]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
24 years agoTreat all interface aliases as local addresses.
Brian Somers [Wed, 9 Aug 2000 09:24:31 +0000 (09:24 +0000)]
Treat all interface aliases as local addresses.

The new rtiparse() function should really be in libc IMHO.... I'm
sick of writing this code :-(

PR: 20381

24 years agoImprove purgedir():
Brian Somers [Wed, 9 Aug 2000 09:23:30 +0000 (09:23 +0000)]
Improve purgedir():

  Fix leading & trailing space handling

  Suggested by: ben

  Handle files beginning with - correctly
  Don't follow symlinks (cd /var/spool/lock; ln -s /. horror)

24 years agoA couple of cosmetic fixes, plus a biggie: When checking the DMA
Nick Sayer [Wed, 9 Aug 2000 07:14:56 +0000 (07:14 +0000)]
A couple of cosmetic fixes, plus a biggie: When checking the DMA
position, channel 1's dma position register must be quiescent. So
the driver will spl, pause the DMA, delay a bit and hold as still as
possible while snapping the picture.

I'm sure there HAS to be a better way to do this, but if there is, it's
not documented.

So far as I can tell, this fixes recording, which means the Solo is open
for business.

24 years agoUse shell matching instead of sed(1) to strip comments and blank lines.
Chris Costello [Wed, 9 Aug 2000 02:03:30 +0000 (02:03 +0000)]
Use shell matching instead of sed(1) to strip comments and blank lines.
Since it doesn't depend on anything in /usr, it should work with a NFS-mounted
/usr partition.

Thanks to Bruce Evans to bringing this to my attention.

24 years agoDon't set flags on the mount structure before all permission checks have
Tor Egge [Wed, 9 Aug 2000 01:57:11 +0000 (01:57 +0000)]
Don't set flags on the mount structure before all permission checks have
been done.

Don't allow multiple mount operations with MNT_UPDATE at the same
time on the same mount point.  When the first mount operation
completed, MNT_UPDATE was cleared in the mount structure, causing
the second to complete as if it was a no-update mount operation
with the following bad side effects:

        - mount structure inserted multiple times onto the mountlist
        - vp->v_mountedhere incorrectly set, causing next namei
          operation walking into the mountpoint to crash with
          a locking against myself panic.

Plug a vnode leak in case vinvalbuf fails.

24 years agoAdded extra paragraph to document the 'rebuild' feature.
Bill Swingle [Wed, 9 Aug 2000 01:54:30 +0000 (01:54 +0000)]
Added extra paragraph to document the 'rebuild' feature.

Reviewed by: msmith

24 years agoFix bug where bundle-level receive statistics were not getting updated.
Archie Cobbs [Wed, 9 Aug 2000 01:43:21 +0000 (01:43 +0000)]
Fix bug where bundle-level receive statistics were not getting updated.

24 years agoMFS: sync with peter's isadma fixup
Cameron Grant [Wed, 9 Aug 2000 01:22:09 +0000 (01:22 +0000)]
MFS: sync with peter's isadma fixup

24 years agoactivate via82c686 audio driver
Cameron Grant [Wed, 9 Aug 2000 01:08:59 +0000 (01:08 +0000)]
activate via82c686 audio driver

24 years agonew driver: via 82c686 chipset audio
Cameron Grant [Wed, 9 Aug 2000 00:55:55 +0000 (00:55 +0000)]
new driver: via 82c686 chipset audio

not personally tested, but i'm assured it works

Submitted by: David Jones <dej@ox.org>
PR: kern/20161

24 years agowarn if hwptr goes backwards
Cameron Grant [Wed, 9 Aug 2000 00:42:00 +0000 (00:42 +0000)]
warn if hwptr goes backwards

24 years agoInitialize *countp to 0 in stub for softdep_flushworklist().
Tor Egge [Wed, 9 Aug 2000 00:41:54 +0000 (00:41 +0000)]
Initialize *countp to 0 in stub for softdep_flushworklist().
This allows ffs_fsync() to break out of a loop that might otherwise
be infinite on kernels compiled without the SOFTUPDATES option.
The observed symptom was a system hang at the first unmount attempt.

24 years agoFix a typo in the last commit so that this compiles.
John Baldwin [Wed, 9 Aug 2000 00:28:09 +0000 (00:28 +0000)]
Fix a typo in the last commit so that this compiles.

24 years agoMF4: fix style(9) whitespace breakage.
Peter Wemm [Tue, 8 Aug 2000 23:53:11 +0000 (23:53 +0000)]
MF4: fix style(9) whitespace breakage.

24 years agoDon't attempt to actually read SAF-TE temperature objects- nobody seems
Matt Jacob [Tue, 8 Aug 2000 23:00:11 +0000 (23:00 +0000)]
Don't attempt to actually read SAF-TE temperature objects- nobody seems
to be obeying the original spec as to what the numeric value means.

Temperature flags are unaffected- these are still the 'pseudo-thermometers'
and overtemp/undertemp warnings will be caught and translated to SES objects
here.

PR: 20475

24 years agoMention the 3ware controllers.
Mike Smith [Tue, 8 Aug 2000 18:54:57 +0000 (18:54 +0000)]
Mention the 3ware controllers.

24 years agoMention the 3ware RAID controllers
Mike Smith [Tue, 8 Aug 2000 18:53:08 +0000 (18:53 +0000)]
Mention the 3ware RAID controllers

24 years agoUpdate to mention support for 6x000 series controllers.
Mike Smith [Tue, 8 Aug 2000 18:50:33 +0000 (18:50 +0000)]
Update to mention support for 6x000 series controllers.

24 years agoo Add POSIX1E_CAPABILITY_EXTATTR_NAME, a string constant for the
Robert Watson [Tue, 8 Aug 2000 18:47:59 +0000 (18:47 +0000)]
o Add POSIX1E_CAPABILITY_EXTATTR_NAME, a string constant for the
  extended attribute name used to store capabilities for binaries in
  FFS.
o Uncomment CAP_SETPCAP because, unfortunately, it is useful when
  developing and testing, as well as in practice for a system not
  fully capability-enabled.
o Synch up prototypes with the implementation, reducing diffs and
  making the patches for capabilities more accessible and less
  dastardly.

Obtained from: TrustedBSD Project

24 years agoo Introduce vn_extattr_{get,set}, wrapper routines for VOP_GETEXTATTR
Robert Watson [Tue, 8 Aug 2000 17:15:32 +0000 (17:15 +0000)]
o Introduce vn_extattr_{get,set}, wrapper routines for VOP_GETEXTATTR
  and VOP_SETEXTATTR to simplify calling from in-kernel consumers,
  such as capability code.  Both accept a vnode (optionally locked,
  with ioflg to indicate that), attribute name, and a buffer + buffer
  length in UIO_SYSSPACE.  Both authorize the call as a kernel request,
  with cred set to NULL for the actual VOP_ calls.

Obtained from: TrustedBSD Project

24 years agoInsert missing word 'that'.
Archie Cobbs [Tue, 8 Aug 2000 16:26:03 +0000 (16:26 +0000)]
Insert missing word 'that'.

24 years agoFix braino in the tsleep call, ata_delayed_attach might be a NULL
Søren Schmidt [Tue, 8 Aug 2000 14:57:36 +0000 (14:57 +0000)]
Fix braino in the tsleep call, ata_delayed_attach might be a NULL
pointer..

24 years ago- add ftp6.fr.freebsd.org
Ollivier Robert [Tue, 8 Aug 2000 14:36:24 +0000 (14:36 +0000)]
- add ftp6.fr.freebsd.org
- correct URL for ftp4 and ftp5.

Submitted by: Benjamin Bayart <bayartb@guinness.domicile.fr> (for ftp6)

24 years agoDon't use find(1) before nfs filesystems have been mounted as
Brian Somers [Tue, 8 Aug 2000 13:30:27 +0000 (13:30 +0000)]
Don't use find(1) before nfs filesystems have been mounted as
it lives in /usr/bin.  Instead, locate files manually.

Note, only *files* under /var/spool/lock are now deleted rather
than everything that's not a directory.  I think this is more
correct, but if anyone disagrees please feel free to change it.

Problem pointed out by: bde

24 years agoMerged from sys/i386/conf/GENERIC revisions 1.266 and 1.267.
Yoshihiro Takahashi [Tue, 8 Aug 2000 11:57:09 +0000 (11:57 +0000)]
Merged from sys/i386/conf/GENERIC revisions 1.266 and 1.267.

24 years agoMerged from sys/i386/conf/GENERIC.hints revision 1.3.
Yoshihiro Takahashi [Tue, 8 Aug 2000 11:56:45 +0000 (11:56 +0000)]
Merged from sys/i386/conf/GENERIC.hints revision 1.3.

24 years agoClarify the guidelines surrounding the use of macros. The patch is
Sheldon Hearn [Tue, 8 Aug 2000 09:54:57 +0000 (09:54 +0000)]
Clarify the guidelines surrounding the use of macros.  The patch is
mostly unrelated to the attributed PR, and the attributed submitter
wasn't so much suggesting the patch for inclusion as providing it
for clarity.

PR: 9869
Submitted by: bde

24 years agoChanged default cursor shape to non-blink mode.
Yoshihiro Takahashi [Tue, 8 Aug 2000 09:28:32 +0000 (09:28 +0000)]
Changed default cursor shape to non-blink mode.

Submitted by: Tomokazu HARADA <tkhara@osk4.3web.ne.jp>

24 years agoFix some whitespace errors.
Kelly Yancey [Tue, 8 Aug 2000 09:27:55 +0000 (09:27 +0000)]
Fix some whitespace errors.

Pointed out by: ps

24 years agoFixed PC-9821 NOTE supports with LINE30 mode.
Yoshihiro Takahashi [Tue, 8 Aug 2000 09:17:15 +0000 (09:17 +0000)]
Fixed PC-9821 NOTE supports with LINE30 mode.

Submitted by: Tomokazu HARADA <tkhara@osk4.3web.ne.jp>

24 years agoReplace -l with .Fl l
Kelly Yancey [Tue, 8 Aug 2000 09:03:48 +0000 (09:03 +0000)]
Replace -l with .Fl l

Pointed out by: sheldonh

24 years agoFix the comments to properly document the PQ_MEDIUMCACHE and
Kelly Yancey [Tue, 8 Aug 2000 08:13:01 +0000 (08:13 +0000)]
Fix the comments to properly document the PQ_MEDIUMCACHE and
PQ_NORMALCACHE options.

PR: 20409
Submitted by: Tony Finch <dot@dotat.at>

24 years agoTypo/spelling fixes.
Robert Nordier [Tue, 8 Aug 2000 07:20:25 +0000 (07:20 +0000)]
Typo/spelling fixes.

24 years agoThe Yamaha CDR100 doesn't seem to like multi-LUN probing.
Kenneth D. Merry [Tue, 8 Aug 2000 06:49:33 +0000 (06:49 +0000)]
The Yamaha CDR100 doesn't seem to like multi-LUN probing.

PR: kern/20347
Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>

24 years agoThis is an overhaul of the mode page handling in camcontrol as well as
Kelly Yancey [Tue, 8 Aug 2000 06:24:17 +0000 (06:24 +0000)]
This is an overhaul of the mode page handling in camcontrol as well as
related patches. These include:
* Mode page editting can be scripted. This involves two
  things: first, if stdin is not a tty, changes are read from
  stdin rather than invoking $EDITOR. Second, and more
  importantly, not all modepage entries must be included in the
  change set. This means that camcontrol can now gracefully handle
  more intrusive editting from the $EDITOR, including removal or
  rearrangement of lines. It also means that you can do stuff
  like:
# echo "WCE: 1" | camcontrol modepage da3 -m 8 -e
# newfs /dev/da3
# echo "WCE: 0" | camcontrol modepage da3 -m 8 -e
* Range-checking on user-supplied input values. modeedit.c now
  uses the field width specifiers to determine the maximum
  allowable value for a field. If the user enters a value larger
  than the maximum, it clips the value to the max and warns the
  user. This also involved patching cam_cmdparse.c to be more
  consistent with regards to the "count" parameter to arg_put
  (previously is was the length of strings and 1 for all integral
  types). The cam_cdbparse(3) man page was also updated to reflect
  the revised semantics.
* In the process, I removed the 64 entry limit on mode pages (not
  that we were even close to hitting that limit). This was a nice
  side-effect of the other changes.
* Technically, the new mode editting functionality allows editting
  of character array entries in mode pages (type 'c' or 'z'),
  however since buff_encode doesn't grok them it is currently
  useless.
* Camcontrol gained two new options related to mode pages: -l and
  -b. The former lists all available mode pages for a given
  device. The latter forces mode page display in binary format
  (the default when no mode page definition was found in
  scsi_modes).
* Added support for mode page names to scsi_modes. Allows names to
  be displayed alongside mode numbers in the mode page
  listing. Updated scsi_modes to use the new functionality. This
  also adds the semicolon into the scsi_modes syntax as an
  optional mode page definition terminator. This is needed to name
  pages without providing a page format definition.
* Updated scsi_all.h to include a structure describing mode page
  headers.
* Added $FreeBSD$ line to scsi_modes.

Inspired by: dwhite
Reviewed by: ken

24 years agoIf using a DB_RECNO, db::put should return the new key if R_IAFTER is
Brian Feldman [Tue, 8 Aug 2000 05:20:57 +0000 (05:20 +0000)]
If using a DB_RECNO, db::put should return the new key if R_IAFTER is
set, not the previous key.

Add $FreeBSD$, not taking this off the vendor branch because it's not on.

24 years ago- Fixed warnings and typo.
Yoshihiro Takahashi [Tue, 8 Aug 2000 04:01:51 +0000 (04:01 +0000)]
- Fixed warnings and typo.
- Recognize slice type 0x24 as FAT (only PC-98).

Submitted by: Nobuyuki Koganemaru <kogane@koganemaru.co.jp>

24 years agokevent() is not a POSIX cancellation point, so _kevent() need not be
Jason Evans [Tue, 8 Aug 2000 00:12:35 +0000 (00:12 +0000)]
kevent() is not a POSIX cancellation point, so _kevent() need not be
defined.

Remove some unnecessary header file inclusions.

24 years agoUpdate link to Making the World tutorial (now in the handbook)
Mark Ovens [Mon, 7 Aug 2000 23:25:22 +0000 (23:25 +0000)]
Update link to Making the World tutorial (now in the handbook)

24 years agoFix bug with -c option
Mark Ovens [Mon, 7 Aug 2000 22:47:46 +0000 (22:47 +0000)]
Fix bug with -c option

PR: 18176
Reviewed by: Bruce A. Mah <bmah@acm.org>

24 years agoFix a bug where we were accessing already free'd memory during node shutdown.
Archie Cobbs [Mon, 7 Aug 2000 22:41:12 +0000 (22:41 +0000)]
Fix a bug where we were accessing already free'd memory during node shutdown.

Detected via: 0xdeadc0de

24 years agoAdd another Bill Paul quote.
Jonathan Lemon [Mon, 7 Aug 2000 20:49:08 +0000 (20:49 +0000)]
Add another Bill Paul quote.

Approved by: jkh

24 years agoAdd a "-i" option ("insecure") which disables the checks for
John Polstra [Mon, 7 Aug 2000 19:12:04 +0000 (19:12 +0000)]
Add a "-i" option ("insecure") which disables the checks for
root ownership, etc.  I will soon commit a companion knob for
"/etc/rc.conf".

Submitted by: Maxime Henrion <mhenrion@cybercable.fr>

24 years agoDocument three new control messages.
Archie Cobbs [Mon, 7 Aug 2000 18:53:42 +0000 (18:53 +0000)]
Document three new control messages.

24 years agoAdd three new control messages to the ng_ether(4) netgraph node type:
Archie Cobbs [Mon, 7 Aug 2000 18:52:26 +0000 (18:52 +0000)]
Add three new control messages to the ng_ether(4) netgraph node type:

    NGM_ETHER_GET_ENADDR: Get the device's Ethernet address
    NGM_ETHER_SET_PROMISC: Enable/disable promiscuous mode
    NGM_ETHER_SET_AUTOSRC: Enable/disable packet source address override

24 years agoFix some warnings. Here are more part-time volatiles - ie: data that
Peter Wemm [Mon, 7 Aug 2000 18:44:05 +0000 (18:44 +0000)]
Fix some warnings.  Here are more part-time volatiles - ie: data that
lives in memory and is sometimes busmastered to/from the controller.
I believe these are all ok.

24 years agoMake the kqueue socket read filter honor the SO_RCVLOWAT value.
Jonathan Lemon [Mon, 7 Aug 2000 17:52:08 +0000 (17:52 +0000)]
Make the kqueue socket read filter honor the SO_RCVLOWAT value.

Spotted by:  "Steve M." <stevem@redlinenetworks.com>

24 years agoClose PR 20438. Make fix for preserving LED settings conditional on
Bill Paul [Mon, 7 Aug 2000 17:03:20 +0000 (17:03 +0000)]
Close PR 20438. Make fix for preserving LED settings conditional on
presence Intel 21143 chip.

24 years agoAdd wrapper for kevent() syscall
Jonathan Lemon [Mon, 7 Aug 2000 16:51:56 +0000 (16:51 +0000)]
Add wrapper for kevent() syscall

Noted as missing by: nicolas.leonard@animaths.com

24 years agoDocument return value of ENOENT for nonexistent/invalid filter entries.
Jonathan Lemon [Mon, 7 Aug 2000 16:47:35 +0000 (16:47 +0000)]
Document return value of ENOENT for nonexistent/invalid filter entries.

24 years agoFix bug with timeout; previously, when attempting to poll the kqueue by
Jonathan Lemon [Mon, 7 Aug 2000 16:45:42 +0000 (16:45 +0000)]
Fix bug with timeout; previously, when attempting to poll the kqueue by
passing a zero-valued timeout, the code would always sleep for one tick.
Change code to avoid calling tsleep if we have no intention of sleeping.

Bring in bugfix from sys_select.c, r1.60 which also applies here.

Modify error handling slightly; passing in an invalid fd will now result
in EBADF returned in the eventlist, while an attempt to change a knote
which does not exist will result in ENOENT being returned.  Previously
such attempts would fail silently without notification.

Pointed out by: nicolas.leonard@animaths.com
        Rick Reed (rr@yahoo-inc.com)

24 years agoTo make compilable without -DINET6.
Hajimu UMEMOTO [Mon, 7 Aug 2000 16:39:33 +0000 (16:39 +0000)]
To make compilable without -DINET6.

PR: bin/20407
Submitted by: Patrick Bihan-Faou <patrick@mindstep.com>

24 years agoMake the update target consistent; both ports and doc are updated
Sheldon Hearn [Mon, 7 Aug 2000 14:35:49 +0000 (14:35 +0000)]
Make the update target consistent; both ports and doc are updated
if their SUPFILE variables are defined _and_ NO_PORTSUPDATE and
NO_DOCUPDATE respectively are not defined.

Previously, only ports was updated and there was no way to prevent
this without undefining its SUPFILE variable.

PR: 17514
Reported by: Udo Erdelhoff <ue@nathan.ruhr.de>

24 years agoAdjust TCP checksum rather than compute it afresh.
Ruslan Ermilov [Mon, 7 Aug 2000 09:51:04 +0000 (09:51 +0000)]
Adjust TCP checksum rather than compute it afresh.

Submitted by: Erik Salander <erik@whistle.com>

24 years agoUse ``diff -w'' for setuid.{to,yester}day comparisons
Brian Somers [Mon, 7 Aug 2000 09:08:35 +0000 (09:08 +0000)]
Use ``diff -w'' for setuid.{to,yester}day comparisons
rather than ``diff -b''.

24 years agoBack this one out until I can fix _all_ of the headers. The headers sure
Brian Feldman [Mon, 7 Aug 2000 03:54:25 +0000 (03:54 +0000)]
Back this one out until I can fix _all_ of the headers.  The headers sure
are shpxed quite nicely.

24 years agoAdd workaround for livelock problem when starting APs.
Tor Egge [Mon, 7 Aug 2000 02:28:37 +0000 (02:28 +0000)]
Add workaround for livelock problem when starting APs.

With more than 1 AP present, an AP could fail to properly release
the mp lock before waiting for smp_started to become nonzero.

With early startup of APs, the BSP could fail to properly release
the mp lock before waiting for smp_started to become nonzero.

24 years agoDon't use kern.disks sysctl on PC-98 because the wd driver doesn't call
Yoshihiro Takahashi [Mon, 7 Aug 2000 02:14:44 +0000 (02:14 +0000)]
Don't use kern.disks sysctl on PC-98 because the wd driver doesn't call
disk_create() function.

24 years agoUpdate the usage string to reflect -n as an option.
Paul Saab [Mon, 7 Aug 2000 01:09:25 +0000 (01:09 +0000)]
Update the usage string to reflect -n as an option.

24 years agoActually make syslogd understand the new -n option by putting it
Paul Saab [Mon, 7 Aug 2000 01:00:01 +0000 (01:00 +0000)]
Actually make syslogd understand the new -n option by putting it
in getopt.

Submitted by: Clive Lin <clive@CirX.ORG>

24 years ago0xA0 = Suspend to disk.
Josef Karthauser [Mon, 7 Aug 2000 00:26:09 +0000 (00:26 +0000)]
0xA0 = Suspend to disk.

24 years agoSave and restore the registers SOF and FRAMENUMBER on suspend. In some
Nick Hibma [Mon, 7 Aug 2000 00:04:53 +0000 (00:04 +0000)]
Save and restore the registers SOF and FRAMENUMBER on suspend. In some
cases the registers are not correctly set on resume.

This solves the problem of USB failing after resuming a machine.

Submitted by: mike+fbsd@medianstrip.net
PR: 18261

24 years agoRegen.
Nick Hibma [Sun, 6 Aug 2000 23:27:36 +0000 (23:27 +0000)]
Regen.

24 years agoNew ids
Nick Hibma [Sun, 6 Aug 2000 23:26:58 +0000 (23:26 +0000)]
New ids

Submitted by: Scott Long
PR: 18897

24 years agoSome new ids, from NetBSD
Nick Hibma [Sun, 6 Aug 2000 23:22:13 +0000 (23:22 +0000)]
Some new ids, from NetBSD

24 years agoType in FreeBSD capitalization
Warner Losh [Sun, 6 Aug 2000 22:18:39 +0000 (22:18 +0000)]
Type in FreeBSD capitalization

24 years agoAdd entry on the installkernel changes as well as more grumpy notes
Warner Losh [Sun, 6 Aug 2000 22:16:34 +0000 (22:16 +0000)]
Add entry on the installkernel changes as well as more grumpy notes
about needing to be on -current to track -current.  Tweak the 4.0 ->
4.x version stuff.

24 years agosync CCR register definitions with creative sources
Cameron Grant [Sun, 6 Aug 2000 20:58:11 +0000 (20:58 +0000)]
sync CCR register definitions with creative sources

24 years agoAdd experimental code for ATA100 support on:
Søren Schmidt [Sun, 6 Aug 2000 19:51:58 +0000 (19:51 +0000)]
Add experimental code for ATA100 support on:

Promise Ultra100 / Fasttrak100
HighPoint HPT370 controllers (fx Abit KA7-100 onboard ctrl, Abit HotRod 100)
Intel ICH2 (Intel 815E based motherboards)

So far I can read >90MB/s on the Promise and the HPT370.
I can write >64MB/s on the promise and >50MB/s on the HPT370 so it seems
writing is still done in ATA66 mode :(
The ICH2 support is untested as of yet...

24 years agoMissed one misspelling in previous commit.
Archie Cobbs [Sun, 6 Aug 2000 19:25:59 +0000 (19:25 +0000)]
Missed one misspelling in previous commit.

24 years agoUse tsleep instead od DELAY in probe when not in boot.
Søren Schmidt [Sun, 6 Aug 2000 19:10:05 +0000 (19:10 +0000)]
Use tsleep instead od DELAY in probe when not in boot.
This fixes the uptil 30s hangs on PCCARD ata device probes.

24 years agoCleaning conflicting module tree. Use modules/3dfx instead.
Coleman Kane [Sun, 6 Aug 2000 19:08:55 +0000 (19:08 +0000)]
Cleaning conflicting module tree. Use modules/3dfx instead.

24 years agoFix clean to remove pci.h
Coleman Kane [Sun, 6 Aug 2000 18:56:46 +0000 (18:56 +0000)]
Fix clean to remove pci.h
Numerous style fixes

Submitted by: Marcel Moolenaar <marcel@cup.hp.com>

24 years agofix the staticy sound issue
Cameron Grant [Sun, 6 Aug 2000 18:10:05 +0000 (18:10 +0000)]
fix the staticy sound issue

use timer instead of per-channel interrupts

do playback like the linux driver - may fix nmi-with-ecc issue

24 years agoRemove sequence checking in the burner support.
Søren Schmidt [Sun, 6 Aug 2000 18:03:42 +0000 (18:03 +0000)]
Remove sequence checking in the burner support.
This was really a leftover from the wormcontrol days, it is no longer
needed when using burncd.

24 years agoBe a bit more restrictive in using multisector transfers.
Søren Schmidt [Sun, 6 Aug 2000 18:01:47 +0000 (18:01 +0000)]
Be a bit more restrictive in using multisector transfers.

24 years agoGood, fixing the header showed incorrect usage of it! #define _KERNEL
Brian Feldman [Sun, 6 Aug 2000 17:03:09 +0000 (17:03 +0000)]
Good, fixing the header showed incorrect usage of it!  #define _KERNEL
here for the include of sys/select.h.

24 years ago- Fixed missing initialization of current device number.
KATO Takenori [Sun, 6 Aug 2000 14:35:37 +0000 (14:35 +0000)]
- Fixed missing initialization of current device number.
- Fixed comment.

Pointed out by: nyan

24 years agoChange the behavior of isa_nmi to log an error message instead of
Paul Saab [Sun, 6 Aug 2000 14:17:21 +0000 (14:17 +0000)]
Change the behavior of isa_nmi to log an error message instead of
panicing and return a status so that we can decide whether to drop
into DDB or panic.  If the status from isa_nmi is true, panic the
kernel based on machdep.panic_on_nmi, otherwise if DDB is
enabled, drop to DDB based on machdep.ddb_on_nmi.

Reviewed by: peter, phk

24 years agoConsistency fix.
Robert Nordier [Sun, 6 Aug 2000 09:19:45 +0000 (09:19 +0000)]
Consistency fix.

24 years agoNone of select.h needs to be exposed to !_KERNEL.
Brian Feldman [Sun, 6 Aug 2000 02:14:52 +0000 (02:14 +0000)]
None of select.h needs to be exposed to !_KERNEL.

24 years agoRemove 3 off by 128 errors.
Tor Egge [Sun, 6 Aug 2000 01:17:39 +0000 (01:17 +0000)]
Remove 3 off by 128 errors.

24 years agoBe more verbose when changing APIC ID on an IO APIC.
Tor Egge [Sun, 6 Aug 2000 00:04:03 +0000 (00:04 +0000)]
Be more verbose when changing APIC ID on an IO APIC.

Don't allow cpu entries in the MP table to contain APIC IDs out of range.

Don't write outside array boundaries if an IO APIC entry in the MP table
contains an APIC ID out of range.

Assign APIC IDs for all IO APICs according to section 3.6.6 in the
Intel MP spec:

  - If the current APIC ID on an IO APIC doesn't conflict with other
    IO APICs or CPUs, that APIC ID should be used.  The copy of the MP
    table must be updated if the corresponding APIC ID in the MP table
    is different.

  - If the current APIC ID was in conflict with other units, the
    corresponding APIC ID specified in the MP table is checked for conflict.

  - If a conflict is still found then fall back to using a new unique ID.
    The copy of the MP table must be updated.

  - IDs out of range is considered to be in conflict.

During these operations, the IO_TO_ID array cannot be used, since any
conflict would have caused information loss.  The array is then corrected,
since all APIC ID conflicts should have been resolved.

PR: 20312, 18919

24 years agoDocument the ``-'' option and minor re-wording in EXAMPLES
Mark Ovens [Sat, 5 Aug 2000 23:03:12 +0000 (23:03 +0000)]
Document the ``-'' option and minor re-wording in EXAMPLES

24 years agoFix misspelling.
Archie Cobbs [Sat, 5 Aug 2000 20:17:04 +0000 (20:17 +0000)]
Fix misspelling.

24 years agoMention the non-English docs as well as ports.
Ben Smithurst [Sat, 5 Aug 2000 18:34:17 +0000 (18:34 +0000)]
Mention the non-English docs as well as ports.

24 years agoFix description of argv[0] passed to interpreted scripts; it's the name of
Ben Smithurst [Sat, 5 Aug 2000 16:38:24 +0000 (16:38 +0000)]
Fix description of argv[0] passed to interpreted scripts; it's the name of
the interpreter, _not_ the argv[0] passed in the original exec() call.

24 years agoFix world-breakage: warnx("..') --> warnx("...")
Alexander Langer [Sat, 5 Aug 2000 15:45:59 +0000 (15:45 +0000)]
Fix world-breakage: warnx("..') --> warnx("...")

Submitted by: Alain Thivillon <Alain.Thivillon@hsc.fr>,
Udo Erdelhoff <ue@nathan.ruhr.de>
Approved by: green

24 years agoDon't call errx() without a format string, to protect against possible
Kris Kennaway [Sat, 5 Aug 2000 06:24:41 +0000 (06:24 +0000)]
Don't call errx() without a format string, to protect against possible
% characters in localized error messages from ipsec_strerror().

Obtained from:  OpenBSD

24 years agoDon't call errx() without a format string, to protect against possible
Kris Kennaway [Sat, 5 Aug 2000 06:21:12 +0000 (06:21 +0000)]
Don't call errx() without a format string, to protect against possible
localized error messages from ipsec_strerror().

Obtained from: OpenBSD

24 years agoDon't call warnx() without a format string (localized error messages
Kris Kennaway [Sat, 5 Aug 2000 06:06:48 +0000 (06:06 +0000)]
Don't call warnx() without a format string (localized error messages
could conceivably cause a crash).

Obtained from: OpenBSD

24 years agoPrevent internal buffer overflow due to expansion of $LANG and/or name
Kris Kennaway [Sat, 5 Aug 2000 04:56:43 +0000 (04:56 +0000)]
Prevent internal buffer overflow due to expansion of $LANG and/or name
argument in catopen().

Reviewed by: chris, -audit

24 years agoArgh! Fix a brainfart of mine. In the old boot0, we relocated ourself
John Baldwin [Fri, 4 Aug 2000 22:37:21 +0000 (22:37 +0000)]
Argh!  Fix a brainfart of mine.  In the old boot0, we relocated ourself
to 0x600 via a 'rep movsw'.  Once that was done, %cx was zero, so we could
simply use 'movb' to update the lower byte of %cx in preparation for
zeroing out the fake partition entry used to boot to other drives via F5.
Well, in the new boot0, we don't actually relocate ourselves, instead it
is easier to create the fake partition entry first and then just use it to
get the BIOS to load all of boot0 into memory at 0x600.  However, since we
aren't doing the relocate code anymore, we don't know that %cx == 0 when
we hit the 'movb' to setup %cx for clearning the fake partition entry.
Thus, if %ch != 0 when the BIOS started boot0, then it would end up zeroing
a lot more memory than just 8 words.  The solution is to do a word move of
$8 into %cx.

Debugging help from: David Wolfskill <dhw@whistle.com>

24 years agoExplain that in order to track -STABLE, you should leave the src-all
Jim Mock [Fri, 4 Aug 2000 22:29:52 +0000 (22:29 +0000)]
Explain that in order to track -STABLE, you should leave the src-all
collection uncommented.

PR: 20398
Submitted by: Jeff Blaine <jeffblaine@mediaone.net>
Reviewed by: jhb

24 years agoRemove everything resembling a comment from "refuse" and put the
John Polstra [Fri, 4 Aug 2000 21:32:52 +0000 (21:32 +0000)]
Remove everything resembling a comment from "refuse" and put the
explanations into a new file "refuse.README".  Some users are simply
copying these files and expecting them to work -- without even
reading them.  I don't want to spend any more time closing bogus
PRs from that.

Also correct an error or two in the patterns.

24 years agoIf caller have privilege to r/w access to /dev/apm, show the
Hajimu UMEMOTO [Fri, 4 Aug 2000 20:57:31 +0000 (20:57 +0000)]
If caller have privilege to r/w access to /dev/apm, show the
info about resume timer and resume on ring indicator, again.

Reviewed by: imp

24 years agoCleanup debug messages and Add some enhancements from linux on
Mitsuru IWASAKI [Fri, 4 Aug 2000 20:28:53 +0000 (20:28 +0000)]
Cleanup debug messages and Add some enhancements from linux on
display control by apm -d.
 - Remove APM_DEBUG to avoid re-build kernel with such a unspported optioin.
 - Introduce new denug flag `debug.apm_debug' which can be controlled by
   sysctl interface and loader by setting like "debug.apm_debug=1", you
   will get debug messages from APM driver.
 - Add some enhancements from linux on display control by apm -d.  I'm
   expecting that we can see some improvements on some laptops where
   apm -d doesn't work correctly so far.

Reviewed by: -mobile and -current folks (no objection)
Suggested by: Susumu WAKABAYASHI <susumu@wakabaya.net>

24 years agoFinally make the module dependencies work. kern_linker.c now allows
Nick Hibma [Fri, 4 Aug 2000 19:05:49 +0000 (19:05 +0000)]
Finally make the module dependencies work. kern_linker.c now allows
modules to depend on modules in the same file (uhub depends on usb) or
even on themselves (usb on usb, makes the define in usb_port.h a lot
less convoluted).

Use ANSI prototypes.

24 years ago*smack* #if 0, not #ifdef 0.
Bill Paul [Fri, 4 Aug 2000 18:17:45 +0000 (18:17 +0000)]
*smack* #if 0, not #ifdef 0.

24 years agoThe check_extport() function appears to have some new code in it that checks
Bill Paul [Fri, 4 Aug 2000 18:09:56 +0000 (18:09 +0000)]
The check_extport() function appears to have some new code in it that checks
the scratch RAM for data normally found in the SEEPROM (presumably in the
event that the SEEPROM is unavailable or can't be read). This code causes
a spontaneous reboot on monster.osd.bsdi.com, which has an embedded aic7880
controller. The problem appears to happen either when it writes to the
SCBPTR port and then reads from the SCB_CONTROL port. Somewhere during
the inb/outb operations, the system has a heart attack and restarts.

This code looks very suspicious, particularly since it has unconditionalized
debug mesages such as "Got here!" and "And it even worked!". With this
block #ifdef'ed out, the machine boots and runs properly. I stronly suggest
that it stay #ifdef'ed out until it's properly tested.