]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoThis commit was generated by cvs2svn to compensate for changes in r69482,
ru [Fri, 1 Dec 2000 15:36:10 +0000 (15:36 +0000)]
This commit was generated by cvs2svn to compensate for changes in r69482,
which included commits to RCS files with non-trunk default branches.

23 years agoFollow POLA and bind the <End> key to the `$' command.
ru [Fri, 1 Dec 2000 15:36:10 +0000 (15:36 +0000)]
Follow POLA and bind the <End> key to the `$' command.

This is committed to the vendor (BOSTIC) branch because
it was accepted (committed) by ``nvi'' maintainers:

On Fri, Dec 01, 2000 at 03:51:14PM +0100, Sven Verdoolaege wrote:
[...]
> RCS file: /b/CVSROOT/vi/cl/cl_bsd.c,v
> Working file: cl_bsd.c
> head: 8.32
> branch:
> locks: strict
> access list:
> symbolic names:
>         nvi1-81-1: 8.31
>         nvi1-81-0: 8.30
>         nvi1-80: 8.29.0.2
> keyword substitution: kv
> total revisions: 32;    selected revisions: 1
> description:
> ----------------------------
> revision 8.32
> date: 2000/12/01 13:56:17;  author: skimo;  state: Exp;  lines: +2 -1
> allow use of <End> key on AT-compatible keyboards (Ruslan Ermilov <ru@FreeBSD.org>)
> =============================================================================
>
> RCS file: /b/CVSROOT/vi/cl/cl_term.c,v
> Working file: cl_term.c
> head: 10.28
> branch:
> locks: strict
> access list:
> symbolic names:
>         nvi1-81-1: 10.27
>         nvi1-81-0: 10.23
>         nvi1-80: 10.23.0.2
> keyword substitution: kv
> total revisions: 47;    selected revisions: 1
> description:
> ----------------------------
> revision 10.28
> date: 2000/12/01 13:56:18;  author: skimo;  state: Exp;  lines: +2 -1
> allow use of <End> key on AT-compatible keyboards (Ruslan Ermilov <ru@FreeBSD.org>)

23 years agoConvert the pcib_{read,write}_config args from signed to unsigned,
gallatin [Fri, 1 Dec 2000 15:27:48 +0000 (15:27 +0000)]
Convert the pcib_{read,write}_config args from signed to unsigned,
like the args to the config space accessors these functions replaced.

This reduces the likelyhood of overflow when the args are used in
macros on the alpha.  This prevents memory management faults when
probing the pci bus on sables, multias and nonames.

Approved by: dfr
Tested by: Bernd Walter <ticso@cicely8.cicely.de>

23 years agoAdd PCI id for the RAID LC2 controller.
jlemon [Fri, 1 Dec 2000 15:26:11 +0000 (15:26 +0000)]
Add PCI id for the RAID LC2 controller.

Pick up correct location for the DEC version of the chip, this
got broken in a previous commit.

23 years agoThis page requires tbl(1).
ru [Fri, 1 Dec 2000 13:51:33 +0000 (13:51 +0000)]
This page requires tbl(1).

23 years agoFix a "#ifdef" where a comparison operator was used.
obrien [Fri, 1 Dec 2000 12:29:00 +0000 (12:29 +0000)]
Fix a "#ifdef" where a comparison operator was used.
GCC 2.96 does not like this.

23 years agoDisconnect sysvipc, since it doesn't build.
des [Fri, 1 Dec 2000 12:26:39 +0000 (12:26 +0000)]
Disconnect sysvipc, since it doesn't build.

23 years agoCatch up to perl-5.6.0.
sheldonh [Fri, 1 Dec 2000 12:11:24 +0000 (12:11 +0000)]
Catch up to perl-5.6.0.

23 years agoRemove duplicate definition.
msmith [Fri, 1 Dec 2000 12:10:06 +0000 (12:10 +0000)]
Remove duplicate definition.

23 years agoUpdate local changes for new hardware register interface.
msmith [Fri, 1 Dec 2000 12:09:18 +0000 (12:09 +0000)]
Update local changes for new hardware register interface.

23 years agoProperly check malloc(3) return values
jedgar [Fri, 1 Dec 2000 12:02:16 +0000 (12:02 +0000)]
Properly check malloc(3) return values

Approved by: ken

23 years agoAdd more comments, fix a typo, mention how to do PPPoUDP using encryption
brian [Fri, 1 Dec 2000 11:52:22 +0000 (11:52 +0000)]
Add more comments, fix a typo, mention how to do PPPoUDP using encryption
to create a VPN.

23 years agoEliminate a few more groff(1) warnings.
ru [Fri, 1 Dec 2000 11:33:55 +0000 (11:33 +0000)]
Eliminate a few more groff(1) warnings.

23 years agoRemove some obsolete comments.
des [Fri, 1 Dec 2000 11:05:15 +0000 (11:05 +0000)]
Remove some obsolete comments.

23 years agoClean up the whitespace encoding code.
des [Fri, 1 Dec 2000 11:04:57 +0000 (11:04 +0000)]
Clean up the whitespace encoding code.

23 years agoAdd SCHEME_HTTPS.
des [Fri, 1 Dec 2000 11:04:43 +0000 (11:04 +0000)]
Add SCHEME_HTTPS.

23 years agoUpdate the 3-June-2000 import delete list to keep mkdeps.[ch] which
obrien [Fri, 1 Dec 2000 11:02:05 +0000 (11:02 +0000)]
Update the 3-June-2000 import delete list to keep mkdeps.[ch] which
is needed by the new C preprocessor.

23 years agoComment an endif.
obrien [Fri, 1 Dec 2000 10:20:17 +0000 (10:20 +0000)]
Comment an endif.

23 years agoAcpiOsMem primitives as required by the new ACPI CA snapshot
msmith [Fri, 1 Dec 2000 10:19:28 +0000 (10:19 +0000)]
AcpiOsMem primitives as required by the new ACPI CA snapshot

23 years agoUpdate to work with the new ACPI CA snapshot.
msmith [Fri, 1 Dec 2000 10:18:57 +0000 (10:18 +0000)]
Update to work with the new ACPI CA snapshot.

 - Use ACPI_PHYSICAL_ADDRESS
 - RSDT -> XSDT
 - FACP -> FADT
 - No APIC table support
 - Don't install a global EC handler; this has bad side-effects
   (it invokes _REG in *all* EC spaces in the namespace!)
 - Check for PCI bus instances already existing before adding them

23 years agoExit if malloc() returns NULL
brian [Fri, 1 Dec 2000 09:59:40 +0000 (09:59 +0000)]
Exit if malloc() returns NULL

Submitted by: Chris Faulhaber <jedgar@fxp.org>

23 years agoWe need support for comparing quad_t's now (ACPI CA uses this).
msmith [Fri, 1 Dec 2000 09:54:17 +0000 (09:54 +0000)]
We need support for comparing quad_t's now (ACPI CA uses this).

23 years agoAdd one new file brought in with the ACPI CA 20001115 import.
msmith [Fri, 1 Dec 2000 09:52:47 +0000 (09:52 +0000)]
Add one new file brought in with the ACPI CA 20001115 import.

23 years agoResolve conflicts from the vendor update.
msmith [Fri, 1 Dec 2000 09:49:35 +0000 (09:49 +0000)]
Resolve conflicts from the vendor update.

23 years agoThe GCC 2.96 snapshots have slightly different rules for finding include
obrien [Fri, 1 Dec 2000 09:39:28 +0000 (09:39 +0000)]
The GCC 2.96 snapshots have slightly different rules for finding include
files.  Mostly -I${.CURDIR} was needed -- especially for YACC generated
files as the new cpp does not look in the ultimate source file
(ie, the .y file)'s directory as told by the "#line" directive.  Some were
misspellings of "-I${.CURDIR}" as "-I.".

23 years agoThis commit was generated by cvs2svn to compensate for changes in r69450,
msmith [Fri, 1 Dec 2000 09:36:25 +0000 (09:36 +0000)]
This commit was generated by cvs2svn to compensate for changes in r69450,
which included commits to RCS files with non-trunk default branches.

23 years agoUpdate to the 20001115 Intel ACPI CA snapshot.
msmith [Fri, 1 Dec 2000 09:36:25 +0000 (09:36 +0000)]
Update to the 20001115 Intel ACPI CA snapshot.

23 years agosysvipc loadable.
alfred [Fri, 1 Dec 2000 08:57:47 +0000 (08:57 +0000)]
sysvipc loadable.

new syscall entry lkmressys - "reserved loadable syscall"

Make syscall_register allow overwriting of such entries (lkmressys).

23 years agoAdd reserved lkmressys keyword. I swear, this script will die the
alfred [Fri, 1 Dec 2000 08:47:54 +0000 (08:47 +0000)]
Add reserved lkmressys keyword.  I swear, this script will die the
next time I need to hack on it.

23 years agomdoc(7) police: fixed errors from the previous revision, keep
ru [Fri, 1 Dec 2000 08:17:15 +0000 (08:17 +0000)]
mdoc(7) police: fixed errors from the previous revision, keep
the synopsis in a nice form, and a few more cosmetic changes.

23 years agoimplement NOSTD syscall type, this creates the syscall args, but sticks
alfred [Fri, 1 Dec 2000 07:40:20 +0000 (07:40 +0000)]
implement NOSTD syscall type, this creates the syscall args, but sticks
a lkmnosys into the sysent table so that SYSCALL_MODULE() works

23 years agoTranslate alfred to english.
alfred [Fri, 1 Dec 2000 06:59:18 +0000 (06:59 +0000)]
Translate alfred to english.

Submitted by: bde

23 years agoThere is no src/contrib-crypto/ anything directory. So don't look for
obrien [Fri, 1 Dec 2000 06:34:44 +0000 (06:34 +0000)]
There is no src/contrib-crypto/ anything directory.  So don't look for
include files in subdirs of it.

23 years agoProtect access to p_stat with sched_lock.
jhb [Fri, 1 Dec 2000 06:34:19 +0000 (06:34 +0000)]
Protect access to p_stat with sched_lock.

23 years agoFix this slightly better by using NON_GPROF_RET instead of duplicating
jhb [Fri, 1 Dec 2000 05:29:21 +0000 (05:29 +0000)]
Fix this slightly better by using NON_GPROF_RET instead of duplicating
hard-coded asm.

Suggested by: bde

23 years ago- Add a hack for "psmintr: out of sync.." This is NOT a fix,
yokota [Fri, 1 Dec 2000 05:26:24 +0000 (05:26 +0000)]
- Add a hack for "psmintr: out of sync.." This is NOT a fix,
  but a hack! Add `flags 0x8000' to the psm driver to enable it.
  The psm driver will try to get out of out-of-sync situation
  by disabling the mouse and immediately enable it again.

If you are seeing this out-of-sync problem because of an
incompetent(?!) KVM switch, this hack will NOT be good
for you.  However, if you are occasionally seeing the
problem because of lost mouse interrupt, this might help.

23 years ago- Slightly rearrnage IntelliMouse Explorer and Logitech
yokota [Fri, 1 Dec 2000 05:24:30 +0000 (05:24 +0000)]
- Slightly rearrnage IntelliMouse Explorer and Logitech
  MouseMan+ identification routines for efficiency.
  No functional change.

23 years agoUse an mp-safe callout for endtsleep.
jake [Fri, 1 Dec 2000 04:55:52 +0000 (04:55 +0000)]
Use an mp-safe callout for endtsleep.

23 years agoUse msleep() instead of mtx_exit()/tsleep() so that we release the lock and
jhb [Fri, 1 Dec 2000 03:43:33 +0000 (03:43 +0000)]
Use msleep() instead of mtx_exit()/tsleep() so that we release the lock and
go to sleep as an "atomic" operation.

23 years agoDon't update p_stat in exit1() to SZOMB until after releasing the allproc
jhb [Fri, 1 Dec 2000 03:42:17 +0000 (03:42 +0000)]
Don't update p_stat in exit1() to SZOMB until after releasing the allproc
lock.  Otherwise, if we block on the backing mutex while releasing the
allproc lock, then when we resume, we will be at SRUN, and we will stay
that way all the way through cpu_exit.  As a result, our parent will never
harvest us.

23 years agoMention the MTU in ``show iface''.
brian [Fri, 1 Dec 2000 03:22:18 +0000 (03:22 +0000)]
Mention the MTU in ``show iface''.

23 years agoUse msleep instead of mtx_exit; tsleep; mtx_enter, which is not safe.
jake [Fri, 1 Dec 2000 02:18:38 +0000 (02:18 +0000)]
Use msleep instead of mtx_exit; tsleep; mtx_enter, which is not safe.

23 years agoChange doreti to take a trapframe instead of an intrframe.
jake [Fri, 1 Dec 2000 02:09:45 +0000 (02:09 +0000)]
Change doreti to take a trapframe instead of an intrframe.
Remove associated pushes of dummy units to convert frame.

Reviewed by: jhb

23 years agoCosmetique: don't use localized `date`
ache [Fri, 1 Dec 2000 01:58:48 +0000 (01:58 +0000)]
Cosmetique: don't use localized `date`

23 years agoSplit the WITNESS and MUTEX_DEBUG options apart so that WITNESS does not
jhb [Fri, 1 Dec 2000 00:10:59 +0000 (00:10 +0000)]
Split the WITNESS and MUTEX_DEBUG options apart so that WITNESS does not
depend on MUTEX_DEBUG.  The MUTEX_DEBUG option turns on extra assertions
and checks to verify that mutexes themselves are implemented properly.
The WITNESS option uses extra checks and diagnostics to verify that other
code is using mutexes properly.

23 years agoKill #define of DETACH_FORCE since it is now in card_if.m
imp [Thu, 30 Nov 2000 23:30:47 +0000 (23:30 +0000)]
Kill #define of DETACH_FORCE since it is now in card_if.m

23 years agoo Make comment match reality, synch code with comment.
rwatson [Thu, 30 Nov 2000 23:14:55 +0000 (23:14 +0000)]
o Make comment match reality, synch code with comment.
o In practice: the comment indicates that all but umask and
  environmental variables of the users login class are applied when su
  occurs, unless -m is used to specify a class.  This was incorrect;
  in practice, the uid, gids, resources, and priority were set, and
  then resources and priority were selectively removed.  This meant
  that some aspects of the user context were not set, including handling
  of login events (wtmp, utmp), as well as the path specified in
  login.conf.
o I changed it so that the behavior is the same, but instead,
  LOGIN_SETALL is used, and appropriate flags are removed, including
  the LOGIN_SETLOGIN and LOGIN_SETPATH entries that were implicitly
  not present before.  I also updated the comment to reflect
  reality, selecting reality as the "correct" behavior.
o This has the practical benefit that as new LOGIN_SET* flags are
  introduced, they are supported by su unless specifically disabled.
  For example, of a LOGIN_SETLABEL flag is introduced to support
  MAC labels determined by the user's login class, then su no longer
  has to be modified.
o It might be desirable to have su use LOGIN_SETPATH depending on
  its command line parameters, as it might or might not be
  considered part of the "environment".

Obtained from: TrustedBSD Project

23 years agoGet rid of that damned ``i4b-L4-i4b_l4_setup_timeout: 975625189: ERROR:
brian [Thu, 30 Nov 2000 23:08:36 +0000 (23:08 +0000)]
Get rid of that damned ``i4b-L4-i4b_l4_setup_timeout: 975625189: ERROR:
idletime[900]+earlyhup[5] > unitlength[60]!'' message.

23 years agoAdd missing && which absense stops 'make world'
ache [Thu, 30 Nov 2000 23:01:43 +0000 (23:01 +0000)]
Add missing && which absense stops 'make world'

23 years agoAdd some comments and $FreeBSD$
brian [Thu, 30 Nov 2000 22:24:51 +0000 (22:24 +0000)]
Add some comments and $FreeBSD$

23 years agoUpdate some misleading comments, empty ``set logout'', and add a
brian [Thu, 30 Nov 2000 22:22:09 +0000 (22:22 +0000)]
Update some misleading comments, empty ``set logout'', and add a
comment describing how to take advantage of telcos that give
free phone calls at certain times.

23 years agologin.conf: we don't want coredumps on picobsd
luigi [Thu, 30 Nov 2000 21:43:37 +0000 (21:43 +0000)]
login.conf: we don't want coredumps on picobsd
rc.conf: make the system recognise the MAC address and assign an IP
    automatically from /etc/hosts (or ask the user)
sshd_config: don't do x11 forwarding.

23 years agoUpgrade to 6.10
ache [Thu, 30 Nov 2000 21:38:37 +0000 (21:38 +0000)]
Upgrade to 6.10

23 years agoremove an undefined reference when no VLAN support is compiled in.
luigi [Thu, 30 Nov 2000 21:35:09 +0000 (21:35 +0000)]
remove an undefined reference when no VLAN support is compiled in.

23 years agoMake crunchgen aware of obj directories. Formerly it had
luigi [Thu, 30 Nov 2000 21:14:54 +0000 (21:14 +0000)]
Make crunchgen aware of obj directories. Formerly it had
/usr/obj hardwired in the code, now you can override it
with a command line option or MAKEOBJDIRPREFIX env. variable.
The above is useful to build picobsd-specific objects in some other
place than /usr/obj

While at it, fix documentation and change a few sprintf -> snprintf.

23 years agoThis file not present in 6.10
ache [Thu, 30 Nov 2000 21:07:50 +0000 (21:07 +0000)]
This file not present in 6.10

23 years agoo Add a comment to exec_check_permissions() to indicate that the
rwatson [Thu, 30 Nov 2000 21:06:05 +0000 (21:06 +0000)]
o Add a comment to exec_check_permissions() to indicate that the
  passed vnode must be locked; this is the case because of calls
  to VOP_GETATTR(), VOP_ACCESS(), and VOP_OPEN().  This becomes
  more of an issue when VOP_ACCESS() gets a bit more complicated,
  which it does when you introduce ACL, Capability, and MAC
  support.

Obtained from: TrustedBSD Project

23 years agoThis commit was generated by cvs2svn to compensate for changes in r69408,
ache [Thu, 30 Nov 2000 21:05:33 +0000 (21:05 +0000)]
This commit was generated by cvs2svn to compensate for changes in r69408,
which included commits to RCS files with non-trunk default branches.

23 years agoInitial import of slightly trimmed tcsh 6.10
ache [Thu, 30 Nov 2000 21:05:33 +0000 (21:05 +0000)]
Initial import of slightly trimmed tcsh 6.10

23 years agoonly call bwillwrite() to stall on IO when dealing with VNODEs otherwise
alfred [Thu, 30 Nov 2000 20:23:14 +0000 (20:23 +0000)]
only call bwillwrite() to stall on IO when dealing with VNODEs otherwise
we will stall on non-disk IO for things like fifos and sockets

23 years agoRevert the previous change to this file. We have to hardcode in the opcode
jhb [Thu, 30 Nov 2000 20:07:49 +0000 (20:07 +0000)]
Revert the previous change to this file.  We have to hardcode in the opcode
for return because we do Evil Things(tm) with a 'ret' macro in
asmacros.h.

Noticed by: markm

23 years ago This is a fix for a problem described in PR kern/19572. It was
alfred [Thu, 30 Nov 2000 20:04:44 +0000 (20:04 +0000)]
This is a fix for a problem described in PR kern/19572. It was
    recently discussed at -hackers. The problem is a null-pointer
    dereference that happens in kern/vfs_lookup.c when accessing ".."
    with a v_mount entry for the current directory vnode of NULL. This
    happens when a volume is forcibly unmounted, and the vnode for a
    working directory in the mounted volume is cleared.

PR: 23191
Submitted by: Thomas Moestl <tmoestl@gmx.net>

23 years agoDocument the mtxname keyword.
jhb [Thu, 30 Nov 2000 19:17:56 +0000 (19:17 +0000)]
Document the mtxname keyword.

Reviewed by: sheldonh

23 years agouse a oppurtunistic locking strategy with the uidinfo structures to avoid
alfred [Thu, 30 Nov 2000 19:15:22 +0000 (19:15 +0000)]
use a oppurtunistic locking strategy with the uidinfo structures to avoid
locking the global hash on each uifree()

make struct uidinfo only visible to the kernel

make uihold() a function rather than a macro to reduce bloat

swap the order of a spl/mutex to maintain consistancy

23 years agoRemove an unused variable.
gibbs [Thu, 30 Nov 2000 19:14:26 +0000 (19:14 +0000)]
Remove an unused variable.

Properly advance to the next image while searching for the ROM image
that contains CIS data.

KNF an if statement.

23 years agomake crfree into a function rather than a macro to avoid bloat because of
alfred [Thu, 30 Nov 2000 19:09:48 +0000 (19:09 +0000)]
make crfree into a function rather than a macro to avoid bloat because of
the mutex aquire/release

reorder struct ucred

23 years agoGet rid of a bogus mtx_exit (it was attempting to release an
mckusick [Thu, 30 Nov 2000 19:09:29 +0000 (19:09 +0000)]
Get rid of a bogus mtx_exit (it was attempting to release an
already released mutex).

Submitted by: "Chris Knight" <chris@aims.com.au>

23 years agoremove unneded sys/ucred.h includes
alfred [Thu, 30 Nov 2000 18:52:32 +0000 (18:52 +0000)]
remove unneded sys/ucred.h includes

23 years agoEliminate groff(1) warnings.
ru [Thu, 30 Nov 2000 18:39:57 +0000 (18:39 +0000)]
Eliminate groff(1) warnings.

23 years agoremove unneded sys/ucred.h include
alfred [Thu, 30 Nov 2000 18:34:08 +0000 (18:34 +0000)]
remove unneded sys/ucred.h include

23 years agoSet the proper bit in the howto flags for a serial console rather than
jhb [Thu, 30 Nov 2000 18:33:59 +0000 (18:33 +0000)]
Set the proper bit in the howto flags for a serial console rather than
setting the index of the bit.  (0xc vs. 0x1000)

23 years agoremove unneeded sys/ucred.h include
alfred [Thu, 30 Nov 2000 18:33:02 +0000 (18:33 +0000)]
remove unneeded sys/ucred.h include

23 years agoEliminate groff(1) warnings.
ru [Thu, 30 Nov 2000 17:06:28 +0000 (17:06 +0000)]
Eliminate groff(1) warnings.

23 years agoAdd support for probing the keyboard from pxeboot which will behave
ps [Thu, 30 Nov 2000 14:55:10 +0000 (14:55 +0000)]
Add support for probing the keyboard from pxeboot which will behave
exactly the same as passing -P to boot2.

Submitted by: jhb

23 years agoFormat string paranoia. This should avoid potential buffer overflows from
will [Thu, 30 Nov 2000 13:56:19 +0000 (13:56 +0000)]
Format string paranoia.  This should avoid potential buffer overflows from
user input (in its ever-broadening definition).

Obtained from: NetBSD

23 years agoAdd more environment variables to be filtered through scrub_env().
asmodai [Thu, 30 Nov 2000 13:14:54 +0000 (13:14 +0000)]
Add more environment variables to be filtered through scrub_env().
Synched from normal telnet.

23 years agoString paranoia fix. Synched from normal telnet.
asmodai [Thu, 30 Nov 2000 13:10:01 +0000 (13:10 +0000)]
String paranoia fix.  Synched from normal telnet.

23 years agoBackout rev 1.2 (this paper is in me(7) not in ms(7) format),
ru [Thu, 30 Nov 2000 13:07:14 +0000 (13:07 +0000)]
Backout rev 1.2 (this paper is in me(7) not in ms(7) format),
eliminate remaining groff(1) warnings.

23 years agoAdd make.conf(5).
nik [Thu, 30 Nov 2000 12:10:18 +0000 (12:10 +0000)]
Add make.conf(5).

23 years agoString paranoia. Merged from regular telnet.
asmodai [Thu, 30 Nov 2000 10:55:25 +0000 (10:55 +0000)]
String paranoia.  Merged from regular telnet.

23 years agoIf the transfer timed out, but we don't know how large the file is supposed
des [Thu, 30 Nov 2000 10:08:22 +0000 (10:08 +0000)]
If the transfer timed out, but we don't know how large the file is supposed
to be, assume it was truncated.

23 years agoThe osf1 utility has not been installed since rev 1.5 of
sheldonh [Thu, 30 Nov 2000 08:20:42 +0000 (08:20 +0000)]
The osf1 utility has not been installed since rev 1.5 of
src/sys/modules/osf1/Makefile, so don't use it here!

23 years agoRetire the osf1(8) utility. The Makefile hasn't installed this critter
sheldonh [Thu, 30 Nov 2000 08:16:19 +0000 (08:16 +0000)]
Retire the osf1(8) utility.  The Makefile hasn't installed this critter
for a while.

Providing shell scripts that do nothing but load a similarly named
kernel loadable module is out of vogue.

23 years agoAdd commentary about shared SASL files versus sendmail compilation.
gshapiro [Thu, 30 Nov 2000 05:34:53 +0000 (05:34 +0000)]
Add commentary about shared SASL files versus sendmail compilation.

Submitted by: Scot W. Hetzel <hetzels@westbend.net>

23 years agoDon't use p->p_sigstk.ss_flags to keep state of whether the
marcel [Thu, 30 Nov 2000 05:23:49 +0000 (05:23 +0000)]
Don't use p->p_sigstk.ss_flags to keep state of whether the
process is on the alternate stack or not. For compatibility
with sigstack(2) state is being updated if such is needed.

We now determine whether the process is on the alternate
stack by looking at its stack pointer. This allows a process
to siglongjmp from a signal handler on the alternate stack
to the place of the sigsetjmp on the normal stack. When
maintaining state, this would have invalidated the state
information and causing a subsequent signal to be delivered
on the normal stack instead of the alternate stack.

PR: 22286

23 years agoIncrease NKPT from 17 to 30. This fixes the 4GB ram boot panic on both
peter [Thu, 30 Nov 2000 01:53:02 +0000 (01:53 +0000)]
Increase NKPT from 17 to 30.  This fixes the 4GB ram boot panic on both
-current and RELENG_4 with GENERIC.

NKPT is the number of initial bootstrap page table pages we create for
the kernel during startup. Once VM is up, we resize it as needed, but
with 4G ram, the size of the vm_page_t structures was pushing it over
the limit.  The fact that trimmed down kernels boot on 4G ram machines
suggests that we were pretty close to the edge.

The "30" is arbitary, but smaller than the 'nkpt' variable on all
machines that I checked.

23 years agoFix up priority propagation:
jhb [Thu, 30 Nov 2000 00:51:16 +0000 (00:51 +0000)]
Fix up priority propagation:
- Use a better test for determining when a process is running.
- Convert some checks to assertions.
- Remove unnecessary tests.
- Save the priority before acquiring a mutex rather than in msleep(9).

23 years agoMake use of the full screen width to display p_comm rather than assuming a
jhb [Wed, 29 Nov 2000 23:03:02 +0000 (23:03 +0000)]
Make use of the full screen width to display p_comm rather than assuming a
hardcoded screen width of 80 chars.

23 years agoLose an extra paragraph break.
dannyboy [Wed, 29 Nov 2000 21:44:46 +0000 (21:44 +0000)]
Lose an extra paragraph break.

PR: 23179
Submitted by: Jimmy Olgeni <olgeni@uli.it>

23 years agoIntroduce a 'mtxname' keyword that displays the current mutex that a
jhb [Wed, 29 Nov 2000 21:09:21 +0000 (21:09 +0000)]
Introduce a 'mtxname' keyword that displays the current mutex that a
process is blocked on or '-'.

23 years agoI didn't maintain the "chs" ordering. Rather the bug was in print_part().
obrien [Wed, 29 Nov 2000 20:22:47 +0000 (20:22 +0000)]
I didn't maintain the "chs" ordering.  Rather the bug was in print_part().

Requested by: bde.

23 years agoDisplay the name of the mutex we are blocked on in the state field. To
jhb [Wed, 29 Nov 2000 20:22:34 +0000 (20:22 +0000)]
Display the name of the mutex we are blocked on in the state field.  To
differentiate mutex names from wait channel names, prefix mutex names with
an asterisk.

Submitted by: Dan Nelson <dnelson@emsphone.com>

23 years agoSet p_mtxname when blocking on a mutex and clear it when waking up.
jhb [Wed, 29 Nov 2000 20:17:15 +0000 (20:17 +0000)]
Set p_mtxname when blocking on a mutex and clear it when waking up.

23 years agoSave a copy of p_mtxname in e_mtxname when creating an eproc.
jhb [Wed, 29 Nov 2000 20:14:50 +0000 (20:14 +0000)]
Save a copy of p_mtxname in e_mtxname when creating an eproc.

23 years ago- Add a p_mtxname field to proc which points to the description of the
jhb [Wed, 29 Nov 2000 20:10:54 +0000 (20:10 +0000)]
- Add a p_mtxname field to proc which points to the description of the
  mutex a process is blocked on or NULL.
- Add a corresponding field e_mtxname to eproc.
- Fix a spelling nit in a comment.

23 years agoThis fixes several problems with CIS as suggested by Justin Gibbs:
jon [Wed, 29 Nov 2000 19:38:25 +0000 (19:38 +0000)]
This fixes several problems with CIS as suggested by Justin Gibbs:
4) The cardbus CIS code treats the CIS_PTR as a mapping register if
   it is mentioned in the CIS.  I don't have a spec handy to understand
   why the CIS_PTR is mentioned in the CIS, but allocating a memory range
   for it is certainly bogus.  My patch ignores bar #6 to prevent the
   mapping.
   [The pccard spec says that BAR 0 and 7 (-1 and 6 in thic case since we
    did a minus one) is "reserved".  The off by 1 error has been fixed.
    also bar=5 is invalid for IO maps, so we check it.]

5) The CIS code allocated duplicate resources to those already found
   by cardbus_add_resources().  The fix is to pass in the bar computed
   from the CIS instead of the particular resource ID for that bar,
   so bus_generic_alloc_resource succeeds in finding the old resource.
   [fixed, also removed superfluous (and incorrect) writing back to the
    PCI config space.]

7) The CIS code seems to use the wrong bit to determine rather a particular
   register mapping is for I/O or memory space.  From looking at the
   two cards I have, it seems TPL_BAR_REG_AS should be 0x10 instead
   of 0x08.  Otherwise, all registers that should be I/O mapped gain
   a second mapping in memory space.
   [Oops, the spec does say 0x10..., fixed]

Submitted by: Justin Gibbs

23 years agoo Add a PATH statement to the beginning of make-localhost, making it
rwatson [Wed, 29 Nov 2000 19:19:06 +0000 (19:19 +0000)]
o Add a PATH statement to the beginning of make-localhost, making it
  work right when the administrator has modified their runtime environment
  in a manner not anticipated by our script.

Requested by: Tom Maher <tardis@ece.cmu.edu>

23 years agoUse an atomic operation with an appropriate memory barrier when releasing
jhb [Wed, 29 Nov 2000 18:41:19 +0000 (18:41 +0000)]
Use an atomic operation with an appropriate memory barrier when releasing
a contested sleep mutex in the case that at least two processes are blocked
on the contested mutex.

23 years agoThe sched_lock mutex goes after the sio mutex in the locking order since
jhb [Wed, 29 Nov 2000 18:38:14 +0000 (18:38 +0000)]
The sched_lock mutex goes after the sio mutex in the locking order since
a software interrupt can be scheduled in the sio interrupt handler while
the sio mutex is held.

23 years agoSave the line number and filename of the last mtx_enter operation for
jhb [Wed, 29 Nov 2000 18:37:01 +0000 (18:37 +0000)]
Save the line number and filename of the last mtx_enter operation for
spin locks.  We already do this for sleep locks.