]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
24 years agoRemove a KASSERT() that has fulfilled its purpose. Note that it did
green [Sun, 24 Oct 1999 08:37:21 +0000 (08:37 +0000)]
Remove a KASSERT() that has fulfilled its purpose.  Note that it did
cause problems by tripping on shutdown (reboot(), not the socket
operation :).  Cause is still uncertain, but the panic isn't really
necessary here.

24 years agoFix compiling PCCARD kernels.
mdodd [Sun, 24 Oct 1999 07:17:26 +0000 (07:17 +0000)]
Fix compiling PCCARD kernels.

This is a hack and I don't like it.

Hopefully the new PCCARD stuff should be in working order in a day or so.

24 years agoModerate the "public" fortunes file somewhat; add the original to
wes [Sun, 24 Oct 1999 06:39:58 +0000 (06:39 +0000)]
Moderate the "public" fortunes file somewhat; add the original to
the -offensive database.  Also provide some whimsy in the $FBSD$
headers.

Suggested by: The Kibbutz squad.

24 years agoCosmetique: use standard prototypes scheme
ache [Sun, 24 Oct 1999 04:47:57 +0000 (04:47 +0000)]
Cosmetique: use standard prototypes scheme
Back out prev change: toupper is more compatible with sort -f

24 years agoCosmetique: use standard prototypes scheme
ache [Sun, 24 Oct 1999 04:41:31 +0000 (04:41 +0000)]
Cosmetique: use standard prototypes scheme
Back out prev. change: toupper is more compatible with sort -f

24 years agotoupper->tolower to match what strcasecmp does
ache [Sun, 24 Oct 1999 04:21:42 +0000 (04:21 +0000)]
toupper->tolower to match what strcasecmp does

24 years agoUse strcoll to provide the same results as sort and comm
ache [Sun, 24 Oct 1999 04:08:15 +0000 (04:08 +0000)]
Use strcoll to provide the same results as sort and comm
Use LINE_MAX for max line size (as comm does)

24 years agoLocalize it
ache [Sun, 24 Oct 1999 03:42:35 +0000 (03:42 +0000)]
Localize it

PR: 11221
Submitted by: Grigoriy Strokin <grg@philol.msu.ru>

24 years ago Adjust the buffer cache to better handle small-memory machines. A
dillon [Sun, 24 Oct 1999 03:27:28 +0000 (03:27 +0000)]
Adjust the buffer cache to better handle small-memory machines.  A
    slightly older version of this code was tested by BDE and I.

    Also fixes a lockup situation when kva gets too fragmented.

    Remove the maxvmiobufspace variable and sysctl, they are no longer
    used.  Also cleanup (remove) #if 0 sections from prior commits.

    This code is more of a hack, but presumably the whole buffer cache
    implementation is going to be rewritten in the next year so it's no
    big deal.

24 years ago Put tty back in, problem was due to cvsup mess on freefall.
dillon [Sun, 24 Oct 1999 03:01:41 +0000 (03:01 +0000)]
Put tty back in, problem was due to cvsup mess on freefall.

24 years ago Put include <netgraph.h> back in - problem was due to a cvsup mess on
dillon [Sun, 24 Oct 1999 02:58:39 +0000 (02:58 +0000)]
Put include <netgraph.h> back in - problem was due to a cvsup mess on
    freefall.  There may also be a real problem w/ buildworld but the fix
    would go somewhere else, not here.

24 years agoAdd commented entry to the lo0 section inviting bridge users to
nsayer [Sun, 24 Oct 1999 00:26:49 +0000 (00:26 +0000)]
Add commented entry to the lo0 section inviting bridge users to
enable ARP on filtering bridges.

24 years agomore fixes, braino's, typo's, etc.
julian [Sat, 23 Oct 1999 22:46:38 +0000 (22:46 +0000)]
more fixes, braino's, typo's, etc.

24 years ago Temporarily remove 'tty' netgraph sub module until it can be made
dillon [Sat, 23 Oct 1999 18:53:46 +0000 (18:53 +0000)]
Temporarily remove 'tty' netgraph sub module until it can be made
    to compile.

24 years agoFix a typo: ``maintain compatible'' -> ``maintain compatibility''
chris [Sat, 23 Oct 1999 18:15:14 +0000 (18:15 +0000)]
Fix a typo:  ``maintain compatible'' -> ``maintain compatibility''

24 years ago Cleanup missing includes, stale includes, and a few printf format
dillon [Sat, 23 Oct 1999 17:45:11 +0000 (17:45 +0000)]
Cleanup missing includes, stale includes, and a few printf format
    inconsistancies.

24 years agoAdd the -F option. This is for forcing restarts with -r by inhibiting
green [Sat, 23 Oct 1999 15:22:53 +0000 (15:22 +0000)]
Add the -F option.  This is for forcing restarts with -r by inhibiting
transmission of the If-Range HTTP header field.

24 years agoA version of the pppoe code ellicits a response from the ISP end
julian [Sat, 23 Oct 1999 15:15:42 +0000 (15:15 +0000)]
A version of the pppoe code ellicits a response from the ISP end
(but still not quite right)

24 years agodang, deleted a line
julian [Sat, 23 Oct 1999 04:52:54 +0000 (04:52 +0000)]
dang, deleted a line

24 years agoNow that Netgraph is in the system there are some cleanups we can do.
julian [Sat, 23 Oct 1999 04:28:11 +0000 (04:28 +0000)]
Now that Netgraph is in the system there are some cleanups we can do.
Also save a slightly closer to completion version of the PPPOE code.

Submitted by: Archie Cobbs <archie@freebsd.org>

24 years agoSmall bugfixes (point not getting marked in one case, string not NUL and
green [Sat, 23 Oct 1999 00:54:58 +0000 (00:54 +0000)]
Small bugfixes (point not getting marked in one case, string not NUL and
wrong size in the other.)

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

24 years agoUse the physical block number, not the logical block number, for I/O
msmith [Fri, 22 Oct 1999 20:55:15 +0000 (20:55 +0000)]
Use the physical block number, not the logical block number, for I/O
operations.  The latter only works where the partion begins at the bottom
of the disk.  Whoops.

Submitted by: Chris Csanady <cc@137.org>

24 years agoremember to initialize mailbox 2 for FC isp bus resets
mjacob [Fri, 22 Oct 1999 17:03:03 +0000 (17:03 +0000)]
remember to initialize mailbox 2 for FC isp bus resets

24 years agoFrom: src/sys/i386/conf/PCCARD
n_hibma [Fri, 22 Oct 1999 04:36:52 +0000 (04:36 +0000)]
From: src/sys/i386/conf/PCCARD

revision 1.21
date: 1999/10/15 17:29:20;  author: imp;  state: Exp;  lines: +3 -3
Reorganize the attachement point for pcic (it was unattached and
floating before).  Attach pccard devices to pcic, one per slot
(although this may change to one per pcic).  pcic is now attached to
isa (to act as a bridge) and pccard is attached to pcic, cbb and
pc98ic (the last two are card bus bridge and the pc98ic version of
pcic, neither of which are in the tree yet).  Move pccard compat code
into pccard/pccard_compat.c.

THIS REQUIRES A CONFIG FILE CHANGE.  You must change your pcic/card
entries to be:
# PCCARD (PCMCIA) support
controller pcic0 at isa?
controller pcic1 at isa?
controller card0

The old system was upside down and this corrects that problem.  It
will make it easier to add support for YENTA pccard/card bus bridges.

Much more cleanup needs to happen before newbus devices can have
pccard attachments.  My previous commit's comments were premature.

Forgotten by: imp

24 years agoAdd missing entries in a structure.
julian [Fri, 22 Oct 1999 04:23:11 +0000 (04:23 +0000)]
Add missing entries in a structure.

24 years agoAdd a location to grog's assesment for historical reasons.
billf [Fri, 22 Oct 1999 00:01:29 +0000 (00:01 +0000)]
Add a location to grog's assesment for historical reasons.

Submitted by: grog

24 years agoSome more old quotations I had sitting around that I shared and were
nsayer [Thu, 21 Oct 1999 23:59:50 +0000 (23:59 +0000)]
Some more old quotations I had sitting around that I shared and were
laughed at.

Reviewed by: the FreeBSDcon terminal room

24 years agoFreeBSDcon Wisdom.
billf [Thu, 21 Oct 1999 23:51:25 +0000 (23:51 +0000)]
FreeBSDcon Wisdom.

Reviewed by: the anonymous members of the FreeBSDcon terminal room.

24 years agoManually create a correct(*) entry in wtmp when logging out a PPPoTCP or
brian [Thu, 21 Oct 1999 21:55:22 +0000 (21:55 +0000)]
Manually create a correct(*) entry in wtmp when logging out a PPPoTCP or
PPPoUDP connection.

(*) This is as correct as ftp and uucp wtmp entries are - that is,
multiple concurrent connections will not record enough information
in wtmp to tell last(1) who was logged in for how long.

24 years agoA few improvements and cleanups.
luoqi [Thu, 21 Oct 1999 21:15:02 +0000 (21:15 +0000)]
A few improvements and cleanups.

24 years agoMake some tweaks to the RealTek driver:
wpaul [Thu, 21 Oct 1999 19:42:03 +0000 (19:42 +0000)]
Make some tweaks to the RealTek driver:

- Fix a bug in rl_rxeof() handler: in the case where the packet wraps
  from the end of the receive buffer back to the beginning, we need to
  insure that at least sizeof(ether_header) bytes make it into the first
  mbuf. If we don't, then doing eh = mtod(m, struct ether_header *)
  loses. To avoid this, we use m_pullup() to suck at least MHLEN -
  RL_ETHER_ALIGN bytes into the first mbuf, which should also help
  small packets fit into a single mbuf.

  Pointed out by: Philip A. Prindeville <philipp@zembu.com>

- Make the transmit threshold autotuning: start off with a small value
  and jack it up when TX underruns are detected.

- Also improve TX error recovery: kick the chip in the head with a
  reset/init sequence to make sure it recovers afer a transmit error.

24 years agoNon-functional changes. Add some comments before I start to forget how
luoqi [Thu, 21 Oct 1999 15:49:25 +0000 (15:49 +0000)]
Non-functional changes. Add some comments before I start to forget how
everything works myself.

24 years agofix typo
julian [Thu, 21 Oct 1999 14:31:32 +0000 (14:31 +0000)]
fix typo

24 years agoCorrectly document the -V and --version options.
mpp [Thu, 21 Oct 1999 11:30:10 +0000 (11:30 +0000)]
Correctly document the -V and --version options.

PR: docs/14417
Submitted by: Greg Lewis <glewis@trc.adelaide.edu.au>

24 years agoFix the duplicate filenames that are the result of using getdents.
marcel [Thu, 21 Oct 1999 09:14:49 +0000 (09:14 +0000)]
Fix the duplicate filenames that are the result of using getdents.

glibc2 defines struct dirent differently than the Linux kernel does.
The getdents function therefore needs to read a heuristically defined
number of kernel dirents to satisfy the request. In case where too
many kernel dirents have been read, the function lseeks on the
directory so that a next call will start with the right dirent. The
offset used in lseeking is the offset-field in the last dirent passed
to the application. This can only mean that the offset-field holds
the offset of the next dirent and not the offset of the dirent itself.

24 years agoResurrect the aic driver.
luoqi [Thu, 21 Oct 1999 09:12:18 +0000 (09:12 +0000)]
Resurrect the aic driver.

24 years agoWhistle's Netgraph link-layer (sometimes more) networking infrastructure.
julian [Thu, 21 Oct 1999 09:06:11 +0000 (09:06 +0000)]
Whistle's Netgraph link-layer (sometimes more) networking infrastructure.
Been in production for 3 years now. Gives Instant Frame relay to if_sr
and if_ar drivers, and PPPOE support soon. See:
ftp://ftp.whistle.com/pub/archie/netgraph/index.html
for on-line manual pages.

Reviewed by: Doug Rabson (dfr@freebsd.org)
Obtained from:  Whistle CVS tree

24 years agoAdd aic CAM driver support.
luoqi [Thu, 21 Oct 1999 09:00:20 +0000 (09:00 +0000)]
Add aic CAM driver support.

24 years agoAdaptec 6260/6360 CAM driver.
luoqi [Thu, 21 Oct 1999 08:56:53 +0000 (08:56 +0000)]
Adaptec 6260/6360 CAM driver.

24 years agoTest commit of first Netgraph file
julian [Thu, 21 Oct 1999 07:27:47 +0000 (07:27 +0000)]
Test commit of first Netgraph file

24 years agoCleanup towards -Wall
julian [Thu, 21 Oct 1999 07:08:56 +0000 (07:08 +0000)]
Cleanup towards -Wall

24 years agoUse ${INSTALL} instead of 'install'.
bp [Thu, 21 Oct 1999 01:50:21 +0000 (01:50 +0000)]
Use ${INSTALL} instead of 'install'.

24 years agoIf we're running ppp -direct over a tcp or udp connection,
brian [Thu, 21 Oct 1999 01:19:23 +0000 (01:19 +0000)]
If we're running ppp -direct over a tcp or udp connection,
record the IP number in the `from' slot and not the tty slot.
We put ``ppp'' in the tty slot - in line with what ftp (and
probably others) does.

24 years agoDon't go from PHASE_NETWORK to PHASE_ESTABLISH when we want to
brian [Thu, 21 Oct 1999 00:19:33 +0000 (00:19 +0000)]
Don't go from PHASE_NETWORK to PHASE_ESTABLISH when we want to
re-open a device.  The fact that we're in NETWORK phase indicates
that there are other links in DATALINK_OPEN and that we don't want
to stop using them.

24 years agofix a serious bug where, on alpha, due to a an int/long type mismatch,
gallatin [Wed, 20 Oct 1999 15:20:00 +0000 (15:20 +0000)]
fix a serious bug where, on alpha, due to a an int/long type mismatch,
the uid arg to use_yp() was getting clobbered by the call to my_yp_match().
This led to a problem where a NIS user could edit root's passwd information.

24 years agoRemove one obsoleted entry from the BUGS section.
ru [Wed, 20 Oct 1999 12:59:35 +0000 (12:59 +0000)]
Remove one obsoleted entry from the BUGS section.

24 years agoAdd ncplist program. That makes minimal set of ncp
bp [Wed, 20 Oct 1999 11:31:02 +0000 (11:31 +0000)]
Add ncplist program. That makes minimal set of ncp
utilities to use NetWare volumes. Other functionality
(ncpsend, ncprint) will be added later.

24 years agoFix termcap % code processing for some terminals.
tg [Wed, 20 Oct 1999 08:52:09 +0000 (08:52 +0000)]
Fix termcap % code processing for some terminals.

Submitted by: Ross Ridge <ross@zooid.guild.org> via
buster@lambda.hh.provi.de (Andreas Burmester)

24 years agoAllow for incoming DNS UDP queries.
ru [Wed, 20 Oct 1999 08:15:13 +0000 (08:15 +0000)]
Allow for incoming DNS UDP queries.

24 years agomergemaster bmake glue.
billf [Wed, 20 Oct 1999 07:33:09 +0000 (07:33 +0000)]
mergemaster bmake glue.

24 years agoThis commit was generated by cvs2svn to compensate for changes in r52400,
billf [Wed, 20 Oct 1999 07:22:52 +0000 (07:22 +0000)]
This commit was generated by cvs2svn to compensate for changes in r52400,
which included commits to RCS files with non-trunk default branches.

24 years agoImport mergemaster into the tree, a product of freebsdcon.
billf [Wed, 20 Oct 1999 07:22:52 +0000 (07:22 +0000)]
Import mergemaster into the tree, a product of freebsdcon.

Submitted by:  Doug Barton <doug@gorean.org>

24 years ago A tentative agreement has been reached in regards to a procedure
dillon [Wed, 20 Oct 1999 06:31:49 +0000 (06:31 +0000)]
A tentative agreement has been reached in regards to a procedure
    to remove 'b'lock devices.  The agreement is, essentially, that
    block devices will be collapsed into character devices as a first
    step (though I don't particularly agree), and raw device names 'rxxx'
    will become simply 'xxx' in devfs in the second step (i.e. no 'rxxx'
    names will exist).  The renaming will not effect the original /dev
    and the expectation is that devfs will eventually (but not immediately)
    become the standard way to access devices in the system.

    If it is determined that a reimplementation of block device access
    characteristics is beneficial, a number of alternatives will
    be possible that do not involve resurrecting the 'b'lock device class.
    For example, an ioctl() that might be made on an open character device
    descriptor or a generic buffered overlay device.

    This commit removes the blockdev disablement sysctl which does not
    apply to the solution that was reached.

24 years agoChange MFS sizes to 2400k since gzip will compact free space anyway.
dwhite [Wed, 20 Oct 1999 04:48:35 +0000 (04:48 +0000)]
Change MFS sizes to 2400k since gzip will compact free space anyway.
Add 'install' to the internal floppy type list.

24 years agoRemove pccard attachment stub, this caused pccard unit 0 to be allocated
peter [Tue, 19 Oct 1999 17:12:28 +0000 (17:12 +0000)]
Remove pccard attachment stub, this caused pccard unit 0 to be allocated
and unusable by the pccard system since pccard doesn't attach to the
nexus any more.  This was stopping my 3c589D from working as pccard unit
0 is used directly for resource allocation and this fails when unit 0
isn't actually attached to anything.

24 years agoAdd the -unit command line switch for specifying the tun device.
brian [Tue, 19 Oct 1999 15:21:09 +0000 (15:21 +0000)]
Add the -unit command line switch for specifying the tun device.
Warn about -alias being depricated (but still allow it).
Don't moan twice about failing to open any tun device.
Fix a diagnostic and add the -quiet switch to the usage message.

24 years agoMake the "machine" target a prequisite to all source files.
dcs [Tue, 19 Oct 1999 03:53:59 +0000 (03:53 +0000)]
Make the "machine" target a prequisite to all source files.

Hinted-at by: bde

24 years agoGive credit where credit's due. The previous commit should have the following
chris [Mon, 18 Oct 1999 23:03:47 +0000 (23:03 +0000)]
Give credit where credit's due.  The previous commit should have the following
information:

PR: docs/14407
Submitted by: Greg Lewis <glewis@trc.adelaide.edu.au>

24 years agoFix a few typos (capitalization and punctuation).
chris [Mon, 18 Oct 1999 22:40:13 +0000 (22:40 +0000)]
Fix a few typos (capitalization and punctuation).

PR: docs/14407
Submitted by: grog

24 years agoAdd ua.koi8-u.kbd
ache [Mon, 18 Oct 1999 22:17:08 +0000 (22:17 +0000)]
Add ua.koi8-u.kbd

Submitted by: Alexey Zelkin <phantom@cris.net>

24 years agoAdd uk_UA.KOI8-U
ache [Mon, 18 Oct 1999 22:04:06 +0000 (22:04 +0000)]
Add uk_UA.KOI8-U

Submitted by: Alexey Zelkin <phantom@cris.net>

24 years agoAdd uk_UA.KOI8-U
ache [Mon, 18 Oct 1999 21:57:33 +0000 (21:57 +0000)]
Add uk_UA.KOI8-U

Submitted by: Alexey Zelkin <phantom@cris.net>

24 years agoadd uk_UA.KOI8-U
ache [Mon, 18 Oct 1999 21:49:02 +0000 (21:49 +0000)]
add uk_UA.KOI8-U

Submitted by: Alexey Zelkin <phantom@cris.net>

24 years agoadd uk_UA.KOI8-U locale dirs
ache [Mon, 18 Oct 1999 20:18:04 +0000 (20:18 +0000)]
add uk_UA.KOI8-U locale dirs

24 years agoIncrease the timeout to be 3*hz. This recalibrates the timeout so that
gallatin [Mon, 18 Oct 1999 17:55:38 +0000 (17:55 +0000)]
Increase the timeout to be 3*hz.  This recalibrates the timeout so that
it has the same value on all platforms.  Previously it was just under
3 seconds on x86 (typically hz<=128) and just under 1/3 of a second on
alpha (typically hz>=1024).  This covers up a race between ad_interrupt()
and ad_timeout() which is being looked into.

reviewd by: sos

24 years agoChange the default for the vfs.bdev_buffered sysctl to zero.
phk [Mon, 18 Oct 1999 16:59:50 +0000 (16:59 +0000)]
Change the default for the vfs.bdev_buffered sysctl to zero.

This means that access to block devices nodes will act the
same as char device nodes for disk-like devices.

If you encounter problems after this, where programs accessing
disks directly fail to operate, please use the following command
to revert to previous behaviour:

        sysctl -w vfs.bdev_buffered=1

And verify that this was indeed the cause of your trouble.

See the mail-archives of the arch@FreeBSD.org list for background.

24 years agoFor emacs key binding clearly describe ^t/^u as "begin of file"/"end of file"
ache [Mon, 18 Oct 1999 16:30:35 +0000 (16:30 +0000)]
For emacs key binding clearly describe ^t/^u as "begin of file"/"end of file"
Old description "start of text"/"end of text" is too confusing assuming
text currently visible on the screen

24 years agoMoved to share
ache [Mon, 18 Oct 1999 14:00:59 +0000 (14:00 +0000)]
Moved to share

24 years agoAdd colldef
ache [Mon, 18 Oct 1999 13:57:10 +0000 (13:57 +0000)]
Add colldef

24 years agoAdd 8859-5, remove BUILD_TOOLS ifdef
ache [Mon, 18 Oct 1999 13:54:51 +0000 (13:54 +0000)]
Add 8859-5, remove BUILD_TOOLS ifdef

24 years agomoved to share
ache [Mon, 18 Oct 1999 13:47:02 +0000 (13:47 +0000)]
moved to share

24 years agoMove sources one directory up
ache [Mon, 18 Oct 1999 13:39:50 +0000 (13:39 +0000)]
Move sources one directory up

24 years agoAppend missing newline to log() message for permanent ARP modification
sheldonh [Mon, 18 Oct 1999 11:56:50 +0000 (11:56 +0000)]
Append missing newline to log() message for permanent ARP modification
attempt warning, which was added in rev 1.48 .

PR: 14371
Submitted by: sec@pi.musin.de (Stefan `Sec` Zehl)

24 years agoAllow suppress printing of lines with zero counts
bp [Mon, 18 Oct 1999 05:45:05 +0000 (05:45 +0000)]
Allow suppress printing of lines with zero counts
for an IPX address family.

PR: 14168

24 years agoThis fixes the problem with SMC NE2000 cards hanging the box on
mdodd [Mon, 18 Oct 1999 04:27:33 +0000 (04:27 +0000)]
This fixes the problem with SMC NE2000 cards hanging the box on
bootup.  Somehow my backout of an abortaive attempt at shared
memory autoconfiguration included this line:

        sc->mem_shared = 1;

Which is fairly important as it turns out.

Since I performed my pre-commit testing on a different box with a generic
NE2000 I didn't catch this.  Pointy hat.

24 years agoFix a few things. Unbogosify a free(), {,UN}MARK with correct args, and
green [Mon, 18 Oct 1999 03:52:20 +0000 (03:52 +0000)]
Fix a few things.  Unbogosify a free(), {,UN}MARK with correct args, and
fix count checks.

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

24 years agosdio_done:
grog [Sun, 17 Oct 1999 23:56:47 +0000 (23:56 +0000)]
sdio_done:
  Set the errors in the correct buffer header.

Inspection-provoked-by: Bernd Walter <ticso@cicely.de>
24 years agolaunch_requests:
grog [Sun, 17 Oct 1999 23:55:10 +0000 (23:55 +0000)]
launch_requests:
  Put splbio protection around the main launch loop.  We've seen cases where
  the bottom half was cutting off the branch on which we're sitting.

Experienced-by: Michael Reifenberger <root@nihil.plaut.de>
24 years agoReverse the sense of the /mnt/boot/defaults dir check. This was causing
dwhite [Sun, 17 Oct 1999 22:42:57 +0000 (22:42 +0000)]
Reverse the sense of the /mnt/boot/defaults dir check.  This was causing
the bootloader to not load the loader.conf and thus the MFS image.  The result
was 'panic: no init'.

24 years agoAdd in inclusion of machine/md_var.h (so alpha_scsi_bus_register or what
mjacob [Sun, 17 Oct 1999 19:03:11 +0000 (19:03 +0000)]
Add in inclusion of machine/md_var.h (so alpha_scsi_bus_register or what
have you is prototyped). Removed code versions in md struct- not used
any more. Allocate transfer dma maps and xflist stuff in mbxdmasetup based
upon isp->isp_maxcmds. Allow for multiple calls to mbxdmasetup (for
isp_reset cases).

24 years agoRemove some target mode stuff. It will get re-introduced in a different
mjacob [Sun, 17 Oct 1999 18:58:22 +0000 (18:58 +0000)]
Remove some target mode stuff. It will get re-introduced in a different
file later. Do some pencil-sharpening types of minor changes. Change
how active commands are remembered (using new inline functions to get
handles, etc..). Now do a GET FIRMWARE STATUS after firing up the f/w as
outgoing mailbox 2 will tell you the f/w's notion of the max commands
that can be supported. Attempt to retrieve loop topology. Add in the
appropriate SWIZZLE/UNSWIZZLE macros calls (this is a no-op on Little
Endian machines but is needed for sparc (on other platforms)). Move
the temp port database we use to find out where things have moved to
after a LIP to the softc and off the kernel stack. Follow Qlogic's
hint and don't bother setting a tag for commands that don't have
this enabled (presumably the f/w will do it's own selection then).
Use an INT_PENDING macro to check for an interrupt. The call to
ISP_DMAFREE now just takes the handle- not the 'handle-1' which was
a layering violation. Use CFGPRINTF in a couple of places to make
things less chatty if not booting verbose, or CAMDEBUG compiles, etc..

24 years agoAdd in isp_debug variable. It defaults to zero unless CAMDEBUG is defined
mjacob [Sun, 17 Oct 1999 18:50:47 +0000 (18:50 +0000)]
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined
where it defaults to one. Change simq width allocation to the max number
of commands supported by the HBA after f/w fires up- not the constant
MAXISPREQUEST value. Do some stylistic changes.

24 years agoRoll platform major && minor (major now tracks FreeBSD major release).
mjacob [Sun, 17 Oct 1999 18:48:16 +0000 (18:48 +0000)]
Roll platform major && minor (major now tracks FreeBSD major release).
Add in null SWIZZLE definitions. Add in CFGPRINTF define. Change default
debug level to refer to an external isp_debug variable. Remove inline
functions as they're now in isp_inline.h and include that file.

24 years agoRoll core version number. Do some stylistic changes. Ensure that
mjacob [Sun, 17 Oct 1999 18:45:31 +0000 (18:45 +0000)]
Roll core version number. Do some stylistic changes. Ensure that
the result queue length is never less than 64. Move (ick) temp port
database used for post-LIP merging off the kernel stack and put it
into the softc. Remove some target mode stuff which will come back
later in a different file. Change how the list of outstanding commands
are stored (now allocated at mailbox setup time to be just enough for
the max for a specific HBA which can vary). Keep a rotating seed of
the last index for this in the softc. Increase the count of active
commands from 10 to 16 bits.

24 years agoadd in an INT_PENDING macro
mjacob [Sun, 17 Oct 1999 18:41:47 +0000 (18:41 +0000)]
add in an INT_PENDING macro

24 years agoRemove target mode definitions (they'll come back later in
mjacob [Sun, 17 Oct 1999 18:41:21 +0000 (18:41 +0000)]
Remove target mode definitions (they'll come back later in
a different file).

24 years agoAdd a file where inline functions for the Qlogic isp cards can go. This reduces
mjacob [Sun, 17 Oct 1999 18:15:05 +0000 (18:15 +0000)]
Add a file where inline functions for the Qlogic isp cards can go. This reduces
duplication in all the platform specific header files.

24 years agoroll 10[24]0 and 1080/12X0 f/w
mjacob [Sun, 17 Oct 1999 18:09:34 +0000 (18:09 +0000)]
roll 10[24]0 and 1080/12X0 f/w

24 years agoquoting << Martin_Blapp
green [Sun, 17 Oct 1999 16:26:58 +0000 (16:26 +0000)]
quoting << Martin_Blapp

- Completly changed the internals of umount(8). We do three
checks now to see if 'argv' is in the mounttable. It they
all fail, we return to main and print a warning.

- fixed the umount mount-order. The checks are rather complex
to do this. Cause umount(8) should also be able to unmount
several devices at once ('umount -a', 'umount -A',
'umount /mnt /mnt2'), the mount-order get's important.
I added checks to mark and unmark already unmounted devices.

-  Various fixes with nfs-unmounts (no rpc-calls were done,
or they were done although there was an existing mount).
Since we allow overlay-mounts, we should also handle them
properly.

- Translate the deprecated nfs-syntax with '@' to ':' like
mount_nfs does. The ':' syntax has now precedence, but '@'
still works.

- 'umount -v' is now fixed for all cases and doesn't print
garbage like two times the mountpoint etc.

- removed non documented and useless umount '-F'.

- hanged nfsmounts can now unmounted 'without' any problems.
I've removed stat() and realpath() checks on the mountpoint.
Instead we just do a realpath() on the basedir of the
mountpath and add the dirname again.
Implemented this as an idea from phk. But there are still
vfs-restrictions if the nfs_mount is busy. If there are
unwritten metadata on a hanged nfs-mount, and we modify
nfs_vfsops.c to not return EBUSY, we get a deadlock :(
The problem has now moved from userland to kernel.

- removed the BUGS part from the umount(8) manpage.

- Converted it to ANSI C (more than 60% of the code have
changed).

Martin_Blapp

Fixed PR's
----------

o [1999/02/03] bin/9893                 NFS umount of regular file impossible

s [1995/11/27] bin/841 stale nfs mounts cannot be umounted

o [1999/08/01] bin/12911 alfred  NFS umounts are not properly done
if just the mountpoint gets umounted

Only partially solved:
----------------------

The problem is now in kernel:

o [1999/04/07] bin/11005 `umount -f' does not work if the
NFS-server is down.

PR: bin/9893 bin/841 bin/12911 bin/11005
Submitted by: Martin Blapp <mb@imp.ch>

24 years agoUse Pa for filenames.
charnier [Sun, 17 Oct 1999 15:57:16 +0000 (15:57 +0000)]
Use Pa for filenames.

24 years agoCorrect use of .Nm.
charnier [Sun, 17 Oct 1999 15:54:50 +0000 (15:54 +0000)]
Correct use of .Nm.

24 years agoXref to warn(3) which is used instead of perror.
charnier [Sun, 17 Oct 1999 15:52:26 +0000 (15:52 +0000)]
Xref to warn(3) which is used instead of perror.

24 years agoadd .Sh DIAGNOSTICS
charnier [Sun, 17 Oct 1999 15:50:19 +0000 (15:50 +0000)]
add .Sh DIAGNOSTICS

24 years agoadd missing .El.
charnier [Sun, 17 Oct 1999 15:46:58 +0000 (15:46 +0000)]
add missing .El.

24 years agoRemove unnecessary includes.
newton [Sun, 17 Oct 1999 14:50:13 +0000 (14:50 +0000)]
Remove unnecessary includes.

phk's script walked through .c and .h files, but some of the ones on
the list are actually derived from sys/svr4/syscalls.master.  Make
the necessary changes here and the others will implicitly follow...

Submitted by: phk

24 years agoRemove unnecessary includes.
newton [Sun, 17 Oct 1999 14:44:48 +0000 (14:44 +0000)]
Remove unnecessary includes.

Submitted by: phk

24 years agoMove data to share
ache [Sun, 17 Oct 1999 13:44:48 +0000 (13:44 +0000)]
Move data to share

24 years agoadd mklocale
ache [Sun, 17 Oct 1999 13:42:03 +0000 (13:42 +0000)]
add mklocale

24 years agorestore 8859-5, remove BUILD_TOOLS define
ache [Sun, 17 Oct 1999 11:00:45 +0000 (11:00 +0000)]
restore 8859-5, remove BUILD_TOOLS define

24 years agoNote that we support the PCI version of the Digi SYNC/570i cards.
jhay [Sun, 17 Oct 1999 09:43:51 +0000 (09:43 +0000)]
Note that we support the PCI version of the Digi SYNC/570i cards.