]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoTCPMSS adjusts all TCP SYN packets, not just outgoing ones.
Brian Somers [Mon, 16 Jul 2001 17:02:10 +0000 (17:02 +0000)]
TCPMSS adjusts all TCP SYN packets, not just outgoing ones.

Pointed out by: ru

23 years agoChange the interrupt handler a bit so that we call the nge_rxeof() handler
Bill Paul [Mon, 16 Jul 2001 16:35:09 +0000 (16:35 +0000)]
Change the interrupt handler a bit so that we call the nge_rxeof() handler
when we get an RX_ERR interrupt rather than the nge_rxeoc() handler. The
rxeoc (end of channel) handler attempts to reinitialize the whole NIC,
which we don't want to do if we only received a bad packet.

23 years agoNote which entries are created by sysinstall(8).
Nik Clayton [Mon, 16 Jul 2001 16:09:21 +0000 (16:09 +0000)]
Note which entries are created by sysinstall(8).

Requested by: ru

23 years agoMFS: Minor style and markup tweaks.
Bruce A. Mah [Mon, 16 Jul 2001 16:05:59 +0000 (16:05 +0000)]
MFS:  Minor style and markup tweaks.

23 years agoAdd an abstract for the README.
Bruce A. Mah [Mon, 16 Jul 2001 15:52:28 +0000 (15:52 +0000)]
Add an abstract for the README.

23 years agoFix style bugs introduced by rev 1.28. No functional changes.
Garrett Wollman [Mon, 16 Jul 2001 15:49:58 +0000 (15:49 +0000)]
Fix style bugs introduced by rev 1.28.  No functional changes.

23 years agoClarify that MAKEOBJDIR and MAKEOBJDIRPREFIX are environment variables.
Ruslan Ermilov [Mon, 16 Jul 2001 15:33:34 +0000 (15:33 +0000)]
Clarify that MAKEOBJDIR and MAKEOBJDIRPREFIX are environment variables.

Submitted by: bde

23 years agomdoc(7) police: Add strncpy() to the NAME section.
Ruslan Ermilov [Mon, 16 Jul 2001 12:47:34 +0000 (12:47 +0000)]
mdoc(7) police: Add strncpy() to the NAME section.

23 years agoGrammar: 'effect' is a noun and 'affect' is a verb.
Sheldon Hearn [Mon, 16 Jul 2001 10:52:19 +0000 (10:52 +0000)]
Grammar: 'effect' is a noun and 'affect' is a verb.

23 years agoSpelling: aquire -> acquire.
Ruslan Ermilov [Mon, 16 Jul 2001 07:37:22 +0000 (07:37 +0000)]
Spelling: aquire -> acquire.

23 years agoManpage for pfind() and zpfind().
Ruslan Ermilov [Mon, 16 Jul 2001 07:04:09 +0000 (07:04 +0000)]
Manpage for pfind() and zpfind().

Submitted by: Evan Sarmiento <kaworu@sektor7.ath.cx>

23 years agoUse PCPU_GET(cpuid) instead of curproc->p_oncpu.
Tor Egge [Mon, 16 Jul 2001 06:00:23 +0000 (06:00 +0000)]
Use PCPU_GET(cpuid) instead of curproc->p_oncpu.
Reviewed by: peter

23 years agoRemove unnecessary #include <stdlib.h>
Kris Kennaway [Mon, 16 Jul 2001 04:48:28 +0000 (04:48 +0000)]
Remove unnecessary #include <stdlib.h>

Obtained from: NetBSD
MFC After: 1 week

23 years agoOops, commit the version which actually works
Kris Kennaway [Mon, 16 Jul 2001 04:04:22 +0000 (04:04 +0000)]
Oops, commit the version which actually works

Obtained from: NetBSD
MFC After: 1 week

23 years agoAvoid a compile-time format string warning
Kris Kennaway [Mon, 16 Jul 2001 03:55:47 +0000 (03:55 +0000)]
Avoid a compile-time format string warning

Obtained from: NetBSD
MFC After: 1 week

23 years agoConnect std{in,out,err} to *something* so that "hooks" scripts that write
David E. O'Brien [Mon, 16 Jul 2001 01:56:48 +0000 (01:56 +0000)]
Connect std{in,out,err} to *something* so that "hooks" scripts that write
to std{err,out} will not spam /etc/resolv.conf.

Ted Lemon fixed the problem in version 3 of the client, but only for the
pre-daemonized case.  Thanks to Brian for pointing that out and helping
to make our future dhclient (v3) better.

Submitted by: brian

23 years agoUse a fixed type for times in on-disk structures for ufs rather than
Peter Wemm [Mon, 16 Jul 2001 00:55:27 +0000 (00:55 +0000)]
Use a fixed type for times in on-disk structures for ufs rather than
something that could potentially change like time_t.

23 years agoAdd myself
Kris Kennaway [Mon, 16 Jul 2001 00:17:48 +0000 (00:17 +0000)]
Add myself

23 years agoNow that config(8) has stopped breaking POLA (mkmakefile.c rev 1.69) by
David E. O'Brien [Sun, 15 Jul 2001 17:56:53 +0000 (17:56 +0000)]
Now that config(8) has stopped breaking POLA (mkmakefile.c rev 1.69) by
always including <KERNEL>.hints if its exist reguardless of the lack of a
"hints" directive in the <KERNEL> file; we can have a real hints file again.

23 years agodo not M_WAITOK in in6_update_ifa(), since this function can be called
Hajimu UMEMOTO [Sun, 15 Jul 2001 14:24:00 +0000 (14:24 +0000)]
do not M_WAITOK in in6_update_ifa(), since this function can be called
under splnet().  (some comment was added by KAME)

PR: 28927
MFC after: 1 week

23 years agoSort options in DESCRIPTION.
Dima Dorfman [Sun, 15 Jul 2001 14:00:19 +0000 (14:00 +0000)]
Sort options in DESCRIPTION.

23 years agoBump date for addition of -D.
Dima Dorfman [Sun, 15 Jul 2001 13:57:06 +0000 (13:57 +0000)]
Bump date for addition of -D.

23 years agoNew release notes: dirhash, new ddb(4) commands, and SSE.
Dima Dorfman [Sun, 15 Jul 2001 12:27:21 +0000 (12:27 +0000)]
New release notes: dirhash, new ddb(4) commands, and SSE.

23 years agoAdd prototypes, staticize, constify, don't shadow global variables and
Dima Dorfman [Sun, 15 Jul 2001 11:03:28 +0000 (11:03 +0000)]
Add prototypes, staticize, constify, don't shadow global variables and
functions, remove an unused function, and silence other assorted
warnings.

Set WARNS=2.

23 years agoSet WARNS=2 on programs which compile cleanly.
Dima Dorfman [Sun, 15 Jul 2001 10:30:46 +0000 (10:30 +0000)]
Set WARNS=2 on programs which compile cleanly.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoRemove whitespace at EOL.
Dima Dorfman [Sun, 15 Jul 2001 08:21:37 +0000 (08:21 +0000)]
Remove whitespace at EOL.

23 years agoRemove whitespace at EOL.
Dima Dorfman [Sun, 15 Jul 2001 08:06:20 +0000 (08:06 +0000)]
Remove whitespace at EOL.

23 years agoRemove whitespace at EOL.
Dima Dorfman [Sun, 15 Jul 2001 07:53:42 +0000 (07:53 +0000)]
Remove whitespace at EOL.

23 years agoRemove whitespace at EOL and add $FreeBSD$.
Dima Dorfman [Sun, 15 Jul 2001 07:48:01 +0000 (07:48 +0000)]
Remove whitespace at EOL and add $FreeBSD$.

23 years agoSet WARNS=2 on libraries that compile cleanly with it.
Dima Dorfman [Sun, 15 Jul 2001 06:46:02 +0000 (06:46 +0000)]
Set WARNS=2 on libraries that compile cleanly with it.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoConstify, add some prototypes, and set WARNS=2.
Dima Dorfman [Sun, 15 Jul 2001 06:20:45 +0000 (06:20 +0000)]
Constify, add some prototypes, and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoConstify, de-register-ify, and set WARNS=2.
Dima Dorfman [Sun, 15 Jul 2001 05:47:47 +0000 (05:47 +0000)]
Constify, de-register-ify, and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoConstify, de-register-ify, __unused-ify, and set WARNS=2.
Dima Dorfman [Sun, 15 Jul 2001 05:46:07 +0000 (05:46 +0000)]
Constify, de-register-ify, __unused-ify, and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoConstify and set WARNS=2.
Dima Dorfman [Sun, 15 Jul 2001 05:44:23 +0000 (05:44 +0000)]
Constify and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoFix a prototype and set WARNS=2.
Dima Dorfman [Sun, 15 Jul 2001 05:41:57 +0000 (05:41 +0000)]
Fix a prototype and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoChange lpd to perform sanity and safety checks on control files as it
Garance A Drosehn [Sun, 15 Jul 2001 05:29:38 +0000 (05:29 +0000)]
Change lpd to perform sanity and safety checks on control files as it
receives them from other hosts.  This is meant to protect from both
nefarious users (which maybe broke into some remote host that we accept
print jobs from), and broken implementations of lpr on other platforms.
This is done by changing recvjob.c to call the new ctl_renametf()
routine in the new common_source/ctlinfo.[ch] files.  This will not
affect jobs coming via lpr on the local machine.

Reviewed by: freebsd-print@bostonradio.org & freebsd-audit
MFC after: 16 days

23 years agoPut module directories one per line so adding new modules creates readable
Brooks Davis [Sun, 15 Jul 2001 04:55:37 +0000 (04:55 +0000)]
Put module directories one per line so adding new modules creates readable
diffs and produces less conflicts with local source trees.

23 years agoThe i386-specific includes in this file were "fixed" by bracketing them with
Benno Rice [Sun, 15 Jul 2001 04:11:51 +0000 (04:11 +0000)]
The i386-specific includes in this file were "fixed" by bracketing them with
#ifndef __alpha__.  Fix this for the rest of the world by turning it into
#ifdef __i386__.

Reviewed by: obrien

23 years agoReplace a call to 'alloca', thus avoiding an error when compiling on
Garance A Drosehn [Sun, 15 Jul 2001 04:10:32 +0000 (04:10 +0000)]
Replace a call to 'alloca', thus avoiding an error when compiling on
freebsd/alpha with -ansi (and on some non-fbsd platforms).  This change
can only affect the access checking of 'lpr -r'.

MFC after: 1 week

23 years agoGet rid of one compile-time warning by changing an 'int' to a 'size_t'.
Garance A Drosehn [Sun, 15 Jul 2001 03:10:54 +0000 (03:10 +0000)]
Get rid of one compile-time warning by changing an 'int' to a 'size_t'.

MFC after: 1 week

23 years agoGet lpr/pac.c to compile without warnings with BDECFLAGS
Garance A Drosehn [Sun, 15 Jul 2001 01:28:56 +0000 (01:28 +0000)]
Get lpr/pac.c to compile without warnings with BDECFLAGS

MFC after: 1 week

23 years agoReplace three subtly-wrong calls to strncpy with calls to strlcpy, and
Garance A Drosehn [Sun, 15 Jul 2001 00:57:18 +0000 (00:57 +0000)]
Replace three subtly-wrong calls to strncpy with calls to strlcpy, and
use (size_t) in a few places to avoid compile-time warnings on alpha.

MFC after: 1 week

23 years agoFix most of the warnings generated by compiling lpr with -Wnon-const-format,
Garance A Drosehn [Sun, 15 Jul 2001 00:09:46 +0000 (00:09 +0000)]
Fix most of the warnings generated by compiling lpr with -Wnon-const-format,
often by just telling gcc that some internal routine is "__printflike"
(work done by Kris Kennaway <kris@FreeBSD.org>).  Also fix the new warnings
which show up once gcc starts checking the "printf-like parameters" passed
to those routines.

MFC after: 1 week

23 years agoTurn off preemption on the Alpha for the time being. Peter and JHB aren't
David E. O'Brien [Sun, 15 Jul 2001 00:03:01 +0000 (00:03 +0000)]
Turn off preemption on the Alpha for the time being.  Peter and JHB aren't
sure when things got so bad (JHB says preemption worked just fine for months
before the AlbertVM commit).  Even post DillionVM locking commit, Miatas
(DEC Personal Workstations) are very fragile -- not making it thru a world
build.  With this patch it does.

Those hacking on SMPng will want to locally back out this commit.  The rest
of us will want to run with it until the SMPng guys figure out the problem(s).

Submitted by: peter

23 years agoFix assorted typos.
Daniel Harris [Sat, 14 Jul 2001 22:41:05 +0000 (22:41 +0000)]
Fix assorted typos.

23 years agoadd -a to usage()
Bill Fumerola [Sat, 14 Jul 2001 21:56:04 +0000 (21:56 +0000)]
add -a to usage()

Submitted by: Ashley Penney <ashp@unloved.org>
MFC after: 3 days

23 years agoChange signal-handling to reset SIGCHLD to SIGDFLT instead of SIG_IGN.
Garance A Drosehn [Sat, 14 Jul 2001 21:49:17 +0000 (21:49 +0000)]
Change signal-handling to reset SIGCHLD to SIGDFLT instead of SIG_IGN.
This fixes a problem with using print filters (if=, of=, etc) that showed
up in -current around June 20th.  That problem initially reported by
Georg-W Koltermann <gwk@sgi.com>, while most of the investigation that
led to this fix was done by Anton Berezin <tobez@FreeBSD.org>.

Reviewed by: freebsd-print@bostonradio.org
MFC after: 1 week

23 years agoFix MCOUNT_ENTER() so it actually compiles in the profiling case.
John Baldwin [Sat, 14 Jul 2001 21:40:53 +0000 (21:40 +0000)]
Fix MCOUNT_ENTER() so it actually compiles in the profiling case.

Pointy hat to: me
Submitted by: Danny J. Zerkel <dzerkel@columbus.rr.com>

23 years agoFix reboot hangs that have happened with multiple processors
Matt Jacob [Sat, 14 Jul 2001 21:37:57 +0000 (21:37 +0000)]
Fix reboot hangs that have happened with multiple processors
on Alpha 4100s.

Basically, if you're halting or you're rebooting, you should
tell all other processors to halt first. Define IPI_HALT- IPI_STOP
is not what we want for this purpose, which will call prom_halt(0)
on receipt.

The processor running the halt or reboot wil send an IPI_HALT to all
other processors, delay a bit, then continue to do what what it was
planning on doing (prom_halt({0|1})).

23 years agoMention the new behaviour of fdisk(8) (i386 only) to select the default
Joerg Wunsch [Sat, 14 Jul 2001 21:21:17 +0000 (21:21 +0000)]
Mention the new behaviour of fdisk(8) (i386 only) to select the default
device.

23 years agoMention new support for Cisco Aironet 350 series adaptors.
Brooks Davis [Sat, 14 Jul 2001 21:17:25 +0000 (21:17 +0000)]
Mention new support for Cisco Aironet 350 series adaptors.

23 years agoHrmpf. nyan's rev 1.25 commit to fdc.h crossed with my removal of it.
Joerg Wunsch [Sat, 14 Jul 2001 20:28:02 +0000 (20:28 +0000)]
Hrmpf.  nyan's rev 1.25 commit to fdc.h crossed with my removal of it.
Back-integrate his (PC98) changes into fd.c.

23 years agoYet another large non-function change.
Joerg Wunsch [Sat, 14 Jul 2001 20:25:01 +0000 (20:25 +0000)]
Yet another large non-function change.

. Integrate fdc.h into fd.c, with the removal of ft(4) there's no longer
  a reason to scatter things across two files.

. Sanitize comments.  Convert them into the style(9)-recommended
  multi-line form, make them sentences where apprpriate, etc.

. Declare all functions on top, and declare them in the order they
  appear in the file.  This order is totally chaotic, but Bruce
  convinced me that reordering the file wouldn't make it better either.

. Kill a `possibly uninitialized' warning (only seen with -O2) in
  fd_read_status().

. Make the comments at return (0|1) statements in fdstate() consistent.

. Nuke a ``keep the compiler happy'' dummy return at the end of fdstate(),
  gcc is smart enough to detect that it would never be reached anyway.

23 years agoRemoved whitespace at end-of-line; no content changes. I simply did
Jens Schweikhardt [Sat, 14 Jul 2001 19:41:16 +0000 (19:41 +0000)]
Removed whitespace at end-of-line; no content changes. I simply did
cd src/share; find man[1-9] -type f|xargs perl -pi -e 's/[ \t]+$//'

BTW, what editors are the culprits? I'm using vim and it shows
me whitespace at EOL in troff files with a thick blue block...

Reviewed by: Silence from cvs diff -b
MFC after: 7 days

23 years agoFix a memory leak in __rpcb_findaddr(), avoid compiler warnings.
Ian Dowse [Sat, 14 Jul 2001 18:18:23 +0000 (18:18 +0000)]
Fix a memory leak in __rpcb_findaddr(), avoid compiler warnings.

Submitted by: Martin Blapp <mb@imp.ch>

23 years agoThis is a selection of essentially cosmetic changes:
Ian Dowse [Sat, 14 Jul 2001 17:46:52 +0000 (17:46 +0000)]
This is a selection of essentially cosmetic changes:
- Use '\0' for a char instead of NULL.
- Explicitly compare against the global `nullstring' to determine if
  a non-NULL uaddr is not malloc'd.
- Remove some unnecessary casting of the argument to free().
- In rpcbproc_callit_com(), move the freeing of m_uaddr to the
  cleanup code at the end of the function.
- To avoid confusion and  possible alignment problems, change
  netbufdup() to allocate the netbuf struct and the sockaddr buffer
  separately, and change netbuffree() accordingly. This makes it
  produce netbufs that are consistent with all other netbufs in
  rpcbind.

23 years agoFix a memory leak in check_bound() by freeing the buffer area of
Ian Dowse [Sat, 14 Jul 2001 17:17:48 +0000 (17:17 +0000)]
Fix a memory leak in check_bound() by freeing the buffer area of
the netbuf before freeing the netbuf structure itself.

23 years agoAdd missing #include <stdio.h>.
Ian Dowse [Sat, 14 Jul 2001 17:10:58 +0000 (17:10 +0000)]
Add missing #include <stdio.h>.

Fix an off-by-one error in logit() when determining if a procedure
number has a known name.

23 years agoSimplify to bitmaskcmp() to use the obvious approach instead of
Ian Dowse [Sat, 14 Jul 2001 17:04:26 +0000 (17:04 +0000)]
Simplify to bitmaskcmp() to use the obvious approach instead of
comparing bit by bit.

Make the logic in in6_fillscopeid() match that in our ifconfig(8):
only set the scope ID if there is one in the address and none in
sin6_scope_id.

Correct a comment in network_init() that didn't make sense; it was
probably never updated after it was pasted from similar code in
addrmerge().

23 years agoUse snprintf instead of sprintf.
Ian Dowse [Sat, 14 Jul 2001 16:54:37 +0000 (16:54 +0000)]
Use snprintf instead of sprintf.

23 years agoAvoid a harmless compiler warning, and add a missing \n to a
Ian Dowse [Sat, 14 Jul 2001 16:32:44 +0000 (16:32 +0000)]
Avoid a harmless compiler warning, and add a missing \n to a
debugging fprintf.

Submitted by:   Martin Blapp <mb@imp.ch>

23 years agoAdd OPIE examples.
Mark Murray [Sat, 14 Jul 2001 08:46:56 +0000 (08:46 +0000)]
Add OPIE examples.

Tidy up columns.

Fix pam_wheel options.

23 years agoUse a better method of getting user credentials to account for
Mark Murray [Sat, 14 Jul 2001 08:42:39 +0000 (08:42 +0000)]
Use a better method of getting user credentials to account for
(legal) UID duplication.

Rename use_uid to auth_as_self for consistency with other modules.

23 years agoUse a better method to get user credentials to account for (legal)
Mark Murray [Sat, 14 Jul 2001 08:38:24 +0000 (08:38 +0000)]
Use a better method to get user credentials to account for (legal)
duplications of UID's in /etc/*passwd.

23 years agoSwitch on SHA1 hashing.
Mark Murray [Sat, 14 Jul 2001 08:34:26 +0000 (08:34 +0000)]
Switch on SHA1 hashing.

23 years agoFix SHA1 hashing.
Mark Murray [Sat, 14 Jul 2001 08:30:54 +0000 (08:30 +0000)]
Fix SHA1 hashing.

23 years agoUsse a better method to get a user's credentials, as uid's may be
Mark Murray [Sat, 14 Jul 2001 08:29:19 +0000 (08:29 +0000)]
Usse a better method to get a user's credentials, as uid's may be
legally duplicated.

Requested by: ache

23 years agoMove the hints gunk to a seperate file. It isn't really part of the
Peter Wemm [Sat, 14 Jul 2001 08:25:18 +0000 (08:25 +0000)]
Move the hints gunk to a seperate file.  It isn't really part of the
newbus structure (no more than subr_rman.c is anyway).

23 years agoFixed conflict with sys/dev/syscons/syscons.h.
Yoshihiro Takahashi [Sat, 14 Jul 2001 08:07:26 +0000 (08:07 +0000)]
Fixed conflict with sys/dev/syscons/syscons.h.

Submitted by: yokota

23 years agoMove fxp to the miibus section where it belongs.
Dima Dorfman [Sat, 14 Jul 2001 06:35:54 +0000 (06:35 +0000)]
Move fxp to the miibus section where it belongs.

23 years agoremember to set the return address in a message when
Julian Elischer [Sat, 14 Jul 2001 05:51:33 +0000 (05:51 +0000)]
remember to set the return address in a message when
sending it along a hook.

PR: 27906
Submitted by: Harti Brandt <brandt@fokus.gmd.de>

no MFC as code is different in 4.x

23 years agoPut on my peril-sensitive sunglasses and remove the POLA-violating
Peter Wemm [Sat, 14 Jul 2001 05:02:40 +0000 (05:02 +0000)]
Put on my peril-sensitive sunglasses and remove the POLA-violating
stealth hints loading.  'make release' has been fixed to not need this
now anyway.  If you want static hints, specify it explicitly.

Hey! Why did it suddenly get so dark??

23 years ago- Refine pc98 supports.
Yoshihiro Takahashi [Sat, 14 Jul 2001 04:19:09 +0000 (04:19 +0000)]
- Refine pc98 supports.
- Use bus_space stuff.
- Rename FDO_* -> FDC_* (obtained from NetBSD/pc98)

23 years agoAdded epson_inw function.
Yoshihiro Takahashi [Sat, 14 Jul 2001 04:14:29 +0000 (04:14 +0000)]
Added epson_inw function.

23 years agoMerged from the following changes.
Yoshihiro Takahashi [Sat, 14 Jul 2001 03:26:02 +0000 (03:26 +0000)]
Merged from the following changes.

sys/dev/syscons/scvgarndr.c revision  1.11
sys/dev/syscons/scvtb.c     revision  1.8
sys/dev/syscons/syscons.c   revisions 1.361, 1.363, 1.364, 1.365 and 1.366
sys/isa/atkbd_isa.c         revision  1.9
sys/isa/syscons_isa.c       revision  1.17

23 years agoMerged from sys/i386/conf/GENERIC revision 1.313.
Yoshihiro Takahashi [Sat, 14 Jul 2001 02:45:53 +0000 (02:45 +0000)]
Merged from sys/i386/conf/GENERIC revision 1.313.

23 years agoAdded (commented out) audio driver.
Yoshihiro Takahashi [Sat, 14 Jul 2001 02:44:51 +0000 (02:44 +0000)]
Added (commented out) audio driver.

23 years agoMerged from sys/conf/options.i386 revision 1.150.
Yoshihiro Takahashi [Sat, 14 Jul 2001 02:43:52 +0000 (02:43 +0000)]
Merged from sys/conf/options.i386 revision 1.150.

23 years agoCatch up with NetBSD/pc98.
Noriaki Mitsunaga [Sat, 14 Jul 2001 00:38:51 +0000 (00:38 +0000)]
Catch up with NetBSD/pc98.
o Much cleanly separate NetBSD(XS) / FreeBSD(CAM) codes.
o Improve tagged queing support (full QTAG).
o Improve quirk support.
o Improve parity error retry.
o Impliment wide negotheation.
o Cmd link support.
o Add copyright of CAM part.
o Change for CAM_NEW_TRAN_CODE.
o Work around for buggy KME UJDCD450.

o stg: add disconnet condition.
o nsp: use suspend I/O.
and more. I thank Honda-san.

conf/options.pc98: add CT_USE_RELOCATE_OFFSET and CT_BUS_WEIGHT
dev/{ct,ncv,nsp,stg}/*_{pccard,isa}.c: add splcam() before calling
        attach/detach functions.

Tested by: bsd-nomads
Obtained from:  NetBSD/pc98

23 years agoGo back to having either static OR dynamic hints, with fallback
Peter Wemm [Sat, 14 Jul 2001 00:23:10 +0000 (00:23 +0000)]
Go back to having either static OR dynamic hints, with fallback
support.  Trying to fix the merged set where dynamic overrode
static was getting more and more complicated by the day.

This should fix the duplicate atkbd, psm, fd* etc in GENERIC.  (which
paniced the alpha, but not the i386)

23 years agoMerge with latest version of the Mylex 6+ driver.
Mike Smith [Sat, 14 Jul 2001 00:12:23 +0000 (00:12 +0000)]
Merge with latest version of the Mylex 6+ driver.

 - All sources are built in a single object, reducing namespace pollution.
 - Kill the ready queue, and handle a busy response to mly_start in callers
   rather than deferring the command.
 - Improve our interaction with CAM:
   - Don't advertise physical channels as SCSI busses by default.
   - use the SIM queue freeze capability rather than queueing CDBs internally.
   - force bus reprobe at module load time.
 - Clean up more resources in mly_free.
 - Tidy up debugging levels.
 - Tidy up handling of events (mostly just code cleanliness).
 - Use explanatory macros for operations on bus/target/channel numbers.

23 years agoRemove the old machine symlink first.
Peter Wemm [Sat, 14 Jul 2001 00:03:05 +0000 (00:03 +0000)]
Remove the old machine symlink first.

23 years agoMAPS is becoming a subscription (fee based) service, warn users
Gregory Neil Shapiro [Fri, 13 Jul 2001 21:51:28 +0000 (21:51 +0000)]
MAPS is becoming a subscription (fee based) service, warn users
ORBS is gone

MFC after: 1 day

23 years agoReturn a locked struct buf from ufsdirhash_lookup() to avoid one
Ian Dowse [Fri, 13 Jul 2001 20:50:38 +0000 (20:50 +0000)]
Return a locked struct buf from ufsdirhash_lookup() to avoid one
extra getblk/brelse sequence for each lookup. We already had this
buf in ufsdirhash_lookup(), so there was no point in brelse'ing it
only to have the caller immediately reaquire the same buffer.

This should make the case of sequential lookups marginally faster;
in my tests, sequential lookups with dirhash enabled are now only
around 1% slower than without dirhash.

23 years agoMarkup fixes: Move <screen></screen> and <programlisting></programlisting>
Bruce A. Mah [Fri, 13 Jul 2001 20:16:43 +0000 (20:16 +0000)]
Markup fixes:  Move <screen></screen> and <programlisting></programlisting>
outside <para></para>.  Some other minor adjustments around <screen></screen>
tags.

Discussed on: -doc (in the context of doc/)

23 years agoFixed double word typo.
Jens Schweikhardt [Fri, 13 Jul 2001 20:11:43 +0000 (20:11 +0000)]
Fixed double word typo.
MFC after: 7 days

23 years agoI really should have just commented out this intire file vs. removing all
David E. O'Brien [Fri, 13 Jul 2001 18:45:06 +0000 (18:45 +0000)]
I really should have just commented out this intire file vs. removing all
the contents.  One may still want the contents as reference for creating
/boot/device.hints.

23 years agoBump net.inet.tcp.sendspace to 32k and net.inet.tcp.recvspace to 65k.
David E. O'Brien [Fri, 13 Jul 2001 18:38:04 +0000 (18:38 +0000)]
Bump net.inet.tcp.sendspace to 32k and net.inet.tcp.recvspace to 65k.
This should help us in nieve benchmark "tests".

It seems a wide number of people think 32k buffers would not cause major
issues, and is in fact in use by many other OS's at this time.  The
receive buffers can be bumped higher as buffers are hardly used and several
research papers indicate that receive buffers rarely use much space at all.

Submitted by: Leo Bicknell <bicknell@ufp.org>
<20010713101107.B9559@ussenterprise.ufp.org>
Agreed to in principle by: dillon (at the 32k level)

23 years agoAdd a few more manpages from Chad.
Ruslan Ermilov [Fri, 13 Jul 2001 18:35:48 +0000 (18:35 +0000)]
Add a few more manpages from Chad.
Fold major.9, minor.9, umajor.9, and uminor.9 into a single device_ids.9.

23 years agoBug fix: When the client connects to a server and Kerberos
Jacques Vidrine [Fri, 13 Jul 2001 18:12:13 +0000 (18:12 +0000)]
Bug  fix: When   the  client  connects   to  a  server   and  Kerberos
authentication is  enabled, the  client effectively ignores  any error
from krb5_rd_rep due to a missing branch.

In  theory  this could  result  in  an  ssh  client using  Kerberos  5
authentication accepting  a spoofed  AP-REP.  I doubt  this is  a real
possiblity, however, because  the AP-REP is passed from  the server to
the client via the SSH  encrypted channel.  Any tampering should cause
the decryption or MAC to fail.

Approved by: green
MFC after: 1 week

23 years agoHints processing is FUBAR'ed right now.
David E. O'Brien [Fri, 13 Jul 2001 17:43:38 +0000 (17:43 +0000)]
Hints processing is FUBAR'ed right now.
By default, we will end up with a duplicate set of hints if people have
a properly populated /boot/device.hints.  So for now, remove the hints
here until Peter revisits the new hints processing from mid-June that
broke Alpha booting.

23 years agoAfter some (long-standing ;-) critics from Bruce, throw away the old
Joerg Wunsch [Fri, 13 Jul 2001 16:48:56 +0000 (16:48 +0000)]
After some (long-standing ;-) critics from Bruce, throw away the old
device search code i introduce nearly six years ago in rev 1.8.  Bruce
suggested to rather use the device name of the root filesystem instead
which is certainly the most sensible default.  Since there are many
possible cases for a root filesystem name (device with and without
slices, consider /dev/vinum/root even though it currently could not
work as such), there's some heuristic using a RE in order to find out
the canonical device name from the mounted name.  This probably won't
quite fit for a NFS root (can't test that right now), but then,
there's hard to find a good default for those machines anyway. ;-)

This unbreaks the functionality of rev 1.2 i once broke in 1.8. :)

23 years agoTweak the Alpha partition warning wording and comment a little.
David E. O'Brien [Fri, 13 Jul 2001 16:45:00 +0000 (16:45 +0000)]
Tweak the Alpha partition warning wording and comment a little.

23 years agoRudamentary attempt to reconize when `a' is not the first partition on the
David E. O'Brien [Fri, 13 Jul 2001 16:37:03 +0000 (16:37 +0000)]
Rudamentary attempt to reconize when `a' is not the first partition on the
Alpha.  (the Alpha will not boot except from the first partition)

PR: 23064
Submitted by: Kees Jan Koster <kjkoster@kjkoster.org>

23 years agoRepeat after me: "when reporting `file not found', perform a little bit of
David E. O'Brien [Fri, 13 Jul 2001 16:30:11 +0000 (16:30 +0000)]
Repeat after me: "when reporting `file not found', perform a little bit of
useful HCI consideration and tell which file was being looked for".

23 years ago- correct two typos
Alexander Langer [Fri, 13 Jul 2001 16:10:39 +0000 (16:10 +0000)]
- correct two typos
- Use NULL instead of 0 for pointers (style(9)!)

Submitted by: Joel Wilsson <siigron@sii.linuxsweden.nu>
MFC after: 37 days

23 years agoPrint the month number properly
Brian Somers [Fri, 13 Jul 2001 15:07:06 +0000 (15:07 +0000)]
Print the month number properly

MFC after: 1 week

23 years agoWhine if we discover an unknown option. This program doesn't use
Dima Dorfman [Fri, 13 Jul 2001 14:10:09 +0000 (14:10 +0000)]
Whine if we discover an unknown option.  This program doesn't use
getopt(3) (and can't be converted without breaking compatibility), and
it's very irritating to have it silently DTWT if one combines options
together (e.g., "-msS domain,server").

23 years agoRecognize the %s format.
Dima Dorfman [Fri, 13 Jul 2001 13:59:24 +0000 (13:59 +0000)]
Recognize the %s format.

Submitted by: Thomas Zenker <thz@lennartz-electronic.de> and
    Maxim Konovalov <maxim@macomnet.ru>, respectively
Reviewed by: -audit

23 years ago`pcn' supports AMD Am79C97x cards, not Am79C79x cards.
Dima Dorfman [Fri, 13 Jul 2001 13:34:08 +0000 (13:34 +0000)]
`pcn' supports AMD Am79C97x cards, not Am79C79x cards.

PR: 28946
Submitted by: Ryuichiro Imura <imura@ryu16.org>