]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agomdoc(7) police: simplify construct.
ru [Thu, 21 Dec 2000 15:31:28 +0000 (15:31 +0000)]
mdoc(7) police: simplify construct.

23 years agoRe-Enable OSD_PRIORITY_GPE. Now 20001215 has been commited.
takawata [Thu, 21 Dec 2000 07:47:43 +0000 (07:47 +0000)]
Re-Enable OSD_PRIORITY_GPE. Now 20001215 has been commited.

23 years agoResolve conflicts from the ACPI CA 20001215 vendor import.
msmith [Thu, 21 Dec 2000 07:04:32 +0000 (07:04 +0000)]
Resolve conflicts from the ACPI CA 20001215 vendor import.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r70243,
msmith [Thu, 21 Dec 2000 06:56:46 +0000 (06:56 +0000)]
This commit was generated by cvs2svn to compensate for changes in r70243,
which included commits to RCS files with non-trunk default branches.

23 years agoUpdate to the Intel 20001215 ACPI CA snapshot.
msmith [Thu, 21 Dec 2000 06:56:46 +0000 (06:56 +0000)]
Update to the Intel 20001215 ACPI CA snapshot.

23 years agoIf the utmp login time is greater than the tty atime, use it to calculate
brian [Thu, 21 Dec 2000 01:30:47 +0000 (01:30 +0000)]
If the utmp login time is greater than the tty atime, use it to calculate
the idle time instead of the atime.

This makes entries for people that have logged in but done nothing
else show up correctly.

Reviewed by: markk@knigma.org

23 years agoAdd commented out entry showing use of console.info feature.
phk [Wed, 20 Dec 2000 22:26:33 +0000 (22:26 +0000)]
Add commented out entry showing use of console.info feature.

23 years agoA last minute brucification resulted in syntax errors in the previous commit.
phk [Wed, 20 Dec 2000 22:07:59 +0000 (22:07 +0000)]
A last minute brucification resulted in syntax errors in the previous commit.

23 years agoReplace logwakeup() with "int msgbuftrigger". There is little
phk [Wed, 20 Dec 2000 21:50:37 +0000 (21:50 +0000)]
Replace logwakeup() with "int msgbuftrigger".  There is little
point in calling a function just to set a flag.

Keep better track of the syslog FAC/PRI code and try to DTRT if
they mingle.

Log all writes to /dev/console to syslog with <console.info>
priority.  The formatting is not preserved, there is no robust,
way of doing it.  (Ideas with patches welcome).

23 years agoDisable my previous committed code for a moment.
iwasaki [Wed, 20 Dec 2000 20:22:47 +0000 (20:22 +0000)]
Disable my previous committed code for a moment.

Note to myself: this needs to be enabled again when newer version of
ACPI is imported.

23 years agoAdd task priority definition for OSD_PRIORITY_GPE in AcpiOsQueueForExecution().
iwasaki [Wed, 20 Dec 2000 19:15:38 +0000 (19:15 +0000)]
Add task priority definition for OSD_PRIORITY_GPE in AcpiOsQueueForExecution().
This is needed to next ACPICA import.

23 years agoJoin tree entries for Linksys PCMPC100 series as follows:
toshi [Wed, 20 Dec 2000 17:13:09 +0000 (17:13 +0000)]
Join tree entries for Linksys PCMPC100 series as follows:

 "Linksys" "EtherFast 10/100 PC Card (PCMPC100)"
 "Linksys" "Etherfast 10/100 PC Card (PCMPC100 V2)"
 "Linksys" "EtherFast 10/100 PC Card (PCMPC100 V3)"

23 years agoFix my previous commit. Remove ed_get_Linksys() function prototypes.
toshi [Wed, 20 Dec 2000 17:10:28 +0000 (17:10 +0000)]
Fix my previous commit. Remove ed_get_Linksys() function prototypes.

23 years agoWhen retrieving the time of day in nanosleep(), store it in the
deischen [Wed, 20 Dec 2000 17:04:12 +0000 (17:04 +0000)]
When retrieving the time of day in nanosleep(), store it in the
global time of day.  This costs us nothing, but is a bit of a hack
to work around a process blocking and not having the time updated
by an ITIMER_PROF signal.

PR: 23679

23 years agoChange priority of procedure queueing.
takawata [Wed, 20 Dec 2000 17:00:33 +0000 (17:00 +0000)]
Change priority of procedure queueing.
This is needed to next ACPICA import.

23 years agoEnable check for pending signals after calling a signal handler.
deischen [Wed, 20 Dec 2000 16:55:57 +0000 (16:55 +0000)]
Enable check for pending signals after calling a signal handler.
Restoration of a threads signal mask after invocation of a signal
handler may allow pending signals to become deliverable.

PR: 23647

23 years agoAdd PCIR_CAP_PTR as a define for the location of the capability
gibbs [Wed, 20 Dec 2000 14:36:52 +0000 (14:36 +0000)]
Add PCIR_CAP_PTR as a define for the location of the capability
pointer of type 0 devices.  This is required by my last aic7xxx change.

23 years agoPrepare for mdoc(7)NG.
ru [Wed, 20 Dec 2000 13:26:01 +0000 (13:26 +0000)]
Prepare for mdoc(7)NG.

23 years agoFix hard sentence break introduced in previous commit.
sheldonh [Wed, 20 Dec 2000 13:23:50 +0000 (13:23 +0000)]
Fix hard sentence break introduced in previous commit.

23 years agoAdd PECOFF (WIN32 Execution file format) support.
takawata [Wed, 20 Dec 2000 12:51:08 +0000 (12:51 +0000)]
Add PECOFF (WIN32 Execution file format) support.
To use it, some dll is needed. And currently, the dll is only for NetBSD.
So one more kernel module is needed.
For more infomation,
http://chiharu.haun.org/peace/ .

Reviewed by: bp

23 years agoRe-enable the lnc driver in GENERIC.
paul [Wed, 20 Dec 2000 11:24:47 +0000 (11:24 +0000)]
Re-enable the lnc driver in GENERIC.

23 years agoRe-commit revision 1.32, which grog incorrectly backed out in revision 1.33.
des [Wed, 20 Dec 2000 11:17:09 +0000 (11:17 +0000)]
Re-commit revision 1.32, which grog incorrectly backed out in revision 1.33.

23 years agoInitialise rcmdlen.
n_hibma [Wed, 20 Dec 2000 09:37:35 +0000 (09:37 +0000)]
Initialise rcmdlen.

Submitted by: Ian Dowse <iedowse@maths.tcd.ie>

23 years agoDisplay pathname of item being rm'ed.
obrien [Wed, 20 Dec 2000 08:31:26 +0000 (08:31 +0000)]
Display pathname of item being rm'ed.

Submitted by: Peter Pentchev <roam@orbitel.bg>

23 years agofix a debugging leftover (min speed=48khz)
cg [Wed, 20 Dec 2000 05:21:17 +0000 (05:21 +0000)]
fix a debugging leftover (min speed=48khz)

23 years agorevive_block: Don't go beyond the end of the stripe when reviving
grog [Wed, 20 Dec 2000 05:18:58 +0000 (05:18 +0000)]
revive_block: Don't go beyond the end of the stripe when reviving
      striped plexes.

Submitted by:   des

Don't lock buffers before calls to sdio, sdio does it by itself.

Submitted by: tegge

parityops: Use correct casts when returning error information.

23 years agobuild_rq_buffer: Note which buffer headers we lock.
grog [Wed, 20 Dec 2000 05:18:09 +0000 (05:18 +0000)]
build_rq_buffer: Note which buffer headers we lock.

sdio: Unlock the buffer if we fail.

Submitted by: tegge

23 years agoRearrange #includes to make more sense. This is still not the reform
grog [Wed, 20 Dec 2000 05:17:29 +0000 (05:17 +0000)]
Rearrange #includes to make more sense.  This is still not the reform
that bde is waiting to see, but at least it works.

23 years agoRename detached plexes and subdisks correctly (off by one error)
grog [Wed, 20 Dec 2000 05:16:46 +0000 (05:16 +0000)]
Rename detached plexes and subdisks correctly (off by one error)

Submitted by: Terry Glanfield <Terry.Glanfield@program-products.co.uk>

23 years agoopen_drive: Add support for more than 32 devices of a particular kind.
grog [Wed, 20 Dec 2000 05:15:50 +0000 (05:15 +0000)]
open_drive:  Add support for more than 32 devices of a particular kind.

Requested by: Bernd Walter <ticso@cicely8.cicely.de>
Cor Bosman <cor@xs4all.net>
Kai Storbeck <kai@xs4all.net>
Joe Greco <jgreco@ns.sol.net>

     Add support for Compaq SMART-2 RAID (idad) as storage
     device for Vinum subdisks.

Reported by: Aaron Hill <hillaa@hotmail.com>

23 years agoForced commit. Grrr. The previous commit contained an incorrect log
grog [Wed, 20 Dec 2000 05:14:37 +0000 (05:14 +0000)]
Forced commit.  Grrr.  The previous commit contained an incorrect log
message.

freerq: When freeing request elements, first unlock them if they're
locked.

sdio_done: Unlock the buffer.

Submitted by: tegge

23 years agogive_plex_to_volume: Recalculate volume size after attaching.
grog [Wed, 20 Dec 2000 05:13:26 +0000 (05:13 +0000)]
give_plex_to_volume:  Recalculate volume size after attaching.

Cosmetics.

23 years agoResolve RAW dependency violation between tbit and adds.
marcel [Wed, 20 Dec 2000 05:12:41 +0000 (05:12 +0000)]
Resolve RAW dependency violation between tbit and adds.

23 years agoAdd flag XFR_BUFLOCKED to identify buffers which have been locked.
grog [Wed, 20 Dec 2000 05:10:08 +0000 (05:10 +0000)]
Add flag XFR_BUFLOCKED to identify buffers which have been locked.
Part of fix to ensure that we unlock buffers we lock.

In principle submitted by: tegge

23 years agoCosmetics.
grog [Wed, 20 Dec 2000 05:08:41 +0000 (05:08 +0000)]
Cosmetics.

Correct location of history file.

Change references from /dev/wd to /dev/ad.

Use more appropriate defaults for stripe size in examples.

23 years agoroughlength: Use correct format strings.
grog [Wed, 20 Dec 2000 05:07:36 +0000 (05:07 +0000)]
roughlength: Use correct format strings.

23 years agoCorrect some comments.
grog [Wed, 20 Dec 2000 05:05:39 +0000 (05:05 +0000)]
Correct some comments.

vinum_attach: Get correct names (were being overwritten).

Submitted by: Terry Glanfield <Terry.Glanfield@program-products.co.uk>

23 years agoIn send_data(), use sendfile() instead of the mmap() algorithm.
dan [Wed, 20 Dec 2000 03:34:54 +0000 (03:34 +0000)]
In send_data(), use sendfile() instead of the mmap() algorithm.

23 years ago ahc_eisa.c:
gibbs [Wed, 20 Dec 2000 01:11:37 +0000 (01:11 +0000)]
ahc_eisa.c:
ahc_pci.c:
Add detach support.

Make use of soft allocated on our behalf by newbus.

For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.

aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.

aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.

Use AHC_* constants instead of hard coded numbers in a
few more places.

Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.

Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.

Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.

Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.

aic7xxx_freebsd.h:
Softc storage for our event handler.

Null implementation for the ahc_platform_flushwork() OSM
callback.  FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().

aic7xxx_inline.h:
Implement new lazy interrupt scheme.  To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed.  If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT.  Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted.  This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion.  This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.

aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.

Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().

Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.

Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.

Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT.  CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective.  The
old code might report a spurious interrupt on some "fast"
chipsets.

export ahc-update_target_msg_request for use by OSM code.

If a target does not respond to our ATN request, clear
it once we move to a non-message phase.  This avoids
sending a MSG_NOOP in some later message out phase.

Use max lun and max target constants instead of
hard-coded values.

Use softc storage built into our device_t under FreeBSD.

Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.

Clean up any tstate/lstate info in ahc_free().

Clear the powerdown state in ahc_reset() so that
registers can be accessed.

Add a preliminary function for pausing the chip and
processing any posted work.

Add a preliminary suspend and resume functions.

aic7xxx.h:
Limit the number of supported luns to 64.  We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.

Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr().  When the flag is not set, we use the
lazy interrupt handling scheme.

Add data structures to store controller state while
we are suspended.

Use constants instead of hard coded values where appropriate.

Correct some harmless "unsigned/signed" conflicts.

aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.

In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.

aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.

Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.

Add rudimentary power mode code so we can put the
controller into the D0 state.  In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.

Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.

In target mode, don't bother sending incremental
CRC data.

23 years agoo Add an example for a large file system.
imp [Tue, 19 Dec 2000 21:55:07 +0000 (21:55 +0000)]
o Add an example for a large file system.
o Remove bug about boot blocks hating non-8k file systems.  This hasn't been
  the case for a long time.

Not Objected to by: hackers, doc

23 years agoFix the PR. Getting a dma channel equal to 0 remains a problem though.
nsouch [Tue, 19 Dec 2000 21:28:17 +0000 (21:28 +0000)]
Fix the PR. Getting a dma channel equal to 0 remains a problem though.

PR: i386/22568

23 years agoDocument IDE on NoName a bit better.
wilko [Tue, 19 Dec 2000 20:13:10 +0000 (20:13 +0000)]
Document IDE on NoName a bit better.

Suggested by: the Dutch FreeBSD Users Group (NLFUG)

23 years agoVarious fixes to make leased line operation more robust. On lcp_up, start
jhay [Tue, 19 Dec 2000 19:08:11 +0000 (19:08 +0000)]
Various fixes to make leased line operation more robust. On lcp_up, start
to negotiate from scratch. Make leased lines survive being put into
loopback mode. Bits and pieces and ideas taken from PRs 11238 and 21771.
Make it a module so that it can be kldloaded. Whitespace cleanup. (Can be
ignored with "cvs diff -b".)

PR: 11238 and 21771 (bits and pieces)

23 years agoPrepare for mdoc(7)NG.
ru [Tue, 19 Dec 2000 16:00:12 +0000 (16:00 +0000)]
Prepare for mdoc(7)NG.

23 years agoPrepare for mdoc(7)NG.
ru [Tue, 19 Dec 2000 15:46:46 +0000 (15:46 +0000)]
Prepare for mdoc(7)NG.

23 years agoFix testing reboot howto flags in acpi_shutdown_final().
iwasaki [Tue, 19 Dec 2000 15:45:11 +0000 (15:45 +0000)]
Fix testing reboot howto flags in acpi_shutdown_final().
This sould make the system power-off correctly where the howto had
more bits set than RB_POWEROFF, e.g. RB_NOSYNC.

Submitted by: Peter Pentchev <roam@orbitel.bg>

23 years agoPrepare for mdoc(7)NG.
ru [Tue, 19 Dec 2000 15:36:48 +0000 (15:36 +0000)]
Prepare for mdoc(7)NG.

23 years agoPrepare for mdoc(7)NG.
ru [Tue, 19 Dec 2000 15:30:26 +0000 (15:30 +0000)]
Prepare for mdoc(7)NG.

23 years agoo Back out 1.39, it was a bad idea. There was, and should be, a
rwatson [Tue, 19 Dec 2000 14:12:46 +0000 (14:12 +0000)]
o Back out 1.39, it was a bad idea.  There was, and should be, a
  distinction between the OS copyright message and the message displayed
  gratuitously to each user at login.  Because, well, they may be
  different, among other things, and boy can a copyright message each
  login consume some screen space.  If people really want to do this,
  they can copy /COPYRIGHT to /etc/COPYRIGHT.

Submitted by: Anders Andersson <anders@codefactory.se>

23 years agoEnable the sym module.
alex [Tue, 19 Dec 2000 12:54:17 +0000 (12:54 +0000)]
Enable the sym module.

23 years agoUse the default value for MFILES
alex [Tue, 19 Dec 2000 12:27:53 +0000 (12:27 +0000)]
Use the default value for MFILES

23 years agoProberly back down DMA modes on the Acer Aladdin.
sos [Tue, 19 Dec 2000 10:37:03 +0000 (10:37 +0000)]
Proberly back down DMA modes on the Acer Aladdin.

23 years agoun-staticize M_AGP so that it can be used in agp*.c
assar [Tue, 19 Dec 2000 08:32:37 +0000 (08:32 +0000)]
un-staticize M_AGP so that it can be used in agp*.c

23 years agoremove pfctlinput
assar [Tue, 19 Dec 2000 08:09:36 +0000 (08:09 +0000)]
remove pfctlinput

23 years agoSeveral small but important fixes for snapshots:
mckusick [Tue, 19 Dec 2000 04:41:09 +0000 (04:41 +0000)]
Several small but important fixes for snapshots:

1) Be more tolerant of missing snapshot files by only trying to decrement
   their reference count if they are registered as active.

2) Fix for snapshots of filesystems with block sizes larger than 8K
   (from Ollivier Robert <roberto@eurocontrol.fr>).

3) Fix to avoid losing last block in snapshot file when calculating blocks
   that need to be copied (from Don Coleman <coleman@coleman.org>).

23 years agoGet rid of spurious check in ffs_truncate for i_size == length
mckusick [Tue, 19 Dec 2000 04:20:13 +0000 (04:20 +0000)]
Get rid of spurious check in ffs_truncate for i_size == length
which fails to set the modification time on the file. The same
check a few lines later takes the correct action.

Submitted by: Ian Dowse <iedowse@maths.tcd.ie>

23 years agotranslate the flags in recvfrom and recvmsg from linux to bsd ones
assar [Tue, 19 Dec 2000 00:24:25 +0000 (00:24 +0000)]
translate the flags in recvfrom and recvmsg from linux to bsd ones

Approved by: marcel

23 years agoFix column alignment (whitespace-only change).
rnordier [Tue, 19 Dec 2000 00:17:36 +0000 (00:17 +0000)]
Fix column alignment (whitespace-only change).

23 years agoFix another sched_sihand -> sched_swi in a KTR trace message.
jhb [Mon, 18 Dec 2000 23:59:34 +0000 (23:59 +0000)]
Fix another sched_sihand -> sched_swi in a KTR trace message.

23 years agoConvert the sio driver to use a spin mutex instead of a s_lock. This is
jhb [Mon, 18 Dec 2000 23:56:16 +0000 (23:56 +0000)]
Convert the sio driver to use a spin mutex instead of a s_lock.  This is
going to hurt sio(4) performance for the time being.  As we get closer to
release and have more of the kernel unlocked we can come back to doing
arcane optimizations to workaround the limitations of the sio hardware.

23 years agoAttempt to read and verify the card's status in wi_stop() before sending it
jhb [Mon, 18 Dec 2000 23:49:56 +0000 (23:49 +0000)]
Attempt to read and verify the card's status in wi_stop() before sending it
the disable command.  On some systems, writing to the card after it has
been ejected causes the machine to hang.

Reviewed by: wpaul

23 years agoNew release notes: spic(4), netgrph(4) updates, sbuf(9), el(4) now
bmah [Mon, 18 Dec 2000 23:15:34 +0000 (23:15 +0000)]
New release notes:  spic(4), netgrph(4) updates, sbuf(9), el(4) now
a module, Kerberos IV fixes, telnet(1) environment fixes, procfs(4)
security fixes, better sysinstall defaults, gcc->2.96, fsck_ffs SIGINFO,
lpr(1)/lpd(8)/syslogd(8)/logger(1) now IPv6-capable.

23 years agoAdd a line specifying the architecture for each RELNOTES.TXT file.
bmah [Mon, 18 Dec 2000 22:50:20 +0000 (22:50 +0000)]
Add a line specifying the architecture for each RELNOTES.TXT file.

23 years agoAdd power state manipulation to the fxp driver. Some people have
wpaul [Mon, 18 Dec 2000 22:06:12 +0000 (22:06 +0000)]
Add power state manipulation to the fxp driver. Some people have
claimed that their Intel NIC is comatose after a warm boot from Windoze.
This is most likely due to the card getting put in the D3 state. This
should bring it back to life.

23 years agoUse pci_get_powerstate()/pci_set_powerstate() which now exists in the
wpaul [Mon, 18 Dec 2000 21:53:05 +0000 (21:53 +0000)]
Use pci_get_powerstate()/pci_set_powerstate() which now exists in the
PCI code. This saves each driver from having to grovel around looking
for the right registers to twiddle.

I should eventually convert the other PCI drivers to do this; for now,
these three are ones which I know need power state handling.

23 years agoAdd a missing period and newline to a message.
phk [Mon, 18 Dec 2000 21:14:25 +0000 (21:14 +0000)]
Add a missing period and newline to a message.

PR: 23334
Submitted by: Rich Morin <rdm@cfcl.com>

23 years agoo Add cardbus support (use pci attachment for now)
imp [Mon, 18 Dec 2000 21:08:53 +0000 (21:08 +0000)]
o Add cardbus support (use pci attachment for now)
<bde>
o Add comments in some places to clarify some points.
o Don't typedef sc_p.  This isn't usually done in the drivers and may
  cause problems in teh future if C goes the C++ route of requiring
  one and only one definition for each and every type.  Instead use
  the current convetion of expanding struct ${1}_softc * inline needed.
o change some comments to be more style(9)-like.
o Define and use DEV2SOFTC to encapsulate storing/getting softc from a
  dev_t.  This also takes care of the missing cast from the examples.
o Define and use DEVICE2SOFTC, similar to DEV2SOFTC for getting the
  softc from a device_t.
</bde>

We still should have this generate foo_{isa,pci,pccard,cardbus,eisa}.c
and foovar.h from templates of some sort, but I was too lazy to do
that in this commit.  I did document it in the comments, however.

Note: bde-like corrections made with the help of my my portable
plastic bde icon.  Results with the real bde may vary with use.

23 years agoAdd if=/etc/issue to the default getty entry.
phk [Mon, 18 Dec 2000 20:56:10 +0000 (20:56 +0000)]
Add if=/etc/issue to the default getty entry.

This does nothing if you don't have an /etc/issue file.

PR: 22331
Submitted by: Rick C. Petty <rick@kiwi-computer.com>

23 years agoRemove a duplicate fortune.
phk [Mon, 18 Dec 2000 20:37:12 +0000 (20:37 +0000)]
Remove a duplicate fortune.

PR: 21629
Submitted by: Sean C. Farley <sean-freebsd@farley.org>

23 years agoRemove the "wd" driver.
phk [Mon, 18 Dec 2000 20:16:52 +0000 (20:16 +0000)]
Remove the "wd" driver.

23 years agoFix floppy drives on machines with lots of RAM.
phk [Mon, 18 Dec 2000 20:12:13 +0000 (20:12 +0000)]
Fix floppy drives on machines with lots of RAM.

The fix works by reverting the ordering of free memory so that the
chances of contig_malloc() succeeding increases.

PR: 23291
Submitted by: Andrew Atrens <atrens@nortel.ca>

23 years agoDivorce the kernel binary ABI version number from the message
julian [Mon, 18 Dec 2000 20:03:32 +0000 (20:03 +0000)]
Divorce the kernel binary ABI version number from the message
format version number. (userland programs should not need to be
recompiled when the netgraph kernel internal ABI is changed.

Also fix modules that don;t handle the fact that a caller may not supply
a return message pointer. (benign at the moment because the calling code
checks, but that will change)

23 years agoAdd note about ed driver changes impacting the Linksys cards.
imp [Mon, 18 Dec 2000 17:25:26 +0000 (17:25 +0000)]
Add note about ed driver changes impacting the Linksys cards.

23 years agoUse "node->ID" for the node's ID, instead of "(long)node".
archie [Mon, 18 Dec 2000 17:18:35 +0000 (17:18 +0000)]
Use "node->ID" for the node's ID, instead of "(long)node".

Reported by: julian

23 years agoUtilise the -p file to crunchgen to specify where the objects should
joe [Mon, 18 Dec 2000 16:28:03 +0000 (16:28 +0000)]
Utilise the -p file to crunchgen to specify where the objects should
be built and remove the previous glue that did the same thing.

23 years agoLinksys Fast Ethernet PCCARD cards supported by the ed driver now
toshi [Mon, 18 Dec 2000 15:28:53 +0000 (15:28 +0000)]
Linksys Fast Ethernet PCCARD cards supported by the ed driver now
require the addition of flag 0x80000 to their config line in
pccard.conf(5).  This flag is not optional.  These Linksys cards will
not be recognized without it.

Reviewed by:    imp, iwasaki

23 years agoPrepare for mdoc(7)NG.
ru [Mon, 18 Dec 2000 15:16:24 +0000 (15:16 +0000)]
Prepare for mdoc(7)NG.

23 years agoPrepare for mdoc(7)NG.
ru [Mon, 18 Dec 2000 14:48:17 +0000 (14:48 +0000)]
Prepare for mdoc(7)NG.

23 years agoIf no device is specified, check the CDROM environment variable before
des [Mon, 18 Dec 2000 14:35:54 +0000 (14:35 +0000)]
If no device is specified, check the CDROM environment variable before
picking the default device.

23 years agoImpossible to see typo.. |= instead of !=
julian [Mon, 18 Dec 2000 13:41:46 +0000 (13:41 +0000)]
Impossible to see typo..   |= instead of !=

23 years agorevert addition of strlcpy/strlcat
assar [Mon, 18 Dec 2000 13:28:12 +0000 (13:28 +0000)]
revert addition of strlcpy/strlcat

23 years agomdoc(7) police: document IPv6 options in the SYNOPSIS.
ru [Mon, 18 Dec 2000 08:42:58 +0000 (08:42 +0000)]
mdoc(7) police: document IPv6 options in the SYNOPSIS.

23 years agomdoc(7) police: document IPv6 options in the SYNOPSIS.
ru [Mon, 18 Dec 2000 08:37:53 +0000 (08:37 +0000)]
mdoc(7) police: document IPv6 options in the SYNOPSIS.

23 years agomdoc(7) police: removed hard sentence breaks, run through spell-checker.
ru [Mon, 18 Dec 2000 08:33:25 +0000 (08:33 +0000)]
mdoc(7) police: removed hard sentence breaks, run through spell-checker.

23 years agoWhitespace. Fix a comment block and an if statement that were wider
jake [Mon, 18 Dec 2000 07:10:04 +0000 (07:10 +0000)]
Whitespace.  Fix a comment block and an if statement that were wider
than 80 characters.

23 years agoUnbreak the modules after the last commit (newpcm kobjification).
green [Mon, 18 Dec 2000 06:39:01 +0000 (06:39 +0000)]
Unbreak the modules after the last commit (newpcm kobjification).

23 years agoForced commit to note that the previous commit message was wrong.
jake [Mon, 18 Dec 2000 06:10:17 +0000 (06:10 +0000)]
Forced commit to note that the previous commit message was wrong.
An exclusive lock is sufficient for a shared assertion to pass,
but not the other way around.

23 years agoAdd an assertion macro for lockmgr locks, LOCKMGR_ASSERT(lkp, what, p).
jake [Mon, 18 Dec 2000 05:50:31 +0000 (05:50 +0000)]
Add an assertion macro for lockmgr locks, LOCKMGR_ASSERT(lkp, what, p).
The what argument is the hold type that assertion acts on.  LK_SHARED
to assert that the process holds a shared, LK_EXCLUSIVE to assert that
the process holds _either_ a shared lock or an exclusive lock.

23 years agoadd strlcpy and strlcat to kernel
assar [Mon, 18 Dec 2000 04:08:58 +0000 (04:08 +0000)]
add strlcpy and strlcat to kernel

23 years agokobjify.
cg [Mon, 18 Dec 2000 01:36:41 +0000 (01:36 +0000)]
kobjify.

this gives us several benefits, including:

* easier extensibility- new optional methods can be added to
  ac97/mixer/channel classes without having to fixup every driver.

* forward compatibility for drivers, provided no new mandatory methods are
  added.

23 years agoMake the comment generated in /etc/pw.conf match reality
dougb [Mon, 18 Dec 2000 01:35:56 +0000 (01:35 +0000)]
Make the comment generated in /etc/pw.conf match reality

PR: misc/23451
Submitted by: Ben Rosengart, ben@narcissus.net

23 years agoadd a stub for softdep_slowdown so that it's possible to build the
assar [Sun, 17 Dec 2000 23:59:56 +0000 (23:59 +0000)]
add a stub for softdep_slowdown so that it's possible to build the
kernel without SOFTUPDATES

23 years agoAvoid a data-consistency race between write() and mmap()
dillon [Sun, 17 Dec 2000 23:57:05 +0000 (23:57 +0000)]
Avoid a data-consistency race between write() and mmap()
by ensuring that newly allocated blocks are zerod.  The
race can occur even in the case where the write covers
the entire block.

Reported by: Sven Berkvens <sven@berkvens.net>, Marc Olzheim <zlo@zlo.nu>

23 years ago* Fix a long line that I introduced in the last commit
dougb [Sun, 17 Dec 2000 23:42:45 +0000 (23:42 +0000)]
* Fix a long line that I introduced in the last commit
* Backslash escape non-alphanumeric chars in the login name so that perl
  doesn't choke on things like '$'.

23 years ago* Add an eval so that ipnat_flags=">/dev/null" works, per the PR
dougb [Sun, 17 Dec 2000 22:14:49 +0000 (22:14 +0000)]
* Add an eval so that ipnat_flags=">/dev/null" works, per the PR
* Do some line length and specify full path cleanups while I'm here

PR: conf/22937
Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>

23 years agoFix mostly harmless typo:
nectar [Sun, 17 Dec 2000 21:10:41 +0000 (21:10 +0000)]
Fix mostly harmless typo:

    if (data);
            free(data);

Discovered by: emacs cc-mode

23 years agoFix bug: a read() on a bpf device which was in non-blocking mode
jdp [Sun, 17 Dec 2000 20:50:22 +0000 (20:50 +0000)]
Fix bug: a read() on a bpf device which was in non-blocking mode
and had no data available returned 0.  Now it returns -1 with errno
set to EWOULDBLOCK (== EAGAIN) as it should.  This fix makes the bpf
device usable in threaded programs.

Reviewed by: bde

23 years agoUse one of the TMPDIR env variable or _PATH_TMP instead of hardcoding
joe [Sun, 17 Dec 2000 19:49:07 +0000 (19:49 +0000)]
Use one of the TMPDIR env variable or _PATH_TMP instead of hardcoding
/tmp.

Submitted by: jedgar

23 years agoAdd a -a option to show the entire message buffer.
phk [Sun, 17 Dec 2000 19:05:17 +0000 (19:05 +0000)]
Add a -a option to show the entire message buffer.

23 years agoo /etc/COPYRIGHT -> /COPYRIGHT, since that's where it's installed.
rwatson [Sun, 17 Dec 2000 18:55:14 +0000 (18:55 +0000)]
o /etc/COPYRIGHT -> /COPYRIGHT, since that's where it's installed.

Submitted by: Anders Andersson <anders@sanyusan.se>