]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoMake the name parameter const char *.
Warner Losh [Sat, 11 Aug 2001 05:16:00 +0000 (05:16 +0000)]
Make the name parameter const char *.

23 years agoSomewhere along the way, configSecurityModerate() lost it's "int"
Robert Watson [Sat, 11 Aug 2001 03:26:52 +0000 (03:26 +0000)]
Somewhere along the way, configSecurityModerate() lost it's "int"
return value.

Spotted by: gratuitous use of diff during MFC process

23 years agoMFCs noted: fsck_msdos(8), kldconfig(8), moused(8) -a.
Bruce A. Mah [Sat, 11 Aug 2001 03:12:38 +0000 (03:12 +0000)]
MFCs noted:  fsck_msdos(8), kldconfig(8), moused(8) -a.

23 years agoNew release note (only about six months late): nmdm(4).
Bruce A. Mah [Sat, 11 Aug 2001 02:49:57 +0000 (02:49 +0000)]
New release note (only about six months late):  nmdm(4).

23 years agoNote: mergemaster stopped being optional a long time ago. Caution against
Warner Losh [Sat, 11 Aug 2001 02:03:51 +0000 (02:03 +0000)]
Note: mergemaster stopped being optional a long time ago.  Caution against
doing it by hand (but provide pointers to those wanting to do it by hand).

23 years agoFor consistency with other man pages, fix first appearance.
Mark Peek [Sat, 11 Aug 2001 00:59:40 +0000 (00:59 +0000)]
For consistency with other man pages, fix first appearance.

23 years agoDon't call errx() with a variable format string
Kris Kennaway [Sat, 11 Aug 2001 00:49:11 +0000 (00:49 +0000)]
Don't call errx() with a variable format string

MFC after: 1 week

23 years agoRemove unused nmdmpoll function.
Mark Peek [Sat, 11 Aug 2001 00:18:15 +0000 (00:18 +0000)]
Remove unused nmdmpoll function.

Approved by: julian

23 years agoo Reduce the number of offered security profiles, as we now have a more
Robert Watson [Fri, 10 Aug 2001 23:57:43 +0000 (23:57 +0000)]
o Reduce the number of offered security profiles, as we now have a more
  conservative default, and actually prompt specifically for inetd rather
  than handling it as a side effect of the security profile.  Update the
  help file to reflect this change.
o Rename "Fascist" to "Extreme" in the source code, to match the names
  presented to the user.
o Remove portmap and inetd from profile management.  Portmap is now
  disabled by default, but automatically turned on if a feature requires
  it (such as NFS, etc).

This is an MFC candidate for 4.4-RELEASE.

Reviewed by: freebsd-arch@FreeBSD.org
Approved by: re@FreeBSD.org
MFC after: 2 days

23 years agoIf we've panic'd already, then just bail in lockmgr rather than blocking or
John Baldwin [Fri, 10 Aug 2001 23:29:15 +0000 (23:29 +0000)]
If we've panic'd already, then just bail in lockmgr rather than blocking or
possibly panic'ing again.

23 years agoMake the protoswitch definitiosn checkable in the same way that
Julian Elischer [Fri, 10 Aug 2001 23:17:22 +0000 (23:17 +0000)]
Make the protoswitch definitiosn checkable in the same way that
cdevsw entries have been for a long time.
Discover that we now have two version sof the same structure.
I will shoot one of them shortly when I figure out why someone thinks
they need it. (And I can prove they don't)
(netinet/ipprotosw.h should GO AWAY)

23 years agoFix some of the GDB linkage setup. The l_name member of the gdb linkage
Bill Paul [Fri, 10 Aug 2001 23:15:13 +0000 (23:15 +0000)]
Fix some of the GDB linkage setup. The l_name member of the gdb linkage
structure is always free()ed yet only sometimes malloc()ed. In particular,
it was simply set to point to l_filename from the a linker_file_t in
link_elf_link_preload_finish(). The l_filename had been malloc()ed inside
the kern_linker.c module and was being free()ed twice: once by
link_elf_unload_file() and again by linker_file_unload(), leading to
a panic.

How to duplicate the problem:

- Pre-load a kernel module from the loader, i.e. if_sis.ko
- Boot system
- Attempt to unload module with kldunload if_sis
- Bewm

The problem here is that the case where the module was loaded with kldload
after system boot would work correctly, so this bug went unnoticed until
I stubbed my toe on it just now. (Also, you can only trip this bug if
you compile a kernel with options DDB, but that's the default now.)

Fix: remember to malloc() a separate copy of the module name for the
l_name member of the gdb linkage structure in three places where the
linkage structure can be initialized.

23 years agoMark some functions as __printflike()
Kris Kennaway [Fri, 10 Aug 2001 23:12:10 +0000 (23:12 +0000)]
Mark some functions as __printflike()

MFC After: 1 week

23 years agoAdd an optimization where we check hte PS_ASTPENDING and PS_NEEDRESCHED
John Baldwin [Fri, 10 Aug 2001 23:00:07 +0000 (23:00 +0000)]
Add an optimization where we check hte PS_ASTPENDING and PS_NEEDRESCHED
flags with interrupts disabled to see if we should call ast() during
doreti.  This was mostly submitted by Bruce, but his original patch did
the looping in ast() in assembly rather than in the ast() function itself.
Once we've actually called into the ast() function, it's cheaper to just
loop inside the function rather than returning from the function,
performing the check, and then calling the function again.  However, we
can optimize the first check to avoid calling the function at all.
Other architectures may choose to implement this optimization if they
wish but it is not required for correct operation.

Submitted by: bde

23 years ago- Close races with signals and other AST's being triggered while we are in
John Baldwin [Fri, 10 Aug 2001 22:53:32 +0000 (22:53 +0000)]
- Close races with signals and other AST's being triggered while we are in
  the process of exiting the kernel.  The ast() function now loops as long
  as the PS_ASTPENDING or PS_NEEDRESCHED flags are set.  It returns with
  preemption disabled so that any further AST's that arrive via an
  interrupt will be delayed until the low-level MD code returns to user
  mode.
- Use u_int's to store the tick counts for profiling purposes so that we
  do not need sched_lock just to read p_sticks.  This also closes a
  problem where the call to addupc_task() could screw up the arithmetic
  due to non-atomic reads of p_sticks.
- Axe need_proftick(), aston(), astoff(), astpending(), need_resched(),
  clear_resched(), and resched_wanted() in favor of direct bit operations
  on p_sflag.
- Fix up locking with sched_lock some.  In addupc_intr(), use sched_lock
  to ensure pr_addr and pr_ticks are updated atomically with setting
  PS_OWEUPC.  In ast() we clear pr_ticks atomically with clearing
  PS_OWEUPC.  We also do not grab the lock just to test a flag.
- Simplify the handling of Giant in ast() slightly.

Reviewed by: bde (mostly)

23 years agoMake witness compile w/o DDB.
John Baldwin [Fri, 10 Aug 2001 22:33:59 +0000 (22:33 +0000)]
Make witness compile w/o DDB.

Reported by: wpaul

23 years agoStyle cleanup.
David E. O'Brien [Fri, 10 Aug 2001 22:31:05 +0000 (22:31 +0000)]
Style cleanup.

23 years agoArbitrarily limit to 64k the number of bytes that can be read at
Ian Dowse [Fri, 10 Aug 2001 22:14:18 +0000 (22:14 +0000)]
Arbitrarily limit to 64k the number of bytes that can be read at
a time using the ogetdirentries() compatibility syscall. This is a
hack to ensure that rediculous values don't get passed to MALLOC().

Reviewed by: kris

23 years agoTypo
Warner Losh [Fri, 10 Aug 2001 21:32:53 +0000 (21:32 +0000)]
Typo

23 years agoClose a </sect1> and add a line of whitespace, discovered during
Bruce A. Mah [Fri, 10 Aug 2001 21:18:35 +0000 (21:18 +0000)]
Close a </sect1> and add a line of whitespace, discovered during
MFC of 1.3->1.4 delta.

23 years agoWork around a race between msleep() and endtsleep() where it was possible
John Baldwin [Fri, 10 Aug 2001 21:08:56 +0000 (21:08 +0000)]
Work around a race between msleep() and endtsleep() where it was possible
for endtsleep() to be executing when msleep() resumed, for endtsleep()
to spin on sched_lock long enough for the other process to loop on
msleep() and sleep again resulting in endtsleep() waking up the "wrong"
msleep.

Obtained from: BSD/OS

23 years agoChange callout_stop() to return an integer. If callout_stop() succeeds in
John Baldwin [Fri, 10 Aug 2001 21:06:59 +0000 (21:06 +0000)]
Change callout_stop() to return an integer.  If callout_stop() succeeds in
removing the callout entry, return 1.  If callout_stop() fails to remove
the callout entry because it is currently executing or has already been
executed, then the function returns 0.  The idea was obtained from BSD/OS,
however, BSD/OS changed untimeout(), and I've just changed callout_stop()
to be more conservative.

Obtained from: BSD/OS

23 years agoStyle nit: covert a couple of if (p_wchan) tests to if (p_wchan != NULL).
John Baldwin [Fri, 10 Aug 2001 20:56:25 +0000 (20:56 +0000)]
Style nit: covert a couple of if (p_wchan) tests to if (p_wchan != NULL).

23 years agoUse .Fn, .Fa, and .Dv where appropriate.
Dima Dorfman [Fri, 10 Aug 2001 20:49:38 +0000 (20:49 +0000)]
Use .Fn, .Fa, and .Dv where appropriate.

23 years agoClean up this module very extensively. Fix the logging, the coding
Mark Murray [Fri, 10 Aug 2001 19:24:34 +0000 (19:24 +0000)]
Clean up this module very extensively. Fix the logging, the coding
standards and the option handling. This module is now much more easy
to maintain as a part of the FreeBSD tree.

23 years agoCode clean up; make logging same as other modules and fix warnings.
Mark Murray [Fri, 10 Aug 2001 19:21:45 +0000 (19:21 +0000)]
Code clean up; make logging same as other modules and fix warnings.

23 years agoGeneral code clean-up. Sort out warnings, and make the warning and
Mark Murray [Fri, 10 Aug 2001 19:18:52 +0000 (19:18 +0000)]
General code clean-up. Sort out warnings, and make the warning and
logging work the same as other modules.

23 years agoSimplify code. Also verbose logging, verbose overridable error reporting.
Mark Murray [Fri, 10 Aug 2001 19:15:48 +0000 (19:15 +0000)]
Simplify code. Also verbose logging, verbose overridable error reporting.

23 years agoVerbose logging, overridable verbose error reporting.
Mark Murray [Fri, 10 Aug 2001 19:12:59 +0000 (19:12 +0000)]
Verbose logging, overridable verbose error reporting.

23 years agoModule clean-up. Verbose logging, Overridable verbose error reporting,
Mark Murray [Fri, 10 Aug 2001 19:10:43 +0000 (19:10 +0000)]
Module clean-up. Verbose logging, Overridable verbose error reporting,
FreeBSD pam_prompt() usage to simplify conversation function usage.

23 years agoVerbosely (overridable) report failure to the user.
Mark Murray [Fri, 10 Aug 2001 19:07:45 +0000 (19:07 +0000)]
Verbosely (overridable) report failure to the user.

23 years agoUse the FreeBSD pam_prompt() interface to the conversation function
Mark Murray [Fri, 10 Aug 2001 19:05:57 +0000 (19:05 +0000)]
Use the FreeBSD pam_prompt() interface to the conversation function
instead of home-rolling it. Clean up debugging code and tidy the
module.

23 years agoVerbosely report errors to the user (overridable), and make sure
Mark Murray [Fri, 10 Aug 2001 19:02:21 +0000 (19:02 +0000)]
Verbosely report errors to the user (overridable), and make sure
that the correct failure mode is reported.

23 years agolibgmp has been superseded by libmp.
Dima Dorfman [Fri, 10 Aug 2001 18:41:56 +0000 (18:41 +0000)]
libgmp has been superseded by libmp.

23 years agoRemove libgmp build infrastructure.
Dima Dorfman [Fri, 10 Aug 2001 18:35:08 +0000 (18:35 +0000)]
Remove libgmp build infrastructure.

23 years agoChange copyright to BSD-style copyright.
Brian Somers [Fri, 10 Aug 2001 17:42:58 +0000 (17:42 +0000)]
Change copyright to BSD-style copyright.

Ok'd by: Gabor Kincses <gabor@acm.org>

23 years agocvs updated to 1.11.1p1
Dima Dorfman [Fri, 10 Aug 2001 17:41:36 +0000 (17:41 +0000)]
cvs updated to 1.11.1p1

23 years agomdoc(7) police: join split punctuation to macro calls.
Ruslan Ermilov [Fri, 10 Aug 2001 17:35:21 +0000 (17:35 +0000)]
mdoc(7) police: join split punctuation to macro calls.

23 years agoDo NOT allocate a 1K buffer on the kernel stack.
Julian Elischer [Fri, 10 Aug 2001 17:32:11 +0000 (17:32 +0000)]
Do NOT allocate a 1K buffer on the kernel stack.

Found by: Smashed u-area and hardware watch points.
MFC after: 1 week

23 years agoInclude string.h for the strlen() prototype to quiet a warning.
John Baldwin [Fri, 10 Aug 2001 16:55:09 +0000 (16:55 +0000)]
Include string.h for the strlen() prototype to quiet a warning.

23 years agomdoc(7) police: fixed the "new sentence" bogons.
Ruslan Ermilov [Fri, 10 Aug 2001 15:03:10 +0000 (15:03 +0000)]
mdoc(7) police: fixed the "new sentence" bogons.

23 years agoFix missing splx().
Kazutaka YOKOTA [Fri, 10 Aug 2001 14:26:48 +0000 (14:26 +0000)]
Fix missing splx().

23 years agoFix broken logic so that this actually works for the superuser.
Mark Murray [Fri, 10 Aug 2001 14:21:58 +0000 (14:21 +0000)]
Fix broken logic so that this actually works for the superuser.
Verbosely log (properly).
Verbosely report errors to the user.

23 years agoRework this to prevent a nasty problem involving different modules'
Mark Murray [Fri, 10 Aug 2001 14:16:47 +0000 (14:16 +0000)]
Rework this to prevent a nasty problem involving different modules'
option interacting with each other.

23 years agoDeclare the new user-error reporting macro.
Mark Murray [Fri, 10 Aug 2001 14:15:00 +0000 (14:15 +0000)]
Declare the new user-error reporting macro.

This is a macro to allow use of the __FILE__ and __FUNCTION__
macros.

23 years agoAdd a routine for providing feedback via the conversation mechanism
Mark Murray [Fri, 10 Aug 2001 14:13:16 +0000 (14:13 +0000)]
Add a routine for providing feedback via the conversation mechanism
(usually to stderr) for user-reportable errors.

23 years agomdoc(7) police: protect trailing full stops of abbreviations
Ruslan Ermilov [Fri, 10 Aug 2001 13:45:36 +0000 (13:45 +0000)]
mdoc(7) police: protect trailing full stops of abbreviations
with a trailing zero-width space: `e.g.\&'.

23 years agoEliminate the hot-spare 'r' in Arrray.
Poul-Henning Kamp [Fri, 10 Aug 2001 11:47:08 +0000 (11:47 +0000)]
Eliminate the hot-spare 'r' in Arrray.

Submitted by: Søren Schrøder <sch@chaos.dk>

23 years agoFixed style bugs (dot `.' at the end of error and warning messages).
Ruslan Ermilov [Fri, 10 Aug 2001 11:46:37 +0000 (11:46 +0000)]
Fixed style bugs (dot `.' at the end of error and warning messages).

Noticed by: bde

23 years agoMarkup nits: use diagnostic type lists for error and warning messages.
Ruslan Ermilov [Fri, 10 Aug 2001 11:41:55 +0000 (11:41 +0000)]
Markup nits: use diagnostic type lists for error and warning messages.

Backout previous revision.  We should not expand plain text xrefs if
they appear in the literal text, e.g. in the error or warning message
of the library function.  (Submitted by: bde)

Moved "out of memory" from warning to errors section.

23 years agoUse some slightly less horrible make magic for doing the version stuff.
Peter Wemm [Fri, 10 Aug 2001 11:24:23 +0000 (11:24 +0000)]
Use some slightly less horrible make magic for doing the version stuff.

23 years agoOne more version related tweak after cvs-1.11 -> 1.11.1p1
Peter Wemm [Fri, 10 Aug 2001 11:18:35 +0000 (11:18 +0000)]
One more version related tweak after cvs-1.11 -> 1.11.1p1

23 years agoArgh, this could not possibly have worked as it existed in the tree.
Peter Wemm [Fri, 10 Aug 2001 11:14:53 +0000 (11:14 +0000)]
Argh, this could not possibly have worked as it existed in the tree.

23 years agoCopy our changes from sccs2rcs.csh to sccs2rcs.in
Peter Wemm [Fri, 10 Aug 2001 11:14:18 +0000 (11:14 +0000)]
Copy our changes from sccs2rcs.csh to sccs2rcs.in

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81440,
Peter Wemm [Fri, 10 Aug 2001 11:04:36 +0000 (11:04 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81440,
which included commits to RCS files with non-trunk default branches.

23 years agoUpdate build rules for .in and tidy up.
Peter Wemm [Fri, 10 Aug 2001 11:03:29 +0000 (11:03 +0000)]
Update build rules for .in and tidy up.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81437,
Peter Wemm [Fri, 10 Aug 2001 11:02:03 +0000 (11:02 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81437,
which included commits to RCS files with non-trunk default branches.

23 years agoThis file is gone in cvs-1.11.1p1 (replaced by log.in)
Peter Wemm [Fri, 10 Aug 2001 11:00:14 +0000 (11:00 +0000)]
This file is gone in cvs-1.11.1p1 (replaced by log.in)

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81434,
Peter Wemm [Fri, 10 Aug 2001 10:55:45 +0000 (10:55 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81434,
which included commits to RCS files with non-trunk default branches.

23 years agomdoc(7) police: add zdestroy() to NAME.
Ruslan Ermilov [Fri, 10 Aug 2001 10:53:10 +0000 (10:53 +0000)]
mdoc(7) police: add zdestroy() to NAME.

23 years agoTools dir is now empty and unused
Peter Wemm [Fri, 10 Aug 2001 10:47:28 +0000 (10:47 +0000)]
Tools dir is now empty and unused

23 years agomdoc(7) police: fix markup and a SYNOPSIS bug from revision 1.24.
Ruslan Ermilov [Fri, 10 Aug 2001 10:45:59 +0000 (10:45 +0000)]
mdoc(7) police: fix markup and a SYNOPSIS bug from revision 1.24.

23 years agomdoc(7) police: fix markup.
Ruslan Ermilov [Fri, 10 Aug 2001 10:43:45 +0000 (10:43 +0000)]
mdoc(7) police: fix markup.

23 years agomdoc(7) police: removed (now gratuitous) .Xo/.Xc brackets.
Ruslan Ermilov [Fri, 10 Aug 2001 10:41:30 +0000 (10:41 +0000)]
mdoc(7) police: removed (now gratuitous) .Xo/.Xc brackets.

23 years agoRe-kill this file. It got whiplash being killed and restored on the
Peter Wemm [Fri, 10 Aug 2001 10:39:29 +0000 (10:39 +0000)]
Re-kill this file.  It got whiplash being killed and restored on the
mainline, it is truely dead now.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81426,
Peter Wemm [Fri, 10 Aug 2001 10:38:11 +0000 (10:38 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81426,
which included commits to RCS files with non-trunk default branches.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81424,
Peter Wemm [Fri, 10 Aug 2001 10:33:41 +0000 (10:33 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81424,
which included commits to RCS files with non-trunk default branches.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81422,
Peter Wemm [Fri, 10 Aug 2001 10:29:04 +0000 (10:29 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81422,
which included commits to RCS files with non-trunk default branches.

23 years agoGenerate version.c on the fly rather than using a generated file.
Peter Wemm [Fri, 10 Aug 2001 10:29:04 +0000 (10:29 +0000)]
Generate version.c on the fly rather than using a generated file.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81420,
Peter Wemm [Fri, 10 Aug 2001 10:22:43 +0000 (10:22 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81420,
which included commits to RCS files with non-trunk default branches.

23 years agoGet the version source from the (more) correct place.
Peter Wemm [Fri, 10 Aug 2001 10:22:43 +0000 (10:22 +0000)]
Get the version source from the (more) correct place.
Remove the obsolete cvsbug.sh script

23 years agomdoc(7) police: spelling, punctuation and markup nits.
Ruslan Ermilov [Fri, 10 Aug 2001 10:15:41 +0000 (10:15 +0000)]
mdoc(7) police: spelling, punctuation and markup nits.

23 years agomdoc(7) police: add xref to intro(2).
Ruslan Ermilov [Fri, 10 Aug 2001 10:11:55 +0000 (10:11 +0000)]
mdoc(7) police: add xref to intro(2).

23 years agomdoc(7) police: compute the exact tag width.
Ruslan Ermilov [Fri, 10 Aug 2001 10:09:00 +0000 (10:09 +0000)]
mdoc(7) police: compute the exact tag width.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81414,
Peter Wemm [Fri, 10 Aug 2001 10:05:43 +0000 (10:05 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81414,
which included commits to RCS files with non-trunk default branches.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81412,
Peter Wemm [Fri, 10 Aug 2001 10:02:36 +0000 (10:02 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81412,
which included commits to RCS files with non-trunk default branches.

23 years agoBah. Previous commit was: update for cvs-1.11 -> 1.11.1p1
Peter Wemm [Fri, 10 Aug 2001 09:57:23 +0000 (09:57 +0000)]
Bah. Previous commit was:  update for cvs-1.11 -> 1.11.1p1
(This was my fat-fingers, not a cvs glitch)

23 years ago*** empty log message ***
Peter Wemm [Fri, 10 Aug 2001 09:55:48 +0000 (09:55 +0000)]
*** empty log message ***

23 years agoAdd annotate.c, remove rtag.c for 1.11.1p1
Peter Wemm [Fri, 10 Aug 2001 09:54:52 +0000 (09:54 +0000)]
Add annotate.c, remove rtag.c for 1.11.1p1

23 years agoTweak for 1.11.1p1
Peter Wemm [Fri, 10 Aug 2001 09:53:53 +0000 (09:53 +0000)]
Tweak for 1.11.1p1

23 years agoInitial merge of cvs-1.11 -> 1.11.1p1 changes onto mainline. There is
Peter Wemm [Fri, 10 Aug 2001 09:53:06 +0000 (09:53 +0000)]
Initial merge of cvs-1.11 -> 1.11.1p1 changes onto mainline.  There is
still some suspicious and/or broken stuff to fix yet.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r81404,
Peter Wemm [Fri, 10 Aug 2001 09:43:22 +0000 (09:43 +0000)]
This commit was generated by cvs2svn to compensate for changes in r81404,
which included commits to RCS files with non-trunk default branches.

23 years agoImport cvs-1.11.1p1 onto vendor branch
Peter Wemm [Fri, 10 Aug 2001 09:43:22 +0000 (09:43 +0000)]
Import cvs-1.11.1p1 onto vendor branch

23 years agoTemporarily turn off cvs builds for the (short) duration of
Peter Wemm [Fri, 10 Aug 2001 09:35:58 +0000 (09:35 +0000)]
Temporarily turn off cvs builds for the (short) duration of
import/verification.

23 years agoDO NOT ALLOCATE 2+K OBJECTS ON THE KERNEL STACK!!!!
Julian Elischer [Fri, 10 Aug 2001 07:50:14 +0000 (07:50 +0000)]
DO NOT ALLOCATE 2+K OBJECTS ON THE KERNEL STACK!!!!
found by: Getting my u-area overwritten

23 years agoAdd manpage for pgfind(9).
Ruslan Ermilov [Fri, 10 Aug 2001 07:19:55 +0000 (07:19 +0000)]
Add manpage for pgfind(9).

PR: docs/29552
Submitted by: Evan Sarmiento <kaworu@sektor7.ath.cx>

23 years ago- Remove asleep(), await(), and M_ASLEEP.
John Baldwin [Fri, 10 Aug 2001 06:56:12 +0000 (06:56 +0000)]
- Remove asleep(), await(), and M_ASLEEP.
- Callers of asleep() and await() have been converted to calling tsleep().
  The only caller outside of M_ASLEEP was the ata driver, which called both
  asleep() and await() with spl-raised, so there was no need for the
  asleep() and await() pair.  M_ASLEEP was unused.

Reviewed by: jasone, peter

23 years ago- Remove asleep(), await(), and M_ASLEEP.
John Baldwin [Fri, 10 Aug 2001 06:45:43 +0000 (06:45 +0000)]
- Remove asleep(), await(), and M_ASLEEP.
- Callers of asleep() and await() have been converted to calling tsleep().
  The only caller outside of M_ASLEEP was the ata driver, which called both
  asleep() and await() with spl-raised, so there was no need for the
  asleep() and await() pair.  M_ASLEEP was unused.

Reviewed by: jasone, peter

23 years ago- Remove asleep(), await(), and M_ASLEEP.
John Baldwin [Fri, 10 Aug 2001 06:37:05 +0000 (06:37 +0000)]
- Remove asleep(), await(), and M_ASLEEP.
- Callers of asleep() and await() have been converted to calling tsleep().
  The only caller outside of M_ASLEEP was the ata driver, which called both
  asleep() and await() with spl-raised, so there was no need for the
  asleep() and await() pair.  M_ASLEEP was unused.

Reviewed by: jasone, peter

23 years agoMove ISA interrupt ISR and timeout routines to pcic from pcic_isa so
Warner Losh [Fri, 10 Aug 2001 06:07:20 +0000 (06:07 +0000)]
Move ISA interrupt ISR and timeout routines to pcic from pcic_isa so
that we can use them in the pci code when we have to fall back to ISA
interrupt routing.

23 years agoType sanity: use uintptr_t * for read_ivar and u_int8_t instead of u_char
Warner Losh [Fri, 10 Aug 2001 06:00:44 +0000 (06:00 +0000)]
Type sanity: use uintptr_t * for read_ivar and u_int8_t instead of u_char

23 years agoRearrange the pcic_irq_type enum (and specifically tag the first one
Warner Losh [Fri, 10 Aug 2001 05:42:08 +0000 (05:42 +0000)]
Rearrange the pcic_irq_type enum (and specifically tag the first one
as being 1) in anticipation of documentation.

23 years agoBump MAXCOMLEN from 16 to 19 to take advantage of 32-bit alignment.
John Baldwin [Fri, 10 Aug 2001 05:34:48 +0000 (05:34 +0000)]
Bump MAXCOMLEN from 16 to 19 to take advantage of 32-bit alignment.

Approved by: peter, jasone

23 years agoCorrect copyright language.
Jake Burkholder [Fri, 10 Aug 2001 04:51:19 +0000 (04:51 +0000)]
Correct copyright language.

23 years agoAdd code to program the tick register and to setup its interrupt handler.
Jake Burkholder [Fri, 10 Aug 2001 04:49:33 +0000 (04:49 +0000)]
Add code to program the tick register and to setup its interrupt handler.

23 years agoAdd early code to support interrupts.
Jake Burkholder [Fri, 10 Aug 2001 04:48:48 +0000 (04:48 +0000)]
Add early code to support interrupts.

23 years agoFake up the frame pointers on a process's initial stack so they can be
Jake Burkholder [Fri, 10 Aug 2001 04:47:14 +0000 (04:47 +0000)]
Fake up the frame pointers on a process's initial stack so they can be
restored correctly from the trapframe.

Submitted by: tmm

23 years agoHandle all types of mmu misses from user mode.
Jake Burkholder [Fri, 10 Aug 2001 04:44:33 +0000 (04:44 +0000)]
Handle all types of mmu misses from user mode.
Pass a context argument to tlb functions.

23 years agoUse the macro for getting the trap type from the trapframe.
Jake Burkholder [Fri, 10 Aug 2001 04:43:31 +0000 (04:43 +0000)]
Use the macro for getting the trap type from the trapframe.
Only set sticks (and acquire sched_lock) on entry from user mode.
Add handlers for all kinds of mmu misses, and for interrupts from
user mode.
Acquire Giant before calling into the vm system so this runs with
invariants.
Try to get the restrictions for page faults on user memory from
kernel mode right.
Only set pcb_onfault and return to the alternate return code if
this is actually a fault on user memory from kernel mode.

23 years agoStore 8 bytes instead of 4 in suword. Use a temporary stack that's known
Jake Burkholder [Fri, 10 Aug 2001 04:39:23 +0000 (04:39 +0000)]
Store 8 bytes instead of 4 in suword.  Use a temporary stack that's known
to be locked in the tlb for calling openfirmware.

Submitted by: tmm