]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years ago o Introduce vm_page_sleep_if_busy() as an eventual replacement for
Alan Cox [Mon, 29 Jul 2002 19:41:22 +0000 (19:41 +0000)]
 o Introduce vm_page_sleep_if_busy() as an eventual replacement for
   vm_page_sleep_busy().  vm_page_sleep_if_busy() uses the page
   queues lock.

22 years agoRemove a XXXKSE comment. the code is no longer a problem..
Julian Elischer [Mon, 29 Jul 2002 18:47:19 +0000 (18:47 +0000)]
Remove a XXXKSE comment. the code is no longer a problem..

22 years agoCreate a new thread state to describe threads that would be ready to run
Julian Elischer [Mon, 29 Jul 2002 18:33:32 +0000 (18:33 +0000)]
Create a new thread state to describe threads that would be ready to run
except for the fact tha they are presently swapped out. Also add a process
flag to indicate that the process has started the struggle to swap
back in. This will be  needed for the case where multiple threads
start the swapin action top a collision. Also add code to stop
a process fropm being swapped out if one of the threads in this
process is actually off running on another CPU.. that might hurt...

Submitted by: Seigo Tanimura <tanimura@r.dl.itc.u-tokyo.ac.jp>

22 years agoCreate a new header <machine/_stdint.h> for storing MD parts of
Mike Barcroft [Mon, 29 Jul 2002 17:41:23 +0000 (17:41 +0000)]
Create a new header <machine/_stdint.h> for storing MD parts of
<stdint.h>.  Previously, parts were defined in <machine/ansi.h> and
<machine/limits.h>.  This resulted in two problems:
  (1) Defining macros in <machine/ansi.h> gets in the way of that
      header only defining types.
  (2) Defining C99 limits in <machine/limits.h> adds pollution to
      <limits.h>.

22 years agoDon't depend on pollution in <machine/limits.h> (by way of
Mike Barcroft [Mon, 29 Jul 2002 17:09:21 +0000 (17:09 +0000)]
Don't depend on pollution in <machine/limits.h> (by way of
<sys/param.h>) for definition of <stdint.h> macros.

22 years agoNew release notes: tr(1) equivalence classes (+MFC), tr(1) -C.
Bruce A. Mah [Mon, 29 Jul 2002 16:01:09 +0000 (16:01 +0000)]
New release notes:  tr(1) equivalence classes (+MFC), tr(1) -C.

MFCs noted:  Various new flags for who(1).

22 years agoConform to RFC 959, Appendix II, when replying
Yaroslav Tykhiy [Mon, 29 Jul 2002 15:54:27 +0000 (15:54 +0000)]
Conform to RFC 959, Appendix II, when replying
to a successful MKD command.

MFC after: 1 week

22 years agoIf we are in hostap mode, do not go into promisc mode. This causes
Warner Losh [Mon, 29 Jul 2002 15:36:22 +0000 (15:36 +0000)]
If we are in hostap mode, do not go into promisc mode.  This causes
problems with the firmware and will result in a) poor performance and
b) the inability to associate certain types of cards (most notibly
cisco).

Idea obtained from OpenBSD, but I implemented it by clearing the
IFF_PROMISC flag rather than the refusing to honor it downstream.

22 years agoWhen translating and the -c option is specified, handle the case where the
Tim J. Robbins [Mon, 29 Jul 2002 14:50:54 +0000 (14:50 +0000)]
When translating and the -c option is specified, handle the case where the
second string argument is more than one character in length in the way
required by SUSv3 (and the way GNU textutils and SVR4 do it).

22 years agoDrop support for COPY, -c has been the default mode of install(1)
Ruslan Ermilov [Mon, 29 Jul 2002 09:40:17 +0000 (09:40 +0000)]
Drop support for COPY, -c has been the default mode of install(1)
for a long time now.

Approved by: bde

22 years agoUse a common way to release locks before exit.
Maxim Konovalov [Mon, 29 Jul 2002 09:01:39 +0000 (09:01 +0000)]
Use a common way to release locks before exit.

Reviewed by: hsu

22 years agoIgnore -C, -p, and -S options of install(1) when used with the -d
Ruslan Ermilov [Mon, 29 Jul 2002 08:51:04 +0000 (08:51 +0000)]
Ignore -C, -p, and -S options of install(1) when used with the -d
option.  Warn about COPY being phased out.  Restore the old method
of always comparing before installing: INSTALL="install -C".

Requested by: bde

22 years ago - Add VBAD to the list of vnodes that are ignored on locking operations.
Jeff Roberson [Mon, 29 Jul 2002 07:45:57 +0000 (07:45 +0000)]
 - Add VBAD to the list of vnodes that are ignored on locking operations.

22 years agoFix typo.
Murray Stokely [Mon, 29 Jul 2002 07:22:26 +0000 (07:22 +0000)]
Fix typo.

PR: docs/41091
MFC after: 1 day

22 years ago - Backout the patch made in revision 1.75 of vfs_mount.c. The vputs here
Jeff Roberson [Mon, 29 Jul 2002 06:26:55 +0000 (06:26 +0000)]
 - Backout the patch made in revision 1.75 of vfs_mount.c.  The vputs here
   were hiding the real problem of the missing unlock in sync_inactive.
 - Add the missing unlock in sync_inactive.

Submitted by: iedowse

22 years ago o Pass VM_ALLOC_WIRED to vm_page_grab() rather than calling vm_page_wire()
Alan Cox [Mon, 29 Jul 2002 05:42:44 +0000 (05:42 +0000)]
 o Pass VM_ALLOC_WIRED to vm_page_grab() rather than calling vm_page_wire()
   in pmap_new_thread(), pmap_pinit(), and vm_proc_new().
 o Lock page queue accesses by vm_page_free() in pmap_object_init_pt().

22 years agoNew release notes: wicontrol(8) -l/-L, traceroute 1.4a12.
Bruce A. Mah [Mon, 29 Jul 2002 05:06:56 +0000 (05:06 +0000)]
New release notes:  wicontrol(8) -l/-L, traceroute 1.4a12.

Updated release note:  tcsh-6.12.

22 years agoFix Makefiles to actually work.
Matt Jacob [Mon, 29 Jul 2002 04:43:10 +0000 (04:43 +0000)]
Fix Makefiles to actually work.

22 years agoForced commit.
Mike Silbersack [Mon, 29 Jul 2002 04:41:30 +0000 (04:41 +0000)]
Forced commit.

Rev 1.134 actually reverts rev 1.132, NOT rev 1.131.

22 years agoRevert rev 1.131.
Mike Silbersack [Mon, 29 Jul 2002 04:32:35 +0000 (04:32 +0000)]
Revert rev 1.131.

1.131 is slightly broken, and I would commit the fix to that here, but it
has been reported that any deviation from the original code is causing
problems with some 82557 chips, causing them to lock hard.

Until those issues have been figured out, going back to the original
code is the best plan.

Frustrated: Silby

22 years agowhitespace commit: fix indentation in fxp_intr_body.
Luigi Rizzo [Mon, 29 Jul 2002 02:48:09 +0000 (02:48 +0000)]
whitespace commit: fix indentation in fxp_intr_body.
This is also done in preparation of a subsequent fix
for the handling of RNR conditions in polling mode.

22 years agoAdd routines needed for high resolution profiling.
Jake Burkholder [Mon, 29 Jul 2002 00:45:13 +0000 (00:45 +0000)]
Add routines needed for high resolution profiling.

22 years agoAdd a symbol for btext.
Jake Burkholder [Mon, 29 Jul 2002 00:42:00 +0000 (00:42 +0000)]
Add a symbol for btext.

22 years agoRemove a stale comment.
Jake Burkholder [Mon, 29 Jul 2002 00:40:48 +0000 (00:40 +0000)]
Remove a stale comment.

22 years agoUse _ALIGN_TEXT. Implement __cyg_profile_func_enter and
Jake Burkholder [Mon, 29 Jul 2002 00:39:46 +0000 (00:39 +0000)]
Use _ALIGN_TEXT.  Implement __cyg_profile_func_enter and
__cyg_profile_func_exit for GUPROF.

22 years agoAdd _ALIGN_DATA and _ALIGN_TEXT macros.
Jake Burkholder [Mon, 29 Jul 2002 00:38:07 +0000 (00:38 +0000)]
Add _ALIGN_DATA and _ALIGN_TEXT macros.

22 years agoRemove some stuff that snuck in last commit.
Jake Burkholder [Mon, 29 Jul 2002 00:37:05 +0000 (00:37 +0000)]
Remove some stuff that snuck in last commit.

22 years agoUse login_getpwclass() instead of login_getclass() so that the root
Tony Finch [Mon, 29 Jul 2002 00:36:24 +0000 (00:36 +0000)]
Use login_getpwclass() instead of login_getclass() so that the root
vs. default login class distinction is made correctly.

PR: 37416
Approved by: des
MFC after: 4 days

22 years agoOK, dont rely on the upper layers handling iosize_max correctly,
Søren Schmidt [Sun, 28 Jul 2002 23:59:00 +0000 (23:59 +0000)]
OK, dont rely on the upper layers handling iosize_max correctly,
instead rely on ATAPI devices ability to do the work instead.

MFC material.

22 years ago o Modify vm_page_grab() to accept VM_ALLOC_WIRED.
Alan Cox [Sun, 28 Jul 2002 23:46:19 +0000 (23:46 +0000)]
 o Modify vm_page_grab() to accept VM_ALLOC_WIRED.

22 years agofetch(1) is WARNS-5 clean (tested on i386 and Alpha)
Dag-Erling Smørgrav [Sun, 28 Jul 2002 21:09:57 +0000 (21:09 +0000)]
fetch(1) is WARNS-5 clean (tested on i386 and Alpha)

22 years agoUse strtol(3) instead of a home-grown version, and fix a "possibly
Dag-Erling Smørgrav [Sun, 28 Jul 2002 21:09:25 +0000 (21:09 +0000)]
Use strtol(3) instead of a home-grown version, and fix a "possibly
unitialized variable" warning.

22 years agoMake a temporary copy of the output data in the generic sysctl handlers
Don Lewis [Sun, 28 Jul 2002 21:06:14 +0000 (21:06 +0000)]
Make a temporary copy of the output data in the generic sysctl handlers
so that the data is less likely to be inconsistent if SYSCTL_OUT() blocks.
If the data is large, wire the output buffer instead.

This is somewhat less than optimal, since the handler could skip the copy
if it knew that the data was static.

If the data is dynamic, we are still not guaranteed to get a consistent
copy since another processor could change the data while the copy is in
progress because the data is not locked.  This problem could be solved if
the generic handlers had the ability to grab the proper lock before the
copy and release it afterwards.

This may duplicate work done in other sysctl handlers in the kernel which
also copy the data, possibly while a lock is held, before calling they call
a generic handler to output the data.  These handlers should probably call
SYSCTL_OUT() directly.

22 years ago o Lock page queue accesses by vm_page_free().
Alan Cox [Sun, 28 Jul 2002 20:13:48 +0000 (20:13 +0000)]
 o Lock page queue accesses by vm_page_free().
 o Apply some style fixes.

22 years agoWire the sysctl output buffer before grabbing any locks to prevent
Don Lewis [Sun, 28 Jul 2002 19:59:31 +0000 (19:59 +0000)]
Wire the sysctl output buffer before grabbing any locks to prevent
SYSCTL_OUT() from blocking while locks are held.  This should
only be done when it would be inconvenient to make a temporary copy of
the data and defer calling SYSCTL_OUT() until after the locks are
released.

22 years agoFix a bug introduced in previous commit. Due to the interaction of the
Jake Burkholder [Sun, 28 Jul 2002 19:15:34 +0000 (19:15 +0000)]
Fix a bug introduced in previous commit.  Due to the interaction of the
direct physical mappings with virtual page colour, we need to flush the
data cache when a page changes colour.  I missed one case which broke
pipes.

22 years ago o Lock page queue accesses by vm_page_free().
Alan Cox [Sun, 28 Jul 2002 19:01:38 +0000 (19:01 +0000)]
 o Lock page queue accesses by vm_page_free().

22 years agoRecognize frames for lcall_syscall and int0x80_syscall.
Peter Wemm [Sun, 28 Jul 2002 18:46:49 +0000 (18:46 +0000)]
Recognize frames for lcall_syscall and int0x80_syscall.

Reminded by: bde

22 years agoUpdate, correct and bring in sync with reality all the i4b kernel
Hellmuth Michaelis [Sun, 28 Jul 2002 18:21:48 +0000 (18:21 +0000)]
Update, correct and bring in sync with reality all the i4b kernel
part manual pages - long overdue ...

22 years agoAdd enough consts to keep everyone happy.
David Malone [Sun, 28 Jul 2002 16:23:28 +0000 (16:23 +0000)]
Add enough consts to keep everyone happy.

22 years agoDon't reuse a const char * when we really want a char *.
David Malone [Sun, 28 Jul 2002 16:17:38 +0000 (16:17 +0000)]
Don't reuse a const char * when we really want a char *.

22 years agoImprove WARNS situation for kdump:
David Malone [Sun, 28 Jul 2002 16:05:07 +0000 (16:05 +0000)]
Improve WARNS situation for kdump:

1) Define _KERNEL while including sys/time.h to get some function prototypes.
2) Add prototypes and ANSIify definitions.
3) Constness changes.
4) Remove register keyword.
5) Actually return a sensible value from main.
6) Make fread_tail take a void * instead of a char *.
7) Avoid a signedness warning by casting to a size_t. Should be safe
   enough 'cos we also check for nonnegativity.
8) Be extra chummy with sigset_t rather than passing a struct to printf
   and pretending it is an int.

22 years agoRevert the previous delta; uintfptr_t needs to be available to
Mike Barcroft [Sun, 28 Jul 2002 15:59:51 +0000 (15:59 +0000)]
Revert the previous delta; uintfptr_t needs to be available to
userland for libc/gmon to compile, so the typedef in <machine/types.h>
isn't good enough.  This is really ugly since we end up with the
actual value which uintfptr_t is typedef'd from, in multiple places.
This is bug for bug compatible with the other FreeBSD architectures.

Noticed by: sparc64 tinderbox

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:50:38 +0000 (15:50 +0000)]
ANSIify function definitions to avoid a warning.

22 years agoThe return value of snprintf should be always nonnegative, so it should
David Malone [Sun, 28 Jul 2002 15:41:15 +0000 (15:41 +0000)]
The return value of snprintf should be always nonnegative, so it should
be safe to cast it to a size_t before comparing it to a sizeof().

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:32:17 +0000 (15:32 +0000)]
ANSIify function definitions to avoid a warning.

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:28:38 +0000 (15:28 +0000)]
ANSIify function definitions to avoid a warning.
Minor constness changes.

22 years agoANSify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:25:15 +0000 (15:25 +0000)]
ANSify function definitions to avoid a warning.

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:22:43 +0000 (15:22 +0000)]
ANSIify function definitions to avoid a warning.

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:20:10 +0000 (15:20 +0000)]
ANSIify function definitions to avoid a warning.

PR: 38930
Submitted by: keramida

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 15:13:17 +0000 (15:13 +0000)]
ANSIify function definitions to avoid a warning.

22 years agoANSIify function definitions to avoid some warnings.
David Malone [Sun, 28 Jul 2002 15:02:24 +0000 (15:02 +0000)]
ANSIify function definitions to avoid some warnings.
Include stdlib.h for exit.

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 14:55:59 +0000 (14:55 +0000)]
ANSIify function definitions to avoid a warning.
Minor constness changes.

22 years agoANSIify function definitions to avoid a warning.
David Malone [Sun, 28 Jul 2002 14:41:26 +0000 (14:41 +0000)]
ANSIify function definitions to avoid a warning.

22 years agoSearch the include path for the argument to -f, to make lives easier for
Juli Mallett [Sun, 28 Jul 2002 13:46:09 +0000 (13:46 +0000)]
Search the include path for the argument to -f, to make lives easier for
those of us who want to figure out how old Jim Mock is, but only want to
type 'calendar -f calendar.freebsd'.  This is done in a way that should
be totally backwards compatible with no noticable differences, at all.

Reviewed by: mux
MFC after: 4 weeks

22 years agoNew release notes: devfs rule subsystem, lock(1) -v
Dima Dorfman [Sun, 28 Jul 2002 08:45:26 +0000 (08:45 +0000)]
New release notes: devfs rule subsystem, lock(1) -v

22 years agoLock page queue accesses by vm_page_free().
Alan Cox [Sun, 28 Jul 2002 08:01:48 +0000 (08:01 +0000)]
Lock page queue accesses by vm_page_free().

22 years agoAdd a -v option that prevents switching virtual terminals while this
Dima Dorfman [Sun, 28 Jul 2002 07:13:53 +0000 (07:13 +0000)]
Add a -v option that prevents switching virtual terminals while this
terminal is locked.  This permits the user to easily lock the entire
console from a single terminal.

22 years agoNuke whitespace at EOL.
Dima Dorfman [Sun, 28 Jul 2002 06:59:40 +0000 (06:59 +0000)]
Nuke whitespace at EOL.

22 years agoCorrect misindentation of DRA_UID.
Dima Dorfman [Sun, 28 Jul 2002 06:57:57 +0000 (06:57 +0000)]
Correct misindentation of DRA_UID.

22 years agoWhat we're actually passing around is an ioctl request, not a command,
Dima Dorfman [Sun, 28 Jul 2002 06:57:13 +0000 (06:57 +0000)]
What we're actually passing around is an ioctl request, not a command,
and it's an unsigned long, not an int.

22 years agoThis compiles cleanly with WARNS=5 on i386 and alpha, so try to make
Dima Dorfman [Sun, 28 Jul 2002 06:52:47 +0000 (06:52 +0000)]
This compiles cleanly with WARNS=5 on i386 and alpha, so try to make
sure that doesn't change.

22 years agoImplement this (quoted from the updated man page): If the first token
Dima Dorfman [Sun, 28 Jul 2002 06:45:30 +0000 (06:45 +0000)]
Implement this (quoted from the updated man page): If the first token
of a rule specification is a single dash (``-''), rules are read from
the standard input and the rest of the specification is ignored.

22 years ago o Lock page queue accesses by vm_page_free().
Alan Cox [Sun, 28 Jul 2002 05:46:47 +0000 (05:46 +0000)]
 o Lock page queue accesses by vm_page_free().
 o Increment cnt.v_dfree inside vm_pageout_page_free() rather than
   at each call.

22 years ago o Lock page queue accesses by vm_page_free().
Alan Cox [Sun, 28 Jul 2002 04:23:03 +0000 (04:23 +0000)]
 o Lock page queue accesses by vm_page_free().

22 years agoUse more standard section names in an attempt to unconfuse new users
Dima Dorfman [Sun, 28 Jul 2002 03:58:12 +0000 (03:58 +0000)]
Use more standard section names in an attempt to unconfuse new users
(and make it easier to find stuff for the old ones).  I originally put
everything under one "RULE SUBSYSTEM" so it didn't look like devfs(8)
is dedicated to the rule subsystem, but since nobody has any idea what
else might go here, it's not worth confusing people just to save a
little time for someone that might add something rule-unrelated to
devfs(8) in a few years.

Suggested by: sheldonh, phk

22 years agoAs of revision 1.38 of make/parse.c, our make(1) will warn too. Note that
Juli Mallett [Sun, 28 Jul 2002 03:57:08 +0000 (03:57 +0000)]
As of revision 1.38 of make/parse.c, our make(1) will warn too.  Note that
this isn't just for the sake of testing behaviour, and that things really
do break if this regression occurs.

22 years agoUnimplement panic(8) by making sure that we don't recurse into a
Dima Dorfman [Sun, 28 Jul 2002 03:52:44 +0000 (03:52 +0000)]
Unimplement panic(8) by making sure that we don't recurse into a
ruleset.  If we do, that means there's a ruleset loop (10 includes 20
include 30 includes 10), which will quickly cause a double fault due
to stack overflow (since "include" is implemented by recursion).
(Previously, we only checked that X didn't include X.)

22 years agoPrint a warning when we are given two scripts for one target. This is neither
Juli Mallett [Sun, 28 Jul 2002 03:52:41 +0000 (03:52 +0000)]
Print a warning when we are given two scripts for one target.  This is neither
as wide-reaching nor intensive as NetBSD's similar, but the warning uses the
same text.

Inspired by: NetBSD

22 years agoStock -current has more than 300 files in /etc, so 255 inodes for the
Dima Dorfman [Sun, 28 Jul 2002 03:41:53 +0000 (03:41 +0000)]
Stock -current has more than 300 files in /etc, so 255 inodes for the
/etc filesystem isn't enough; consequently, add "-i 4096" to the newfs
command for /etc.  This results in 1022 inodes, which should be enough
for the forseeable future (although I don't know why we would ever
have more than 1000 files in a default /etc).

Silence by: -current

22 years agoAdd a period to the end of the "starting" message to be consistent
Dima Dorfman [Sun, 28 Jul 2002 03:38:10 +0000 (03:38 +0000)]
Add a period to the end of the "starting" message to be consistent
with the rest of the output during a boot.

22 years agoCalculate checksums correctly when LSRR is in effect by passing the
Bill Fenner [Sun, 28 Jul 2002 02:34:39 +0000 (02:34 +0000)]
Calculate checksums correctly when LSRR is in effect by passing the
 IP header and protocol header seperately.
Also calculate TCP checksums.

Submitted by: orion

22 years agoUpdate for traceroute 1.4a12
Bill Fenner [Sun, 28 Jul 2002 02:27:07 +0000 (02:27 +0000)]
Update for traceroute 1.4a12

22 years agoMerge 1.4a12
Bill Fenner [Sun, 28 Jul 2002 02:26:31 +0000 (02:26 +0000)]
Merge 1.4a12

22 years agoThis commit was generated by cvs2svn to compensate for changes in r100784,
Bill Fenner [Sun, 28 Jul 2002 02:24:33 +0000 (02:24 +0000)]
This commit was generated by cvs2svn to compensate for changes in r100784,
which included commits to RCS files with non-trunk default branches.

22 years agoVirgin import of LBL traceroute v1.4a12
Bill Fenner [Sun, 28 Jul 2002 02:24:33 +0000 (02:24 +0000)]
Virgin import of LBL traceroute v1.4a12

22 years agoAdd declarations for btext and etext.
Jake Burkholder [Sun, 28 Jul 2002 01:01:14 +0000 (01:01 +0000)]
Add declarations for btext and etext.

22 years agoClarify J.S. Bach's life: he was only born once, and he died in Leipzig.
Greg Lehey [Sun, 28 Jul 2002 00:42:02 +0000 (00:42 +0000)]
Clarify J.S. Bach's life: he was only born once, and he died in Leipzig.
Add entry for Antonio Vivaldi's death.

22 years agoUnwind the syscall_with_err_pushed tweak that jake did some time back.
Peter Wemm [Sun, 28 Jul 2002 00:27:51 +0000 (00:27 +0000)]
Unwind the syscall_with_err_pushed tweak that jake did some time back.

OK'ed by: jake

22 years agouintfptr_t has moved to machine/types.h.
Jake Burkholder [Sat, 27 Jul 2002 23:36:51 +0000 (23:36 +0000)]
uintfptr_t has moved to machine/types.h.

22 years ago o Require that the page queues lock is held on entry to vm_pageout_clean()
Alan Cox [Sat, 27 Jul 2002 23:20:32 +0000 (23:20 +0000)]
 o Require that the page queues lock is held on entry to vm_pageout_clean()
   and vm_pageout_flush().
 o Acquire the page queues lock before calling vm_pageout_clean()
   or vm_pageout_flush().

22 years agoMake usage match options.
David Malone [Sat, 27 Jul 2002 23:19:25 +0000 (23:19 +0000)]
Make usage match options.

Approved by: rwatson

22 years agoMake bn_name a const char * to quieten some warnings.
David Malone [Sat, 27 Jul 2002 23:16:49 +0000 (23:16 +0000)]
Make bn_name a const char * to quieten some warnings.

Approved by: bp (long ago)

22 years agoMake spxnames a const char * to quieten some warnings in netstat.
David Malone [Sat, 27 Jul 2002 23:15:08 +0000 (23:15 +0000)]
Make spxnames a const char * to quieten some warnings in netstat.

22 years agoIf a socket is disconnected for some reason (like a TCP connection
David Malone [Sat, 27 Jul 2002 23:06:52 +0000 (23:06 +0000)]
If a socket is disconnected for some reason (like a TCP connection
not responding) then drop any data on the outgoing queue in
soisdisconnected because there is no way to get it to its destination
any longer.

The only objection to this patch I got on -net was from Terry, who
wasn't sure that the condition in question could arise, so I provided
some example code.

22 years agoMake test check the tv_nsec part of a struct stat when comparing
David Malone [Sat, 27 Jul 2002 22:53:44 +0000 (22:53 +0000)]
Make test check the tv_nsec part of a struct stat when comparing
the mtimes of a file. (This is probably only useful if you have
vfs.timestamp_precision set to something nonzero).

PR: 39163
Submitted by: Hal Burch <hburch@lumeta.com>
MFC after: 2 weeks

22 years ago- Fixup whitespace after previous commit.
John Baldwin [Sat, 27 Jul 2002 22:15:42 +0000 (22:15 +0000)]
- Fixup whitespace after previous commit.
- To minimize whitespace changes, remove a test that didn't define
  _CPUCFLAGS if both NO_CPU_CFLAGS and NO_CPU_COPTFLAGS were defined
  since it is redundant (we don't use _CPUCFLAGS if those are defined).

22 years agoIf there is not a CPUTYPE defined by default, then allow for _CPUCFLAGS
John Baldwin [Sat, 27 Jul 2002 22:04:05 +0000 (22:04 +0000)]
If there is not a CPUTYPE defined by default, then allow for _CPUCFLAGS
to tune for more advanced processors while still supporting the minimum
processor in an architecture.  We can do this with the '-mtune=' option
to gcc for alpha, sparc64, and powerpc and with the mis-named '-mcpu='
option for i386.

This defaults to tuning i386 builds for i686 machines though not using
any instructions that aren't found on an 80386.  For alpha it defaults
to tuning for an EV5.

Approved by: peter
Peril sensitive sunglasses borrowed from: peter

22 years agoImplement a direct mapped address region, like alpha and ia64. This
Jake Burkholder [Sat, 27 Jul 2002 21:57:38 +0000 (21:57 +0000)]
Implement a direct mapped address region, like alpha and ia64.  This
basically maps all of physical memory 1:1 to a range of virtual addresses
outside of normal kva.  The advantage of doing this instead of accessing
phsyical addresses directly is that memory accesses will go through the
data cache, and will participate in the normal cache coherency algorithm
for invalidating lines in our own and in other cpus' data caches.  So
we don't have to flush the cache manually or send IPIs to do so on other
cpus.  Also, since the mappings never change, we don't have to flush them
from the tlb manually.
This makes pmap_copy_page and pmap_zero_page MP safe, allowing the idle
zero proc to run outside of giant.

Inspired by: ia64

22 years agoFix location and name of if_an_pci.c in comment.
Bill Fenner [Sat, 27 Jul 2002 21:28:40 +0000 (21:28 +0000)]
Fix location and name of if_an_pci.c in comment.

22 years agoDocument DOCRELEASETAG and PORTSRELEASETAG.
John Baldwin [Sat, 27 Jul 2002 21:09:29 +0000 (21:09 +0000)]
Document DOCRELEASETAG and PORTSRELEASETAG.

22 years agoFix spacing for -P (policy) examples.
Bill Fenner [Sat, 27 Jul 2002 21:06:06 +0000 (21:06 +0000)]
Fix spacing for -P (policy) examples.

22 years agoDocument PREFETCHDISTFILES.
John Baldwin [Sat, 27 Jul 2002 20:04:48 +0000 (20:04 +0000)]
Document PREFETCHDISTFILES.

22 years agoDocument KERNEL_FLAGS and WORLD_FLAGS.
John Baldwin [Sat, 27 Jul 2002 20:00:39 +0000 (20:00 +0000)]
Document KERNEL_FLAGS and WORLD_FLAGS.

22 years agoReserve VCACHEDLABEL vnode flag for use by the TrustedBSD MAC
Robert Watson [Sat, 27 Jul 2002 19:56:27 +0000 (19:56 +0000)]
Reserve VCACHEDLABEL vnode flag for use by the TrustedBSD MAC
implementation.  This flag will indicate that the security label
in the vnode is currently valid, and therefore doesn't need to
be refreshed before an access control decision can be made.  Most
file systems (or stdvops) will set this flag after they load the
MAC label from disk the first time to prevent redundant disk I/O;
some file synthetic file systems (procfs, for example) may not.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoRemote socheckproc(), which was removed when p_can*() was introduced
Robert Watson [Sat, 27 Jul 2002 19:54:15 +0000 (19:54 +0000)]
Remote socheckproc(), which was removed when p_can*() was introduced
ages ago.  The prototype was missed.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoSlight whitespace cleanup. Whitespace sync to MAC tree.
Robert Watson [Sat, 27 Jul 2002 19:53:02 +0000 (19:53 +0000)]
Slight whitespace cleanup.  Whitespace sync to MAC tree.

22 years ago- Move BUILDNAME up before CHROOTDIR.
John Baldwin [Sat, 27 Jul 2002 19:52:35 +0000 (19:52 +0000)]
- Move BUILDNAME up before CHROOTDIR.
- RELEASETAG is an optional variable, not a required one.

22 years agoKernel options for Mandatory Access Control (MAC).
Robert Watson [Sat, 27 Jul 2002 19:50:28 +0000 (19:50 +0000)]
Kernel options for Mandatory Access Control (MAC).

MAC support will be merged into the main tree over the next week in
reasonable size chunks; much more to follow.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoIf we are building docs with this release and PREFETCHDISTFILES is defined
John Baldwin [Sat, 27 Jul 2002 19:43:52 +0000 (19:43 +0000)]
If we are building docs with this release and PREFETCHDISTFILES is defined
then download the distfiles for the ports needed to build the docs before
we enter the chroot environment.  This is useful since often times releases
get in a funk trying to download distfiles in the chroot.

Approved by: re (murray, bmah)
MFC after: 5 days