]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoKeep track of negotiated transfer parameters for each initiator<->target
Justin T. Gibbs [Fri, 5 Mar 1999 23:35:48 +0000 (23:35 +0000)]
Keep track of negotiated transfer parameters for each initiator<->target
connection.

Clean up support for devices featuring the multiple target SCSI ID feature.
On aic7890/91/96/97 chips, we can now assume the target role on multiple
target ids simultaneously.  Although these chips also have sufficient
instruction space to hold to support the initiator and target role at the
same time, the initiator role is currently disabled as it will conflict
(chip design restriction) with the multi-tid feature.  I'll probably add
a nob to enable the initiator (there-by disabling multi-tid) some time
in the future.

Return queue full or busy, depending on the tagged nature of the incoming
request, if our command input queue fills up in host memeory.

Deal with accept target I/O resource shortages.

If we get an underrun on a transaction that wasn't supposed to transmit
any data, don't attempt to print out the S/G list.  The code would
run until hitting a non-present page. (oops)

25 years agoModify to deal with changes in the storage of user negotiation settings.
Justin T. Gibbs [Fri, 5 Mar 1999 23:28:42 +0000 (23:28 +0000)]
Modify to deal with changes in the storage of user negotiation settings.

25 years agoAdd untested code to support the aha2930U2W.
Justin T. Gibbs [Fri, 5 Mar 1999 23:28:36 +0000 (23:28 +0000)]
Add untested code to support the aha2930U2W.

Modify to deal with changes in the storage of user negotiation settings.

25 years agoImplement the control device for the "targ" target mode processor target
Justin T. Gibbs [Fri, 5 Mar 1999 23:25:11 +0000 (23:25 +0000)]
Implement the control device for the "targ" target mode processor target
emulator so that instances can be dynamically added and removed from the
system.

Properly reference count peripheral instances so they are cleaned up when
destroyed by the control device.

Set a timeout for test unit ready commands.  Before it was uninitialized
and could cause us to drop off the bus when no real timeout had occurred.

25 years agoSilence errors.
Justin T. Gibbs [Fri, 5 Mar 1999 23:21:28 +0000 (23:21 +0000)]
Silence errors.

Reduce the number of accept target I/O entries we allow to better
exercise the resource shortage handling code in controller drivers.

25 years agoSilence complaints about synchronize cache requests that fail with illegal
Justin T. Gibbs [Fri, 5 Mar 1999 23:20:20 +0000 (23:20 +0000)]
Silence complaints about synchronize cache requests that fail with illegal
request.

25 years agoxpt_done has no reason to take a volatile ccb *.
Justin T. Gibbs [Fri, 5 Mar 1999 23:18:56 +0000 (23:18 +0000)]
xpt_done has no reason to take a volatile ccb *.

25 years agoProtect access to the EDT with splcam() to protect against corruption
Justin T. Gibbs [Fri, 5 Mar 1999 23:18:16 +0000 (23:18 +0000)]
Protect access to the EDT with splcam() to protect against corruption
caused by temporary EDT allocations performed by controller drivers in
their interrupt routiens.

Reference count bus entries in the EDT in preparation for support for
dynamic controller arrival and departure.

Have children of the EDT hold references to their parents.

Correct routing of the XPT_IMMED_NOTIFY ccb type for use in
target mode applications.

Fix a few cases where the generation count for EDT data members was
not being updated when a modification occurred.

splcam() problem Noticed by: Tor Egge <tegge@FreeBSD.org>

25 years agoAdd constants and macros for dealing with XPT opcode group types.
Justin T. Gibbs [Fri, 5 Mar 1999 23:13:20 +0000 (23:13 +0000)]
Add constants and macros for dealing with XPT opcode group types.

25 years agoAllocate a target mode instance to handle our target mode session, clean
Justin T. Gibbs [Fri, 5 Mar 1999 23:12:02 +0000 (23:12 +0000)]
Allocate a target mode instance to handle our target mode session, clean
it up on exit.  The address for attaching the emulator (path, target id,
lun) is now specified on the command line.  Some attempt at cathing
signals and cleaning up target mode instances is now made.

25 years agoAdd targ.ctl, a device node for allocating and configuring instances
Justin T. Gibbs [Fri, 5 Mar 1999 23:08:45 +0000 (23:08 +0000)]
Add targ.ctl, a device node for allocating and configuring instances
of the "targ" processor target target mode emulator.

25 years agoFixed bitrot in the types of fusword() and susword(). swords will
Bruce Evans [Fri, 5 Mar 1999 19:27:22 +0000 (19:27 +0000)]
Fixed bitrot in the types of fusword() and susword().  swords will
probably always be 16 bits if they exist at all, and fusword() and
susword() are only used in i386 code, so there aren't any portability
functions with them.

25 years agoFixed prototype for fetchParseURL().
Bruce Evans [Fri, 5 Mar 1999 18:45:32 +0000 (18:45 +0000)]
Fixed prototype for fetchParseURL().

25 years agoFixed bitrot in synopsis (some const poisoning hadn't reached here).
Bruce Evans [Fri, 5 Mar 1999 18:43:00 +0000 (18:43 +0000)]
Fixed bitrot in synopsis (some const poisoning hadn't reached here).

25 years agoUse __i386__ instead of i386 in an ifdef.
Bruce Evans [Fri, 5 Mar 1999 18:15:05 +0000 (18:15 +0000)]
Use __i386__ instead of i386 in an ifdef.

25 years agoDon't use u_long in the synopsis, since u_long is not part of the kvm
Bruce Evans [Fri, 5 Mar 1999 17:32:53 +0000 (17:32 +0000)]
Don't use u_long in the synopsis, since u_long is not part of the kvm
interface.

25 years agoFixed missing header in synopsis (<devstat.h> unfortunately depends on
Bruce Evans [Fri, 5 Mar 1999 17:14:05 +0000 (17:14 +0000)]
Fixed missing header in synopsis (<devstat.h> unfortunately depends on
<sys/dkstat.h>).

25 years agoFixed missing header in synopsis (<camlib.h> includes half the universe
Bruce Evans [Fri, 5 Mar 1999 17:11:37 +0000 (17:11 +0000)]
Fixed missing header in synopsis (<camlib.h> includes half the universe
but not <stdio.h>).

25 years agoFixed bitrot in synopsis (some const poisoning had not reached here).
Bruce Evans [Fri, 5 Mar 1999 17:08:23 +0000 (17:08 +0000)]
Fixed bitrot in synopsis (some const poisoning had not reached here).

25 years agoThe magic "no-cpu" cpu number is 0xff. Don't misrepresent cpu
Bruce Evans [Fri, 5 Mar 1999 16:38:13 +0000 (16:38 +0000)]
The magic "no-cpu" cpu number is 0xff.  Don't misrepresent cpu
numbers as chars or use bogus casts in an attempt to unmisrepresnt
them.  In top, don't assume that 0xff is the only negative cpu
number when cpu numbers are (mis)represented.

25 years agoFixed bitrot in synopsis (some const poisoning hadn't reached here).
Bruce Evans [Fri, 5 Mar 1999 15:50:07 +0000 (15:50 +0000)]
Fixed bitrot in synopsis (some const poisoning hadn't reached here).

25 years agoDon't bogotify the synopsis by attempting to describe err_set_file() there.
Bruce Evans [Fri, 5 Mar 1999 15:36:23 +0000 (15:36 +0000)]
Don't bogotify the synopsis by attempting to describe err_set_file() there.

Fixed some disorder.

25 years agoFixed references to unmount(2) specified as umount.
Alexander Langer [Fri, 5 Mar 1999 15:16:31 +0000 (15:16 +0000)]
Fixed references to unmount(2) specified as umount.

Submitted by: Markus Friedl <markus.friedl@informatik.uni-erlangen.de>
via OpenBSD

25 years agoDon't use backslash-newline to split the too-long line for
Bruce Evans [Fri, 5 Mar 1999 14:43:29 +0000 (14:43 +0000)]
Don't use backslash-newline to split the too-long line for
".Fn dialog_ftree ...", since my synopsis checker doesn't support it
and no other too-long .Fn lines are split.

25 years agoFixed disordering and incoinsistent style in previous commit.
Bruce Evans [Fri, 5 Mar 1999 13:01:22 +0000 (13:01 +0000)]
Fixed disordering and incoinsistent style in previous commit.

25 years agoThe pseudocode in the synopsis didn't come close to actually
Bruce Evans [Fri, 5 Mar 1999 12:56:37 +0000 (12:56 +0000)]
The pseudocode in the synopsis didn't come close to actually
compiling, since <stdio.h> correctly doesn't declare off_t although
the pseudo-prototypes for the new fseeko() and ftello() functions
use it.  Handle this like the corresponding problem for va_list
versus the vprintf() family.

Fixed some English errors.

25 years agoMoved kernel declarations inside the KERNEL ifdef, and removed
Bruce Evans [Fri, 5 Mar 1999 11:25:31 +0000 (11:25 +0000)]
Moved kernel declarations inside the KERNEL ifdef, and removed
include of <sys/queue.h> in the !KERNEL case.  The prerequisites
for <ufs/ufs/quota.h> were broken in Lite2 by converting some of
the kernel declarations to use queue macros without including
<sys/queue.h>.  <sys/queue.h> was included in applications in
/usr/src instead.  We polluted this file instead of merging the
changes in the applications.

Include <sys/queue.h> in the KERNEL case, and forward-declare all
structs that are used in prototypes, so that this file is almost
self-sufficient even in the kernel.

Obtained from: mostly from NetBSD

25 years agoDon't use undocumented markup "{}". Use 32767 instead of LINK_MAX to
Bruce Evans [Fri, 5 Mar 1999 10:39:50 +0000 (10:39 +0000)]
Don't use undocumented markup "{}".  Use 32767 instead of LINK_MAX to
give bug for bug compatibility with intro.2.

25 years agoBacked out most of rev.1.4. I didn't submit it; I only submitted a
Bruce Evans [Fri, 5 Mar 1999 10:29:34 +0000 (10:29 +0000)]
Backed out most of rev.1.4.  I didn't submit it; I only submitted a
request for it something like it.  It was poorly worded and too
far from both POSIX wording and normal (mal)practice by referring to
sysconf(_SC_NGROUPS_MAX) instead of {NGROUPS_MAX} or NGROUPS.  POSIX.1
uses curly braces to mark up "symbolic constants or limits [that may
be] defined in certain headers".  Since we don't document this markup,
don't use it.  Just use NGROUPS_MAX.

25 years agoUse sysctlbyname() instead of sysctl().
Poul-Henning Kamp [Fri, 5 Mar 1999 10:01:24 +0000 (10:01 +0000)]
Use sysctlbyname() instead of sysctl().

25 years agoForgot to add $Id$ and change date in previous commit.
Bruce Evans [Fri, 5 Mar 1999 09:44:59 +0000 (09:44 +0000)]
Forgot to add $Id$ and change date in previous commit.

25 years agoNow all actual probing of both ATA & ATAPI devices are done after
Søren Schmidt [Fri, 5 Mar 1999 09:43:30 +0000 (09:43 +0000)]
Now all actual probing of both ATA & ATAPI devices are done after
interrupts are enabled, this kills the last "unwanted interrupts"
(and there is no ugly hacks like in the old driver to avoid them).
COmmand interrupt devices are now supported, this applies mostly
to older CDROM's and apparently also the ZIP.

Fixed problems:

    Number of total sectors wrong on some older drives.
        Fixed by not using the LBA size unless we know its valid.

There has also been more general code clenaups, some reorgs also.

25 years agoChanged the type of quotactl()'s 4th arg from `char *' to `void *'
Bruce Evans [Fri, 5 Mar 1999 09:28:33 +0000 (09:28 +0000)]
Changed the type of quotactl()'s 4th arg from `char *' to `void *'
so that non-sloppy applications can call it without using disgusting
casts to avoid warnings.  The 4th arg is sort of varargs -- it must
sometimes represent a filename, sometimes a struct pointer, and is
sometimes unused.  The arg type is still caddr_t in the kernel.

Obtained from: mostly from NetBSD

25 years agoFix an SMP-only TLB invalidation bug. Specifically, disable
Alan Cox [Fri, 5 Mar 1999 08:05:44 +0000 (08:05 +0000)]
Fix an SMP-only TLB invalidation bug.  Specifically, disable
a TLB invalidation optimization that won't work given the
limitations of our current SMP support.

This patch should be applied to -stable ASAP.

Thanks to John Capo <jc@irbs.com>,
          Steve Kargl <sgk@troutmask.apl.washington.edu>, and
          Chuck Robey <chuckr@mat.net>
for testing.

25 years agoexec_aout_imgact should lock the vm_map before calling vm_map_insert.
Alan Cox [Thu, 4 Mar 1999 18:04:40 +0000 (18:04 +0000)]
exec_aout_imgact should lock the vm_map before calling vm_map_insert.

Reviewed by: Matthew Dillon <dillon@apollo.backplane.com>,
"John S. Dyson" <dyson@iquest.net>, and
David Greenman <dg@root.com>

25 years agoExtend the ``set redial'' command to allow incremental
Brian Somers [Thu, 4 Mar 1999 17:42:15 +0000 (17:42 +0000)]
Extend the ``set redial'' command to allow incremental
redial timeouts.

25 years agoMake loader be able to load 1.2MB fd on machines w/ 1.44MB fdd.
KATO Takenori [Thu, 4 Mar 1999 16:38:12 +0000 (16:38 +0000)]
Make loader be able to load 1.2MB fd on machines w/ 1.44MB fdd.

Sumbmitted by: IMAI Takeshi <take-i@ceres.dti.ne.jp>

25 years agoAdded an entry for 1.44MB fd.
KATO Takenori [Thu, 4 Mar 1999 16:34:52 +0000 (16:34 +0000)]
Added an entry for 1.44MB fd.

Submitted by: IMAI Takeshi <take-i@ceres.dti.ne.jp>

25 years agoOops, forgot to read boot.config file.
KATO Takenori [Thu, 4 Mar 1999 16:31:55 +0000 (16:31 +0000)]
Oops, forgot to read boot.config file.

Submitted by: IMAI Takeshi <take-i@ceres.dti.ne.jp>

25 years agoAdopt syscons's recent change from key_t to keyent_t (this part of
Joerg Wunsch [Thu, 4 Mar 1999 10:50:36 +0000 (10:50 +0000)]
Adopt syscons's recent change from key_t to keyent_t (this part of
pcvt is plug-compatible to syscons to facilitate an identical
VT-switching interface), so PR misc/1247 can be closed.

25 years agoWhen SCSI IDs are not contiguous, the loader could not find correct
KATO Takenori [Thu, 4 Mar 1999 10:48:14 +0000 (10:48 +0000)]
When SCSI IDs are not contiguous, the loader could not find correct
unit number for da device.  This commit fixes this problem.

Submitted by: IMAI Takeshi <take-i@ceres.dti.ne.jp>

25 years agoboot2 doesn't need to say 'Can't find file boot.config.'
KATO Takenori [Thu, 4 Mar 1999 10:45:06 +0000 (10:45 +0000)]
boot2 doesn't need to say 'Can't find file boot.config.'

Submitted by: IMAI Takeshi <take-i@ceres.dti.ne.jp>

25 years agoFix character code of backslash. PC98 should use 0xfc instead of 0x5c.
KATO Takenori [Thu, 4 Mar 1999 10:44:01 +0000 (10:44 +0000)]
Fix character code of backslash.  PC98 should use 0xfc instead of 0x5c.

Submitted by: IMAI Takeshi <take-i@ceres.dti.ne.jp>

25 years agoSync with sys/i386/isa/sio.c revision 1.229.
KATO Takenori [Thu, 4 Mar 1999 10:37:30 +0000 (10:37 +0000)]
Sync with sys/i386/isa/sio.c revision 1.229.

25 years agoSync with sys/i386/conf/files.i386 revision 1.225.
KATO Takenori [Thu, 4 Mar 1999 10:36:27 +0000 (10:36 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.225.

25 years agoAdd the atapi fd driver (LS120 & ZIP drive support)
Søren Schmidt [Thu, 4 Mar 1999 09:53:19 +0000 (09:53 +0000)]
Add the atapi fd driver (LS120 & ZIP drive support)

25 years agoarprequest() allocates an mbuf with m_gethdr() but does not initialize
Bill Paul [Thu, 4 Mar 1999 04:03:57 +0000 (04:03 +0000)]
arprequest() allocates an mbuf with m_gethdr() but does not initialize
m->m_pkthdr.rcvif to NULL. Bad arprequest(). No biscuit.

25 years agoCorrect some ntohl/htonl bogons in the netmask handling.
Brian Somers [Wed, 3 Mar 1999 23:00:41 +0000 (23:00 +0000)]
Correct some ntohl/htonl bogons in the netmask handling.
This was pretty harmless as netmasks on a POINTOPOINT
interface are pretty much ignored, but it looked funny.

Mention the configured netmask in ``show ipcp''.

Describe in more detail what a proxy arp entry is.

25 years agoAdded driver to support ATAPI floppies ie LS-120 & ZIP drives.
Søren Schmidt [Wed, 3 Mar 1999 21:10:29 +0000 (21:10 +0000)]
Added driver to support ATAPI floppies ie LS-120 & ZIP drives.

Added "options ATA_STATIC_ID" that wires ATA disks like the old wd driver.

Fixed problems:

Dont use more sectors/intr than the drive supports.
Fix announce of > 8.4G disks.
Dont call ad_interrupt/ad_transfer when no disks config'd.
Use the right page# for CDR write mode params.
Fix breakage when no PCI support in kernel.
Implement DEVFS stuff.

General code clenaup.

25 years agoThe tunable parameter for the scheduler quantum was inverted.
Julian Elischer [Wed, 3 Mar 1999 18:15:29 +0000 (18:15 +0000)]
The tunable parameter for the scheduler quantum was inverted.
Higher numbers led to smaller quanta.
In discussion with BDE, change this parameter to be in uSecs
to make it machine independent,
and limit it to non zero multiples of 'tick' (rounding down).
Also make the variabel globally available so that the present function that
returns its value (used for posix scheduling I believe) can go away.

Submitted by: Bruce Evans <bde@freebsd.org>

25 years agoChange the comment to sendmail_flags which has scared so many people
Poul-Henning Kamp [Wed, 3 Mar 1999 15:04:33 +0000 (15:04 +0000)]
Change the comment to sendmail_flags which has scared so many people
from removing -bd to something more sensible.

25 years agoFixed null pointer panic for misconfigured master ports.
Bruce Evans [Wed, 3 Mar 1999 14:38:07 +0000 (14:38 +0000)]
Fixed null pointer panic for misconfigured master ports.

PR: i386/10273
Submitted by: Stephen Kiernan <sab@yin.vegamuse.org>

25 years agoFix the style of my previous commit.
Joerg Wunsch [Wed, 3 Mar 1999 14:02:50 +0000 (14:02 +0000)]
Fix the style of my previous commit.

Submitted by: The Style Police[tm]

25 years agoIncrease the vector that's used to look for free pty's. We go through
Joerg Wunsch [Wed, 3 Mar 1999 12:08:49 +0000 (12:08 +0000)]
Increase the vector that's used to look for free pty's.  We go through
/dev/ttypv right now, but window(1) was only looking up to ttypf,
causing an `Out of pseudo-terminals' when i just tried on freefall.

25 years agoAdded FE_8BIT_SUPPORT into the list. The fe driver includes opt_fe.h.
KATO Takenori [Wed, 3 Mar 1999 10:40:27 +0000 (10:40 +0000)]
Added FE_8BIT_SUPPORT into the list.  The fe driver includes opt_fe.h.

Submitted by: Takahashi Yoshihiro <nyan@dd.catv.ne.jp>

25 years agoThese were missing from the previous commit.
Warner Losh [Wed, 3 Mar 1999 07:08:03 +0000 (07:08 +0000)]
These were missing from the previous commit.

25 years agoBring in initial libc support for mips. These files were taken from
Warner Losh [Wed, 3 Mar 1999 07:06:17 +0000 (07:06 +0000)]
Bring in initial libc support for mips.  These files were taken from
the OpenBSD tree and should be considered preliminary.  They are here
to facilitate building of the tree.

25 years agoSlight cleanup of code resurected for union mounts..
Julian Elischer [Wed, 3 Mar 1999 02:35:51 +0000 (02:35 +0000)]
Slight cleanup of code resurected for union mounts..
Submitted by: Tony Finch <dot@dotat.at>

25 years agoTo avoid a conflict for the vm_map's lock with vm_fault, release
Alan Cox [Tue, 2 Mar 1999 22:55:02 +0000 (22:55 +0000)]
To avoid a conflict for the vm_map's lock with vm_fault, release
the read lock around the subyte operations in mincore.  After the lock is
reacquired, use the map's timestamp to determine if we need to restart
the scan.

25 years agoadd more MLINKS
Andrey A. Chernov [Tue, 2 Mar 1999 22:53:24 +0000 (22:53 +0000)]
add more MLINKS

25 years agoChange SAVECONFIG ioctl: it now writes an int parameter.
Greg Lehey [Tue, 2 Mar 1999 22:00:19 +0000 (22:00 +0000)]
Change SAVECONFIG ioctl: it now writes an int parameter.

25 years agoMake comment match code.
Julian Elischer [Tue, 2 Mar 1999 21:23:38 +0000 (21:23 +0000)]
Make comment match code.

25 years agoBetter logging when the Geometry register probe fails.
Warner Losh [Tue, 2 Mar 1999 20:56:07 +0000 (20:56 +0000)]
Better logging when the Geometry register probe fails.

25 years agoRemove inapropriate use of VOP_ISLOCKED()
Julian Elischer [Tue, 2 Mar 1999 20:26:39 +0000 (20:26 +0000)]
Remove inapropriate use of VOP_ISLOCKED()
This produced races resulting in panics and filesystem corruptions
under some circumstances.

Reviewed by: luoqi chen <luoqi@freebsd.org>
Reviewed by: Kirk McKusick <mckusick@mckusick.com>
Submitted by: Matt Dillon <dillon@freebsd.org>

25 years agoAdd new kernel option:
Semen Ustimenko [Tue, 2 Mar 1999 20:26:02 +0000 (20:26 +0000)]
Add new kernel option:

options NTFS

25 years agoMake window(1) actually work again. It has been broken for quite some
Joerg Wunsch [Tue, 2 Mar 1999 19:08:09 +0000 (19:08 +0000)]
Make window(1) actually work again.  It has been broken for quite some
time now.

For whatever reason, the kernel seems to have generated SIGIOs
previously without an initial fcntl(...,F_SETOWN), but does no longer.
This caused window(1) to wait indefinitely for input.

Also, undo rev 1.3 of wwspawn.c, it was not well-thought, and
apparently not even tested at all.  The blindly (even in a nonsensical
place like the comment on top of the function) applied replacement of
vfork() by fork() totally ignored that window(1) *does* abuse the
feature of vfork() where a modification of the parent's address space
is possible (in this case, to notify the parent of an erred exec*).
Also, with vfork(), it is guaranteed that the parent is only woken up
after the exec*() happened, where the replacement by fork() made the
parent to almost always become runnable again before the child, in
which case the parent simply told `subprocess died'.  Unfortunately,
working around _this_ seems to be a lot more of redesign work compared
to little gained value, so i think relying on the specifics of vfork()
is the simpler way.

Submitted by: Philipp Mergenthaler <un1i@rz.uni-karlsruhe.de>

25 years agoCorrect casts in vtophys and avtophys to be vm_offset_t.
David Greenman [Tue, 2 Mar 1999 16:20:39 +0000 (16:20 +0000)]
Correct casts in vtophys and avtophys to be vm_offset_t.

25 years agoDon't treat empty lines as end-of-file in include(). (I can't
Daniel C. Sobral [Tue, 2 Mar 1999 16:16:57 +0000 (16:16 +0000)]
Don't treat empty lines as end-of-file in include(). (I can't
believe this bug is still here... :-( )

25 years agoPut back the bktr update to generate /dev/vbi
Roger Hardiman [Tue, 2 Mar 1999 16:04:27 +0000 (16:04 +0000)]
Put back the bktr update to generate /dev/vbi
which I managed to delete yesterday.

25 years agoMerge with sys/i386/conf/files.i386 revision 1.224.
KATO Takenori [Tue, 2 Mar 1999 14:03:21 +0000 (14:03 +0000)]
Merge with sys/i386/conf/files.i386 revision 1.224.

25 years agoSync with sys/i386/conf/Makefile.i386 revision 1.140.
KATO Takenori [Tue, 2 Mar 1999 14:01:33 +0000 (14:01 +0000)]
Sync with sys/i386/conf/Makefile.i386 revision 1.140.

25 years agoRestored rev.1.2, which was clobbered by rev.1.4 (don't print "Making
Bruce Evans [Tue, 2 Mar 1999 13:04:00 +0000 (13:04 +0000)]
Restored rev.1.2, which was clobbered by rev.1.4 (don't print "Making
info file ..." except in verbose mode).

25 years agoFix for LINE30 option. This option was not tested under new console
KATO Takenori [Tue, 2 Mar 1999 12:34:24 +0000 (12:34 +0000)]
Fix for LINE30 option.  This option was not tested under new console
driver.

Submitted by: Takahashi Yoshihiro <nyan@dd.catv.ne.jp>

25 years agoTypo fix.
Joseph Koshy [Tue, 2 Mar 1999 07:27:01 +0000 (07:27 +0000)]
Typo fix.

PR: docs/10348
Submitted by: Chris Costello <chris@holly.dyndns.org>

25 years agoChange some .Ar macros to .Pa.
Greg Lehey [Tue, 2 Mar 1999 07:03:30 +0000 (07:03 +0000)]
Change some .Ar macros to .Pa.

Describe /dev/vinum/control*

Describe drive "referenced" state.

Remove warning about kldunload; it seems to work now.

Still more descriptions of how to debug things.

25 years agoDocument "stop" command with no parameters.
Greg Lehey [Tue, 2 Mar 1999 07:02:29 +0000 (07:02 +0000)]
Document "stop" command with no parameters.

Document "printconfig" and "saveconfig" commands.

Add warning about running "create" multiple times.

25 years agoTake definitions of VINUMMOD and WRONGMOD from v.c.
Greg Lehey [Tue, 2 Mar 1999 07:02:09 +0000 (07:02 +0000)]
Take definitions of VINUMMOD and WRONGMOD from v.c.

25 years agoMove definitions of VINUMMOD and WRONGMOD to vext.h.
Greg Lehey [Tue, 2 Mar 1999 07:01:51 +0000 (07:01 +0000)]
Move definitions of VINUMMOD and WRONGMOD to vext.h.

Wait4 zombies.

make_devices: Don't try if the /dev directory is mounted read-only.
      Create daemon superdevice /dev/vinum/controld.

Format a couple of multiline comments conformant with style(9).

25 years agoChange the format of the lists to reflect loss of pid field.
Greg Lehey [Tue, 2 Mar 1999 07:01:26 +0000 (07:01 +0000)]
Change the format of the lists to reflect loss of pid field.
Show open state as a flag.

vinum_ldi: Show available space in short form listing.

25 years agoresetconfig: don't log the "obliterated" message, the kernel does that
Greg Lehey [Tue, 2 Mar 1999 07:00:57 +0000 (07:00 +0000)]
resetconfig: don't log the "obliterated" message, the kernel does that
     for us.

     Rebuild the (almost empty) /dev/vinum directory.

vinum_start:  remove superflous "read" parameter when starting with no
      parameters.

vinum_stop:  without an argument, stop Vinum and remove the kld if
     it's idle.

vinum_saveconfig: New command to save configuration.

25 years agoCorrect some diagrams
Greg Lehey [Tue, 2 Mar 1999 06:58:09 +0000 (06:58 +0000)]
Correct some diagrams

Remove pid field from struct sd, struct plex and struct volume

Add flags VF_STOPPING, VF_OPEN and VF_DAEMONOPEN

25 years agosizespec: allow negative values (which signal error conditions).
Greg Lehey [Tue, 2 Mar 1999 06:57:48 +0000 (06:57 +0000)]
sizespec: allow negative values (which signal error conditions).
  catch missing (NULL) parameters.

25 years agoChange the name of drive state "uninit" to "referenced": it only
Greg Lehey [Tue, 2 Mar 1999 06:57:30 +0000 (06:57 +0000)]
Change the name of drive state "uninit" to "referenced": it only
occurs if we have a dangling pointer to a drive.

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:57:08 +0000 (06:57 +0000)]
Change printf() calls to log() calls.

set_sd_state: allow stop command to stop a reborn subdisk.

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:56:39 +0000 (06:56 +0000)]
Change printf() calls to log() calls.

25 years agoAdd keywords "size", synonym for "length", and "max".
Greg Lehey [Tue, 2 Mar 1999 06:56:05 +0000 (06:56 +0000)]
Add keywords "size", synonym for "length", and "max".

Check for NULL name parameters.

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:55:44 +0000 (06:55 +0000)]
Change printf() calls to log() calls.

expand_table: Zero out newly allocated tables

25 years agoAdd keywords "size", synonym for "length", and "max".
Greg Lehey [Tue, 2 Mar 1999 06:55:18 +0000 (06:55 +0000)]
Add keywords "size", synonym for "length", and "max".

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:54:57 +0000 (06:54 +0000)]
Change printf() calls to log() calls.

Change VINUM_SAVECONFIG: it now requires a parameter.  0 means
"configuration updates are finished, please save", and 1 means "please
just save the config".  This second meaning is invoked by the new
"saveconfig" command to vinum(8).

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:54:30 +0000 (06:54 +0000)]
Change printf() calls to log() calls.

Recognize "referenced" drives by the lack of a slash in the device
name, not by a NUL character.

vinum_scandisk: return error indication (ENOENT if we can't find any
vinum drive, otherwise 0).

VINUM_SAVECONFIG: change parameters.

25 years agoNew function return_drive_space
Greg Lehey [Tue, 2 Mar 1999 06:53:42 +0000 (06:53 +0000)]
New function return_drive_space

Add definitions of vinum_inactive and free_vinum, formerly local to
vinum.c

vinum_scandisk now returns int (error number)

25 years agochange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:53:15 +0000 (06:53 +0000)]
change printf() calls to log() calls.

Don't save config while we're reading it from disk.

Change the way we handle the daemon: if we can't communicate with it
for 1 second (which is possible), start a new one.  The daemon saves
its pid in daemonpid; on each iteration of the main loop the daemon
checks whether it's still in favour.  If not, it silently exits.
Also, when trying to communicate with the daemon, check daemonpid
first.  If it's set to 0, don't even try.

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:52:47 +0000 (06:52 +0000)]
Change printf() calls to log() calls.

Rename the VF_KERNELOP to VF_DISKCONFIG and checkkernel () to
checkdiskconfig (), which better describes their function.

Disable configuration updates if we have an error reading in the
configuration.  This stops a "shoot-in-foot" problem where a mistake
can cause the configuration to be obliterated.

Tidy up some messages, which included superfluous \ns.

Recognize RAID-5 configuration information even in the non-RAID-5
version.  This fixes shoot-in-foot problems where starting the wrong
version of vinum would kill RAID-5 plexes.

Recognize drives that have been referenced, but for which no physical
location is known.  This is part of a modification which will
ultimately allow incrementally reading configurations.  Such drives
will have a device name "unknown".

New function return_drive_space () returns space to a drive.
Previously this was part of free_sd ().

give_sd_to_drive: don't do it if the subdisk needs more space than the
drive has available.

config_sd: if reading config from disk, accept plex offset, drive
offset and length specs of -1 to indicate error conditions.

parse_config: return ENOENT if the "read" command doesn't find any
drives.

remove_sd_entry: don't do it, even by force, if it's open.

If the size of a striped or RAID-5 plex is not an integral multiple of
the stripe size, trim the size until it is.

reinstate update_volume_config, which had atrophied, to recalculate
the size of a volume if a plex has shrunk due to stripe size
considerations.

25 years agoChange printf() calls to log() calls.
Greg Lehey [Tue, 2 Mar 1999 06:52:11 +0000 (06:52 +0000)]
Change printf() calls to log() calls.

vinumattach: Zero out tables after allocating them

Modify procedure at unload: if a vinum(8) has the superdev open, don't
close down.  If only the daemon has it open, send the daemon a stop
request and wait for it to close the superdev, then unload.

In order to do this, create a second superdev which is opened by the
daemon.  The open and close routines set a different bit in
vinum_conf.flags; otherwise the treatment is identical.

Remove opencount field in vol structure; replace by a flag bit, since
we can't count the number of opens.

Remove dead LKM grunge.

25 years agoChange the name of drive state "uninit" to "referenced": it only
Greg Lehey [Tue, 2 Mar 1999 06:51:48 +0000 (06:51 +0000)]
Change the name of drive state "uninit" to "referenced": it only
occurs if we have a dangling pointer to a drive.

25 years agoReorganize locking to avoid holding the lock during calls to bdwrite
Kirk McKusick [Tue, 2 Mar 1999 06:38:07 +0000 (06:38 +0000)]
Reorganize locking to avoid holding the lock during calls to bdwrite
and brelse (which may sleep in some systems).

Obtained from: Matthew Dillon <dillon@apollo.backplane.com>

25 years agoPrint actual density code as well as string for density- I dunno about others,
Matt Jacob [Tue, 2 Mar 1999 06:27:59 +0000 (06:27 +0000)]
Print actual density code as well as string for density- I dunno about others,
but I sure remember 0x15 easier than 'ECMA 17'. Also handle density codes
0 (default) and 0x7f ('same') as special cases.

25 years agoRemove the last of the share map code: struct vm_map::is_main_map.
Alan Cox [Tue, 2 Mar 1999 05:43:18 +0000 (05:43 +0000)]
Remove the last of the share map code: struct vm_map::is_main_map.

Reviewed by: Matthew Dillon <dillon@apollo.backplane.com>

25 years agoMerge patch to ufs_vnops.c's ufs_rename to the copy of ufs_rename that
Warner Losh [Tue, 2 Mar 1999 05:31:47 +0000 (05:31 +0000)]
Merge patch to ufs_vnops.c's ufs_rename to the copy of ufs_rename that
lives in ext2_vnops.c for ext2fs.  Also remove cast from comparision.
Bruce pointed out that it was bogus since we'd force a signed
comparision when we really wanted an unsigned comparison.

25 years agoFix reference to rpc.yppasswdd.
Guy Helmer [Tue, 2 Mar 1999 04:14:53 +0000 (04:14 +0000)]
Fix reference to rpc.yppasswdd.

PR: docs/10171