]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoAdd infrastructure for Cologne Chip HFC-S PCI driver.
hm [Thu, 26 Oct 2000 07:00:05 +0000 (07:00 +0000)]
Add infrastructure for Cologne Chip HFC-S PCI driver.

23 years agoAdd infrastructure for Cologne Chip HFC-S PCI driver.
hm [Thu, 26 Oct 2000 06:58:43 +0000 (06:58 +0000)]
Add infrastructure for Cologne Chip HFC-S PCI driver.
Increment isdn4bsd step.

23 years agoMake the software HDLC decoder work again.
hm [Thu, 26 Oct 2000 06:31:18 +0000 (06:31 +0000)]
Make the software HDLC decoder work again.
Submitted by: Hans Petter Selasky <hselasky@c2i.net>

23 years agoAlmost a year exactly from when I first imported mergemaster at freebsdcon,
billf [Thu, 26 Oct 2000 05:36:35 +0000 (05:36 +0000)]
Almost a year exactly from when I first imported mergemaster at freebsdcon,
remove myself as co-maintainer and let Doug handle all the fun.

23 years agohandle AFMT_QUERY right
cg [Thu, 26 Oct 2000 01:34:54 +0000 (01:34 +0000)]
handle AFMT_QUERY right

23 years agoClose PR22152 and PR19511 -- correct the naming of a variable
luigi [Thu, 26 Oct 2000 00:16:12 +0000 (00:16 +0000)]
Close PR22152 and PR19511 -- correct the naming of a variable

23 years agoSet the DC_TX_INTR_ALWAYS and DC_TX_STORENFWD flags for the Davicom
wpaul [Wed, 25 Oct 2000 23:46:31 +0000 (23:46 +0000)]
Set the DC_TX_INTR_ALWAYS and DC_TX_STORENFWD flags for the Davicom
DM9100/DM9102 chips. Do not set DC_TX_ONE. The DC_TX_USE_TX_INTR flag
causes dc_encap() to set the 'interrupt on TX completion' bit only
once every 64 packets. This is an attempt to reduce the number
of interrupts generated by the chip. You're supposed to get a 'no more
TX buffers left' interrupt once you hit the last packet whether you
ask for one or not, however it seems the Davicom chip doesn't generate
this interrupt, or at least it doesn't generate it under the same
circumstances. The result is that if you transmit n packets, where
n is less than 64, and then wait 5 seconds, you'll get a watchdog
timeout whether you want one or not. The DC_TX_INTR_ALWAYS causes
dc_encap() to request an interrupt for every frame.

I'm still waiting on confirmation from a couple of users to see if this
fixes their problems with the Davicom DM9102 before I merge this into
-stable, but this fixed the problem for me in my own testing so I'm
willing to make the change to -current right away.

23 years agoAdd a missing SK_UNLOCK() to sk_attach_xmac().
wpaul [Wed, 25 Oct 2000 23:36:45 +0000 (23:36 +0000)]
Add a missing SK_UNLOCK() to sk_attach_xmac().

23 years agoMove the call to extend_heap() from main to start so that if our BSS
dfr [Wed, 25 Oct 2000 23:36:01 +0000 (23:36 +0000)]
Move the call to extend_heap() from main to start so that if our BSS
expands beyond the limit we will extend the address space before trying
to zero the BSS. This should give us plenty of headroom for modest
expansion of the loader.

23 years agoDon't build start.S as part of libalpha.a - its built specially.
dfr [Wed, 25 Oct 2000 23:30:04 +0000 (23:30 +0000)]
Don't build start.S as part of libalpha.a - its built specially.

23 years agoDon't indirect through a null pointer if we run out of kernel filenames.
dfr [Wed, 25 Oct 2000 23:28:19 +0000 (23:28 +0000)]
Don't indirect through a null pointer if we run out of kernel filenames.

23 years agoMake a few functions inline to save space.
dfr [Wed, 25 Oct 2000 23:24:43 +0000 (23:24 +0000)]
Make a few functions inline to save space.

23 years agoBring cpuinfo closer to what it looks like in Linux 2.2.
des [Wed, 25 Oct 2000 22:38:23 +0000 (22:38 +0000)]
Bring cpuinfo closer to what it looks like in Linux 2.2.

Submitted by: R Bradford Jones <brad@kazrak.com>

23 years agoAdd /proc/<pid>/status and /proc/<pid>/stat (the latter being mostly
des [Wed, 25 Oct 2000 22:12:59 +0000 (22:12 +0000)]
Add /proc/<pid>/status and /proc/<pid>/stat (the latter being mostly
zeroes for the time being).

Prompted by: Nathan Boeger <nathan@khmere.com>

23 years ago- Add atomic_cmpset_{acq_,rel_,}_long
jhb [Wed, 25 Oct 2000 21:56:16 +0000 (21:56 +0000)]
- Add atomic_cmpset_{acq_,rel_,}_long
- Add in atomic operations for 8-bit, 16-bit, and 32-bit integers

23 years agoThis driver doesn't have a software interrupt handler, so don't attempt to
jhb [Wed, 25 Oct 2000 21:29:36 +0000 (21:29 +0000)]
This driver doesn't have a software interrupt handler, so don't attempt to
schedule a non-existant handler to run.

23 years agoCatch up to the new software interrupt code.
jhb [Wed, 25 Oct 2000 21:24:08 +0000 (21:24 +0000)]
Catch up to the new software interrupt code.

Noticed by: phk

23 years ago- Catch up to new software interrupt code.
jhb [Wed, 25 Oct 2000 21:13:42 +0000 (21:13 +0000)]
- Catch up to new software interrupt code.
- Add a missing curly brace.

Noticed by: phk

23 years agoNote the MFC of ATA support for new chipsets and tagged queueing,
bmah [Wed, 25 Oct 2000 21:04:09 +0000 (21:04 +0000)]
Note the MFC of ATA support for new chipsets and tagged queueing,
ATA raid support (+MFC), snc(4) driver added (+MFC), ich(4) driver
added (+MFC), IFS, libisc now added, fix to netstat(1) "in use"
percentage (+MFC).

23 years ago- Remove unneeded machine/ipl.h #include.
jhb [Wed, 25 Oct 2000 20:38:08 +0000 (20:38 +0000)]
- Remove unneeded machine/ipl.h #include.
- Change the softintr() macro to do nothing on FreeBSD.  Previously,
  this macro would set a bit in spending and schedule the softinterrupt
  thread to run.  However, the bs driver never actually registers a
  a software interrupt handler, so all this work achieved nothing.  From
  the code it is not clear what exactly the softintr() macro is actually
  supposed to be doing.  It looks like it is supposed to be possibly
  running the hardware interrupt handler maybe?  This handler is only
  present in the #ifdef __NetBSD__ code however.  I have no idea how this
  driver handles interrupts at all, but at least it compiles now.

23 years agoCatch up to the new swi code.
jhb [Wed, 25 Oct 2000 20:21:42 +0000 (20:21 +0000)]
Catch up to the new swi code.

Noticed by: phk

23 years agoMerge from vendor branch.
wollman [Wed, 25 Oct 2000 19:36:49 +0000 (19:36 +0000)]
Merge from vendor branch.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r67575,
wollman [Wed, 25 Oct 2000 19:33:50 +0000 (19:33 +0000)]
This commit was generated by cvs2svn to compensate for changes in r67575,
which included commits to RCS files with non-trunk default branches.

23 years ago../changes
wollman [Wed, 25 Oct 2000 19:33:50 +0000 (19:33 +0000)]
../changes

23 years agoUtilize the "-o" (origin) option recently added to pkg_create(1)
bmah [Wed, 25 Oct 2000 17:10:20 +0000 (17:10 +0000)]
Utilize the "-o" (origin) option recently added to pkg_create(1)
and pkg_info(1), for more accurate determination of the
"up-to-date-ness" of installed packages.

23 years agoDo not include uscanner in the generated kernel config. Also, remove
n_hibma [Wed, 25 Oct 2000 15:21:19 +0000 (15:21 +0000)]
Do not include uscanner in the generated kernel config. Also, remove
ugen, uhid, ulpt from small as well.

Suggested by: obrien

23 years agoadd Xircom CreditCard adapters and workalikes(xe driver)
sanpei [Wed, 25 Oct 2000 15:20:10 +0000 (15:20 +0000)]
add Xircom CreditCard adapters and workalikes(xe driver)

23 years agoSlowly add more functionality as I learn more about newbus etc.
julian [Wed, 25 Oct 2000 15:08:11 +0000 (15:08 +0000)]
Slowly add more functionality as I learn more about newbus etc.

23 years agoMake pthread_kill() know about temporary signal handlers installed
deischen [Wed, 25 Oct 2000 11:46:07 +0000 (11:46 +0000)]
Make pthread_kill() know about temporary signal handlers installed
by sigwait().  This prevents a signal from being sent to the process
when there are no application installed signal handlers.

Correct a typo in sigwait (foo -> foo[i]).

23 years agoTrim right column messages to display correctly in 80-column terminal.
kuriyama [Wed, 25 Oct 2000 11:36:31 +0000 (11:36 +0000)]
Trim right column messages to display correctly in 80-column terminal.

Reviewed by: jkh

23 years agoWe now keep the ip_id field in network byte order all the
ru [Wed, 25 Oct 2000 10:56:41 +0000 (10:56 +0000)]
We now keep the ip_id field in network byte order all the
time, so there is no need to make the distinction between
ip_output() and ip_input() cases.

Reviewed by: silence on freebsd-net

23 years agoFast interrupts have no associated process, therefore do not try
ps [Wed, 25 Oct 2000 10:40:20 +0000 (10:40 +0000)]
Fast interrupts have no associated process, therefore do not try
and schedule it.  This fixes booting machines with broken MP tables.

23 years agoThe USB scanner driver. To be used together with SANE.
n_hibma [Wed, 25 Oct 2000 10:34:38 +0000 (10:34 +0000)]
The USB scanner driver. To be used together with SANE.

23 years agoRegen.
n_hibma [Wed, 25 Oct 2000 10:14:44 +0000 (10:14 +0000)]
Regen.

23 years agoSync with NetBSD
n_hibma [Wed, 25 Oct 2000 10:13:47 +0000 (10:13 +0000)]
Sync with NetBSD

23 years agoAdd the ATA_ENABLE_TAGS option
sos [Wed, 25 Oct 2000 08:25:37 +0000 (08:25 +0000)]
Add the ATA_ENABLE_TAGS option

23 years agoMajor update to the 'twe' driver.
msmith [Wed, 25 Oct 2000 06:59:06 +0000 (06:59 +0000)]
Major update to the 'twe' driver.

 - Layout reorganisation to enhance portability.  The driver now has
   a relatively MI 'core' and a FreeBSD-specific layer over the top.
   Since the NetBSD people have already done their own port, this is
   largely just to help me with the BSD/OS port.

 - Request ID allocation changed to improve performance (I'd been
   considering switching to this approach after having failed to come
   up with a better way to dynamically allocate request IDs, and seeing
   Andy Doran use it in the NetBSD port of the driver convinced me
   that I was wasting my time doing it any other way).  Now we just
   allocate all the requests up front.

 - Maximum request count bumped back to 255 after characterisation
   of a firmware issue (off-by-one causing it to crash with 256
   outstanding commands).

 - Control interface implemented.  This allows 3ware's '3dm' utility to
   talk to the controller.  3dm will be available from 3ware shortly.

 - Controller soft-reset feature added; if the controller signals a
   firmware or protocol error, the controller will be reset and all
   outstanding commands will be retried.

23 years ago- Overhaul the software interrupt code to use interrupt threads for each
jhb [Wed, 25 Oct 2000 05:19:40 +0000 (05:19 +0000)]
- Overhaul the software interrupt code to use interrupt threads for each
  type of software interrupt.  Roughly, what used to be a bit in spending
  now maps to a swi thread.  Each thread can have multiple handlers, just
  like a hardware interrupt thread.
- Instead of using a bitmask of pending interrupts, we schedule the specific
  software interrupt thread to run, so spending, NSWI, and the shandlers
  array are no longer needed.  We can now have an arbitrary number of
  software interrupt threads.  When you register a software interrupt
  thread via sinthand_add(), you get back a struct intrhand that you pass
  to sched_swi() when you wish to schedule your swi thread to run.
- Convert the name of 'struct intrec' to 'struct intrhand' as it is a bit
  more intuitive.  Also, prefix all the members of struct intrhand with
  'ih_'.
- Make swi_net() a MI function since there is now no point in it being
  MD.

Submitted by: cp

23 years agoGet rid of ridiculous ISP_PVS macro. Instead, just set an
mjacob [Wed, 25 Oct 2000 04:42:46 +0000 (04:42 +0000)]
Get rid of ridiculous ISP_PVS macro. Instead, just set an
ISP_SMPLOCK define based on the previous 5.4 major/minor release
define of PVS- because this allows us to turn it off easier.

23 years agoWhoops! Forgot to commit this when I committed the other (turnin on locks)
mjacob [Wed, 25 Oct 2000 04:40:49 +0000 (04:40 +0000)]
Whoops! Forgot to commit this when I committed the other (turnin on locks)
change. Sorry about that.

23 years agoQuite some warnings.
jhb [Wed, 25 Oct 2000 04:37:54 +0000 (04:37 +0000)]
Quite some warnings.

23 years agoFail the 'buildkernel' target if not even one of the specified kernel configuration
jkoshy [Wed, 25 Oct 2000 04:31:32 +0000 (04:31 +0000)]
Fail the 'buildkernel' target if not even one of the specified kernel configuration
files was found.

Reviewed by: obrien

23 years agofix bogus cast in osf1_getrlimit/osf1_setrlimit
gallatin [Wed, 25 Oct 2000 00:37:34 +0000 (00:37 +0000)]
fix bogus cast in osf1_getrlimit/osf1_setrlimit

23 years agoImplement atomic_{set,clear,add,subtract}_{acq_,rel_,}_ptr()
jhb [Wed, 25 Oct 2000 00:16:38 +0000 (00:16 +0000)]
Implement atomic_{set,clear,add,subtract}_{acq_,rel_,}_ptr()

23 years agoOnly use 1 set of memory barrier operations with the atomic_*_{acq,rel}_ptr
jhb [Wed, 25 Oct 2000 00:15:21 +0000 (00:15 +0000)]
Only use 1 set of memory barrier operations with the atomic_*_{acq,rel}_ptr
functions.

23 years agoteach the osf1_getsysinfo() function about a few more fields
gallatin [Wed, 25 Oct 2000 00:14:11 +0000 (00:14 +0000)]
teach the osf1_getsysinfo() function about a few more fields

submitted by: Jim.Pirzyk@disney.com
PR: alpha/22263

23 years ago- Forward declare struct mtx instead of #include'ing sys/mutex.h
jhb [Wed, 25 Oct 2000 00:07:09 +0000 (00:07 +0000)]
- Forward declare struct mtx instead of #include'ing sys/mutex.h

23 years ago- Catch a machine/mutex.h -> sys/mutex.h I somehow missed.
jhb [Wed, 25 Oct 2000 00:04:16 +0000 (00:04 +0000)]
- Catch a machine/mutex.h -> sys/mutex.h I somehow missed.
- Close a small race condition.  The sched_lock mutex protects
  p->p_stat as well as the run queues.  Another CPU could change p_stat
  of the process while we are waiting for the lock, and we would end up
  scheduling a process that isn't runnable.

23 years ago- Make the eventhandler_mutex mutex a private variable in
jhb [Wed, 25 Oct 2000 00:01:39 +0000 (00:01 +0000)]
- Make the eventhandler_mutex mutex a private variable in
  subr_eventhandler.c
- Move the extra #include's in sys/eventhandler.h to be protected by
  the #ifndef SYS_EVENTHANDLER/#endif

23 years agoEliminate some bitrot (nonexisting member variable names).
tegge [Tue, 24 Oct 2000 23:33:01 +0000 (23:33 +0000)]
Eliminate some bitrot (nonexisting member variable names).
Don't use curproc when a proc pointer is available.

23 years agoStyle fixes.
tegge [Tue, 24 Oct 2000 22:40:18 +0000 (22:40 +0000)]
Style fixes.

23 years agoConvert the USB ethernet drivers to use mutexes. Also convert
wpaul [Tue, 24 Oct 2000 22:38:54 +0000 (22:38 +0000)]
Convert the USB ethernet drivers to use mutexes. Also convert
usb_ethersubr.c. This module maintains two queues for packets which
are each protected with one mutex. These are all the changes I can
do for now. Removing the USBD_NO_TSLEEP flag doesn't work yet: when
I tried it, the system would usually freeze up after a NIC had been
operating for a while. The usb_ethersubr module itself ought to
go away; this is the next thing I need to test.

23 years agoMake RPC timeout message more readable.
tegge [Tue, 24 Oct 2000 22:37:55 +0000 (22:37 +0000)]
Make RPC timeout message more readable.
Supply proc pointer to sosend.

23 years agoAdd a relnote about large swap areas not panicking the system.
bmah [Tue, 24 Oct 2000 22:10:23 +0000 (22:10 +0000)]
Add a relnote about large swap areas not panicking the system.

Submitted by: grog

23 years agoBuild the ISC library as libisc. This library comes as part of the
archie [Tue, 24 Oct 2000 20:10:25 +0000 (20:10 +0000)]
Build the ISC library as libisc. This library comes as part of the
bind distribution, but until now was not being built as a separate
entity. For documentation, see these man pages:

assertions(3), eventlib(3), heap(3), logging(3), memcluster(3), tree(3).

Reviewed by: jdp

23 years ago* Various fixes to breakage introduced by the atomic and mutex reorgs.
dfr [Tue, 24 Oct 2000 19:54:38 +0000 (19:54 +0000)]
* Various fixes to breakage introduced by the atomic and mutex reorgs.
* Fixes to the signal delivery code. Not quite right yet.

I would have preferred to wait until I have signal delivery actually
working but the current kernel in CVS doesn't build.

23 years agoWe were trying to dike out `apm0', but of course devices do not have
obrien [Tue, 24 Oct 2000 19:08:40 +0000 (19:08 +0000)]
We were trying to dike out `apm0', but of course devices do not have
unit numbers in GENERIC any more.  So `apm' was acutally being compiled
in.

23 years agoRemove /boot/boot[12] from the MFS installation floppies.
obrien [Tue, 24 Oct 2000 19:05:39 +0000 (19:05 +0000)]
Remove /boot/boot[12] from the MFS installation floppies.
This saves us 8KB.

Submitted by: Makoto MATSUSHITA <matusita@jp.FreeBSD.org>
Reviewed by: jhb

23 years agoDike out more for the i386 release build: NCR SCSI card; anchient WDC
obrien [Tue, 24 Oct 2000 18:48:16 +0000 (18:48 +0000)]
Dike out more for the i386 release build:  NCR SCSI card; anchient WDC
SCSI card (should it ever find its way into GENERIC); LPT (we don't need
to print during install time); the parallel 'geek' port; generic USB
driver (thus some attached USB devices will not be detected and thus the
user may wonder what is going on, we couldn't do anything with the device
if only ugen attached to it anyway and we are getting very, very low on
available space; USB "Human Interface Devices" as we don't do anything
with them during installation; and USB printers (same argument as LPT).

23 years agoFix two typos in comments.
dwmalone [Tue, 24 Oct 2000 18:47:57 +0000 (18:47 +0000)]
Fix two typos in comments.

PR: 22268
Submitted by: Daniel S. Lewart <d-lewart@uiuc.edu>

23 years agoDocument that pci_device_override requires a recent SRM rev on Miata.
wilko [Tue, 24 Oct 2000 18:45:42 +0000 (18:45 +0000)]
Document that pci_device_override requires a recent SRM rev on Miata.

Submitted by: Ralph Schreyer <schreyer@th.physik.uni-bonn.de>

23 years agoChange archie's and my email addresses from Whistle.com to FreeBSD.org
julian [Tue, 24 Oct 2000 18:01:45 +0000 (18:01 +0000)]
Change archie's and my email addresses from Whistle.com to FreeBSD.org

23 years agoCatch up on some release notes: write combining for crash dumps,
bmah [Tue, 24 Oct 2000 17:59:28 +0000 (17:59 +0000)]
Catch up on some release notes:  write combining for crash dumps,
bktr(4) update to 2.17, MFC new make(1) variable modifiers, note
the addition of style.perl.7.

23 years agoSince neither archie nor I work at Whistle any more, change our email
julian [Tue, 24 Oct 2000 17:32:45 +0000 (17:32 +0000)]
Since neither archie nor I work at Whistle any more, change our email
addresses to be the more usefu @freebsd.org ones
so we can keep getting bug-reports.
- man pages to follow..

23 years agoFirst effort at bringing these up-to-date.
julian [Tue, 24 Oct 2000 16:45:58 +0000 (16:45 +0000)]
First effort at bringing these up-to-date.
This creates a skeleton ISA device driver.
I don't pretend that it's fully correct or even opitimal
but it at least creates (and compiles) a 'clean' ISA driver.

Hopefully PCI/PCCARD/etc. support will be added when I understand it.
Unlike the old version this just creates a module. The old one tried to
create a new kernel with the driver to be tested.

23 years agoFix up the build for the STARTTLS version of sendmail (again). This method
gshapiro [Tue, 24 Oct 2000 16:04:56 +0000 (16:04 +0000)]
Fix up the build for the STARTTLS version of sendmail (again).  This method
mimics that of tcpdump in that for normal builds, sendmail will only be
built once.  For 'make release', it is built once for the bin dist and
once for the crypto dist.  This method also removes the need for two separate
Makefiles (which could become out of sync).

Suggested by: bde
Assisted by: kris

23 years agoOops better ad the ar device to the all target, people might want to
sos [Tue, 24 Oct 2000 15:40:20 +0000 (15:40 +0000)]
Oops better ad the ar device to the all target, people might want to
install on these...

23 years agoDescribe %c better
ache [Tue, 24 Oct 2000 15:37:48 +0000 (15:37 +0000)]
Describe %c better

Submitted by: ru (with modifications)

23 years agoAdd the ATA RAID (ar) device
sos [Tue, 24 Oct 2000 14:31:24 +0000 (14:31 +0000)]
Add the ATA RAID (ar) device

23 years agoAdd $FreeBSD$
phk [Tue, 24 Oct 2000 13:54:31 +0000 (13:54 +0000)]
Add $FreeBSD$

Rename local offsetof() macro to boffsetof() to avoid clashing with
the offsetof() from <stddef.h>

23 years agoCorrect the description of a few options to reflect that commands and/or
mpp [Tue, 24 Oct 2000 13:54:17 +0000 (13:54 +0000)]
Correct the description of a few options to reflect that commands and/or
the results that they actualy use/generate.

PR: docs/22267
Submitted by: SUZUKI Koichi <koich@cac.co.jp>

23 years agoFix the problem with DMA mode not working on Aladdin chips.
sos [Tue, 24 Oct 2000 13:50:22 +0000 (13:50 +0000)]
Fix the problem with DMA mode not working on Aladdin chips.
Amasing in how many ways Acer has screwed up that chip.

23 years agoAdd MASTER_SITE_MOZILLA and MASTER_SITE_XEMACS.
knu [Tue, 24 Oct 2000 12:37:52 +0000 (12:37 +0000)]
Add MASTER_SITE_MOZILLA and MASTER_SITE_XEMACS.

23 years agoAdd "NO_CXX" knob to match src/gnu/usr.bin/cc/Makefile.
obrien [Tue, 24 Oct 2000 11:25:28 +0000 (11:25 +0000)]
Add "NO_CXX" knob to match src/gnu/usr.bin/cc/Makefile.

23 years agoAdjust comments
obrien [Tue, 24 Oct 2000 10:49:56 +0000 (10:49 +0000)]
Adjust comments
Submitted by: bde

Add ISO C99's long long type limits.
Reviewed by: bde

23 years ago* Update comments
obrien [Tue, 24 Oct 2000 10:48:50 +0000 (10:48 +0000)]
* Update comments
* convert decimal constants to hex
Submitted by: bde

* Add ISO-C99 long long limits

23 years agoProblem to avoid processes getting stuck in "vmopar". From Ian's
dwmalone [Tue, 24 Oct 2000 10:13:36 +0000 (10:13 +0000)]
Problem to avoid processes getting stuck in "vmopar". From Ian's
mail:

The problem seems to originate with NFS's postop_attr
information that is returned with a read or write RPC.
Within a vm_fault context, the code cannot deal with
vnode_pager_setsize() shrinking a vnode.

The workaround in the patch below stops the nfsm_postop_attr()
macro from ever shrinking a vnode. If the new size in the
postop_attr information is smaller, then it just sets the
nfsnode n_attrstamp to 0 to stop the wrong size getting
used in the future. This change only affects postop_attr
attributes; the nfsm_loadattr() macro works as normal.

The change is implemented by adding a new argument to
nfs_loadattrcache() called 'dontshrink'. When this is
non-zero, nfs_loadattrcache() will never reduce the
vnode/nfsnode size; instead it zeros n_attrstamp.

There remain other was processes can get stuck in vmopar.

Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
Reviewed by: dillon
Tested by: Vadim Belman <voland@lflat.org>

23 years agoThe -o is the default for "short" display, not the -h.
ru [Tue, 24 Oct 2000 08:47:07 +0000 (08:47 +0000)]
The -o is the default for "short" display, not the -h.

23 years agoAdded the CAM-ified wds driver for the ancient WD7000 SCSI card.
babkin [Tue, 24 Oct 2000 03:44:31 +0000 (03:44 +0000)]
Added the CAM-ified wds driver for the ancient WD7000 SCSI card.
Last time it was present in FreeBSD 3.x, before CAM.

Reviewed by: gibbs
Approved by: gibbs

23 years agoAdded lines for the wds driver.
babkin [Tue, 24 Oct 2000 03:38:28 +0000 (03:38 +0000)]
Added lines for the wds driver.

Approved by: gibbs

23 years agoAdd some very crude man pages to try to induce folk to work on them.
markm [Tue, 24 Oct 2000 03:33:24 +0000 (03:33 +0000)]
Add some very crude man pages to try to induce folk to work on them.

Agreed to work on them: jasone

23 years agoCast block number to off_t to avoid possible overflow bugs.
jwd [Tue, 24 Oct 2000 03:28:59 +0000 (03:28 +0000)]
Cast block number to off_t to avoid possible overflow bugs.

Pointed out by: Bruce Evans <bde@zeta.org.au>

23 years agoThe write combining code in revision 1.30 needs a few additional
jwd [Tue, 24 Oct 2000 00:08:30 +0000 (00:08 +0000)]
The write combining code in revision 1.30 needs a few additional
touch ups.  The cache needs to be flushed against block
reads, and a final flush at process termination to force the
backup superblocks to disk.

I believe this will allow 'make release' to complete.

Submitted by: Tor.Egge@fast.no

23 years agoDon't dink with interrupts in vm_page_zero_idle(). This code assumed it
jhb [Mon, 23 Oct 2000 23:32:36 +0000 (23:32 +0000)]
Don't dink with interrupts in vm_page_zero_idle().  This code assumed it
was being called with interrupts disabled, when it was actually being called
with them enabled.

Pointed out by: tegge

23 years agoGive correct results for SYSCTL_LONG arrays when sizeof(int) != sizeof(long)
gallatin [Mon, 23 Oct 2000 21:04:18 +0000 (21:04 +0000)]
Give correct results for SYSCTL_LONG arrays when sizeof(int) != sizeof(long)
This fixes unaligned access on alpha for, eg, sysctl kern.ipc.mbtypes.

23 years agoCURTHD now defines in globals.h
mjacob [Mon, 23 Oct 2000 18:39:30 +0000 (18:39 +0000)]
CURTHD now defines in globals.h

23 years agoMove bogus proc reference stuff into <machine/globals.h>. There is no
mjacob [Mon, 23 Oct 2000 18:36:03 +0000 (18:36 +0000)]
Move bogus proc reference stuff into <machine/globals.h>. There is no
more include file including <sys/proc.h>, but there still is this wonky
and (causes warnings on i386) reference in globals.h.

CURTHD is now defined in <machine/globals.h> as well. The correct thing
to do is provide a platform function for this.

23 years agoPut back inclusion of proc.h so that alpha kernels (at the very least)
mjacob [Mon, 23 Oct 2000 15:34:44 +0000 (15:34 +0000)]
Put back inclusion of proc.h so that alpha kernels (at the very least)
will compile again. I can't quite see where this was a recursive inclusion.
We probably need to do something to fix the alpha, but let's not break it
in the interim- it's broken enough.

23 years agoFix %c
ache [Mon, 23 Oct 2000 14:31:27 +0000 (14:31 +0000)]
Fix %c

23 years agoAdd PC-Card/ISA SCSI host adpater drivers from NetBSD/pc98
non [Mon, 23 Oct 2000 12:55:51 +0000 (12:55 +0000)]
Add PC-Card/ISA SCSI host adpater drivers from NetBSD/pc98
(a NetBSD port for NEC PC-98x1 machines). They are ncv for NCR 53C500,
nsp for Workbit Ninja SCSI-3, and stg for TMC 18C30 and 18C50.

I thank NetBSD/pc98 and bsd-nomads people.

Obtained from: NetBSD/pc98

23 years ago-Wall cleanup.
ru [Mon, 23 Oct 2000 12:18:34 +0000 (12:18 +0000)]
-Wall cleanup.

PR: 22187

23 years agoGenerate LOTS of warnings to remind the SMPng crew to fix the curproc
phk [Mon, 23 Oct 2000 09:22:18 +0000 (09:22 +0000)]
Generate LOTS of warnings to remind the SMPng crew to fix the curproc
UP/SMP issue.

23 years agoDo not recursively include <sys/proc.h>
phk [Mon, 23 Oct 2000 09:14:20 +0000 (09:14 +0000)]
Do not recursively include <sys/proc.h>

23 years agoApply style(9). [best I could]
obrien [Mon, 23 Oct 2000 08:46:25 +0000 (08:46 +0000)]
Apply style(9).  [best I could]

23 years agoApply style(9).
obrien [Mon, 23 Oct 2000 08:36:59 +0000 (08:36 +0000)]
Apply style(9).
(best I could)

23 years agoGet rid of the last traces of ACTUALLY_LKM_NOT_KERNEL
phk [Mon, 23 Oct 2000 08:35:41 +0000 (08:35 +0000)]
Get rid of the last traces of ACTUALLY_LKM_NOT_KERNEL

23 years agoSort the #define's.
obrien [Mon, 23 Oct 2000 08:29:04 +0000 (08:29 +0000)]
Sort the #define's.

23 years agoMinimal set of diffs from the i386 bits to the AlphaLinux.
obrien [Mon, 23 Oct 2000 08:15:58 +0000 (08:15 +0000)]
Minimal set of diffs from the i386 bits to the AlphaLinux.

23 years agoRe-order the #define's into a more logical one.
obrien [Mon, 23 Oct 2000 08:13:19 +0000 (08:13 +0000)]
Re-order the #define's into a more logical one.

23 years agobe careful on mbuf overrun on ctlinput.
itojun [Mon, 23 Oct 2000 07:11:01 +0000 (07:11 +0000)]
be careful on mbuf overrun on ctlinput.
short icmp6 packet may be able to panic the kernel.
sync with kame.