dragonfly.git
19 years agoalias rm to rm -I by default in interactive shells. rm -I is a safer, more
Matthew Dillon [Wed, 6 Oct 2004 06:31:39 +0000 (06:31 +0000)]
alias rm to rm -I by default in interactive shells.  rm -I is a safer, more
friendly rm that will not let you shoot yourself in the foot yet which is
not as annoying as rm -i. [ Experienced sysops can always remove the aliases ].

This only effects fresh installs.

19 years agoDo not run getty on ttyd0 by default. The installer's version of ttys already
Matthew Dillon [Wed, 6 Oct 2004 06:24:11 +0000 (06:24 +0000)]
Do not run getty on ttyd0 by default.  The installer's version of ttys already
turns this off but we might as well do it in the master copy too.

19 years agoAdd the -I option to rm. This option asks for confirmation just once if
Matthew Dillon [Wed, 6 Oct 2004 06:16:43 +0000 (06:16 +0000)]
Add the -I option to rm.  This option asks for confirmation just once if
recursively removing directories or if more then 3 files are listed in the
command line.  It is designed as a better safety net for new users then
the far more annoying -i.

After: reading the bikeshed on the freebsd groups and thinking up a better way.

19 years agoUpdate ncptrace.c to handle DragonFly_Stable and HEAD.
Matthew Dillon [Wed, 6 Oct 2004 05:13:20 +0000 (05:13 +0000)]
Update ncptrace.c to handle DragonFly_Stable and HEAD.
Add ttyinfo.c (does a sanity check on the kernel's clist data)

19 years agoRegister ips device with devstat subsystem so that {io,sys,vm}stat
YONETANI Tomokazu [Wed, 6 Oct 2004 02:12:31 +0000 (02:12 +0000)]
Register ips device with devstat subsystem so that {io,sys,vm}stat
show statistics for ServeRAID device.

Requested-by: Toma Bor¹ônar
19 years agoVFS messaging/interfacing work stage 7e/99: More firming up of stage 7.
Matthew Dillon [Tue, 5 Oct 2004 07:57:41 +0000 (07:57 +0000)]
VFS messaging/interfacing work stage 7e/99: More firming up of stage 7.

Fix the linux emulation code for [l]stat(), it was not properly
disposing of the nlookupdata structure.

Fix chroot()'s use of the new api, it was horribly broken.

Cleanup cache_alloc().  Rewrite __getcwd() and vn_fullpath() to use newapi
namecache data.  Cleanup nlookup().  Fix bugs in nlookup() related to
stacked mount points.  Fix a bug related to VFS_ROOT() mount errors.

Linux-bugs-reported-by: Bartek Stalewski, walt
19 years agoVFS messaging/interfacing work stage 7d/99: More firming up of stage 7.
Matthew Dillon [Tue, 5 Oct 2004 03:24:35 +0000 (03:24 +0000)]
VFS messaging/interfacing work stage 7d/99: More firming up of stage 7.

Additional work to deal with old-api/new-api issues.  Cut more stuff
out of the old-api's cache_enter() routine to deal with deadlocks, at
the cost of some performance loss (temporary until the VFS's start using
the new APIs).  Change UFS and NFS to not purge whole directories in
*_rename() and *_rmdir().

Add some minor breakage to the API which will not be fixed until the VFS's
get new rename implementations - renaming a directory in which a process
has chdir'd will create problems for that process.  This doesn't happen
normally anyway so this temporary breakage should not cause any significant
problems.

Bug-reports-by: walt, Sascha Wildner, others
19 years agoVFS messaging/interfacing work stage 7c/99: More firming up of stage 7.
Matthew Dillon [Mon, 4 Oct 2004 09:20:43 +0000 (09:20 +0000)]
VFS messaging/interfacing work stage 7c/99: More firming up of stage 7.

Make a minor change to cache_alloc()'s semantics by having it returned a
ref'd locked ncp.  Make the same change to cache_allocroot().

Fix a bug in the resolver that could deadlock a namecache and vnode lock by
unlocking the namecache node before locking the vnode.  This hack is only
needed until the old API has been removed.

Get rid of the old API compat code that tried to create dummy ncp's for
directory vnodes that did not have one and then merge old and new, it got
too messy.  The idea was to be able to cache things in the middle of the
topology and then hook the topology up later when we knew the path.

Instead, the old API simply does not enter things into the namecache
if the parent directory vnode has not namecache structure associated
with it.  This maintains the unbroken topology that we want to guarentee
at the cost of the old API functions potentially not caching as much
information.  [ since kern_open(), [f]chdir(), and [f]stat() execute the
new API the loss of efficiency in the old API should not be noticeable ].

When cache_enter() re-enters the same vnode, it was zaping the old namecache
structure unnecessarily and causing problems with directories.  It no longer
does this.

NFS CHANGE: get rid of the va_ctime vs n_ctime test.  This test was supposed
to detect server-side directory changes and purge the directory, but half the
time it purged the directory anyway.  The new API requires that the directory
topology remain intact so only purge if the directory is found not to exist.
This change may have to be revisited later.

NOTE: cache_nlookup() is sensitive to the directory topology when traversing
".." directories.  If a directory in the middle of the topology is purged
it can cause the new API lookup to fail with EXDEV.

Bug-reports-by: drhodus, various other people.
19 years agoOnly include opt_vmpage.h if _KERNEL is defined.
Matthew Dillon [Mon, 4 Oct 2004 09:05:26 +0000 (09:05 +0000)]
Only include opt_vmpage.h if _KERNEL is defined.

19 years agoAdd a debugging utility which dumps that kernel's namecache topology.
Matthew Dillon [Sun, 3 Oct 2004 06:12:34 +0000 (06:12 +0000)]
Add a debugging utility which dumps that kernel's namecache topology.

19 years agoMake patch's -b option match the old patch's -b option, at least for now,
Matthew Dillon [Sat, 2 Oct 2004 05:52:06 +0000 (05:52 +0000)]
Make patch's -b option match the old patch's -b option, at least for now,
so ports don't break so badly.

Patch-by: Sascha Wildner <saw@online.de>
19 years agoVFS messaging/interfacing work stage 7b/99: More firming up of stage 7.
Matthew Dillon [Sat, 2 Oct 2004 03:18:28 +0000 (03:18 +0000)]
VFS messaging/interfacing work stage 7b/99: More firming up of stage 7.

(1) Enhance cache_resolve() to go up the directory chain as far as necessary
    to resolve the chain.  Previously I wimped out and returned an error.

(2) Be sure not to use the parent of a mount point to obtain the vnode
    operations vector for a child of a mount point (which resides on a
    different filesystem!).

Generally speaking the namecache directory chain should contain resolved
vnodes due to the fact that the vnode associated with a namecache entry is
held if any children exist, preventing the vnode from being recycled.
However, the NFS client code as originally written wimps out and does
wholesale namecache flushing of directories when it isn't sure about the
state of things (which is quite often, especially when you are rm'ing
files), and this breaks that assumption and causes some intermediate NFS
directory nodes to revert back into an 'unresolved' state.  This will
eventually be fixed, but not right now.

Add a nc_mount pointer to the namecache structure.  For the moment this is
only used to get at the mount point associated with a NCF_MOUNTPT namecache
node (whether resolved or unresolved), making it easier for us to resolve
the vnode.  But eventually it will be used as the basis for obtaining the
v_ops for (new stlye) VOP calls on an unresolved namecache node, saving us
a few indirections so I don't consider it a hack.

Bugs-and-cores-by: drhodus
19 years agoEnforce directory creation ordering for subdirectories to fix a make -j
Matthew Dillon [Sat, 2 Oct 2004 03:08:07 +0000 (03:08 +0000)]
Enforce directory creation ordering for subdirectories to fix a make -j
problem.

19 years agoUnsetting an undefined variable or function is not an error.
Matthew Dillon [Fri, 1 Oct 2004 20:38:40 +0000 (20:38 +0000)]
Unsetting an undefined variable or function is not an error.

Submitted-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
Take-from: FreeBSD

19 years agoRemove ATTY.
Matthew Dillon [Fri, 1 Oct 2004 20:30:25 +0000 (20:30 +0000)]
Remove ATTY.

Taken-from: FreeBSD

19 years agoVFS messaging/interfacing work stage 7a/99: Firm up stage 7 a bit by
Matthew Dillon [Fri, 1 Oct 2004 07:08:23 +0000 (07:08 +0000)]
VFS messaging/interfacing work stage 7a/99:  Firm up stage 7 a bit by
automatically vhold()ing any vnodes associated with a locked namecache
record.  This will prevent the vnode from being recycled (which would
unexpectedly cause the namecache entry to revert to an unresolved state
even while it is locked).

Previously only vnodes associated with namecache records with children
were being held (in order to guarentee that the vnode associated with the
parent directory of a locked leaf namecache record would stick around without
having to lock the parent directory's namecache record).

The point here being that we want to have to only lock and pass a single
namecache record to definitively identify an object in the namespace,
rather then the combination of directory and leaf nodes (and other junk)
that the old API required.  For example, the old RENAME API required 6
arguments (fdvp, fvp, fcnp, tdvp, tvp, tcnp) while the new one will only
require two arguments (fncp, tncp).

19 years agoVFS messaging/interfacing work stage 7/99. BEGIN DESTABILIZATION!
Matthew Dillon [Thu, 30 Sep 2004 19:00:29 +0000 (19:00 +0000)]
VFS messaging/interfacing work stage 7/99.  BEGIN DESTABILIZATION!

Implement the infrastructure required to allow us to begin switching to the
new nlookup() VFS API.

filedesc->fd_ncdir, fd_nrdir, fd_njdir

    File descriptors (associated with processes) now record the
    namecache pointer related to the current directory, root directory,
    and jail directory, in addition to the vnode pointers.  These
    pointers are used as the basis for the new path lookup code
    (nlookup() and friends).

file->f_ncp

    File pointers may now have a referenced+unlocked namecache
    pointer associated with them.  All fp's representing directories
    have this attached.  This allows fchdir() to properly record
    the ncp in fdp->fd_ncdir and friends.

mount->mnt_ncp

    The namecache topology for crossing a mount point works as
    follows: when looking up a path element which is a mount point,
    cache_nlookup() will locate the ncp for the vnode-under the
    mount point.  mount->mnt_ncp represents the root of the mount,
    that is the vnode-over.  nlookup() detects the mount point and
    accesses mount->mnt_ncp to skip past the vnode-under.  When going
    backwards (..), nlookup() detects the case and skips backwards.

    The ncp linkages are: ncp->ncp->ncp[vnode_under]->ncp[vnode_over].
    That is, when going forwards or backwards nlookup must explicitly
    skip over the double-ncp when crossing a mount point.  This allows
    us to keep the namecache topology intact across mount points.

NEW CACHE level API functions:

cache_get() Reference and lock a namecache entry
cache_put() Dereference and unlock a namecache entry
cache_lock() lock an already-referenced namecache entry
cache_unlock() unlock a lockednamecache entry

    NOTE: namecache locks are exclusive and recursive.  These are
    the 'namespace' locks that we will be using to guarentee namespace
    operations such as in a CREATE, RENAME, or REMOVE.

vfs_cache_setroot()  Set the new system-wide root directory
cache_allocroot()    System bootstrap helper function to allocate
      the root namecache node.

cache_resolve() Resolve a NCF_UNRESOLVED namecache node.  The
namecache node should be locked on call.

cache_setvp() (resolver) associate a VP or create a negative
cache entry representation for a namecache
pointer and clear NCF_UNRESOLVED.  The
namecache node should be locked on call.

cache_setunresolved() Revert a resolved namecache entry back to an
unresolved state, disassociating any vnode
but leaving the topology intact.  The
namecache node should be locked on call.

cache_vget() Obtain the locked+refd vnode related to
a namecache entry, resolving the entry if
necessary.  Return ENOENT if the entry
represents a negative cache hit.

cache_vref() Obtained a refd (not locked) vnode related to
a namecache entry, as above.

cache_nlookup() The new namecache lookup routine.  This routine
does a lookup and allocates a new namecache
node (into an unresolved state) if necessary.
Returns a namecache record whether or not
the item can be found and whether or not it
represents a positive or negative hit.

cache_lookup() OLD API CODE DEPRECATED, but must be maintained
until everything has been converted over.
cache_enter() OLD API CODE DEPRECATED, but must be maintained
until everything has been converted over.

NEW default VOPs

vop_noresolve() Implements a namecache resolver for VFSs
which are still using the old VOP_LOOKUP/
VOP_CACHEDLOOKUP API (which is all of them
still).

VOP_LOOKUP OLD API CODE DEPRECATED, but must be maintained
until everything has been converted over.
VOP_CACHEDLOOKUP OLD API CODE DEPRECATED, but must be maintained
until everything has been converted over.

NEW PATHNAME LOOKUP CODE

nlookup_init() Similar to NDINIT, initialize a nlookupdata
structure for nlookup() and nlookup_done().

nlookup() Lookup a path.  Unlike the old namei/lookup
code the new lookup code does not do any
fancy pre-disposition of the cache for
create/delete, it simply looks up the requested
path and returns the appropriate locked
namecache pointer.  The caller can obtain the
vnode and directory vnode, as applicable, from
the one namecache structure that is returned.

Access checks are done on directories leading
up to the result but not done on the returned
namecache node.

nlookup_done() Mandatory routine to cleanup a nlookupdata
structure after it has been initialized and
all operations have been completed on it.

nlookup_simple() (in progress) all-in-one wrapped new lookup.

nlookup_mp() helper call for resolving a mount point's
glue NCP.  hackish, will be cleaned up later.

nreadsymlink() helper call to resolve a symlink.  Note that
the namecache does not yet cache symlink data
but the intention is to eventually do so to
avoid having to do VFS ops to get the data.

naccess() Perform access checks on a namecache node
given a mode and cred.

naccess_va() Perform access cheks on a vattr given a
mode and cred.

Begin switching VFS operations from using namei to using nlookup.
In this batch:

* mount  (install mnt_ncp for cross-mount-point handling in
nlookup, simplify the vfs_mount() API to no longer
pass a nameidata structure)
* [l]stat (use nlookup)
* [f]chdir (use nlookup, use recorded f_ncp)
* [f]chroot (use nlookup, use recorded f_ncp)

19 years agoAttempt to make the boot code operate in a more deterministic fashion.
Matthew Dillon [Thu, 30 Sep 2004 18:32:00 +0000 (18:32 +0000)]
Attempt to make the boot code operate in a more deterministic fashion.
Have it always try EDD (linear block number) mode first, and only fall back
to cylinder/head/sector if EDD fails.  Before it would always try CHS
mode for cylinders < 1024.

19 years agoCorrect the Polish locales per the submission of Bodek <bodek@blurp.org>
Jeroen Ruigrok/asmodai [Thu, 30 Sep 2004 12:46:03 +0000 (12:46 +0000)]
Correct the Polish locales per the submission of Bodek <bodek@blurp.org>
to FreeBSD.

- Month and weekday names should begin with lowercase letter.
- First part of long month names should use genitive.
- Use more proper shortcuts, leaving the first 3 letters is not always
  correct.

19 years agoFix the code for the nxt != 0 case and use the assembler backend.
Joerg Sonnenberger [Thu, 30 Sep 2004 10:21:07 +0000 (10:21 +0000)]
Fix the code for the nxt != 0 case and use the assembler backend.

19 years agoAdd CTASSERT, a facility for simple compile time assertions. This is useful
Joerg Sonnenberger [Thu, 30 Sep 2004 10:18:07 +0000 (10:18 +0000)]
Add CTASSERT, a facility for simple compile time assertions. This is useful
to ensure that certain structures keep there size.

Obtained-from: FreeBSD

19 years agoFix the setup for initial processing. Always use argv[0] and correctly
Joerg Sonnenberger [Thu, 30 Sep 2004 10:08:46 +0000 (10:08 +0000)]
Fix the setup for initial processing. Always use argv[0] and correctly
decrement argc for the argc!=1 case first.

Reported-by: Andreas Hauser <andy@splashground.de>
19 years agoSwitch patch(1) from GNU to BSD version.
Joerg Sonnenberger [Tue, 28 Sep 2004 20:56:38 +0000 (20:56 +0000)]
Switch patch(1) from GNU to BSD version.

19 years agoDon't use patch -b, the flag has different meanings for BSD and GNU patch.
Joerg Sonnenberger [Tue, 28 Sep 2004 19:34:18 +0000 (19:34 +0000)]
Don't use patch -b, the flag has different meanings for BSD and GNU patch.
Instead use --suffix=, which is the same for both.

In-discussion-with: Sascha Wildner

19 years agoMake the BSD patch WARNS=6 clean
Joerg Sonnenberger [Tue, 28 Sep 2004 19:09:50 +0000 (19:09 +0000)]
Make the BSD patch WARNS=6 clean
- constify string operations if possible
- rename some variables to avoid namespace clashes
- don't cast the signal functions to sig_t), it's unnecessary

Submitted-by: Sascha Wildner <saw@online.de>
Instead of i_ptr[++iline][0] = '\0', I've used a special empty char array
instead. I'm not sure if i_ptr[++iline] points to a valid memory location
under all circumstances and this is a safer bet.

19 years agoWe now have pidfile in libutil, update kerberos5 accordingly.
Joerg Sonnenberger [Tue, 28 Sep 2004 18:17:37 +0000 (18:17 +0000)]
We now have pidfile in libutil, update kerberos5 accordingly.

Submitted-by: Pete Carah <pete@altadena.net>
19 years agoCorrect BASE_LOCALEDIR to ..
Jeroen Ruigrok/asmodai [Tue, 28 Sep 2004 16:30:13 +0000 (16:30 +0000)]
Correct BASE_LOCALEDIR to ..

Reminded by: YONETANI Tomoka

19 years agoEVENTHANDLER_REGISTER uses the name of the event as string and therefore
Joerg Sonnenberger [Tue, 28 Sep 2004 16:22:41 +0000 (16:22 +0000)]
EVENTHANDLER_REGISTER uses the name of the event as string and therefore
the compiler couldn't detect the wrong event names. I used
ifnet_attach_event and ifnet_detach_event instead of ifnet_arrival_event
and ifnet_departure_event. This seemed to work for PF as module because
all existing interfaces are added by hand first.

Problem-reported-by: Aggelos Economopoulos <aoiko@cc.ece.ntua.gr>
19 years agoIncorporate Hye-Shik's work for numeric UTF-8 definitions.
Jeroen Ruigrok/asmodai [Tue, 28 Sep 2004 09:18:55 +0000 (09:18 +0000)]
Incorporate Hye-Shik's work for numeric UTF-8 definitions.

Provided by: Hye-Shik Chang <perky@FreeBSD.org>

19 years agoAdd Hye-Shik's UTF-8 message definitions.
Jeroen Ruigrok/asmodai [Tue, 28 Sep 2004 08:53:32 +0000 (08:53 +0000)]
Add Hye-Shik's UTF-8 message definitions.

Provided by: Hye-Shik Chang <perky@FreeBSD.org>

19 years agoVFS messaging/interfacing work stage 6/99. Populate and maintain the
Matthew Dillon [Tue, 28 Sep 2004 00:25:34 +0000 (00:25 +0000)]
VFS messaging/interfacing work stage 6/99.  Populate and maintain the
namecache pointers previously attached to struct filedesc, giving the new
lookup code a base from which to work.

Implement the new lookup API (it is not yet being used by anything) and
augment the namecache API to handle the new functions, in particular
adding cache_setvp() to resolve an unresolved namecache entry into a
positive or negative hit and set various flags.  Note that we do not yet
cache symlink data but we could very easily.

The new API is greatly simplified.  Basically nlookups need only returned
a locked namecache pointer (guarenteeing namespace atomicy).  Related
vnodes are not locked.  Both the leaf and governing directory vnodes can
be extracted from the returned namecache pointer.  namecache pointers may
also represent negative hits, which means that their namespace locking
feature serves to reserve a filename that has not yet been created (e.g.
open+create, rename).

The kernel is still using the old API as of this commit.  This commit is
primarily introducing the management infrastructure required to actually
start writing code to use the new API.

VOP_RESOLVE() has been added, along with a default function which falls back
to VOP_LOOKUP()/VOP_CACHEDLOOKUP().  This VOP function is not yet being used
as of this commit.  This VOP will be responsible for taking an unresolved
but locked namecache structure (hence the namespace is locked), and actually
does the directory lookup.  But unlike the far more complex
VOP_LOOKUP()/VOP_CACHEDLOOKUP() API the VOP_RESOLVE() API only needs to
attach a vnode (or NULL if the entry does not exist) to the passed-in
namecache structure.  It is likely that timeouts, e.g. for NFS, will also
be attached via this API.

This commit does not implement any of the cache-coherency infrastructure
but keeps this future requirement in mind in its design.

19 years agoAdd Hye-Shik's UTF monetary defitions.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 20:02:21 +0000 (20:02 +0000)]
Add Hye-Shik's UTF monetary defitions.

This requires some reworking which will follow.

Provided by: Hye-Shik Chang <perky@FreeBSD.org>

19 years agoAdd zh_CN.GBK.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 19:39:33 +0000 (19:39 +0000)]
Add zh_CN.GBK.

19 years agoFix consistent mistake: CP1252 -> CP1251.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 19:37:31 +0000 (19:37 +0000)]
Fix consistent mistake: CP1252 -> CP1251.

19 years agoAdd collation definition for UTF-8.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 19:30:50 +0000 (19:30 +0000)]
Add collation definition for UTF-8.

Synchronised partially with FreeBSD.
Decided upon following Hye-Shik's original plan with a separate la_LN.UTF-8
even though it duplicates US-ASCII's contents, it is, however, much more
clear this way.

Provided by: Hye-Shik Chang <perky@FreeBSD.org>

19 years agoAdd la_LN.UTF-8.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 19:27:03 +0000 (19:27 +0000)]
Add la_LN.UTF-8.

19 years agoBe a coward and add ja_JP.EUC back in for now.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 18:43:58 +0000 (18:43 +0000)]
Be a coward and add ja_JP.EUC back in for now.

19 years agoAdd UTF-8 time definitions.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 14:44:33 +0000 (14:44 +0000)]
Add UTF-8 time definitions.

Synchronise Makefile with some FreeBSD changes made by Tim J. Robbins.

Provided by: Hye-Shik Chang <perky@FreeBSD.org>

19 years agoAdd en_GB.UTF-8.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 14:26:14 +0000 (14:26 +0000)]
Add en_GB.UTF-8.

19 years agoPrepare for the locale additions.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 12:54:03 +0000 (12:54 +0000)]
Prepare for the locale additions.

19 years agoAdd the Unicode 3.2 ctypes table.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 06:03:48 +0000 (06:03 +0000)]
Add the Unicode 3.2 ctypes table.

Provided by: Hye-Shik Chang <perky@FreeBSD.org>

19 years agoChange to use USB_PRODUCT_LINKSYS2_USB200M.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 05:57:58 +0000 (05:57 +0000)]
Change to use USB_PRODUCT_LINKSYS2_USB200M.

19 years agoChange SYNOPSYS to SYNOPSYS2 and regenerate.
Jeroen Ruigrok/asmodai [Mon, 27 Sep 2004 05:30:16 +0000 (05:30 +0000)]
Change SYNOPSYS to SYNOPSYS2 and regenerate.

19 years agoMinor cleanups to TIMER_USE_1 (no real fixes or anything).
Matthew Dillon [Mon, 27 Sep 2004 04:04:09 +0000 (04:04 +0000)]
Minor cleanups to TIMER_USE_1 (no real fixes or anything).

19 years agoVFS messaging/interfacing work stage 5b/99. More cleanups, remove the
Matthew Dillon [Sun, 26 Sep 2004 20:14:21 +0000 (20:14 +0000)]
VFS messaging/interfacing work stage 5b/99.  More cleanups, remove the
(unused) ni_ncp and ni_dncp from struct nameidata.  A new structure will
be used for the new API.

19 years agoClarify an element in the BUGS section related to the sticky bit.
Matthew Dillon [Sun, 26 Sep 2004 19:16:41 +0000 (19:16 +0000)]
Clarify an element in the BUGS section related to the sticky bit.

Brought-to-attention-by: =?ISO-8859-1?Q?Jonas_Sundstr=F6m?= <jonas@kirilla.com>
19 years agoRegenerate.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 18:18:54 +0000 (18:18 +0000)]
Regenerate.

19 years agoAdd some more vendors and some HP devices.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 18:18:13 +0000 (18:18 +0000)]
Add some more vendors and some HP devices.

19 years agoBe consistent in the white space usage.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 17:58:02 +0000 (17:58 +0000)]
Be consistent in the white space usage.

19 years agoBump to WARNS 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 17:40:17 +0000 (17:40 +0000)]
Bump to WARNS 6.

Reminded by: Sascha Wildner

19 years agoRegenerate.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 17:03:27 +0000 (17:03 +0000)]
Regenerate.

19 years agoAdd Wacom Graphire 3.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 17:02:37 +0000 (17:02 +0000)]
Add Wacom Graphire 3.

19 years agoForced commit:
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:32:47 +0000 (16:32 +0000)]
Forced commit:

Submitted by: Sascha Wildner

19 years agoBump WARNS to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:32:12 +0000 (16:32 +0000)]
Bump WARNS to 6.

Put external declarations where they are supposed to go.

19 years agoBump to WARNS 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:29:49 +0000 (16:29 +0000)]
Bump to WARNS 6.
Use tmpsin instead of sin to avoid conflicts.

Submitted by: Sascha Wildner

19 years agoFix author name with .An macro. Add missing period.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:25:42 +0000 (16:25 +0000)]
Fix author name with .An macro.  Add missing period.

Submitted by: Sascha Wildner

19 years agoRegenerate.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:14:17 +0000 (16:14 +0000)]
Regenerate.

19 years agotsptype is an array of character constants, declare it as such.
Joerg Sonnenberger [Sun, 26 Sep 2004 16:09:48 +0000 (16:09 +0000)]
tsptype is an array of character constants, declare it as such.

Submitted-by: Sascha Wildner
19 years agoBump WARNS to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:09:13 +0000 (16:09 +0000)]
Bump WARNS to 6.
Move WARNS under PROG in test.
Remove commented MLINKS in varsym, it is automatically handled through PROG.

19 years agoBump WARNS to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 16:03:39 +0000 (16:03 +0000)]
Bump WARNS to 6.
Get rid of SRCS, rm.c is implied through PROG.

19 years agoAdd WARNS, set to 0.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:58:51 +0000 (15:58 +0000)]
Add WARNS, set to 0.

19 years agoBump WARNS to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:58:32 +0000 (15:58 +0000)]
Bump WARNS to 6.

19 years agoAdd, if not present, WARNS and set to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:56:47 +0000 (15:56 +0000)]
Add, if not present, WARNS and set to 6.

19 years ago- make process() static and take const arguments
Joerg Sonnenberger [Sun, 26 Sep 2004 15:56:13 +0000 (15:56 +0000)]
- make process() static and take const arguments
- use the exit code to indicate a problem while processing any of the input
- mark as WARNS=6 clean

Submitted-by: Liam J. Foy
- reduce the special casing of argc == 1 and switch to a while loop

19 years agoBump hostname to WARNS 6 by initialising silen.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:55:37 +0000 (15:55 +0000)]
Bump hostname to WARNS 6 by initialising silen.

19 years agoMake ls compile under WARNS 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:53:25 +0000 (15:53 +0000)]
Make ls compile under WARNS 6.

Submitted by: Sascha Wildner

19 years agoSet NCURSES_CONST to const.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:43:40 +0000 (15:43 +0000)]
Set NCURSES_CONST to const.

Submitted by: Sascha Wildner

19 years agoAdd WARNS and set to 5.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:42:09 +0000 (15:42 +0000)]
Add WARNS and set to 5.

19 years agoSet WARNS to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:41:49 +0000 (15:41 +0000)]
Set WARNS to 6.

19 years agoAdd, if not already present, WARNS and set to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:38:44 +0000 (15:38 +0000)]
Add, if not already present, WARNS and set to 6.

19 years agoAdd WARNS and set to 3.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:34:16 +0000 (15:34 +0000)]
Add WARNS and set to 3.

19 years agoAdd WARNS and set to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:32:56 +0000 (15:32 +0000)]
Add WARNS and set to 6.

19 years agoAdd WARNS, set to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:31:15 +0000 (15:31 +0000)]
Add WARNS, set to 6.

19 years agoBump WARNS to 6.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:30:45 +0000 (15:30 +0000)]
Bump WARNS to 6.

19 years agoMake WARNS ?= instead of = per all the other Makefiles.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:29:11 +0000 (15:29 +0000)]
Make WARNS ?= instead of = per all the other Makefiles.

19 years agoAdd WARNS, set to 3.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:28:02 +0000 (15:28 +0000)]
Add WARNS, set to 3.

19 years agoFix spammage introduced by dillon's commit in r1.5.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:18:29 +0000 (15:18 +0000)]
Fix spammage introduced by dillon's commit in r1.5.

19 years agoSynchronise with FreeBSD-CURRENT as of 2004-09-26.
Jeroen Ruigrok/asmodai [Sun, 26 Sep 2004 15:10:26 +0000 (15:10 +0000)]
Synchronise with FreeBSD-CURRENT as of 2004-09-26.

In a number of cases I used normal English alphabetical sorting instead of
where FreeBSD placed their entries.

19 years agoVFS messaging/interfacing work stage 5/99. Start work on the new
Matthew Dillon [Sun, 26 Sep 2004 06:00:10 +0000 (06:00 +0000)]
VFS messaging/interfacing work stage 5/99.  Start work on the new
namecache/lookup API.  The core of this API will be the concept of a
'locked namespace' rather then a 'locked vnode' for namespace centric
operations like CREATE, DELETE, and RENAME.   The namecache will no longer
be optional for a VFS but instead will become centric to all kernel
namespace operations.

This commit is mostly non-functional.  It removes an extension I had made
to VOP_LOOKUP/VOP_CACHEDLOOKUP and starts adding support functions for the
new cache_nclookup() API.  The work being backed out was originally intended
to shepard the new work, but the new interface is now so different (and far
more simplified) then the original that it's actually better to create a
new VOP for it instead of augmenting existing VOPs.

19 years agoThe wrong vendor-id was used when adding Adaptec ServeRAID Adapter support.
Matthew Dillon [Sun, 26 Sep 2004 05:30:24 +0000 (05:30 +0000)]
The wrong vendor-id was used when adding Adaptec ServeRAID Adapter support.

Submitted-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
19 years agoGet rid of some conditionalized code which the pmap invalidation API took
Matthew Dillon [Sun, 26 Sep 2004 01:53:51 +0000 (01:53 +0000)]
Get rid of some conditionalized code which the pmap invalidation API took
over from long ago and is no longer used.

19 years agoDocument the additional vendors applicable to the 'sk' driver.
Matthew Dillon [Sun, 26 Sep 2004 01:52:48 +0000 (01:52 +0000)]
Document the additional vendors applicable to the 'sk' driver.

19 years agoVFS messaging/interfacing work stage 5/99. Start work on the new
Matthew Dillon [Sun, 26 Sep 2004 01:25:52 +0000 (01:25 +0000)]
VFS messaging/interfacing work stage 5/99.  Start work on the new
namecache/lookup API.  The core of this API will be the concept of a
'locked namespace' rather then a 'locked vnode' for namespace centric
operations like CREATE, DELETE, and RENAME.   The namecache will no longer
be optional for a VFS but instead will become centric to all kernel
namespace operations.

This commit is mostly non-functional.  It removes an extension I had made
to VOP_LOOKUP/VOP_CACHEDLOOKUP and starts adding support functions for the
new cache_nclookup() API.  The work being backed out was originally intended
to shepard the new work, but the new interface is now so different (and far
more simplified) then the original that it's actually better to create a
new VOP for it instead of augmenting existing VOPs.

19 years agoOnly assume a numeric group id if the ENTIRE string is numeric.
Matthew Dillon [Sat, 25 Sep 2004 20:38:21 +0000 (20:38 +0000)]
Only assume a numeric group id if the ENTIRE string is numeric.

Reported-by: "Martin P. Hellwig" <mhellwig@xs4all.nl>
19 years agoAllow ip_output to be called with rt=NULL by making the FAST_IPSEC code
Joerg Sonnenberger [Sat, 25 Sep 2004 17:01:16 +0000 (17:01 +0000)]
Allow ip_output to be called with rt=NULL by making the FAST_IPSEC code
included by default.

Obtained-from: FreeBSD netinet/ip_output.c, 1.201

This fixes a panic from PF.

Reported-by: Aggelos Economopoulos via IRC
19 years agoFrom FreeBSD:
Joerg Sonnenberger [Sat, 25 Sep 2004 15:20:33 +0000 (15:20 +0000)]
From FreeBSD:
Introduce an M_ASSERTPKTHDR() macro which performs the very common task
of asserting that an mbuf has a packet header.  Use it instead of hand-
rolled versions wherever applicable.

Submitted by: Hiten Pandya <hiten@unixdaemons.com>

19 years agoFix a bug where DragonFly's nat was closing TCP connections every 10 minutes.
Matthew Dillon [Sat, 25 Sep 2004 03:42:58 +0000 (03:42 +0000)]
Fix a bug where DragonFly's nat was closing TCP connections every 10 minutes.
This was due to the TCPS state not being properly initialized (our TCPS_CLOSED
state is not 0).

Submitted-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
19 years agoAdd GPL-free patch. This is the extended version of Larry Wall's original
Joerg Sonnenberger [Fri, 24 Sep 2004 18:44:28 +0000 (18:44 +0000)]
Add GPL-free patch. This is the extended version of Larry Wall's original
patch.

Obtained-from: OpenBSD
Submitted-by: Sascha Wildner <saw@online.de>
19 years agoFix typo
Joerg Sonnenberger [Fri, 24 Sep 2004 12:39:25 +0000 (12:39 +0000)]
Fix typo

19 years agoThe inode size must be unsigned-extended to a quad, not sign-extended, so
Matthew Dillon [Fri, 24 Sep 2004 09:18:03 +0000 (09:18 +0000)]
The inode size must be unsigned-extended to a quad, not sign-extended, so
the size of files in the 2-4GB range on DVDs are properly reported.

Taken-from: Jun Kuriyama <kuriyama@imgsrc.co.jp> post on freebsd-current@

19 years agoAdd support for the LinkSys EG1032 GigE PCI card, and add support for
Matthew Dillon [Thu, 23 Sep 2004 23:18:01 +0000 (23:18 +0000)]
Add support for the LinkSys EG1032 GigE PCI card, and add support for
the D-Link DGE-530T GigE card.  These cards are commonly available
off-the-shelf.  These are just SK-9521/Marvell chipsets with new vendor
PCI IDs.

Taken-from: FreeBSD-5

19 years agoWhile I'm here, convert other strcpy -> strlcpy
Scott Ullrich [Thu, 23 Sep 2004 21:39:08 +0000 (21:39 +0000)]
While I'm here, convert other strcpy -> strlcpy

Reviewed-by: Matthew Dillon

19 years agoConvert strcpy -> strlcpy and convert strcpy and strcat -> snprintf
Scott Ullrich [Thu, 23 Sep 2004 20:20:59 +0000 (20:20 +0000)]
Convert strcpy -> strlcpy and convert strcpy and strcat -> snprintf

Submitted-by: Christer Öberg ( christer.oberg _AT_ ornsat.com )

19 years agoAdd the -L and -l options to install. -L allows an alternative /etc directory
Matthew Dillon [Thu, 23 Sep 2004 19:13:51 +0000 (19:13 +0000)]
Add the -L and -l options to install.  -L allows an alternative /etc directory
path to be specified and causes install to search LOPT/group and
LOPT/master.passwd for group and user names.  If the -l option is also
specified, install will fallback to system files if the group or user name
cannot be found in the alternative /etc directory.

This is intended to allow us to ensure that the correct uid's and gid's are
used in a release build and to allow new uid's and gid's to be used during
an installworld before they have necessarily been added to the system
password and group files.

19 years agoBYTE_ORDER ==> _BYTE_ORDER
Joerg Sonnenberger [Thu, 23 Sep 2004 16:46:02 +0000 (16:46 +0000)]
BYTE_ORDER ==> _BYTE_ORDER

19 years agoChange the conditionals to use the protected _BYTE_ORDER defines.
Joerg Sonnenberger [Thu, 23 Sep 2004 16:44:32 +0000 (16:44 +0000)]
Change the conditionals to use the protected _BYTE_ORDER defines.

19 years agoTypo.
Joerg Sonnenberger [Thu, 23 Sep 2004 16:13:21 +0000 (16:13 +0000)]
Typo.

19 years agoUse the protected names for BYTE_ORDER and co, the others are not defined
Joerg Sonnenberger [Thu, 23 Sep 2004 16:11:47 +0000 (16:11 +0000)]
Use the protected names for BYTE_ORDER and co, the others are not defined
by default from machine/endian.h anymore. Also change the implicit
!LITTLE_ENDIAN == BIG_ENDIAN where it is used.

19 years agoUpdate FireWire device nodes.
Hidetoshi Shimokawa [Thu, 23 Sep 2004 14:30:16 +0000 (14:30 +0000)]
Update FireWire device nodes.

19 years agoAdd dcons(4).
Hidetoshi Shimokawa [Thu, 23 Sep 2004 14:22:48 +0000 (14:22 +0000)]
Add dcons(4).