]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoDutch calendar.
grog [Sat, 9 Jun 2007 05:41:18 +0000 (05:41 +0000)]
Dutch calendar.

Submitteed by: edwin@

17 years agoPhysical memory regions can be larger than INT_MAX. Change size1
marcel [Sat, 9 Jun 2007 01:19:08 +0000 (01:19 +0000)]
Physical memory regions can be larger than INT_MAX. Change size1
from an int to a long to avoid printing negative byte and page
counts.

17 years agoCleanup messages printed on attach. Since the description gets set to
imp [Fri, 8 Jun 2007 22:25:09 +0000 (22:25 +0000)]
Cleanup messages printed on attach.  Since the description gets set to
what we print, don't print it anymore.  And don't compute it anymore.
And don't malloc/free memory for it anymore.  While I'm here, prefer
device_printf where appropriate.

17 years agoDon't cast the command argument to ether_ioctl() to an int since its not an
dwhite [Fri, 8 Jun 2007 22:00:56 +0000 (22:00 +0000)]
Don't cast the command argument to ether_ioctl() to an int since its not an
int anymore. This was causing all sorts of bad behavior when booting a system
with an nve interface present.

17 years agoRemove the MUTEX_WAKE_ALL option and make it the default behaviour for our
attilio [Fri, 8 Jun 2007 21:36:52 +0000 (21:36 +0000)]
Remove the MUTEX_WAKE_ALL option and make it the default behaviour for our
mutexes.
Currently we alredy force MUTEX_WAKE_ALL beacause of some problems with the
!MUTEX_WAKE_ALL case (unavioidable priority inversion).

17 years agoEnable AUDIT by default in the GENERIC kernel, allowing security event
rwatson [Fri, 8 Jun 2007 20:29:07 +0000 (20:29 +0000)]
Enable AUDIT by default in the GENERIC kernel, allowing security event
auditing to be turned on without a kernel recompile, just an rc.conf
option.

Approved by: re (kensmith)
Obtained from: TrustedBSD Project

17 years agoAdd my copyright.
marcel [Fri, 8 Jun 2007 16:20:03 +0000 (16:20 +0000)]
Add my copyright.

Requested by: pjd@

17 years agoReplace a constant with an already defined symbolic name for it.
yar [Fri, 8 Jun 2007 13:43:28 +0000 (13:43 +0000)]
Replace a constant with an already defined symbolic name for it.

Tested with: md5(1)

17 years agoAdd a sysctl for the purge run interval so that it can
yar [Fri, 8 Jun 2007 13:35:51 +0000 (13:35 +0000)]
Add a sysctl for the purge run interval so that it can
be tuned along with the rest of hostcache parameters.
The new sysctl name is `net.inet.tcp.hostcache.prune'.

17 years agoReset dc->paddr and dc->reset if we cannot read configuration ROM.
simokawa [Fri, 8 Jun 2007 12:58:06 +0000 (12:58 +0000)]
Reset dc->paddr and dc->reset if we cannot read configuration ROM.

17 years agoCorrect the definition of PFIL_HOOKED() so that it compares
gallatin [Fri, 8 Jun 2007 12:43:25 +0000 (12:43 +0000)]
Correct the definition of PFIL_HOOKED() so that it compares
the value of ph_nhooks to zero, not the address.  This removes
extranious calls to pfil_run_hooks (and an rw lock) from the
network stack's critical path when no pfil hooks are active.

Reviewed by: csjp
Sponsored by: Myricom Inc.

17 years ago- Reduce number of atomic operations needed to be implemented in asm by
pjd [Fri, 8 Jun 2007 12:35:47 +0000 (12:35 +0000)]
- Reduce number of atomic operations needed to be implemented in asm by
  implementing some of them using existing ones.
- Allow to compile ZFS on all archs and use atomic operations surrounded
  by global mutex on archs we don't have or can't have all atomic
  operations needed by ZFS.

17 years agoMissing atomic operations for ZFS/ia64.
pjd [Fri, 8 Jun 2007 12:26:30 +0000 (12:26 +0000)]
Missing atomic operations for ZFS/ia64.

Submitted by: marcel

17 years agoDouble the WITNESS and DIAGNOSTIC benchmark warnings right before we
phk [Fri, 8 Jun 2007 11:47:36 +0000 (11:47 +0000)]
Double the WITNESS and DIAGNOSTIC benchmark warnings right before we
go into userland to improve the chances of people noticing them.

17 years ago- RTO was not being initialized to 0, thus the rtt calculation
rrs [Fri, 8 Jun 2007 10:57:11 +0000 (10:57 +0000)]
- RTO was not being initialized to 0, thus the rtt calculation
  algoritm would not go through the proper initialization.
- The initialization was incorrect as well, causing problems in
  sat networks with > 1sec RTT
- Get rid of magic numbers in RTT calculations.

17 years agoTimestamp after sent.
simokawa [Fri, 8 Jun 2007 09:04:30 +0000 (09:04 +0000)]
Timestamp after sent.

17 years agoDon't invalidate dcons buffer on shutdown.
simokawa [Fri, 8 Jun 2007 08:23:14 +0000 (08:23 +0000)]
Don't invalidate dcons buffer on shutdown.
We would like to keep connection after halt.

17 years agoFix a race after a bus reset.
simokawa [Fri, 8 Jun 2007 07:53:59 +0000 (07:53 +0000)]
Fix a race after a bus reset.
- We are in FWBUSINIT state just after SID interrupt.
- Do not pass normal xfers before bus probe is done.

17 years agoIn getblk(), before gbincore(), use BO_LOCK directly when locking
delphij [Fri, 8 Jun 2007 07:05:08 +0000 (07:05 +0000)]
In getblk(), before gbincore(), use BO_LOCK directly when locking
the bufobj, rather than using VI_LOCK, like what was done with
revision 1.453.

17 years agoIncrease inputFS buffer. awk will not accept some old scripts
delphij [Fri, 8 Jun 2007 05:44:12 +0000 (05:44 +0000)]
Increase inputFS buffer.  awk will not accept some old scripts
otherwise.

17 years agoClean up escape sequence handling and add support for
simokawa [Fri, 8 Jun 2007 05:26:11 +0000 (05:26 +0000)]
Clean up escape sequence handling and add support for
resetting target and suspending dconschat.

17 years agoSync with other platforms: add kluge to use contigmalloc when the
marcel [Fri, 8 Jun 2007 04:46:50 +0000 (04:46 +0000)]
Sync with other platforms: add kluge to use contigmalloc when the
alignment is larger than the size and print a diagnostic when we
didn't satisfy the alignment.

17 years agoAdd the address of IDT in the configuration ROM. (i386/amd64 only)
simokawa [Fri, 8 Jun 2007 04:33:25 +0000 (04:33 +0000)]
Add the address of IDT in the configuration ROM. (i386/amd64 only)

A change to dconschat(8) will follow so that it can bomb
this address over FireWire to reset a wedged system.

Though this method is just a hack and far from perfection,
it should be useful if you don't want to go machine room
just to reset or to power-cycle a machine without
remote-managed power supply.  And much better than doing:
# fwcontrol -m target-eui64
# dd if=/dev/zero of=/dev/fwmem0.2 bs=1m

17 years agogcc 4.2 thinks that tupleid is uninitialized. Or might be used
imp [Fri, 8 Jun 2007 04:03:57 +0000 (04:03 +0000)]
gcc 4.2 thinks that tupleid is uninitialized.  Or might be used
uninitialized.  It gets passed into other routines that initialize
it...  Cope by initializing.

Submitted by: mjacob

17 years agoReplace breakpoint() with kdb_enter().
simokawa [Fri, 8 Jun 2007 03:05:57 +0000 (03:05 +0000)]
Replace breakpoint() with kdb_enter().

17 years ago- Fixed a problem that caused autonegotiation failures.
davidch [Fri, 8 Jun 2007 02:34:44 +0000 (02:34 +0000)]
- Fixed a problem that caused autonegotiation failures.

Submitted by: tor.egge@cvsup.no.freebsd.org
MFC after: 4 weeks

17 years agoSet warning level to 2.
davidxu [Fri, 8 Jun 2007 02:21:13 +0000 (02:21 +0000)]
Set warning level to 2.

17 years agoPropagae volatile qualifier.
mjacob [Fri, 8 Jun 2007 01:54:22 +0000 (01:54 +0000)]
Propagae volatile qualifier.

17 years agoFix preprocessor code to check for a symbol being defined prior to checking
mjacob [Fri, 8 Jun 2007 01:49:04 +0000 (01:49 +0000)]
Fix preprocessor code to check for a symbol being defined prior to checking
for non-zero.

17 years agoCarry volatile type in cast so gcc 4.2 will be happy.
mjacob [Fri, 8 Jun 2007 01:48:23 +0000 (01:48 +0000)]
Carry volatile type in cast so gcc 4.2 will be happy.

17 years agoQuiet GCC 4.2 warning.
mjacob [Fri, 8 Jun 2007 01:39:04 +0000 (01:39 +0000)]
Quiet GCC 4.2 warning.

17 years agoRemove the __inline qualifier from a function which is in fact not an
mjacob [Fri, 8 Jun 2007 01:37:47 +0000 (01:37 +0000)]
Remove the __inline qualifier from a function which is in fact not an
inline but instead a module wide function, thus quieting a GCC 4.2 warning.

17 years agoInclude now unused var within #if 0 where it come back if the other
mjacob [Fri, 8 Jun 2007 01:34:04 +0000 (01:34 +0000)]
Include now unused var within #if 0 where it come back if the other
#if 0 code comes back- quiets gcc 4.2

17 years agoRemove assignment to uninitialized variable that wasn't then used anyway.
mjacob [Fri, 8 Jun 2007 01:21:20 +0000 (01:21 +0000)]
Remove assignment to uninitialized variable that wasn't then used anyway.

17 years agoPoll bus resets on FireWire while kdb/gdb is active.
simokawa [Fri, 8 Jun 2007 00:54:44 +0000 (00:54 +0000)]
Poll bus resets on FireWire while kdb/gdb is active.

Now, it's safe to call the fwohci interrupt(polling) routine while ddb/gdb
is active. After this change, a dcons connnection over FireWire can survive
bus resets even in kernel debugger.

This means that it is not too late to plug a FireWire cable after a panic
to investigate the problem.

Actually there is a small window(between a jump to kernel from loader and
initialization of dcons_crom) in which no one can take care of a bus reset.
Except that window, firewire console should keep working
from loader to reboot even with a panic and a bus reset.
(as far as you enable LOADER_FIREWIRE_SUPPORT)

17 years agoMove per-process audit state from a pointer in the proc structure to
rwatson [Thu, 7 Jun 2007 22:27:15 +0000 (22:27 +0000)]
Move per-process audit state from a pointer in the proc structure to
embedded storage in struct ucred.  This allows audit state to be cached
with the thread, avoiding locking operations with each system call, and
makes it available in asynchronous execution contexts, such as deep in
the network stack or VFS.

Reviewed by: csjp
Approved by: re (kensmith)
Obtained from: TrustedBSD Project

17 years agoThere's no nobounce_dmamap on arm.
cognet [Thu, 7 Jun 2007 21:51:09 +0000 (21:51 +0000)]
There's no nobounce_dmamap on arm.

17 years agoIn tcp_hc_insert() we may have the case where we have hit the global
andre [Thu, 7 Jun 2007 21:41:50 +0000 (21:41 +0000)]
In tcp_hc_insert() we may have the case where we have hit the global
cache size limit but this bucket row is empty.  Normally we want to
recycle the oldest entry in the bucket row.  If there isn't any the
TAILQ_REMOVE leads to a panic by trying to remove a non-existing
element.  Fix this by just returning NULL and failing the insert.
This is not a problem as the TCP hostache is only advisory.

Submitted by: jhb

17 years ago- Remove unused variable from create_thread().
jhb [Thu, 7 Jun 2007 19:45:19 +0000 (19:45 +0000)]
- Remove unused variable from create_thread().
- Move kern_thr_*() prototype to <sys/syscallsubr.h> where all the other
  kern_*() prototypes live.

17 years agoRemove remaining references to pc_curtid missed in previous commit.
marcel [Thu, 7 Jun 2007 18:36:58 +0000 (18:36 +0000)]
Remove remaining references to pc_curtid missed in previous commit.

17 years agoEliminate pmap_install(), which was used to wrap pmap_switch() and
marcel [Thu, 7 Jun 2007 16:04:23 +0000 (16:04 +0000)]
Eliminate pmap_install(), which was used to wrap pmap_switch() and
grab sched_lock. This would serialize calls to pmap_switch from
cpu_switch(). With the introduction of thread_lock, this is not
possible anymore, because thread_lock is not a single lock. It
varies.  Secondly and most importantly, it's not needed at all. The
only requirement for pmap_switch() is that it's not preempted
while in the middle of updating the CPU and PCPU. In other words,
it's a critical region. No locking required.

17 years agoFix off-by-one error (introduced in r1.60) that had the effect of
bmah [Thu, 7 Jun 2007 15:04:30 +0000 (15:04 +0000)]
Fix off-by-one error (introduced in r1.60) that had the effect of
disallowing a read of exactly MAXPHYS bytes.

Reviewed by: des, rdivacky
MFC after: 1 week
Sponsored by: nCircle Network Security

17 years agoAdd a tunable hw.firewire.phydma_enable.
simokawa [Thu, 7 Jun 2007 13:20:48 +0000 (13:20 +0000)]
Add a tunable hw.firewire.phydma_enable.

This is enabled by default.  It should be disabled for
those who are uneasy with peeking/poking from FireWire.

Please note sbp(4) and dcons(4) over FireWire need
this feature.

17 years agoAdd heuristics for smooth reconnection.
simokawa [Thu, 7 Jun 2007 12:29:33 +0000 (12:29 +0000)]
Add heuristics for smooth reconnection.

17 years agoRemove compat ifdefs for version < 500014 from the rest of the USB files.
brueffer [Thu, 7 Jun 2007 09:29:37 +0000 (09:29 +0000)]
Remove compat ifdefs for version < 500014 from the rest of the USB files.

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

17 years ago"-b /boot/mbr" is redundant, /boot/mbr is the default boot code for fdisk(8).
pav [Thu, 7 Jun 2007 07:43:04 +0000 (07:43 +0000)]
"-b /boot/mbr" is redundant, /boot/mbr is the default boot code for fdisk(8).

Pointed out by: ru

17 years agoBump __FreeBSD_version for the change of the vn_open/VOP_OPEN file index
kib [Thu, 7 Jun 2007 07:32:15 +0000 (07:32 +0000)]
Bump __FreeBSD_version for the change of the vn_open/VOP_OPEN file index
argument to struct file *.

Requested by: Csaba Henk <csaba-ml at creo hu>

17 years agoCalculate the correct PCI BAR for the Timedia based serial cards. The
peter [Thu, 7 Jun 2007 06:28:48 +0000 (06:28 +0000)]
Calculate the correct PCI BAR for the Timedia based serial cards.  The
Linux equivalent gives BAR offsets relative to the implied base of 0x10.
Our PUC_CFG_GET_RID functions have to add the base offset themselves.

17 years agoFix spaces introduced in last commit
imp [Thu, 7 Jun 2007 06:05:53 +0000 (06:05 +0000)]
Fix spaces introduced in last commit

17 years agoNew Features:
davidch [Thu, 7 Jun 2007 02:23:56 +0000 (02:23 +0000)]
New Features:
- Added 2.5G support for BCM5708S.

MFC after: 4 weeks

17 years agoNew features:
davidch [Thu, 7 Jun 2007 02:21:38 +0000 (02:21 +0000)]
New features:
- Moved BCM5706S/5708S SerDes support to brgphy (since they are not technically
  TBI interfaces)
- Added 2.5G support for BCM5708S

Comments:
Since this driver is shared with bge I tested several available controllers
supported by bge and all worked as expected, however the list was not
exhaustive.  Need wider testing.

MFC after: 4 weeks

17 years agoFix compiling error.
davidxu [Thu, 7 Jun 2007 01:53:29 +0000 (01:53 +0000)]
Fix compiling error.

17 years agoThe maxsegsz of a dma tag created in de(4) is TULIP_DATA_PER_DESC bytes.
yongari [Thu, 7 Jun 2007 00:28:47 +0000 (00:28 +0000)]
The maxsegsz of a dma tag created in de(4) is TULIP_DATA_PER_DESC bytes.
In Rx path it allocates a new mbuf with m_getcl(9) so the length of
the mbuf is MCLBYTES which is greater than a segment size specified by
the dma tag. This segment size mismatch caused a voluntary panic.
Fix the panic by settting the mbuf length to TULIP_DATA_PER_DESC.

Reported by: Arne H Juul <arnej AT yahoo-inc DOT com>
Tested by: Arne H Juul <arnej AT yahoo-inc DOT com>

17 years ago - PCPU_ADD is no longer spelled with LAZY_ in the middle.
jeff [Wed, 6 Jun 2007 23:23:47 +0000 (23:23 +0000)]
 - PCPU_ADD is no longer spelled with LAZY_ in the middle.

Submitted by: attilio

17 years agoTeach the bridge wrapper how to handle the filter+ithread case.
piso [Wed, 6 Jun 2007 22:19:23 +0000 (22:19 +0000)]
Teach the bridge wrapper how to handle the filter+ithread case.

Reviewed by: marius

17 years ago-axe p_ih from struct puc_port cause it was useless
piso [Wed, 6 Jun 2007 22:17:01 +0000 (22:17 +0000)]
-axe p_ih from struct puc_port cause it was useless
-correctly check for filter only handler

Reviewed by: marcel
Tested by: marcel

17 years agoCorrectly print SEQ and IRS in the corresponding log message in
andre [Wed, 6 Jun 2007 22:10:12 +0000 (22:10 +0000)]
Correctly print SEQ and IRS in the corresponding log message in
syncache_expand().

17 years ago- Revert previous revision, it was incorrect
pav [Wed, 6 Jun 2007 21:28:50 +0000 (21:28 +0000)]
- Revert previous revision, it was incorrect
- Add an example using fdisk instead

Pointed out by: ru
Submitted by: Warren Block <wblock@wonkity.com>
MFC after: 3 days

17 years agoSpecify which Dlink chip is supported (D-Link DUB-E100 ver A) because
remko [Wed, 6 Jun 2007 19:27:10 +0000 (19:27 +0000)]
Specify which Dlink chip is supported (D-Link DUB-E100 ver A) because
others (D-Link DUB-E100 ver B1) are not.

PR: docs/109664
Submitted by: Martin Nilsson<martin at mullet dot se>
Patch by: asmodai
MFC after: 1 week

17 years agoDocument getaddrinfo(3)'s AI_ADDRCONFIG
remko [Wed, 6 Jun 2007 19:24:02 +0000 (19:24 +0000)]
Document getaddrinfo(3)'s AI_ADDRCONFIG

PR: docs/78357
Submitted by: Matthias Andree <matthias dot andree at gmx dot de>
Patch by: asmodai
MFC after: 1 week

17 years agoFix a typo to make this file compile
gallatin [Wed, 6 Jun 2007 18:33:38 +0000 (18:33 +0000)]
Fix a typo to make this file compile

17 years agoupdate copyrights to 2007 and convert to be 2-clause bsd-only
sam [Wed, 6 Jun 2007 15:49:16 +0000 (15:49 +0000)]
update copyrights to 2007 and convert to be 2-clause bsd-only

17 years agoMFp4: MPSAFE firewire stack.
simokawa [Wed, 6 Jun 2007 14:31:36 +0000 (14:31 +0000)]
MFp4: MPSAFE firewire stack.
- lock its own locks and drop Giant.
- create its own taskqueue thread.
- split interrupt routine
- use interrupt filter as a fast interrupt.
- run watchdog timer in taskqueue so that it should be
   serialized with the bottom half.
- add extra sanity check for transaction labels.
   disable ad-hoc workaround for unknown tlabels.
- add sleep/wakeup synchronization primitives
- don't reset OHCI in fwohci_stop()

17 years agoDo not leak lock in the case of EEXIST error.
glebius [Wed, 6 Jun 2007 14:21:49 +0000 (14:21 +0000)]
Do not leak lock in the case of EEXIST error.

PR: kern/92776
Submitted by: Ed Schouten <Ed.Schouten tunix.nl>

17 years agoMFi386: revision 1.657
nyan [Wed, 6 Jun 2007 13:04:15 +0000 (13:04 +0000)]
MFi386: revision 1.657

  Backout experimental adaptive-spin umtx code.

17 years ago- Sort the headers per style(9)
rafan [Wed, 6 Jun 2007 11:14:30 +0000 (11:14 +0000)]
- Sort the headers per style(9)

Reminded by: yar
Approved by: delphij (mentor, implicit)

17 years agoCorrect typo.
ceri [Wed, 6 Jun 2007 11:12:56 +0000 (11:12 +0000)]
Correct typo.

17 years agoUpdate rum driver that needs wlan(4).
kevlo [Wed, 6 Jun 2007 07:58:03 +0000 (07:58 +0000)]
Update rum driver that needs wlan(4).

17 years agoBackout experimental adaptive-spin umtx code.
davidxu [Wed, 6 Jun 2007 07:35:08 +0000 (07:35 +0000)]
Backout experimental adaptive-spin umtx code.

17 years agoAdd support IC Plus IP101 10/100 PHY that is found on nVidia network
yongari [Wed, 6 Jun 2007 07:07:23 +0000 (07:07 +0000)]
Add support IC Plus IP101 10/100 PHY that is found on nVidia network
adapters.

Submitted by: Shigeaki Tagashira < shigeaki AT se DOT hiroshima-u DOT ac DOT jp >

17 years agoAdd IC Plus IP101 PHY
yongari [Wed, 6 Jun 2007 07:05:02 +0000 (07:05 +0000)]
Add IC Plus IP101 PHY

17 years agoAdd support Vitesse VSC8601 PHY that is found on nVidia network
yongari [Wed, 6 Jun 2007 06:55:49 +0000 (06:55 +0000)]
Add support Vitesse VSC8601 PHY that is found on nVidia network
adapters.

Submitted by: Shigeaki Tagashira < shigeaki AT se DOT hiroshima-u DOT ac DOT jp >
Tested by: Yuri Pankov < yuri.pankov AT gmail DOT com>,
Rainer Hurling <rhurlin AT gwdg DOT de >

17 years agoAdd OUI for Vitesse Semiconductor.
yongari [Wed, 6 Jun 2007 06:53:40 +0000 (06:53 +0000)]
Add OUI for Vitesse Semiconductor.
Add Vitesse VSC8601 PHY.

17 years agoFix the compile. Band-aid until it is worked out how to use the context
grehan [Wed, 6 Jun 2007 06:01:56 +0000 (06:01 +0000)]
Fix the compile. Band-aid until it is worked out how to use the context
switch api on ppc.

17 years agoPrefix unknown (i.e. un-aliased) partition types with '!'. This is
marcel [Wed, 6 Jun 2007 05:06:14 +0000 (05:06 +0000)]
Prefix unknown (i.e. un-aliased) partition types with '!'. This is
how they had to be given with ctlreq.

17 years agoCall sbuf_finish() before sbuf_data() and sbuf_len().
marcel [Wed, 6 Jun 2007 05:01:41 +0000 (05:01 +0000)]
Call sbuf_finish() before sbuf_data() and sbuf_len().

17 years agocopyright updates:
sam [Wed, 6 Jun 2007 04:56:04 +0000 (04:56 +0000)]
copyright updates:
o update to include 2007
o switch back to a 2-clause bsd-only license

Reviewed by: onoe

17 years agoInclude <sys/sched.h> for sched_throw().
marcel [Wed, 6 Jun 2007 04:44:19 +0000 (04:44 +0000)]
Include <sys/sched.h> for sched_throw().

17 years ago - Placing the 'volatile' on the right side of the * in the td_lock
jeff [Wed, 6 Jun 2007 03:40:47 +0000 (03:40 +0000)]
 - Placing the 'volatile' on the right side of the * in the td_lock
   declaration removes the need for __DEVOLATILE().

Pointed out by: tegge

17 years agoMention 'k' option of gunzip.
delphij [Wed, 6 Jun 2007 02:51:54 +0000 (02:51 +0000)]
Mention 'k' option of gunzip.

Pointed out by: ighighi gmail.com

17 years ago- Fix crash with ncurses on FreeBSD/amd64 CURRENT by including missing
rafan [Wed, 6 Jun 2007 02:23:03 +0000 (02:23 +0000)]
- Fix crash with ncurses on FreeBSD/amd64 CURRENT by including missing
  header <term.h>, which is required by both tputs() and tgoto().

  Without this header, nvi crashes with the call to tputs() at line 114.
  This is because the first argument of tputs() is a pointer, and the
  returned pointer from tgoto() is casted to an integer by compiler.

  For some reasons, this only crashes on amd64+CURRENT. On 6.x system,
  this does not happen.

PR: 101248
Reported by: yar, infofarmer, Roland Smith
Tested by: infofarmer, delphij, rafan
Approved by: delphij (mentor)

17 years ago- Fixes a case where doing a sysctl would leave locks held
rrs [Wed, 6 Jun 2007 00:40:41 +0000 (00:40 +0000)]
- Fixes a case where doing a sysctl would leave locks held
  when coping out association data.
- Fixes a small bug that prevented the SCTP_UNORDERED indication
  from going up to the app on a recv in the sinfo_flags field.

17 years agoRemove the special atomic.h case for arm, and allow it to use
dougb [Tue, 5 Jun 2007 22:17:16 +0000 (22:17 +0000)]
Remove the special atomic.h case for arm, and allow it to use
the platform specific file that imp provided.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r170349,
dougb [Tue, 5 Jun 2007 22:15:38 +0000 (22:15 +0000)]
This commit was generated by cvs2svn to compensate for changes in r170349,
which included commits to RCS files with non-trunk default branches.

17 years agoAdd a custom atomic.h file which implements the C versions of the
dougb [Tue, 5 Jun 2007 22:15:38 +0000 (22:15 +0000)]
Add a custom atomic.h file which implements the C versions of the
code we already have assembly versions of.

Written by: imp

17 years agoAdd more IDs for the uftdi driver. Slight tweaks to patch by me.
imp [Tue, 5 Jun 2007 21:06:17 +0000 (21:06 +0000)]
Add more IDs for the uftdi driver.  Slight tweaks to patch by me.

Submitted by:  Thorsten Trampisch
PR: 113384

17 years agobump date for 1.20
imp [Tue, 5 Jun 2007 20:53:18 +0000 (20:53 +0000)]
bump date for 1.20

17 years agoAdd note about condvars not working with spin locks.
imp [Tue, 5 Jun 2007 20:45:10 +0000 (20:45 +0000)]
Add note about condvars not working with spin locks.

17 years ago- Do triple reads on reset register to detect read register bug. 2 reads
ariff [Tue, 5 Jun 2007 20:30:16 +0000 (20:30 +0000)]
- Do triple reads on reset register to detect read register bug. 2 reads
  seems not enough to verify its consistencies.
- Define AC97_MIXER_SIZE as SOUND_MIXER_NRDEVICES (25), since we
  don't need more than that. Stop doing wild and random guess about
  its size since we're stricly bound to it.

17 years agoFix (enable) phone out for laptops with ALC655, specifically
ariff [Tue, 5 Jun 2007 20:12:40 +0000 (20:12 +0000)]
Fix (enable) phone out for laptops with ALC655, specifically
for Amilo Pro V2055.

PR: kern/113101
Submitted by: konrad@egipt-medytacje.pl
MFC after: 3 days

17 years agoMove a warning under bootverbose as no machines that trigger it have ended
jhb [Tue, 5 Jun 2007 18:57:48 +0000 (18:57 +0000)]
Move a warning under bootverbose as no machines that trigger it have ended
up being broken.

17 years agoFix a problem with not-preemptive kernels caming from mis-merging of
attilio [Tue, 5 Jun 2007 18:57:09 +0000 (18:57 +0000)]
Fix a problem with not-preemptive kernels caming from mis-merging of
existing code with the new thread_lock patch.
This also cleans up a bit unlock operation for mutexes.

Approved by: jhb, jeff(mentor)

17 years agoMFp4: When querying the operating condition of SD cards (using the
imp [Tue, 5 Jun 2007 17:04:44 +0000 (17:04 +0000)]
MFp4: When querying the operating condition of SD cards (using the
application specific SEND_OP_COND (CMD55 + ACMD41), go ahead and allow
100 tries.  This gives a timeout of a second rather than the ~100ms
the old style produces.

I've had one old 16MB SD card which needs the extra time.  I've now
had reports from the field that other cards need this too.

Originally done at BSDcan 2007 while waiting to give my embedding
madness minitalk.

17 years agoUpdate release notes: awk is updated to 1 May 2007 release.
rafan [Tue, 5 Jun 2007 15:38:54 +0000 (15:38 +0000)]
Update release notes: awk is updated to 1 May 2007 release.

Approved by: delphij (mentor)

17 years ago- Update for 2007/05/01 import
rafan [Tue, 5 Jun 2007 15:35:05 +0000 (15:35 +0000)]
- Update for 2007/05/01 import

Approved by: delphij (mentor)
Nodded by: ru
Tested by: make universe

17 years ago- Update for 2007/05/01 import.
rafan [Tue, 5 Jun 2007 15:34:40 +0000 (15:34 +0000)]
- Update for 2007/05/01 import.

Approved by: delphij (mentor)
Nodded by: ru

17 years agoThis commit was generated by cvs2svn to compensate for changes in r170331,
rafan [Tue, 5 Jun 2007 15:33:51 +0000 (15:33 +0000)]
This commit was generated by cvs2svn to compensate for changes in r170331,
which included commits to RCS files with non-trunk default branches.

17 years agoVendor import of bwk's 01-May-2007 release.
rafan [Tue, 5 Jun 2007 15:33:51 +0000 (15:33 +0000)]
Vendor import of bwk's 01-May-2007 release.

Approved by: delphij (mentor)
Nodded by: ru
Tested by: make universe

17 years agoUse pmap_change_attr() to setup a write combine attribute for our
gallatin [Tue, 5 Jun 2007 15:02:14 +0000 (15:02 +0000)]
Use pmap_change_attr() to setup a write combine attribute for our
device memory, rather than relying on the less reliable MTRR method
used by mem_range_attr_set().

Glanced at by: jhb

17 years agoUpdated release notes: less was updated to v403.
delphij [Tue, 5 Jun 2007 14:55:15 +0000 (14:55 +0000)]
Updated release notes: less was updated to v403.