]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoThere seem to be two messages that were added with soft-updates
Poul-Henning Kamp [Tue, 4 Aug 1998 09:19:03 +0000 (09:19 +0000)]
There seem to be two messages that were added with soft-updates
support, which need a final "\n". I only observed one line of
mangled output, but I think there is another one which suffers
from the same problem, and thus I provide a patch that covers
both.

PR: 7483
Reviewed by: phk
Submitted by: Stefan Esser <se@FreeBSD.org>

26 years agoImplemented a more sophisticated mechanism for handling transmitter
David Greenman [Tue, 4 Aug 1998 08:53:12 +0000 (08:53 +0000)]
Implemented a more sophisticated mechanism for handling transmitter
interrupts which now defers them until the transmit queue if filled
up with completed buffers. This has two advantages: first, it reduces
the number of transmitter interrupts to just 1/120th of the rate
that they occured previously, and two, running down many buffers
at once has much improved cache effects.

26 years agoFix this thing to work properly with multiple tlan adapters; increment
Bill Paul [Tue, 4 Aug 1998 01:38:52 +0000 (01:38 +0000)]
Fix this thing to work properly with multiple tlan adapters; increment
unit count correctly.

Problem reported by Larry Baird <lab@gta.com>.

26 years agoCorrect an irritating documentation error.
Mike Smith [Tue, 4 Aug 1998 00:39:45 +0000 (00:39 +0000)]
Correct an irritating documentation error.
(get_diskinfo is unnecessarily obfuscated)

26 years agoCopy in the nfs_diskless structure if NFS_ROOT is defined. A previous
Mike Smith [Mon, 3 Aug 1998 21:31:32 +0000 (21:31 +0000)]
Copy in the nfs_diskless structure if NFS_ROOT is defined.  A previous
change to include nfs_root.h precluded NFS from being defined.
Submitted by: Parag Patel <parag@cgt.com>

26 years agokeep more up to date..
Julian Elischer [Mon, 3 Aug 1998 19:18:08 +0000 (19:18 +0000)]
keep more up to date..
fixit uses the same crunch.conf as normal generation
upgrade boot floppy to 1.44MB so that it fits

26 years agoMajor ppbus updates from the author.
Mike Smith [Mon, 3 Aug 1998 19:14:33 +0000 (19:14 +0000)]
Major ppbus updates from the author.

 - ppbus now supports PLIP via the if_plip driver
 - ieee1284 infrastructure added, including parallel-port PnP
 - port microsequencer added, for scripting the sort of port I/O
   that is common with parallel devices without endless calls up and down
   through the driver structure.
 - improved bus ownership behaviour among the ppbus-using drivers.
 - improved I/O chipset feature detection

The vpo driver is now implemented using the microsequencer, leading to
some performance improvements as well as providing an extensive example
of its use.

Reviewed by: msmith
Submitted by: Nicolas Souchu <Nicolas.Souchu@prism.uvsq.fr>

26 years agoFixed bitrot in prototypes in synopsis.
Bruce Evans [Mon, 3 Aug 1998 17:52:27 +0000 (17:52 +0000)]
Fixed bitrot in prototypes in synopsis.

26 years agoFIxed missing include in synopsis.
Bruce Evans [Mon, 3 Aug 1998 17:29:44 +0000 (17:29 +0000)]
FIxed missing include in synopsis.

26 years agoFIxed bitrot in prototype in synopsis.
Bruce Evans [Mon, 3 Aug 1998 17:27:41 +0000 (17:27 +0000)]
FIxed bitrot in prototype in synopsis.

26 years agoUse explicitly sized types when digging through packet headers.
Doug Rabson [Mon, 3 Aug 1998 17:23:37 +0000 (17:23 +0000)]
Use explicitly sized types when digging through packet headers.

Reviewed by: Julian Elischer <julian@whistle.com>

26 years agoChanged prototype in synopsis to match prototype in <pthread.h>.
Bruce Evans [Mon, 3 Aug 1998 16:54:51 +0000 (16:54 +0000)]
Changed prototype in synopsis to match prototype in <pthread.h>.

26 years agoFixed missing header in synopsis.
Bruce Evans [Mon, 3 Aug 1998 16:29:31 +0000 (16:29 +0000)]
Fixed missing header in synopsis.

26 years agoisnetworktty -> isnettty. ttyent.h has already gone through 3
Bruce Evans [Mon, 3 Aug 1998 15:35:41 +0000 (15:35 +0000)]
isnetworktty -> isnettty.  ttyent.h has already gone through 3
revisions to match the confusing spelling in getttyent.c (1 to
break it to match the man page and 1 in each of 2 branches to fix
it).  This function seems to be orphaned and unused.

26 years agoAdded the `saver' key definition. The saver key will activate the screen
Kazutaka YOKOTA [Mon, 3 Aug 1998 11:33:22 +0000 (11:33 +0000)]
Added the `saver' key definition.  The saver key will activate the screen
saver if one is loaded.

26 years ago1. Reorganized screen saver related code so that both the LKM screen
Kazutaka YOKOTA [Mon, 3 Aug 1998 11:30:45 +0000 (11:30 +0000)]
1. Reorganized screen saver related code so that both the LKM screen
saver and splash screen can all work properly with syscons.  Note that
the splash screen option (SC_SPLASH_SCREEN) does not work yet, as it
requires additional code from msmith.

- Reorganized the splash screen code to match the latest development
  in this area.
- Delay screen switch in `switch_scr()' until the screen saver is
  stopped, if one is running,
- Start the screen saver immediately, if any, when the `saver' key is
  pressed. (There will be another commit for `kbdcontrol' to support
  this keyword in the keymap file.)
- Do not always stop the screen saver when mouse-related ioctls
  are called.  Stop it only if the mouse is moved or buttons are
  clicked; don't stop it if any other mouse ioctls are called.

2. Added provision to write userland screen savers.  (Contact me if you
are interested in writing one.)

- Added CONS_IDLE, CONS_SAVERMODE, and CONS_SAVERSTART ioctls to
  support userland screen savers.

3. Some code clean-ups.

26 years agoFixed garbage after comment after #endif. Fixed reversed sense of same
Bruce Evans [Mon, 3 Aug 1998 10:59:02 +0000 (10:59 +0000)]
Fixed garbage after comment after #endif.  Fixed reversed sense of same
comment.

26 years agoSync with sys/i386/isa/syscons.h revision 1.38.
KATO Takenori [Mon, 3 Aug 1998 10:50:57 +0000 (10:50 +0000)]
Sync with sys/i386/isa/syscons.h revision 1.38.

26 years agoSync with sys/i386/isa/syscons.c revision up to 1.271.
KATO Takenori [Mon, 3 Aug 1998 10:50:27 +0000 (10:50 +0000)]
Sync with sys/i386/isa/syscons.c revision up to 1.271.

26 years agoSync with sys/i386/i386/userconfig.c revision 1.106.
KATO Takenori [Mon, 3 Aug 1998 10:49:18 +0000 (10:49 +0000)]
Sync with sys/i386/i386/userconfig.c revision 1.106.

26 years ago- When the system is shut down, switch to the vty0 if possible.
Kazutaka YOKOTA [Mon, 3 Aug 1998 09:18:58 +0000 (09:18 +0000)]
- When the system is shut down, switch to the vty0 if possible.
- Don't try to ring bell when system is going down. Beacuse the clock
  code is about to be stopped, the timeout routine won't be called
  anymore.

26 years agoFix the bug which always reallocated the cut buffer whenever
Kazutaka YOKOTA [Mon, 3 Aug 1998 09:17:06 +0000 (09:17 +0000)]
Fix the bug which always reallocated the cut buffer whenever
the screen mode is changed even if another vty has larger size.
Reallocate the buffer only when the new screen size is larger than
the current cut buffer size.

26 years agoDon't accept the blank time value, if it is too big.
Kazutaka YOKOTA [Mon, 3 Aug 1998 09:15:36 +0000 (09:15 +0000)]
Don't accept the blank time value, if it is too big.

PR: bin/6188

26 years ago- Set bell pitch in hertz and duration in msecs.
Kazutaka YOKOTA [Mon, 3 Aug 1998 09:12:39 +0000 (09:12 +0000)]
- Set bell pitch in hertz and duration in msecs.
  The change corresponds to syscons.c 1.268.

PR: bin/6037
Submitted by: Kouichi Hirabayashi (kh@eve.mogami-wire.co.jp)

26 years ago- Add new bell types: "quiet.normal" and "quiet.visual".
Kazutaka YOKOTA [Mon, 3 Aug 1998 09:09:35 +0000 (09:09 +0000)]
- Add new bell types: "quiet.normal" and "quiet.visual".
  When bell is of "quiet" types, the console won't ring (or flush)
  if the ringing process is in a background vty.
  PR: i386/2853

- Modify the escape sequence 'ESC[=%d;%dB' so that bell pitch and
  duration are set in hertz and msecs by kbdcontrol(1).
  There will be a corresponding kbdcontrol patch.
  PR: bin/6037
  Submitted by: Kouichi Hirabayashi (kh@eve.mogami-wire.co.jp)

26 years agoFixed building -current under 2.2.6 using `make world'. Moved some
Bruce Evans [Mon, 3 Aug 1998 08:28:14 +0000 (08:28 +0000)]
Fixed building -current under 2.2.6 using `make world'.  Moved some
recently added definitions from sys.mk to bsd.own.mk.  Include the
src-relative bsd.own.mk in src/Makefile to pick up all new definitions.
Don't check that MACHINE_ARCH is defined in src/Makefile, since it is
(and should have been) guaranteed to be defined.

26 years agoUse errx() instead of err() for malloc failures. -Wall. Do not dot
Philippe Charnier [Mon, 3 Aug 1998 06:44:46 +0000 (06:44 +0000)]
Use errx() instead of err() for malloc failures. -Wall. Do not dot
terminate errx() string. Remove unused #includes. Use .Tn for NFS.

26 years agoDocument -n (soft-update) flag.
Philippe Charnier [Mon, 3 Aug 1998 06:41:20 +0000 (06:41 +0000)]
Document -n (soft-update) flag.
Add rcsid, remove unused #includes. Sync usage() and SYNOPSIS.

26 years ago.Nm swapon -> .Nm.
Philippe Charnier [Mon, 3 Aug 1998 06:35:01 +0000 (06:35 +0000)]
.Nm swapon -> .Nm.
Sort #includes. Add rcsid.

26 years agoCorrect use of .Nm and .Ev.
Philippe Charnier [Mon, 3 Aug 1998 06:32:57 +0000 (06:32 +0000)]
Correct use of .Nm and .Ev.
Add rcsid, rmove unused #include. Spelling.

26 years agoTypo: i. e. -> i.e.
Philippe Charnier [Mon, 3 Aug 1998 06:24:59 +0000 (06:24 +0000)]
Typo: i. e. -> i.e.
Remove unused #include.

26 years agoCorrect use of .Nm. Add rcsid. Do not dot terminate err(3) strings. Spelling.
Philippe Charnier [Mon, 3 Aug 1998 06:22:43 +0000 (06:22 +0000)]
Correct use of .Nm. Add rcsid. Do not dot terminate err(3) strings. Spelling.

26 years agoCorrect use of .Nm, add .An.
Philippe Charnier [Mon, 3 Aug 1998 06:19:47 +0000 (06:19 +0000)]
Correct use of .Nm, add .An.

26 years agoThis commit was generated by cvs2svn to compensate for changes in r38032,
Peter Wemm [Mon, 3 Aug 1998 05:56:20 +0000 (05:56 +0000)]
This commit was generated by cvs2svn to compensate for changes in r38032,
which included commits to RCS files with non-trunk default branches.

26 years agoImport sendmail-8.9.1 (slightly trimmed) onto a fresh branch under
Peter Wemm [Mon, 3 Aug 1998 05:56:20 +0000 (05:56 +0000)]
Import sendmail-8.9.1 (slightly trimmed) onto a fresh branch under
src/contrib as per various discussions.  I will copy across our changes
and then point the Makefiles across once the dust has settled..

26 years agoIgnore (effectively) SIGTSTP if we're already in the process of tstoping.
Tim Vanderhoek [Mon, 3 Aug 1998 03:11:09 +0000 (03:11 +0000)]
Ignore (effectively) SIGTSTP if we're already in the process of tstoping.

26 years agoUpdates for the ThunderLAN driver:
Bill Paul [Mon, 3 Aug 1998 01:33:12 +0000 (01:33 +0000)]
Updates for the ThunderLAN driver:

- probe for PHYs by checking the BMSR (phy status) register instead
  of the vendor ID register.

- fix the autonegotiation routine so that it figures out the autonegotiated
  modes correctly.

- add tweaks to support the Olicom OC-2326 now that I've actually had
  a chance to test one

o Olicom appears to encode the ethernet address in the EEPROM
  in 16-bit chunks in network byte order. If we detect an
  Olicom card (based on the PCI vendor ID), byte-swap the station
  address accordingly.

  XXX The Linux driver does not do this. I find this odd since
  the README from the Linux driver indicates that patches to
  support the Olicom cards came from somebody at Olicom; you'd
  think if anyone would get that right, it'd be them. Regardless,
  I accepted the word of the disgnoatic program that came bundled
  with the card as gospel and fixed the attach routine to make
  the station address match what it says.

o The version of the 2326 card that I got for testing is a
  strange beast: the card does not look like the picture on
  the box in which it was packed. For one thing, the picture
  shows what looks like an external NS 83840A PHY, but the
  actual card doesn't have one. The card has a TNETE100APCM
  chip, which appears to have not only the usual internal
  tlan 10Mbps PHY at MII address 32, but also a 10/100 PHY
  at MII address 0. Curiously, this PHY's vendor and device ID
  registers always return 0x0000. I suspect that this is
  a mutant version of the ThunderLAN chip with 100Mbps support.
  This combination behaves a little strangely and required the
  following changes:

- The internal PHY has to be enabled in tl_softreset().
- The internal PHY doesn't seem to come to life after
  detecting the 100Mbps PHY unless it's reset twice.
- If you want to use 100Mbps modes, you have to isolate
  the internal PHY.
- If you want to use 10Mbps modes, you have to un-isolate
  the internal PHY.

The latter two changes are handled at the end of tl_init(): if
the PHY vendor ID is 0x0000 (which should not be possible if we
have a real external PHY), then tl_init() forces the internal
PHY's BMCR register to the proper values.

26 years agoThe pthreads standard has been published. Change:
Alexander Langer [Mon, 3 Aug 1998 00:58:37 +0000 (00:58 +0000)]
The pthreads standard has been published.  Change:
   ...is expected to conform to IEEE (``POSIX'') Std 1003.1c when it is
   published.
to:
   ...conforms to ISO/IEC 9945-1 ANSI/IEEE (``POSIX'') Std 1003.1 Second
   Edition 1996-07-12.

Discussed with: jb

26 years agoA style fix for my previous commit.
Alexander Langer [Sun, 2 Aug 1998 23:07:25 +0000 (23:07 +0000)]
A style fix for my previous commit.

26 years agoAdd missing "else", without whom the prev. commit is a null change.
Tim Vanderhoek [Sun, 2 Aug 1998 22:47:11 +0000 (22:47 +0000)]
Add missing "else", without whom the prev. commit is a null change.
Fix a 2nd level indentation style bug.

26 years agoFixed a race condition during the first lock/trylock of a statically
Alexander Langer [Sun, 2 Aug 1998 17:04:25 +0000 (17:04 +0000)]
Fixed a race condition during the first lock/trylock of a statically
initialized mutex.  Statically initialized mutexes are actually
initialized at first use (pthread_mutex_lock/pthread_mutex_trylock).
To prevent concurrent initialization by multiple threads, all
static initializations are now serialized by a spinlock.

Reviewed by: jb

26 years agoFixed printf format errors.
Bruce Evans [Sun, 2 Aug 1998 16:44:18 +0000 (16:44 +0000)]
Fixed printf format errors.

26 years agoFixed printf format errors.
Bruce Evans [Sun, 2 Aug 1998 16:06:35 +0000 (16:06 +0000)]
Fixed printf format errors.

26 years agoFixed printf format errors.
Bruce Evans [Sun, 2 Aug 1998 15:18:45 +0000 (15:18 +0000)]
Fixed printf format errors.

26 years agoFixed printf format errors. gcc should have a macro HOST_PTR_PRINTF_TYPE
Bruce Evans [Sun, 2 Aug 1998 14:57:00 +0000 (14:57 +0000)]
Fixed printf format errors.  gcc should have a macro HOST_PTR_PRINTF_TYPE
to go with its HOST_PTR_PRINTF[_FORMAT], since if %p is wrong for the
format then `void *' is probably wrong for the type.

26 years agoFixed printf format errors.
Bruce Evans [Sun, 2 Aug 1998 14:41:34 +0000 (14:41 +0000)]
Fixed printf format errors.

26 years agoFixed printf format warnings for non-errors. This change is null unless
Bruce Evans [Sun, 2 Aug 1998 14:35:35 +0000 (14:35 +0000)]
Fixed printf format warnings for non-errors.  This change is null unless
a system header defines a macro __printf0like() using the new printf0
format attribute.  uucp's internal ulog() function isn't actually
printf-like but uucp normally declares it as such.

26 years agoFixed printf format errors (time_t is not necessarily long).
Bruce Evans [Sun, 2 Aug 1998 14:25:24 +0000 (14:25 +0000)]
Fixed printf format errors (time_t is not necessarily long).

26 years agoOops, the SRCDIR=${.OBJDIR} hack is still necessary, (since bsd.doc.mk
Bruce Evans [Sun, 2 Aug 1998 14:21:35 +0000 (14:21 +0000)]
Oops, the SRCDIR=${.OBJDIR} hack is still necessary, (since bsd.doc.mk
has to cd to ${SRCDIR} to support inclusion messes elsewhere).
Fortunately the previous commit forgot to remove it.

26 years agoCleaned up. tmac.srefs was in both SRCS and EXTRAS. The
Bruce Evans [Sun, 2 Aug 1998 14:07:27 +0000 (14:07 +0000)]
Cleaned up.  tmac.srefs was in both SRCS and EXTRAS.  The
SRCDIR=${.OBDDIR} hack is now unnecessary.  It was never necessary
to use so many explicit paths to ${.OBJDIR} and ${.CURDIR}.

26 years agoDon't use `::' to break dependencies. Cleaned some nearby things.
Bruce Evans [Sun, 2 Aug 1998 14:00:39 +0000 (14:00 +0000)]
Don't use `::' to break dependencies.  Cleaned some nearby things.

26 years agoUse a different hack in the ${DFILE} rule to keep indirect dependencies
Bruce Evans [Sun, 2 Aug 1998 13:50:23 +0000 (13:50 +0000)]
Use a different hack in the ${DFILE} rule to keep indirect dependencies
(mostly for includes) separate from direct dependencies (so that ${.ALLSRC}
can be used to find full paths to the sources for the direct dependencies
only).  The `::' hack just forgot the indirect dependendencies.  This
broke building doc/usd/13.viref with `make -jN' - the index got corrupted
by being built twice concurrently.

Cleaned up the ${DFILE} rule.  There was a .else clause with dead code in
it following a .else clause (make accepts this bad syntax).  ${.ALLSRC}
now works in the USE_SOELIMPP case.  Some client Makefiles no longer need
the SRCDIR=${.OBJDIR} hack.

26 years agoPR: 7469
Brian Somers [Sun, 2 Aug 1998 13:01:16 +0000 (13:01 +0000)]
PR: 7469
Be careful that the current or next prompt in the list
that we're iterating through doesn't get changed by
descriptor_Read().

26 years agoAdd the ISP Qlogic SCSI card to the list of known devices.
Gary Palmer [Sun, 2 Aug 1998 09:32:52 +0000 (09:32 +0000)]
Add the ISP Qlogic SCSI card to the list of known devices.

26 years agoMalformed conditional if MACHINE_ARCH is not defined. From rev 1.195.
Wolfram Schneider [Sun, 2 Aug 1998 09:24:59 +0000 (09:24 +0000)]
Malformed conditional if MACHINE_ARCH is not defined. From rev 1.195.

26 years agoOn receipt of a sig 15, do a full bundle_Close() even if
Brian Somers [Sun, 2 Aug 1998 06:56:40 +0000 (06:56 +0000)]
On receipt of a sig 15, do a full bundle_Close() even if
our current phase is PHASE_DEAD as it's possible that there
are links stuck in OPENING waiting for a redial timer.

26 years agoAdded 82555B support for media status.
David Greenman [Sun, 2 Aug 1998 00:33:38 +0000 (00:33 +0000)]
Added 82555B support for media status.

26 years agoReordered fxp_softc for optimal cacheline behavior.
David Greenman [Sun, 2 Aug 1998 00:29:15 +0000 (00:29 +0000)]
Reordered fxp_softc for optimal cacheline behavior.

26 years agoKilled unused variable/assignment.
David Greenman [Sun, 2 Aug 1998 00:28:45 +0000 (00:28 +0000)]
Killed unused variable/assignment.

26 years agoIntroduce KRB5_HOME variable.
Dima Ruban [Sat, 1 Aug 1998 19:55:32 +0000 (19:55 +0000)]
Introduce KRB5_HOME variable.

26 years agoAvoid trying to malloc > (1<<32) bytes of memory due to an arithmetic
Doug Rabson [Sat, 1 Aug 1998 18:03:28 +0000 (18:03 +0000)]
Avoid trying to malloc > (1<<32) bytes of memory due to an arithmetic
underflow on the alpha.

26 years agoAdded iso646.h as defined by the Single UNIX Specification, version 2.
Alexander Langer [Sat, 1 Aug 1998 13:42:12 +0000 (13:42 +0000)]
Added iso646.h as defined by the Single UNIX Specification, version 2.

26 years agoSync with sys/i386/boot/netboot/main.c revision 1.23.
KATO Takenori [Sat, 1 Aug 1998 12:20:59 +0000 (12:20 +0000)]
Sync with sys/i386/boot/netboot/main.c revision 1.23.

26 years agoFix reference to /usr/share/misc/iso3166.
Dag-Erling Smørgrav [Sat, 1 Aug 1998 09:41:09 +0000 (09:41 +0000)]
Fix reference to /usr/share/misc/iso3166.

PR: misc/7459

26 years agoIf we get an ENOBUFS from the network, it's normally transient network
Peter Wemm [Sat, 1 Aug 1998 09:04:02 +0000 (09:04 +0000)]
If we get an ENOBUFS from the network, it's normally transient network
interface congestion (eg: nfs over a ppp link, etc).  Don't log these
for UDP mounts, and don't cause syscalls to fail with EINTR.
This stops the 'nfs send error 55' warnings.

If the error is because the system is really hosed, this is the least
of your problems...

26 years agoFix a compile error if IPFIREWALL_FORWARD active without IPDIVERT.
Peter Wemm [Sat, 1 Aug 1998 08:44:33 +0000 (08:44 +0000)]
Fix a compile error if IPFIREWALL_FORWARD active without IPDIVERT.

26 years agoWhen we receive an LCP config REQ in Opened state,
Brian Somers [Sat, 1 Aug 1998 01:02:41 +0000 (01:02 +0000)]
When we receive an LCP config REQ in Opened state,
do TLD *before* processing the config request as
TLD initialises the peers LCP values.

It's strange that an IRC isn't required here - but
I'll bow to the wisdom of the rfc.

26 years agoCosmetic: Diagnostic wording (Unknown -> Unexpected)
Brian Somers [Sat, 1 Aug 1998 01:02:12 +0000 (01:02 +0000)]
Cosmetic: Diagnostic wording (Unknown -> Unexpected)

26 years agoAlways dial immediately on ``open'', ``dial'' and ``call''.
Brian Somers [Fri, 31 Jul 1998 19:50:24 +0000 (19:50 +0000)]
Always dial immediately on ``open'', ``dial'' and ``call''.
We don't need a ``!''.

26 years agoCosmetic change to driver registration.
Doug Rabson [Fri, 31 Jul 1998 09:20:35 +0000 (09:20 +0000)]
Cosmetic change to driver registration.

26 years agoMake it possible to have a driver registered with more than one parent bus.
Doug Rabson [Fri, 31 Jul 1998 09:18:52 +0000 (09:18 +0000)]
Make it possible to have a driver registered with more than one parent bus.

26 years agoOnly perform byte/word accesses in the correct EV56 memory space.
Doug Rabson [Fri, 31 Jul 1998 09:17:51 +0000 (09:17 +0000)]
Only perform byte/word accesses in the correct EV56 memory space.
Sparse macros have moved to <machine/swiz.h>.
Fix sparse memory access so that it actually works as intended.
Tidy up sparse configuration access slightly.

26 years agoAdd definitions of EV56 INT1/INT2/INT4/INT8 memory spaces.
Doug Rabson [Fri, 31 Jul 1998 09:14:49 +0000 (09:14 +0000)]
Add definitions of EV56 INT1/INT2/INT4/INT8 memory spaces.

26 years agoMerge with i386/isa/diskslice_machdep.c 1.30
Doug Rabson [Fri, 31 Jul 1998 09:13:25 +0000 (09:13 +0000)]
Merge with i386/isa/diskslice_machdep.c 1.30

26 years agoI've put together man pages for the pthread_cleanup, pthread_cond, and
Poul-Henning Kamp [Fri, 31 Jul 1998 09:09:19 +0000 (09:09 +0000)]
I've put together man pages for the pthread_cleanup, pthread_cond, and
pthread_mutex routines. I've also tweaked pthread_create.3 to point to
pthread_cleanup_push(3) and pthread_cleanup_pop(3).

PR: 7450
Submitted by: Brian Cully <shmit@kublai.com>

26 years agoTFTP_ACK goes to port 69 (default tftp port) and tftpd says 'Timeout'.
Poul-Henning Kamp [Fri, 31 Jul 1998 09:03:15 +0000 (09:03 +0000)]
TFTP_ACK goes to port 69 (default tftp port) and tftpd says 'Timeout'.
It will be sent to source port of received packet.

PR: 7442
Reviewed by: phk
Submitted by: Sergey Shkonda <serg@bcs.zp.ua>

26 years agoIn sys/scsi/st.c, st_close() does not check for errors from calling
Poul-Henning Kamp [Fri, 31 Jul 1998 09:00:39 +0000 (09:00 +0000)]
In sys/scsi/st.c, st_close() does not check for errors from calling
st_write_filemarks().  This means that it is possible to write a file
on the tape for which all the writes and the close return without
errors, but the all bits did not make it onto the tape.

PR: 741
Reviewed by: phk
Submitted by: Andrew Heybey <ath@niksun.com>

26 years agondc hardcodes 'named' instead of using $named_program from rc.conf
Poul-Henning Kamp [Fri, 31 Jul 1998 08:47:16 +0000 (08:47 +0000)]
ndc hardcodes 'named' instead of using $named_program from rc.conf

PR: 7425
Reviewed by: phk
Submitted by: frf <frf@xocolatl.com>

26 years agoBacked out rev.1.31. It was a workaround for bugs in rev.1.4 of
Bruce Evans [Fri, 31 Jul 1998 07:22:31 +0000 (07:22 +0000)]
Backed out rev.1.31.  It was a workaround for bugs in rev.1.4 of
libc/gen/getpass.c.  The old behaviour of blocking SIGINT and not
changing SIGQUIT was restored in rev.1.5 of getpass.c.  The change
here completely restores the old behaviour of not supporting killing
login with keyboard signals (only) at the password prompt.  There
is no reason to support this, since login can be exited normally
by typing a couple of ^D's.  Login certainly shouldn't dump core
in response to user input.  Previously, SIGQUIT killed login
immediately but SIGINT killed it only after the password was
entered.

PR: 7444

26 years agoFix coredump when two signals are sent in rapid succession.
Tim Vanderhoek [Fri, 31 Jul 1998 04:05:11 +0000 (04:05 +0000)]
Fix coredump when two signals are sent in rapid succession.
PR: bin/5721
Submitted by: Oliver Fromme <oliver.fromme@heim3.tu-clausthal.de>

Also, add "volatile" to a variable modified by signal handlers (coincidentally,
the same variable involved in the above fix, although this isn't related
to the reported problem).

26 years agoFixed printf format errors.
Bruce Evans [Thu, 30 Jul 1998 17:40:45 +0000 (17:40 +0000)]
Fixed printf format errors.

26 years agoFixed printf format errors.
Bruce Evans [Thu, 30 Jul 1998 17:12:39 +0000 (17:12 +0000)]
Fixed printf format errors.

26 years agoUse the slice interfaces for SCSI cdroms. This centralizes handling
Bruce Evans [Thu, 30 Jul 1998 17:05:10 +0000 (17:05 +0000)]
Use the slice interfaces for SCSI cdroms.  This centralizes handling
of some disk ioctls and uniformizes bounds checking in the strategy
routine.  EOF handling got fixed as a side effect.  The changes
are similar to old changes for SCSI disks, except slices and labels
are not searched for on the disk and the in-core label has a few
more details (mostly just for backwards compatibility).  Bugs in
the in-core label had to be fixed to get dsopen() to accept it.
The slice interfaces had to support large sectors for all this to
work.

26 years agoAdded a flags arg to dsopen() and updated drivers. The DSO_ONESLICE
Bruce Evans [Thu, 30 Jul 1998 15:16:06 +0000 (15:16 +0000)]
Added a flags arg to dsopen() and updated drivers.  The DSO_ONESLICE
and DSO_NOLABELS flags prevent searching for slices and labels
respectively.  Current drivers don't set these flags.  When
DSO_NOLABELS is set, the in-core label for the whole disk is cloned
to create an in-core label for each slice.  This gives the correct
result (a good in-core label for the compatibility slice) if
DSO_ONESLICE is set or only one slice is found, but usually gives
broken labels otherwise, so DSO_ONESLICE should be set if DSO_NOLABELS
is set.

26 years agoSync with sys/i386/isa/fd.c revision 1.120.
KATO Takenori [Thu, 30 Jul 1998 09:01:12 +0000 (09:01 +0000)]
Sync with sys/i386/isa/fd.c revision 1.120.

26 years agoChange load address to 0xfffffc0000300000 to help support AS200.
Doug Rabson [Thu, 30 Jul 1998 08:12:14 +0000 (08:12 +0000)]
Change load address to 0xfffffc0000300000 to help support AS200.

26 years agoRemove spurious printf.
Doug Rabson [Thu, 30 Jul 1998 08:10:26 +0000 (08:10 +0000)]
Remove spurious printf.

26 years agoDocument behaviour of "-" and "--" on the command line.
Joseph Koshy [Thu, 30 Jul 1998 04:47:56 +0000 (04:47 +0000)]
Document behaviour of "-" and "--" on the command line.

PR: docs/5399

26 years agoDocument meanings of flags to the -r and -s options.
Joseph Koshy [Thu, 30 Jul 1998 04:17:12 +0000 (04:17 +0000)]
Document meanings of flags to the -r and -s options.

26 years agoadd anti-panic workaround from chris radek (cradek@in221.inetnebr.com)
Julian Elischer [Thu, 30 Jul 1998 03:22:52 +0000 (03:22 +0000)]
add anti-panic workaround from chris radek (cradek@in221.inetnebr.com)
Not sure why this is needed but but does stop crashes.

26 years agoTypo fix: teh --> (the|they)
Alexander Langer [Thu, 30 Jul 1998 02:27:41 +0000 (02:27 +0000)]
Typo fix: teh --> (the|they)

26 years agoRemove ppp 3000/tcp. Ppp has nothing in common with port 3000.
Brian Somers [Wed, 29 Jul 1998 19:30:44 +0000 (19:30 +0000)]
Remove ppp 3000/tcp.  Ppp has nothing in common with port 3000.

26 years agoAdd singlehand dvorak keyboard layouts, one for right hand use and
Søren Schmidt [Wed, 29 Jul 1998 18:49:45 +0000 (18:49 +0000)]
Add singlehand dvorak keyboard layouts, one for right hand use and
one for lefthand use.

Submitted by: "Chris Csanady" <cc@tarsier.ca.sandia.gov>

26 years agoFixed sign extension bugs awoken by changing speed_t to an unsigned
Bruce Evans [Wed, 29 Jul 1998 18:48:20 +0000 (18:48 +0000)]
Fixed sign extension bugs awoken by changing speed_t to an unsigned
type.  19200, 1200 and other relatively uninteresting speeds were
broken.

Submitted by: Rob Mallory <rmallory@qualcomm.com>

26 years agoOnly access an int for READU/WRITEU since that is what ptrace is declared to
Doug Rabson [Wed, 29 Jul 1998 18:41:30 +0000 (18:41 +0000)]
Only access an int for READU/WRITEU since that is what ptrace is declared to
return.

26 years agoDefault to FreeBSD if no brand detected. This makes life easier when
Doug Rabson [Wed, 29 Jul 1998 18:39:35 +0000 (18:39 +0000)]
Default to FreeBSD if no brand detected.  This makes life easier when
bootstrapping from NetBSD/alpha.

26 years agoFix msgbuf so that it actually works properly.
Doug Rabson [Wed, 29 Jul 1998 18:36:29 +0000 (18:36 +0000)]
Fix msgbuf so that it actually works properly.

26 years agoAllow an optional ``!'' in the open, dial & call commands.
Brian Somers [Wed, 29 Jul 1998 18:21:17 +0000 (18:21 +0000)]
Allow an optional ``!'' in the open, dial & call commands.
When used, the redial timer is ignored and the modem is
opened immediately.

26 years agoUse tcsetattr TCSANOW instead of TCSADRAIN so that dedicated mode doesn't
Brian Somers [Wed, 29 Jul 1998 18:20:53 +0000 (18:20 +0000)]
Use tcsetattr TCSANOW instead of TCSADRAIN so that dedicated mode doesn't
end up blocking indefinitely when there's no carrier.

26 years agoDon't try to dereference a pointer to freed memory.
Dag-Erling Smørgrav [Wed, 29 Jul 1998 17:50:06 +0000 (17:50 +0000)]
Don't try to dereference a pointer to freed memory.

PR: bin/7393