports committer.

Approved by: adamw (mentor)

19 years agoDot the i's:
marcel [Tue, 12 Apr 2005 18:50:57 +0000 (18:50 +0000)]
Dot the i's:
1  Move the debug.clock_adjust_* sysctls to debug.clock.adjust_* to
   make it easier to get only the clock statistics.
2  Make the sysctls read-only [suggested by Marius].
3  When determining the new clock adjustment, we checked for an error
   either larger than 12.5% or smaller than 12.5%. We left out an error
   of exactly 12.5%. For errors larger than 12.5% we adjust the clock
   reload value in such a way that the next clock interrupt would be
   early (as in premature). For errors less than 12.5% we stopped the
   adjustment.
   The current algorithm doesn't benefit from excluding an error of
   exactly 12.5%. Change the code to stop adjusting the clock if the
   error is *not* larger than 12.5% [suggested by Marius].

Discussed with: marius@

19 years agohonor new IEEE80211_KEY_GROUP key flag
sam [Tue, 12 Apr 2005 17:56:43 +0000 (17:56 +0000)]
honor new IEEE80211_KEY_GROUP key flag

Reviewed by: Tai-hwa Liang

19 years agoRevise crypto api lightly to improve group key handling:
sam [Tue, 12 Apr 2005 17:55:13 +0000 (17:55 +0000)]
Revise crypto api lightly to improve group key handling:
o don't pre-assign key index to the global key table entries so device
  has a chance to decide what to use
o make ieee80211_crypto_newkey take the desired flags as an argument
  instead of wacking the key structure directly; this eliminates a
  bunch of code warts
o add a new flag IEEE80211_KEY_GROUP to indicate a key is a WPA Group
  key so devices don't need to guess (temporarily add this flag in the
  ioctl code until we can get wpa_supplicant+hostapd updated)
o shuffle IEEE80211_KEY_* bits to move flags used internally to the high
  nibble of the flags word

Reviewed by: Tai-hwa Liang

19 years agoUnbreak the powerpc build by fixing some ATA constants that were renamed.
ssouhlal [Tue, 12 Apr 2005 15:59:54 +0000 (15:59 +0000)]
Unbreak the powerpc build by fixing some ATA constants that were renamed.

Approved by: grehan (mentor)

19 years agoBig cleanup of resource code for pccard. Once coventry noticed
imp [Tue, 12 Apr 2005 15:25:31 +0000 (15:25 +0000)]
Big cleanup of resource code for pccard.  Once coventry noticed
problems here, it became clear we were being too complex.

o Don't keep track of resources in two places
o Use resource_list_purge instead of rolling our own
o Just reassign the ownership of the resource, rather than freeing it
  and reallocating it.
o Fix compile problems when sizeof(u_long) != sizeof(int)

19 years agoOn amd64 int64_t != long long.
sos [Tue, 12 Apr 2005 15:24:00 +0000 (15:24 +0000)]
On amd64 int64_t != long long.

19 years agoDocument that dumpdev may be set to AUTO to dump to the first appropriate
thomas [Tue, 12 Apr 2005 15:21:51 +0000 (15:21 +0000)]
Document that dumpdev may be set to AUTO to dump to the first appropriate
swap device listed in /etc/fstab.

19 years agoresource_list_purge: release the resources in this list, and purge the
imp [Tue, 12 Apr 2005 15:20:36 +0000 (15:20 +0000)]
resource_list_purge: release the resources in this list, and purge the
elements of this list (eg, reset it).

Man page to follow

19 years agoFix minor discrepancy between documentation and implementation: the
thomas [Tue, 12 Apr 2005 15:16:32 +0000 (15:16 +0000)]
Fix minor discrepancy between documentation and implementation: the
destination port is incremented for each packet sent to the same hop.

MFC after: 1 week

19 years ago(handler): When exitting upon an abnormal signal, yppush_exit() should not
thomas [Tue, 12 Apr 2005 15:02:57 +0000 (15:02 +0000)]
(handler): When exitting upon an abnormal signal, yppush_exit() should not
attempt to clear its pending jobs list, as this could trigger another
signal, and cause an infinite recursion. What yppush_exit() tests
in order to determine whether to flush pending jobs is the
yppush_joblist chained list, so this is what needs to be cleared
in that case (not the yppush_jobs counter).

Reviewed by: audit@ (no objection)
MFC after: 2 weeks

19 years agoAdding a brief description for pirtool.
avatar [Tue, 12 Apr 2005 13:52:55 +0000 (13:52 +0000)]
Adding a brief description for pirtool.

Reviewed by: sam (mentor)

19 years agoAutogenerate hardware notes for snd_als4000.4, snd_fm801.4 and snd_via8233.4.
brueffer [Tue, 12 Apr 2005 13:33:20 +0000 (13:33 +0000)]
Autogenerate hardware notes for snd_als4000.4, snd_fm801.4 and snd_via8233.4.

19 years agoBetter use the right name for the VIA software RAID.
sos [Tue, 12 Apr 2005 13:29:47 +0000 (13:29 +0000)]
Better use the right name for the VIA software RAID.

19 years agoRegen:
sheldonh [Tue, 12 Apr 2005 13:03:42 +0000 (13:03 +0000)]
Regen:

* Hart: rev 517 of pcidevs.txt (2005-03-21)
* Boemler:      vendors.txt (2005-04-04)

Approved by: re (kensmith)

19 years agoAdd support for VIA Tech metadata as used on thier SATA parts.
sos [Tue, 12 Apr 2005 12:25:27 +0000 (12:25 +0000)]
Add support for VIA Tech metadata as used on thier SATA parts.

19 years agoSilence a warning on systems without carp(4).
ru [Tue, 12 Apr 2005 08:47:04 +0000 (08:47 +0000)]
Silence a warning on systems without carp(4).

19 years agoProtect against recursive labels creation in simlar way as it is done
pjd [Tue, 12 Apr 2005 08:14:15 +0000 (08:14 +0000)]
Protect against recursive labels creation in simlar way as it is done
in BSD and MBR classes, ie. if provider below us uses the same metadata,
don't create labels based on the metadata.
This allows to create labels on geoms with rank != 1 without hacks.

Tested by: Chris Elsworth <chris@shagged.org> on sparc64
OK'ed by: phk
MFC after: 2 weeks

19 years agorman_set_device() seems to have been omitted by mistake. Implement it.
imp [Tue, 12 Apr 2005 06:21:59 +0000 (06:21 +0000)]
rman_set_device() seems to have been omitted by mistake.  Implement it.

19 years agoCleanup of resource allocation code after having my attention focused on
imp [Tue, 12 Apr 2005 06:00:06 +0000 (06:00 +0000)]
Cleanup of resource allocation code after having my attention focused on
this code:
o rid is stored in the resource, so don't bother keeping track of it here.
o Implement memory space
o Don't try to activate 'memory card' CFEs.  This is type memory, as opposed
  to the memory resource.

19 years ago - Remove unused include.
jeff [Tue, 12 Apr 2005 05:45:58 +0000 (05:45 +0000)]
 - Remove unused include.

19 years ago - Differentiate two UPGRADE panics so I have a better idea of what's going
jeff [Tue, 12 Apr 2005 05:43:03 +0000 (05:43 +0000)]
 - Differentiate two UPGRADE panics so I have a better idea of what's going
   on here.

19 years agoUse return value of resource_list_add to avoid a second
imp [Tue, 12 Apr 2005 04:30:35 +0000 (04:30 +0000)]
Use return value of resource_list_add to avoid a second
resource_list_find.  Check to make sure that rle is not NULL and panic
if it is (but it appears that resource_list_add already panics, so I'm
not entirely sure it is necessary now).

Add a test to make sure we have a interrupt resource when we're
disabling it.  This is also a cannot happen, but the extra care
shoudln't hurt.

Found by: Coventry tool via sam@

19 years agoReturn the resource created/found in resource_list_add to avoid an extra
imp [Tue, 12 Apr 2005 04:22:17 +0000 (04:22 +0000)]
Return the resource created/found in resource_list_add to avoid an extra
resouce_list_find in some places.

Suggested by: sam
Found by: Coventry Analysis tool.