]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoUpdate for tcpdump 3.7.1
fenner [Fri, 21 Jun 2002 00:52:33 +0000 (00:52 +0000)]
Update for tcpdump 3.7.1

22 years agoMerge tcpdump 3.7.1
fenner [Fri, 21 Jun 2002 00:49:02 +0000 (00:49 +0000)]
Merge tcpdump 3.7.1

MFC after: 2 weeks

22 years agoThis commit was generated by cvs2svn to compensate for changes in r98524,
fenner [Fri, 21 Jun 2002 00:43:23 +0000 (00:43 +0000)]
This commit was generated by cvs2svn to compensate for changes in r98524,
which included commits to RCS files with non-trunk default branches.

22 years agoImport tcpdump 3.7.1, from
fenner [Fri, 21 Jun 2002 00:43:23 +0000 (00:43 +0000)]
Import tcpdump 3.7.1, from
 http://www.tcpdump.org/release/tcpdump-3.7.1.tar.gz

22 years agoRelease note added: rc.d.
bmah [Thu, 20 Jun 2002 23:44:18 +0000 (23:44 +0000)]
Release note added:  rc.d.

MFCs noted:  fold(1) -b/-s, pwd(1) -L, renice(8) -n, unexpand(1) -t,
xargs(1) -I/-L.

22 years agoModified release note: ncurses 5.2-20020615, added version number for
bmah [Thu, 20 Jun 2002 22:50:15 +0000 (22:50 +0000)]
Modified release note:  ncurses 5.2-20020615, added version number for
lukemftp.

Deleted release note:  gawk.

MFC noted:  lukemftpd.

22 years agoAdd some #define's for mbuf annotations.
luigi [Thu, 20 Jun 2002 21:29:55 +0000 (21:29 +0000)]
Add some #define's for mbuf annotations.

As the comment in the code says, eventually there will be a proper
data structure (e.g NetBSD's struct m_tag) to store chains of
annotations, and mbuf-handling procedures will handle these chains
in the correct way.

Right now, these chains do not exist, and we just use the constants
defined here to implement simple ad-hoc solutions to remove some global
variables used so far to pass around informations about packets
being processed.

Global variables are not only ugly and make the code unreadable, they
also prevent from using parallelism in network stack processing.

(the 3-days MFC only refers to this commit, i.e. the PACKET_TAG_*
constants; the full mechanism will be committed and MFC'ed on a
longer timescale).

MFC after: 3 days

22 years agoProvide a way to keep out old ("historic") release notes from bloating
bmah [Thu, 20 Jun 2002 21:19:49 +0000 (21:19 +0000)]
Provide a way to keep out old ("historic") release notes from bloating
the size of the rendered release notes.  By default, any element with
a role="historic" attribute won't get rendered in the output unless
the INCLUDE_HISTORIC Makefile variable is defined.

22 years agoAdd a few files which are needed to build "passwd" now that
luigi [Thu, 20 Jun 2002 21:17:33 +0000 (21:17 +0000)]
Add a few files which are needed to build "passwd" now that
the standard version of these files have been PAMified (and axed).

22 years agoRemove GAWK. I removed it from the build 2 weeks ago and the world
obrien [Thu, 20 Jun 2002 20:54:54 +0000 (20:54 +0000)]
Remove GAWK.  I removed it from the build 2 weeks ago and the world
did not end.  So finish the deed.

22 years ago{f,s}usword -> {f,s}uword16. Implement {f,s}uword32.
jake [Thu, 20 Jun 2002 20:41:57 +0000 (20:41 +0000)]
{f,s}usword -> {f,s}uword16.  Implement {f,s}uword32.

Requested by: peter

22 years agoChange the way we internally store the mount options to
mux [Thu, 20 Jun 2002 20:03:42 +0000 (20:03 +0000)]
Change the way we internally store the mount options to
a linked list.  This is to allow the merging of the mount
options in the MNT_UPDATE case, as the current data structure
is unsuitable for this.

There are no functional differences in this commit.

Reviewed by: phk

22 years agoFix a bug fixed by NetBSD in revision 1.42 of parse.c by christos which caused
jmallett [Thu, 20 Jun 2002 19:51:13 +0000 (19:51 +0000)]
Fix a bug fixed by NetBSD in revision 1.42 of parse.c by christos which caused
an example Makefile I was showing someone just last night to report a variable
as being recursive.

Obtained from: NetBSD

22 years agoMore unsigned char casts to isupper(3).
jmallett [Thu, 20 Jun 2002 19:45:44 +0000 (19:45 +0000)]
More unsigned char casts to isupper(3).

Obtained from: NetBSD

22 years agoMerge ncurses 5.2-20020615 changes onto mainline.
peter [Thu, 20 Jun 2002 19:44:40 +0000 (19:44 +0000)]
Merge ncurses 5.2-20020615 changes onto mainline.

22 years agoProvide a heuristic for RCS conflicts.
jmallett [Thu, 20 Jun 2002 19:44:34 +0000 (19:44 +0000)]
Provide a heuristic for RCS conflicts.

Obtained from: NetBSD

22 years agoThis commit was generated by cvs2svn to compensate for changes in r98503,
peter [Thu, 20 Jun 2002 19:42:51 +0000 (19:42 +0000)]
This commit was generated by cvs2svn to compensate for changes in r98503,
which included commits to RCS files with non-trunk default branches.

22 years agoImport ncurses 5.2-20020615 to fix a first-column display bug.
peter [Thu, 20 Jun 2002 19:42:51 +0000 (19:42 +0000)]
Import ncurses 5.2-20020615 to fix a first-column display bug.

22 years agoLook for wildcards, and balanced meta-characters such as { and } like NetBSD
jmallett [Thu, 20 Jun 2002 19:41:21 +0000 (19:41 +0000)]
Look for wildcards, and balanced meta-characters such as { and } like NetBSD
does.

Obtained from: NetBSD

22 years agoExpand a buffer to reduce diffs to NetBSD.
jmallett [Thu, 20 Jun 2002 19:31:55 +0000 (19:31 +0000)]
Expand a buffer to reduce diffs to NetBSD.

Obtained from: NetBSD

22 years agoDiff reduction for great justice against NetBSD, cast to unsigned char when
jmallett [Thu, 20 Jun 2002 19:28:00 +0000 (19:28 +0000)]
Diff reduction for great justice against NetBSD, cast to unsigned char when
passing an argument to isspace(3).

22 years agoImplement SO_NOSIGPIPE option for sockets. This allows one to request that
alfred [Thu, 20 Jun 2002 18:52:54 +0000 (18:52 +0000)]
Implement SO_NOSIGPIPE option for sockets.  This allows one to request that
an EPIPE error return not generate SIGPIPE on sockets.

Submitted by: lioux
Inspired by: Darwin

22 years ago o Remove an incorrect cast from obreak(). This cast would,
alc [Thu, 20 Jun 2002 18:38:28 +0000 (18:38 +0000)]
 o Remove an incorrect cast from obreak().  This cast would,
   for example, break an sbrk(>=4GB) on 64-bit architectures
   even if the resource limit allowed it.
 o Correct an off-by-one error.
 o Correct a spelling error in a comment.
 o Reorder an && expression so that the commonly FALSE expression
   comes first.

Submitted by: bde (bullets 1 and 2)

22 years agoDon't leak resources if fdcheckstd() fails during exec.
alfred [Thu, 20 Jun 2002 17:27:28 +0000 (17:27 +0000)]
Don't leak resources if fdcheckstd() fails during exec.

Submitted by: Mike Makonnen <makonnen@pacbell.net>

22 years agoDon't try to decode old-style options if the argv[1] begins with `-' and the
sobomax [Thu, 20 Jun 2002 14:55:53 +0000 (14:55 +0000)]
Don't try to decode old-style options if the argv[1] begins with `-' and the
second character represents some option taking an argument. This fixes
problem when ps(1) is invoked for examply as follows:

$ ps -Ufoobar1234

the above example results in option string being interpreted as
-U foobarp1234 - note extra `p'.

Reported by: Vladimir Sotnikov <vovan@kyivstar.net>
MFC after: 2 weeks

22 years agoDisplay the mutex name in the ^T status line if the selected thread
iedowse [Thu, 20 Jun 2002 14:03:36 +0000 (14:03 +0000)]
Display the mutex name in the ^T status line if the selected thread
is blocked on a mutex. Prepend a '*' to distinguish this case as
is done in top(1).

22 years agoWhitespace fixes to line-up EOL escapes whose right-most extraints changed in
jmallett [Thu, 20 Jun 2002 11:17:46 +0000 (11:17 +0000)]
Whitespace fixes to line-up EOL escapes whose right-most extraints changed in
the previous revision due to the biggest line changing.

22 years agoCast sp to uintptr_t when doing a compare of it to STACKMAX which is a size_t.
jmallett [Thu, 20 Jun 2002 11:14:54 +0000 (11:14 +0000)]
Cast sp to uintptr_t when doing a compare of it to STACKMAX which is a size_t.

This messes up some indentation in mdef.h for some macros.

22 years agoIt seems the hard lock on m4(1) is viewed as superfluous. Okay. It's gone.
jmallett [Thu, 20 Jun 2002 10:37:43 +0000 (10:37 +0000)]
It seems the hard lock on m4(1) is viewed as superfluous.  Okay.  It's gone.

I'm sick of waiting on OpenBSD to make a number of changes anyway.  If someone
else wants to take over fixing m4(1), I'll be glad to make them aware of the
existing issues.

22 years agoQuick fix for the type of the bitmap in sigset_t. It was an array of
bde [Thu, 20 Jun 2002 09:04:33 +0000 (09:04 +0000)]
Quick fix for the type of the bitmap in sigset_t.  It was an array of
4 u_ints but needs to be an array of 4 uint32_t's to work, at least
if unsigned ints have less than 32 bits.  It should be a non-array of
1 uint128_t on 128-bit machines, especially if u_int has 128 bits.
The headers that declare uint32_t (actually __uint32_t) are intentionally
not included here since this header should only be included by other
headers.

Fixed some style bugs (space instead of tab after #ifndef and #endif).

22 years agoAnonymize the "portmap" program to get better compatibility with
dougb [Thu, 20 Jun 2002 08:49:44 +0000 (08:49 +0000)]
Anonymize the "portmap" program to get better compatibility with
rpcbind in -current.

Submitted by: Alexander Kabaev <ak03@gte.com>

22 years agoUpdate an 'XXX what is this?' type comment about suswintr and fuswintr.
peter [Thu, 20 Jun 2002 07:40:53 +0000 (07:40 +0000)]
Update an 'XXX what is this?' type comment about suswintr and fuswintr.
These are 16 bit short values used only by the profiling code.

22 years agoStyle: put static qualifier on definition of static functions.
tjr [Thu, 20 Jun 2002 07:36:22 +0000 (07:36 +0000)]
Style: put static qualifier on definition of static functions.

22 years agoUse suword16/fuword16 instead of susword/fusword - this has two different
peter [Thu, 20 Jun 2002 07:23:08 +0000 (07:23 +0000)]
Use suword16/fuword16 instead of susword/fusword - this has two different
definitions so far.. 16 bit on x86 and appears to be 32 bit on sparc64.
Be explicit to avoid suprises.

22 years agoBogusness may have happened using a variable assignment here before, but
jmallett [Thu, 20 Jun 2002 07:15:55 +0000 (07:15 +0000)]
Bogusness may have happened using a variable assignment here before, but
proper parens mean that fd is always set by open(2) [in any part of C],
and so we can accurately check for it returning -1, without feeling like
we need to initialise fd to -1 in its declaration.

In other words, fix a stylistic/bogus nit.

22 years agoDeorbit suibyte(). It was only used for split address space systems
peter [Thu, 20 Jun 2002 07:13:35 +0000 (07:13 +0000)]
Deorbit suibyte().  It was only used for split address space systems
for supporting UIO_USERISPACE (ie: it wasn't used).

22 years agotime(1) does not use the shell to execute the specified utility, remove
tjr [Thu, 20 Jun 2002 07:12:34 +0000 (07:12 +0000)]
time(1) does not use the shell to execute the specified utility, remove
sentence stating it did.

22 years agoRemove references to UIO_USERISPACE which no longer exists.
peter [Thu, 20 Jun 2002 07:09:24 +0000 (07:09 +0000)]
Remove references to UIO_USERISPACE which no longer exists.

22 years agoRemove UIO_USERISPACE - we do not support any split instruction/data
peter [Thu, 20 Jun 2002 07:08:43 +0000 (07:08 +0000)]
Remove UIO_USERISPACE - we do not support any split instruction/data
address space machines (eg: pdp-11) and are not likely to ever do so.
Nothing in our kernel sets this.

22 years agoMake it obvious that command line arguments may be specified for the utility
tjr [Thu, 20 Jun 2002 07:07:00 +0000 (07:07 +0000)]
Make it obvious that command line arguments may be specified for the utility
that is to be invoked on the command line. Use "utility" instead of "command"
in manual page and usage message for consistency with POSIX.

22 years agoThat's right, you can shove your xargs(1) issues in my direction, and I'm
jmallett [Thu, 20 Jun 2002 07:06:21 +0000 (07:06 +0000)]
That's right, you can shove your xargs(1) issues in my direction, and I'm
also glad to review changes to it.

22 years agoia32 %edx return comes from td_retval[1], not td_retval[0]
peter [Thu, 20 Jun 2002 06:47:44 +0000 (06:47 +0000)]
ia32 %edx return comes from td_retval[1], not td_retval[0]

Obtained from: dfr

22 years agoUse suword32/64 and fuword32/64 like elsewhere instead of inventing
peter [Thu, 20 Jun 2002 06:45:43 +0000 (06:45 +0000)]
Use suword32/64 and fuword32/64 like elsewhere instead of inventing
suhword/fuhword.

22 years agoAdd boot_serial and boot_multicons variables to set RB_SERIAL and
peter [Thu, 20 Jun 2002 06:29:42 +0000 (06:29 +0000)]
Add boot_serial and boot_multicons variables to set RB_SERIAL and
RB_MULTIPLE since this seems to be the easiest way to add these flags
for non-forth loaders etc.

22 years agopanic rather than fault and explode if we fail to contigmalloc a kernel
peter [Thu, 20 Jun 2002 06:25:22 +0000 (06:25 +0000)]
panic rather than fault and explode if we fail to contigmalloc a kernel
stack.  This is still bad(TM), but at least we have a clue when we get
hit when contigmalloc fails.

22 years agoUse the canonical pmap_{new,dispose,swapin,swapout}_proc() functions,
peter [Thu, 20 Jun 2002 06:23:40 +0000 (06:23 +0000)]
Use the canonical pmap_{new,dispose,swapin,swapout}_proc() functions,
in this case cut/pasted from sparc64 instead of messing with
contigmalloc where it is not needed.

22 years agoMove the "- 1" into the RQB_FFS(mask) macro itself so that
peter [Thu, 20 Jun 2002 06:21:20 +0000 (06:21 +0000)]
Move the "- 1" into the RQB_FFS(mask) macro itself so that
implementations can provide a base zero ffs function if they wish.
This changes
  #define RQB_FFS(mask) (ffs64(mask))
  foo = RQB_FFS(mask) - 1;
to
  #define RQB_FFS(mask) (ffs64(mask) - 1)
  foo = RQB_FFS(mask);
On some platforms we can get the "- 1" for free, eg: those that use the
C code for ffs64().

Reviewed by: jake (in principle)

22 years agoUse size_t consistently and complete some uncompleted code resulting in a
jmallett [Thu, 20 Jun 2002 06:00:51 +0000 (06:00 +0000)]
Use size_t consistently and complete some uncompleted code resulting in a
memory leak by assigning and freeing a variable appropriately as well as
keeping track of the amount of allocated ram properly.

MFC after: 1 month

22 years agoA function which takes no arguments has an argument list spelled (void) in the
jmallett [Thu, 20 Jun 2002 05:35:40 +0000 (05:35 +0000)]
A function which takes no arguments has an argument list spelled (void) in the
world of ANSI C.

22 years agoWe have a place for extern declarations of global variables in ed.h, do not
jmallett [Thu, 20 Jun 2002 05:28:02 +0000 (05:28 +0000)]
We have a place for extern declarations of global variables in ed.h, do not
use main() to do it locally.

22 years agoRemove two unused variables.
jmallett [Thu, 20 Jun 2002 05:21:43 +0000 (05:21 +0000)]
Remove two unused variables.

22 years agoMinor const cleanup.
jmallett [Thu, 20 Jun 2002 05:20:50 +0000 (05:20 +0000)]
Minor const cleanup.

Don't discard qualifiers we don't need to discard.

22 years agoAdd a warning regarding the SENDMAIL_*_MC make.conf variable values.
gshapiro [Thu, 20 Jun 2002 04:33:06 +0000 (04:33 +0000)]
Add a warning regarding the SENDMAIL_*_MC make.conf variable values.
Using /etc/mail/sendmail.mc will create /etc/mail/sendmail.cf during
a buildworld.

PR: misc/39397
MFC after: 3 days

22 years agoGrrr, make the test for embedded variables in the left-hand-side actually do
jmallett [Thu, 20 Jun 2002 03:08:20 +0000 (03:08 +0000)]
Grrr, make the test for embedded variables in the left-hand-side actually do
the right thing in every case.  Yuck.

22 years ago o Acquire and release the vm_map lock instead of Giant in obreak().
alc [Thu, 20 Jun 2002 02:04:55 +0000 (02:04 +0000)]
 o Acquire and release the vm_map lock instead of Giant in obreak().
   Consequently, use vm_map_insert() and vm_map_delete(), which expect
   the vm_map to be locked, instead of vm_map_find() and vm_map_remove(),
   which do not.

22 years agoSolve the 'unregistered netisr 18' information notice with a sledgehammer.
peter [Thu, 20 Jun 2002 01:27:40 +0000 (01:27 +0000)]
Solve the 'unregistered netisr 18' information notice with a sledgehammer.
Register the ISR early, but do not actually kick off the timer until we
see some activity.  This still saves us from running the arp timers on
a system with no network cards.

22 years ago- Move the computation of pflags out of the page allocation loop in
jeff [Wed, 19 Jun 2002 23:49:57 +0000 (23:49 +0000)]
- Move the computation of pflags out of the page allocation loop in
  kmem_malloc()
- zero fill pages if PG_ZERO bit is not set after allocation in kmem_malloc()

Suggested by: alc, jake

22 years ago- Remove the lock(9) protecting the kernel linker system.
arr [Wed, 19 Jun 2002 21:25:59 +0000 (21:25 +0000)]
- Remove the lock(9) protecting the kernel linker system.
- Added a mutex, kld_mtx, to protect the kernel_linker system.  Note that
  while ``classes'' is global (to that file), it is only read only after
  SI_SUB_KLD, SI_ORDER_ANY.
- Add a SYSINIT to flip a flag that disallows class registration after
  SI_SUB_KLD, SI_ORDER_ANY.

Idea for ``classes'' read only by: jake
Reviewed by: jake

22 years ago- Remove bogus use of kmem_alloc that was inherited from the old zone
jeff [Wed, 19 Jun 2002 20:49:44 +0000 (20:49 +0000)]
- Remove bogus use of kmem_alloc that was inherited from the old zone
  allocator.
- Properly set M_ZERO when talking to the back end page allocators for
  non malloc zones.  This forces us to zero fill pages when they are first
  brought into a cache.
- Properly handle M_ZERO in uma_zalloc_internal.  This fixes a problem where
  per cpu buckets weren't always getting zeroed.

22 years agoTeach kmem_malloc about M_ZERO.
jeff [Wed, 19 Jun 2002 20:47:18 +0000 (20:47 +0000)]
Teach kmem_malloc about M_ZERO.

22 years agoCorrect spelling of 'supplied'.
robert [Wed, 19 Jun 2002 20:44:48 +0000 (20:44 +0000)]
Correct spelling of 'supplied'.

PR: misc/39528

22 years agoWe don't use this any more.
des [Wed, 19 Jun 2002 20:01:25 +0000 (20:01 +0000)]
We don't use this any more.

Sponsored by: DARPA, NAI Labs

22 years agoEnable OPIE for sshd and telnetd. I thought I'd done this a long time
des [Wed, 19 Jun 2002 20:00:43 +0000 (20:00 +0000)]
Enable OPIE for sshd and telnetd.  I thought I'd done this a long time
ago...

Sponsored by: DARPA, NAI Labs

22 years agoMake locate.updatedb tell about the security risk when it is run as root.
eivind [Wed, 19 Jun 2002 19:22:18 +0000 (19:22 +0000)]
Make locate.updatedb tell about the security risk when it is run as root.

22 years agoChange spelling of `u_char' to `unsigned char' to avoid requiring
mike [Wed, 19 Jun 2002 19:05:41 +0000 (19:05 +0000)]
Change spelling of `u_char' to `unsigned char' to avoid requiring
<sys/types.h> as a prerequisite.

22 years agoAdd a test for what was fixed in revision 1.27 and 1.28 of make(1)'s var.c,
jmallett [Wed, 19 Jun 2002 18:57:04 +0000 (18:57 +0000)]
Add a test for what was fixed in revision 1.27 and 1.28 of make(1)'s var.c,
expansion of embedded variables in the left-hand-side of an assignment
expression, using the simplest case - hiding recursion using nil-expanded
variables.

22 years agoFix a memory leak from previous commit by freeing the possibly expanded
jmallett [Wed, 19 Jun 2002 17:39:36 +0000 (17:39 +0000)]
Fix a memory leak from previous commit by freeing the possibly expanded
string at the first opportunity, being sure to now always allocate the
new string from VarPossiblyExpand.  Oops.

22 years agoAdd better mediaopt support for ibss and friends.
imp [Wed, 19 Jun 2002 17:37:34 +0000 (17:37 +0000)]
Add better mediaopt support for ibss and friends.

Now the driver is closer to matching the wi man page.

Submitted by: jhay (who obtained it from OpenBSD).

22 years agoPossibly expand the variable name's embedded variables before using it, as
jmallett [Wed, 19 Jun 2002 17:23:08 +0000 (17:23 +0000)]
Possibly expand the variable name's embedded variables before using it, as
seen (somewhat) in NetBSD.  This catches a few extra recursion cases that
could be hidden by expanding a NIL variable causing an existing variable to
be returned (which caused infinite looping and climbing memory usage in at
least one case).

Obtained from: NetBSD (in principle)

22 years agoSet PATH statically, and use 'env -i' to kick chroot(8). As a result,
matusita [Wed, 19 Jun 2002 15:06:38 +0000 (15:06 +0000)]
Set PATH statically, and use 'env -i' to kick chroot(8).  As a result,
the second buildworld environment is fully isolated from parent's
environment variables.

Tested virtually on: snapshots.jp.FreeBSD.org
Approved by: arch@ (silently)
MFC after: 2 weeks

22 years agoMFi386: revisions from 1.342 to 1.344
nyan [Wed, 19 Jun 2002 13:47:53 +0000 (13:47 +0000)]
MFi386: revisions from 1.342 to 1.344

22 years agoBackout previous change and merge from sys/dev/sio/sio.c revision 1.375.
nyan [Wed, 19 Jun 2002 13:20:20 +0000 (13:20 +0000)]
Backout previous change and merge from sys/dev/sio/sio.c revision 1.375.

22 years agoMerged from sys/isa/fd.c revision 1.233.
nyan [Wed, 19 Jun 2002 13:17:43 +0000 (13:17 +0000)]
Merged from sys/isa/fd.c revision 1.233.

22 years agoUse si_iosize_max to tell the upper layers not to use more
sos [Wed, 19 Jun 2002 12:41:05 +0000 (12:41 +0000)]
Use si_iosize_max to tell the upper layers not to use more
than 32K chunks on ZIP drives instead of deblocking it in
the driver.

22 years agoAdd yet another (older) Promise chip
sos [Wed, 19 Jun 2002 12:26:20 +0000 (12:26 +0000)]
Add yet another (older) Promise chip

22 years agoBring documentation on CDPATH and its effects on cd(1) back into sync with
tjr [Wed, 19 Jun 2002 12:04:22 +0000 (12:04 +0000)]
Bring documentation on CDPATH and its effects on cd(1) back into sync with
reality (and POSIX): current directory isn't searched unless CDPATH has
a "." element or is unset.

PR: 38442
Submitted by: oleg dashevskii <be9@be9.ru>
MFC after: 1 week

22 years agoFix duplicate % in %b format introduced in rev 1.22.
tjr [Wed, 19 Jun 2002 09:42:20 +0000 (09:42 +0000)]
Fix duplicate % in %b format introduced in rev 1.22.

22 years agoIn rev 1.72 a situation related to write/mmap was fixed which could result
dillon [Wed, 19 Jun 2002 09:39:41 +0000 (09:39 +0000)]
In rev 1.72 a situation related to write/mmap was fixed which could result
in a user process gaining visibility into the 'old' contents of a filesystem
block.  There were two cases:  (1) when uiomove() fails (user process issues
illegal write), and (2) when uiomove() overlaps a mmap() of the same file at
the same offset (fault -> recursive buffer I/O reads contents of old block).

Unfortunately 1.72 also had the unintended effect of forcing the filesystem
to do a read-before-write in the case of a full-block-write (non append case),
e.g. 'dd if=/dev/zero of=test.dat bs=1m count=256 conv=notrunc'.  This
destroys performance.. not only is a read forced for every write, but
clustering breaks as well.

The solution is to clear the buffer manually in the full-block case rather
then asking BALLOC to do it (BALLOC issues the read-before-write).  In the
partial-block case we want BALLOC to do it because the read-before-write
is necessary.  This patch should greatly improve database and news-feed
server performance.

Found by: MKI <mki@mozone.net>
MFC after: 3 days

22 years agoLet printf(1) tell the difference between zero width/precision and
tjr [Wed, 19 Jun 2002 09:24:30 +0000 (09:24 +0000)]
Let printf(1) tell the difference between zero width/precision and
unspecified width/precision.

PR: 39116
Submitted by: Egil Brendsdal <egilb@ife.no>
MFC after: 1 week

22 years agoRemove the compat bits for the mis-aligned struct disklabel on alpha,
phk [Wed, 19 Jun 2002 08:37:02 +0000 (08:37 +0000)]
Remove the compat bits for the mis-aligned struct disklabel on alpha,
people got three times longer than I promised.

Sponsored by: DARPA & NAI Labs.

22 years agoDon't try to dereference conn when we know it's NULL.
des [Wed, 19 Jun 2002 08:36:00 +0000 (08:36 +0000)]
Don't try to dereference conn when we know it's NULL.

22 years agoAllow format strings containing "%%" to be reused.
tjr [Wed, 19 Jun 2002 08:18:37 +0000 (08:18 +0000)]
Allow format strings containing "%%" to be reused.

PR: 39116
Submitted by: Egil Brendsdal <egilb@ife.no>
MFC after: 1 week

22 years agoAllow `%' to be written out with an octal escape (\45 or \045).
tjr [Wed, 19 Jun 2002 08:16:14 +0000 (08:16 +0000)]
Allow `%' to be written out with an octal escape (\45 or \045).

PR: 39116
Submitted by: Egil Brendsdal <egilb@ife.no>
MFC after: 1 week

22 years agoIndicate that env(1) allows you to supply arguments to the utility it
tjr [Wed, 19 Jun 2002 07:09:44 +0000 (07:09 +0000)]
Indicate that env(1) allows you to supply arguments to the utility it
executes in the usage() message and manual page. Use "utility" instead of
"command" in both places to emphasise that shell builtins etc. will not work,
and to be consistent with the terminology used by POSIX.

PR: 39210
Submitted by: Danny J. Zerkel <dzerkel@columbus.rr.com>
MFC after: 1 week

22 years agoSquish the "could sleep with process lock" messages caused by calling
alfred [Wed, 19 Jun 2002 06:39:25 +0000 (06:39 +0000)]
Squish the "could sleep with process lock" messages caused by calling
uifind() with a proc lock held.

change_ruid() and change_euid() have been modified to take a uidinfo
structure which will be pre-allocated by callers, they will then
call uihold() on the uidinfo structure so that the caller's logic
is simplified.

This allows one to call uifind() before locking the proc struct and
thereby avoid a potential blocking allocation with the proc lock
held.

This may need revisiting, perhaps keeping a spare uidinfo allocated
per process to handle this situation or re-examining if the proc
lock needs to be held over the entire operation of changing real
or effective user id.

Submitted by: Don Lewis <dl-freebsd@catspoiler.org>

22 years agoGuard definitions for use with C++ code.
obrien [Wed, 19 Jun 2002 06:04:37 +0000 (06:04 +0000)]
Guard definitions for use with C++ code.

Submitted by: Ed Hall <edhall@yahoo-inc.com>

22 years ago o Replace GIANT_REQUIRED in vm_object_coalesce() by the acquisition and
alc [Wed, 19 Jun 2002 06:02:03 +0000 (06:02 +0000)]
 o Replace GIANT_REQUIRED in vm_object_coalesce() by the acquisition and
   release of Giant.
 o Reduce the scope of GIANT_REQUIRED in vm_map_insert().

These changes will enable us to remove the acquisition and release
of Giant from obreak().

22 years agoDisconnect the docs until we figure out if there are any with v3.
obrien [Wed, 19 Jun 2002 05:38:45 +0000 (05:38 +0000)]
Disconnect the docs until we figure out if there are any with v3.

22 years agoAnother good suggestion from Bruce, only create links if the
dougb [Wed, 19 Jun 2002 05:35:16 +0000 (05:35 +0000)]
Another good suggestion from Bruce, only create links if the
file doesn't exist already.

22 years agoDon't convert a single space before a tab stop into a tab when the
tjr [Wed, 19 Jun 2002 01:45:03 +0000 (01:45 +0000)]
Don't convert a single space before a tab stop into a tab when the
-i option is used.

22 years agofix whitespace botch in previous commit.
billf [Wed, 19 Jun 2002 01:23:54 +0000 (01:23 +0000)]
fix whitespace botch in previous commit.

22 years agoRemoved unneeded files.
pdeuskar [Tue, 18 Jun 2002 22:51:26 +0000 (22:51 +0000)]
Removed unneeded files.
if_em_fxhw.[c,h] and if_em_phy.[c,h]
have been merged into one [c,h] file.

MFC after: 3 days

22 years agosetsugid() touches p->p_flag so assert that the proc is locked.
alfred [Tue, 18 Jun 2002 22:41:35 +0000 (22:41 +0000)]
setsugid() touches p->p_flag so assert that the proc is locked.

22 years agoA node that creates a device entry in /dev (yay devfs)
julian [Tue, 18 Jun 2002 21:32:33 +0000 (21:32 +0000)]
A node that creates a device entry in /dev (yay devfs)
so that /dev/mumble can be the entrypoint to some networking graph,
e.g. a tunnel or a remote tape drive or whatever...

Not fully tested (by me) yet.

Submitted by: Mark Santcroos <marks@ripe.net>
MFC after: 3 weeks

22 years agoMake the speed used by gdb over serial settable in the kernel configuration.
n_hibma [Tue, 18 Jun 2002 21:30:37 +0000 (21:30 +0000)]
Make the speed used by gdb over serial settable in the kernel configuration.

This facilitates the use in circumstances where you are using a serial
console as well. GDB doesn't support anything higher than 9600 baud (19k2
if you are lucky), but the console does.

22 years agoFix a typo in the named startup options
gordon [Tue, 18 Jun 2002 19:42:37 +0000 (19:42 +0000)]
Fix a typo in the named startup options

Submitted by: sheldonh@

22 years ago o Remove LK_CANRECURSE from the vm_map lock.
alc [Tue, 18 Jun 2002 18:31:35 +0000 (18:31 +0000)]
 o Remove LK_CANRECURSE from the vm_map lock.

22 years agoFreeBSD is one of the OS's that does not require the use of libio with Gcc 3.1.
obrien [Tue, 18 Jun 2002 18:13:14 +0000 (18:13 +0000)]
FreeBSD is one of the OS's that does not require the use of libio with Gcc 3.1.

22 years agoFix style and wording bugs introduced in my last commit.
chris [Tue, 18 Jun 2002 08:55:17 +0000 (08:55 +0000)]
Fix style and wording bugs introduced in my last commit.

Sponsored by: DARPA, NAI Labs

22 years agoDo vidcontrol -m on after starting moused for a usb mouse.
dougb [Tue, 18 Jun 2002 07:50:16 +0000 (07:50 +0000)]
Do vidcontrol -m on after starting moused for a usb mouse.

PR: conf/39125
Submitted by: Mike Makonnen <makonnen@pacbell.net>
Not objected to by: -current and -hackers