dragonfly.git
14 years agoVFS messaging/interfacing work stage 4/99. This stage goes a long ways
Matthew Dillon [Sat, 28 Aug 2004 19:02:35 +0000 (19:02 +0000)]
VFS messaging/interfacing work stage 4/99.  This stage goes a long ways
towards allowing us to move the vnode locking into a kernel layer.  It
gets rid of a lot of cruft from FreeBSD-4.  FreeBSD-5 has done some of this
stuff too (such as changing the default locking to stdlock from nolock),
but DragonFly is going further.

* Consolidate vnode locks into the vnode structure, add an embedded v_lock,
  and getting rid of both v_vnlock and v_data based head-of-structure locks.

* Change the default vops to use a standard vnode lock rather then a fake
  non-lock.

* Get rid of vop_nolock() and friends, we no longer support non-locking
  vnodes.

* Get rid of vop_sharedlock(), we no longer support non standard shared-only
  locks (only NFS was using it and the mount-crossing lookup code should
  now prevent races to root from dead NFS volumes).

* Integrate lock initialization into getnewvnode().  We do not yet
  incorporate automatically locking into getnewvnode().  getnewvnode()
  now has two additional arguments, lktimeout and lkflags, for lock
  structure initialization.

* Change the sync vnode lock from nolock to stdlock.  This may require more
  tuning down the line.  Fix various sync_inactive() to properly unlock
  the lock as per the VOP API.

* Properly flag the 'rename' vop operation regarding required tdvp and tvp
  unlocks (the flags are only used by nullfs).

* Get rid of all inode-embedded vnode locks

* Remove manual lockinit and use new getnewvnode() args instead.
  Lock the vnode prior to doing anything that might block in
  order to avoid synclist access before the vnode has been properly
  initialize.

* Generally change inode hash insertion to also check
  for a hash collision and return failure if it occurs,
  rather then doing (often non-atomic) relookups and
  other checks.  These sorts of collisions can occur
  if a vnode is being destroyed at the same time a new
  vnode is being created from an inode.  A new vnode is
  not generally accessible, except by the sync code (from
  the mountlist) until it's underlying inode has been hashed
  so dealing with a hash collision should be as simple as
  throwing away the vnode with a vput().

* Do not initialize a new vnode's v_data until after
  the associated inode has been successfully added to
  the hash, and make the xxx_inactive() and xxx_reclaim()
  code friendly towards vnodes with a NULL v_data.

* NFS now uses standard locks rather then shared-only locks.

* PROCFS now uses standard locks rather then non-locks, and PROCFS's
  lookup code now understands VOP lookup semantics.  PROCFS now uses
  a real hash table for its node search rather then a single singly-linked
  list (which should better scale to systems with thousands of processes).

* NULLFS should now properly handle lookup() and rename() locks.  NULLFS's
  node handling code has been rewritten.  NULLFS's bypass code now understands
  vnode unlocks (rename case).

* UFS no longer needs the ffs_inode_hash_lock hacks.  It now uses the new
  collision-on-hash-add methodology.   This will speed up UFS when operating
  on lots of small files (reported by David Rhodus).

14 years agoSince ip_input() truncates the packet to ip->ip_len prior to entering the
Matthew Dillon [Sat, 28 Aug 2004 18:33:03 +0000 (18:33 +0000)]
Since ip_input() truncates the packet to ip->ip_len prior to entering the
protocol stack, ip_demux must incorporate an ip_len check in its tcp/udp
prechecks to avoid an assertion in the tcp/udp stacks if the packet is
malformed.

This is a temporary hack until Jeff and I can come up with a better way
to do per-protocol mbuf checks.  Basically the issue involved is that we
want to pull up the entire tcp/udp/etc... header in order to be able to
trivially demux an IP packet (choose which protocol thread to route it to).
Since we have to m_pullup the packet in the demux we would rather do all
header length checks in the demux as well so as not have to repeat them in
the protocol code.

Reported-by: Sven Willenberger <sven@dmv.com>
In-discussion-with: Jeffrey Hsu

14 years agoGet rid of COPY, its functionality has been superseded by install -C.
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 17:57:51 +0000 (17:57 +0000)]
Get rid of COPY, its functionality has been superseded by install -C.

Inspired by: a FreeBSD commit message

14 years agoSync VESA support with FreeBSD-CURRENT, adding support for cards that
Matthew Dillon [Sat, 28 Aug 2004 17:21:22 +0000 (17:21 +0000)]
Sync VESA support with FreeBSD-CURRENT, adding support for cards that
identify themselves as 'VIA CLE266' and fixing an ATI Radeon 7000
issue with the NON-VGA bit being set for all of its SVGA mods,
plus other minor changes.

Submitted-by: Sascha Wildner <saw@online.de>
Radeon-testing-by: Sarunas Vancevicius
14 years agoCorrect typo: mv -> nv.
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 16:32:42 +0000 (16:32 +0000)]
Correct typo: mv -> nv.

14 years agoUpdate zoneinfo database with the latest information.
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 16:30:41 +0000 (16:30 +0000)]
Update zoneinfo database with the latest information.

Theory: 7.13 -> 7.15
asia: 7.71 -> 7.75
australasia: 7.68 -> 7.69
backward: 7.23 -> 7.26
europe: 7.83 -> 7.88
factory: use zic, not tzsetup
leapseconds: 7.15 -> 7.17
northamerica: 7.62 -> 7.69
southamerica: 7.46 -> 7.52
zone.tab: 1.28 -> 1.30

14 years agoAdd updated Lithuanian locale (lt_LT-ISO8859-13).
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 15:17:37 +0000 (15:17 +0000)]
Add updated Lithuanian locale (lt_LT-ISO8859-13).

Bug: 8
Submitted by: Edmondas Girkantas <eg@zemaitija.net>

Taken from: FreeBSD
Originally submitted by: Kestutis Paulikas <kestas@elen.ktu.lt>
PR 44268

14 years agoAdd support for Nforce onboard ethernet. This is the content of
Joerg Sonnenberger [Sat, 28 Aug 2004 15:08:02 +0000 (15:08 +0000)]
Add support for Nforce onboard ethernet. This is the content of
net/nvnet.

Wrapper-written-by: Quinton Dolon
In-correspondenc-with: Nick Triantos (NVidia)

14 years agoChange rcvar from "mixer" to `set_rcvar`.
Jeroen Ruigrok/asmodai [Sat, 28 Aug 2004 14:15:18 +0000 (14:15 +0000)]
Change rcvar from "mixer" to `set_rcvar`.
This allows mixer to be properly started when doing:

# /etc/rc.d/mixer start

14 years agoTCPS_CLOSED is no longer 0 in DragonFly. Because ipfilter was assuming
Matthew Dillon [Sat, 28 Aug 2004 07:27:02 +0000 (07:27 +0000)]
TCPS_CLOSED is no longer 0 in DragonFly.  Because ipfilter was assuming
that TCPS_CLOSED was 0, the tcp keep state table was never properly
initialized and ipfilter considered most tcp packets to be bad.  Change
ip_state.c to properly initialize the starting state for new tcp keep
state entries.

Problems reported by: Toma<9E> Bor<9A>tna <tomaz.borstnar@over.net>,
Bernhard Valenti <bernhard.valenti@gmx.net>,
Peter Kadau <peter.kadau@tuebingen.mpg.de>

14 years agoRework the wording in a different way. For send() and sendto() change
Jeroen Ruigrok/asmodai [Fri, 27 Aug 2004 21:55:16 +0000 (21:55 +0000)]
Rework the wording in a different way.  For send() and sendto() change
msg in the argument list to msgbuf and note in the text that this is,
surprise, a pointer to a buffer holding the message.
This should be more self-describing and less confusing with the msg mention
in sendmsg, which is a totally different thing.

Discussed with: eirikn

14 years agoRemove a redundant call to gettimeofday().
Jeroen Ruigrok/asmodai [Fri, 27 Aug 2004 21:27:41 +0000 (21:27 +0000)]
Remove a redundant call to gettimeofday().

Noticed by: eirikn

14 years agoClarify the 's' and 'msg' arguments and note how send() and sendto()'s msg
Jeroen Ruigrok/asmodai [Fri, 27 Aug 2004 21:23:32 +0000 (21:23 +0000)]
Clarify the 's' and 'msg' arguments and note how send() and sendto()'s msg
argument is different from sendmsg()'s.  (Even though it later in
uipc_syscalls.c gets reintegrated.)

14 years agoAdd a commented out STRIP variable to show people how to make sure installed
Jeroen Ruigrok/asmodai [Fri, 27 Aug 2004 18:30:37 +0000 (18:30 +0000)]
Add a commented out STRIP variable to show people how to make sure installed
files keep their debugging symbols.

14 years agoAdd a temporary hack to avoid the local files to be picked up.
Jeroen Ruigrok/asmodai [Fri, 27 Aug 2004 18:07:49 +0000 (18:07 +0000)]
Add a temporary hack to avoid the local files to be picked up.

14 years agoFix linker set creation for GCC 3.4 with -funit-at-a-time.
Joerg Sonnenberger [Fri, 27 Aug 2004 12:08:41 +0000 (12:08 +0000)]
Fix linker set creation for GCC 3.4 with -funit-at-a-time.

GCC 3.4 optimizes away global statics, which are referenced. Define a new
macro __used, which expands to attribute((used)) and tells the compiler that
a certain static indead referenced. For older versions and other compilers
it defaults to the meaning of __unused to avoid warnings.

14 years agoBring in FreeBSD/1.218.
Matthew Dillon [Thu, 26 Aug 2004 21:21:46 +0000 (21:21 +0000)]
Bring in FreeBSD/1.218.
>date: 2004/05/11 19:14:44;  author: maxim;  state: Exp;  lines: +1 -1
>o Calculate a number of bytes to copy (cnt) correctly:
>
>  +----+-+-+-+-+----+----+- - - - - - - - - - - -  -+----+
>  |    | |C| | |    |    |                          |    |
>  | IP |N|O|L|P|    | IP |                          | IP |
>  | #1 |O|D|E|T|    | #2 |                          | #n |
>  |    |P|E|N|R|    |    |                          |    |
>  +----+-+-+-+-+----+----+- - - - - - - - - - - -  -+----+
>               ^    ^<---- cnt - (IPOPT_MINOFF - 1) ---->|
>               |    |
>src            |    +-- cp[IPOPT_OFF + 1] + sizeof(struct in_addr)
>               |
>dst            +-- cp[IPOPT_OFF + 1]
>
>PR:             kern/66386
>Submitted by:   Andrei Iltchenko

14 years agoBring in FreeBSD/1.214 - UC Regent's advertising clause removal per
Matthew Dillon [Thu, 26 Aug 2004 20:59:07 +0000 (20:59 +0000)]
Bring in FreeBSD/1.214 - UC Regent's advertising clause removal per
per letter dated July 22, 1999 (see /usr/src/COPYRIGHT).

14 years agoMerge FreeBSD/1.212 and FreeBSD/1.213. These only appear to have an
Matthew Dillon [Thu, 26 Aug 2004 20:57:02 +0000 (20:57 +0000)]
Merge FreeBSD/1.212 and FreeBSD/1.213.  These only appear to have an
effect when multi-cast routing is enabled.

Suggested-by: David Rhodus <sdrhodus@gmail.com>
1.213:
>date: 2004/04/07 10:01:38;  author: ru;  state: Exp;  lines: +8 -8
>Fixed a bug in previous revision: compute the payload checksum before
>we convert ip_len into a network byte order; in_delayed_cksum() still
>expects it in host byte order.
>
>The symtom was the ``in_cksum_skip: out of data by %d'' complaints
>from the kernel.
>
>To add to the previous commit log.  These fixes make tcpdump(1) happy
>by not complaining about UDP/TCP checksum being bad for looped back
>IP multicast when multicast router is deactivated.
>Reported by:    Vsevolod Lobko

1.212:
>date: 2004/03/25 08:46:27;  author: ru;  state: Exp;  lines: +3 -13
>Untangle IP multicast routing interaction with delayed payload checksums.
>
>Compute the payload checksum for a locally originated IP multicast where
>God intended, in ip_mloopback(), rather than doing it in ip_output() and
>only when multicast router is active.  This is more correct as we do not
>fool ip_input() that the packet has the correct payload checksum when in
>fact it does not (when multicast router is inactive).  This is also more
>efficient if we don't join the multicast group we send to, thus allowing
>the hardware to checksum the payload.

14 years agoTurn off the getty on ttyd0 by default to avoid certain machines from
Matthew Dillon [Thu, 26 Aug 2004 20:32:00 +0000 (20:32 +0000)]
Turn off the getty on ttyd0 by default to avoid certain machines from
crashing.

14 years agoCorrect an inaccurate statement. According to my testing, cpdup never
Chris Pressey [Wed, 25 Aug 2004 22:53:07 +0000 (22:53 +0000)]
Correct an inaccurate statement.  According to my testing, cpdup never
crosses mountpoints in either the source or the destination.

14 years agoVFS messaging/interfacing work stage 3/99: Bring in the journaling
Matthew Dillon [Wed, 25 Aug 2004 19:14:40 +0000 (19:14 +0000)]
VFS messaging/interfacing work stage 3/99:  Bring in the journaling
(and other) facilities infrastructure.  This is not yet operational and
will probably change, but is being brought in now to serve as a foil to
ensure that the namespace and range locking work is properly integrated with
the intended journaling and cache coherency infrastructure.

This commit also reserves some space in the vop_ops args structures to
reduce pain later on.

Since no flags are yet set in vop_ops this commit should not result in
any operational changes.

14 years agoGet rid of dfly/fbsd4/fbsd5 checks for the ntohl() return type. We are
Matthew Dillon [Wed, 25 Aug 2004 19:02:42 +0000 (19:02 +0000)]
Get rid of dfly/fbsd4/fbsd5 checks for the ntohl() return type.  We are
now compatible with fbsd5.  This fixes a compile warning.

14 years agoCorrect a mistake in the last commit that caused usage() to seg-fault,
Matthew Dillon [Wed, 25 Aug 2004 16:07:18 +0000 (16:07 +0000)]
Correct a mistake in the last commit that caused usage() to seg-fault,
and document -j in the manual page options summary.

Submitted-by: Maxim Konovalov
14 years agoUpdate installer packages to 1.1.1, fixing a couple of minor bugs:
Chris Pressey [Wed, 25 Aug 2004 03:12:14 +0000 (03:12 +0000)]
Update installer packages to 1.1.1, fixing a couple of minor bugs:
ability to install bootblocks on an unformatted disk drive, and
ability to install /usr as a plain directory in the / partition.

14 years agoOutput an error message if the open fails.
Matthew Dillon [Wed, 25 Aug 2004 01:56:49 +0000 (01:56 +0000)]
Output an error message if the open fails.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoCleanup various type-o's in comments.
Matthew Dillon [Wed, 25 Aug 2004 01:53:39 +0000 (01:53 +0000)]
Cleanup various type-o's in comments.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoMinor cleanups. Document -j in usage.
Matthew Dillon [Wed, 25 Aug 2004 01:43:43 +0000 (01:43 +0000)]
Minor cleanups.  Document -j in usage.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoCheck for a mkstemps() failure, generate a proper warning if the fopen()
Matthew Dillon [Wed, 25 Aug 2004 01:42:26 +0000 (01:42 +0000)]
Check for a mkstemps() failure, generate a proper warning if the fopen()
fails.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoDocument the unorthordox use of getopt().
Matthew Dillon [Wed, 25 Aug 2004 01:40:23 +0000 (01:40 +0000)]
Document the unorthordox use of getopt().

14 years agoMinor cleanups. Also, change various exit(10) codes to exit(1).
Matthew Dillon [Wed, 25 Aug 2004 01:38:50 +0000 (01:38 +0000)]
Minor cleanups.  Also, change various exit(10) codes to exit(1).

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoMinor cleanups.
Matthew Dillon [Wed, 25 Aug 2004 01:23:15 +0000 (01:23 +0000)]
Minor cleanups.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoMinor cleanups.
Matthew Dillon [Wed, 25 Aug 2004 01:15:38 +0000 (01:15 +0000)]
Minor cleanups.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agoMinor cleanups, no operational changes other then to add an error message if
Matthew Dillon [Wed, 25 Aug 2004 01:05:29 +0000 (01:05 +0000)]
Minor cleanups, no operational changes other then to add an error message if
fdopen() fails.

Submitted-by: "Douwe Kiela" <virtus@wanadoo.nl>
14 years agodoingdirectory is really a boolean, use an int rather then ino_t and cast
Matthew Dillon [Wed, 25 Aug 2004 00:05:11 +0000 (00:05 +0000)]
doingdirectory is really a boolean, use an int rather then ino_t and cast
it properly for the one case where it needs to be cast.  Document the special
case.  Cast -1 to (daddr_t) when checking for daddr_t special cases.

14 years agoM_EXT_CLUSTER was not being properly inherited in m_copym(), m_copypacket(),
Matthew Dillon [Tue, 24 Aug 2004 21:55:47 +0000 (21:55 +0000)]
M_EXT_CLUSTER was not being properly inherited in m_copym(), m_copypacket(),
and m_split(), resulting in potentially unnecessary extra copying.

14 years agoFix a SFBUF memory leak in sendfile(). We were not properly tracking
Matthew Dillon [Tue, 24 Aug 2004 21:53:41 +0000 (21:53 +0000)]
Fix a SFBUF memory leak in sendfile().  We were not properly tracking
references which would lead to SFBUFs not getting freed when two or more
sendfile()'s are operating on the same file at the same time (e.g. parallel
ftp downloads of the same file).

Get rid of the sf_buf->aux1 and aux2 hacks for sendfile.

Add a sysctl to allow the number of free SFBUFs to be monitored.

14 years agoThe VFS work has made vnode_if.awk obsolete.
Matthew Dillon [Tue, 24 Aug 2004 21:16:14 +0000 (21:16 +0000)]
The VFS work has made vnode_if.awk obsolete.

14 years agoHook pkill up to the build.
Chris Pressey [Tue, 24 Aug 2004 20:52:45 +0000 (20:52 +0000)]
Hook pkill up to the build.

14 years agoClearly I need to wakeup all the way before starting to make changes.
David Rhodus [Tue, 24 Aug 2004 16:32:11 +0000 (16:32 +0000)]
Clearly I need to wakeup all the way before starting to make changes.
Fix missing closing comment.

Noted by Sven Willenberger.

14 years agoSpelling.
David Rhodus [Tue, 24 Aug 2004 16:08:35 +0000 (16:08 +0000)]
Spelling.

Noted by eirikn.

14 years agoAdd a few notes.
David Rhodus [Tue, 24 Aug 2004 14:01:57 +0000 (14:01 +0000)]
Add a few notes.

14 years agoMinor cleanups. Make a type usage correction. Remove unneeded cast when
David Rhodus [Tue, 24 Aug 2004 13:52:59 +0000 (13:52 +0000)]
Minor cleanups. Make a type usage correction. Remove unneeded cast when
using struct bp.

14 years agoCheck error return value when creating the IPC service.
David Rhodus [Tue, 24 Aug 2004 13:45:54 +0000 (13:45 +0000)]
Check error return value when creating the IPC service.

14 years agoFix compilation of !PCI config.
Joerg Sonnenberger [Tue, 24 Aug 2004 08:15:22 +0000 (08:15 +0000)]
Fix compilation of !PCI config.

14 years agoUse sys/types.h here, since we don't want to restrict ourselves to the safe
Joerg Sonnenberger [Tue, 24 Aug 2004 06:10:57 +0000 (06:10 +0000)]
Use sys/types.h here, since we don't want to restrict ourselves to the safe
types (__) here.

14 years agoAdd missing extension return value for __byte_swap32_var() in the case
Matthew Dillon [Mon, 23 Aug 2004 21:16:38 +0000 (21:16 +0000)]
Add missing extension return value for __byte_swap32_var() in the case
where I386_CPU is not defined.  This missing value was causing a buildkernel
compile-time error.

14 years agoRemove pre-FreeBSD4 compability code.
Joerg Sonnenberger [Mon, 23 Aug 2004 16:13:03 +0000 (16:13 +0000)]
Remove pre-FreeBSD4 compability code.

14 years agoImprove the endian support for DragonFly by providing functions to convert
Joerg Sonnenberger [Mon, 23 Aug 2004 16:03:44 +0000 (16:03 +0000)]
Improve the endian support for DragonFly by providing functions to convert
16/32/64 bit variables and parts of byte streams between host order and
big / little endian.

Reduce the namespace pollution for normal usage, e.g. of sys/types.h.
The duplication of ntohl and friends in arpa/inet.h and sys/param.h comes
from byteorder(3) showing sys/param.h as header file and SUS requiring
arpa/inet.h to provide this functions.

Remove local hacks in acpidump and usb_port.h.

The prototype of ntohl and friends is in sync with SUS, warnings e.g. for
implicit sign casts or use of int/long will be handled separately.

Written-by: Hitten Pandya and Joerg Sonnenberger
Inspired-by: FreeBSD

14 years agoRemove ASR_MEASURE_PERFORMANCE, it doesn't work anyway.
Joerg Sonnenberger [Mon, 23 Aug 2004 15:11:44 +0000 (15:11 +0000)]
Remove ASR_MEASURE_PERFORMANCE, it doesn't work anyway.

Inspired-by: FreeBSD
14 years agorev 1.35:
Joerg Sonnenberger [Mon, 23 Aug 2004 09:39:29 +0000 (09:39 +0000)]
rev 1.35:
  Pass a correct lowaddr to bus_dma_tag_create(), lnc(4) cards can only
  deal with 24-bit addresses.  While the two other attachments, namely
  isa and cbus, do it properly, the PCI attachment was passing
  BUS_SPACE_MAXADDR instead of BUS_SPACE_MAXADDR_24BIT.  This bug
  became apparent with the new contigmalloc() code.

Obtained-from: FreeBSD

14 years agosys/types.h ==> sys/param.h for endian macros
Joerg Sonnenberger [Fri, 20 Aug 2004 01:21:36 +0000 (01:21 +0000)]
sys/types.h ==> sys/param.h for endian macros

14 years ago- include sys/param.h for endian macros
Joerg Sonnenberger [Fri, 20 Aug 2004 00:47:37 +0000 (00:47 +0000)]
- include sys/param.h for endian macros

14 years agosys/types.h ==> sys/param.h for endian macros
Joerg Sonnenberger [Fri, 20 Aug 2004 00:43:00 +0000 (00:43 +0000)]
sys/types.h ==> sys/param.h for endian macros

14 years ago- include sys/param.h for endian macros
Joerg Sonnenberger [Fri, 20 Aug 2004 00:08:17 +0000 (00:08 +0000)]
- include sys/param.h for endian macros

14 years ago- include sys/param.h for endian macros
Joerg Sonnenberger [Thu, 19 Aug 2004 23:57:46 +0000 (23:57 +0000)]
- include sys/param.h for endian macros

14 years agoAdd the standard DragonFly copyright with attribution to the author (which is
Matthew Dillon [Thu, 19 Aug 2004 23:57:02 +0000 (23:57 +0000)]
Add the standard DragonFly copyright with attribution to the author (which is
me).

Missing-copyright-noticed-by: Joerg Sonnenberger <joerg@britannica.bec.de>
14 years agoInclude language describing the prefered method for recognizing authors as
Matthew Dillon [Thu, 19 Aug 2004 23:53:39 +0000 (23:53 +0000)]
Include language describing the prefered method for recognizing authors as
part of the copyright and license section.  Split major elements of the
copyright file with '--' to make it more readable.

14 years ago- include sys/param.h for endian macros
Joerg Sonnenberger [Thu, 19 Aug 2004 23:48:16 +0000 (23:48 +0000)]
- include sys/param.h for endian macros

14 years ago- include sys/param.h for endian macros
Joerg Sonnenberger [Thu, 19 Aug 2004 23:41:07 +0000 (23:41 +0000)]
- include sys/param.h for endian macros

14 years ago- include sys/param.h for ntohl
Joerg Sonnenberger [Thu, 19 Aug 2004 23:40:15 +0000 (23:40 +0000)]
- include sys/param.h for ntohl
- include string.h for strcpy
- remove local define of PAGE_SIZE, use the version from machine/param.h

14 years agosys/types.h ==> sys/param.h for endian macros
Joerg Sonnenberger [Thu, 19 Aug 2004 23:35:45 +0000 (23:35 +0000)]
sys/types.h ==> sys/param.h for endian macros

14 years agosys/types.h ==> sys/param.h for endian macros
Joerg Sonnenberger [Thu, 19 Aug 2004 23:32:02 +0000 (23:32 +0000)]
sys/types.h ==> sys/param.h for endian macros

14 years agoFix prototype for signal handlers
Joerg Sonnenberger [Thu, 19 Aug 2004 23:26:12 +0000 (23:26 +0000)]
Fix prototype for signal handlers

14 years agoHide prototyp for loginit if ACULOG is false.
Joerg Sonnenberger [Thu, 19 Aug 2004 23:21:12 +0000 (23:21 +0000)]
Hide prototyp for loginit if ACULOG is false.

14 years agoFix spurious warning
Joerg Sonnenberger [Thu, 19 Aug 2004 23:12:10 +0000 (23:12 +0000)]
Fix spurious warning

14 years agosys/types.h ==> sys/param.h for ntohl
Joerg Sonnenberger [Thu, 19 Aug 2004 21:38:30 +0000 (21:38 +0000)]
sys/types.h ==> sys/param.h for ntohl

14 years ago- use WARNS?= 6 instead of a hard-wired list. -pedantic doesn't really work
Joerg Sonnenberger [Thu, 19 Aug 2004 21:36:46 +0000 (21:36 +0000)]
- use WARNS?= 6 instead of a hard-wired list. -pedantic doesn't really work
  anyway (no long long)
- use sys/param.h for nhtl in kgzcmp.c
- don't cast away const in kgzip.c, correct the usage of malloc'd space
  instead

14 years agoUse sys/param.h instead of sys/types.h for endian macros.
Joerg Sonnenberger [Thu, 19 Aug 2004 21:25:58 +0000 (21:25 +0000)]
Use sys/param.h instead of sys/types.h for endian macros.

This is in sync with the man page and results in a warning with the coming
endian patch.

14 years agoAdd implemenation of splay tree and red-black tree.
Joerg Sonnenberger [Thu, 19 Aug 2004 20:38:33 +0000 (20:38 +0000)]
Add implemenation of splay tree and red-black tree.

Obtained-from: NetBSD

14 years agoRemove the advertising clause where possible as per the directive from
Matthew Dillon [Thu, 19 Aug 2004 17:36:42 +0000 (17:36 +0000)]
Remove the advertising clause where possible as per the directive from
UC Berkeley (see /usr/src/COPYRIGHT) and with Ken Smith's permission
for mv.c.

>From:     Ken Smith <kensmith@cse.Buffalo.EDU>
>Subject:  mv.c license change
>To:       Matthew Dillon <dillon@apollo.backplane.com>
>Date:     Thu, 19 Aug 2004 13:30:28 -0400
>
>Yes, you have my permission to remove the advertising clause from
>the license agreement in /usr/src/bin/mv/mv.c.
>
>--
>                                                Ken Smith
>- From there to here, from here to      |       kensmith@cse.buffalo.edu
>  there, funny things are everywhere.   |
>                      - Theodore Geisel |

14 years agoPlug in missing brelse calls as to fix a bug in the FFS reload code.
David Rhodus [Thu, 19 Aug 2004 14:42:46 +0000 (14:42 +0000)]
Plug in missing brelse calls as to fix a bug in the FFS reload code.

14 years agoUpdate to reflect DragonFly reality.
Jeroen Ruigrok/asmodai [Thu, 19 Aug 2004 09:36:47 +0000 (09:36 +0000)]
Update to reflect DragonFly reality.

14 years agoAdd the long overdue ehci(4) manual page.
Jeroen Ruigrok/asmodai [Thu, 19 Aug 2004 09:34:55 +0000 (09:34 +0000)]
Add the long overdue ehci(4) manual page.

Obtained from: NetBSD's v1.11

14 years agoAdd the vop_ops for ntfs before obtaining the root vnode(s) rather then
Matthew Dillon [Thu, 19 Aug 2004 00:30:07 +0000 (00:30 +0000)]
Add the vop_ops for ntfs before obtaining the root vnode(s) rather then
after.

Reported-by: "David W. Chapman Jr." <dwcjr@inethouston.net>
14 years agoThe -D__FreeBSD__ must be -D__FreeBSD__=4 or sendmail will not be properly
Matthew Dillon [Wed, 18 Aug 2004 16:55:21 +0000 (16:55 +0000)]
The -D__FreeBSD__ must be -D__FreeBSD__=4 or sendmail will not be properly
configured to use features like setproctitle().

14 years agoMake the buildkernel and nativekernel targets completely wipe and regenerate
Matthew Dillon [Wed, 18 Aug 2004 16:37:15 +0000 (16:37 +0000)]
Make the buildkernel and nativekernel targets completely wipe and regenerate
the kernel object hierarchy.

Add a quickkernel target which does not wipe the object hieracrhy and which
skips the make depend step.  quickkernel can be run on kernels previously
built with buildkernel or nativekernel.

This brings the operation of our kernel building targets in-line with the
operation of our world building targets.

14 years agoProperly free the temporary sf_buf in uiomove_fromphys() if a copyin/copyout
Matthew Dillon [Wed, 18 Aug 2004 09:16:16 +0000 (09:16 +0000)]
Properly free the temporary sf_buf in uiomove_fromphys() if a copyin/copyout
fails.  Such failures commonly occur when pipes are broken by a ^C.

14 years ago[test format strings for cvs server config]
Matthew Dillon [Wed, 18 Aug 2004 09:10:32 +0000 (09:10 +0000)]
[test format strings for cvs server config]

14 years agotest
Matthew Dillon [Wed, 18 Aug 2004 09:07:06 +0000 (09:07 +0000)]
test

14 years ago*** empty log message ***
Matthew Dillon [Wed, 18 Aug 2004 09:04:29 +0000 (09:04 +0000)]
*** empty log message ***

14 years ago*** empty log message ***
Matthew Dillon [Wed, 18 Aug 2004 09:02:08 +0000 (09:02 +0000)]
*** empty log message ***

14 years agobleh
Matthew Dillon [Wed, 18 Aug 2004 08:56:53 +0000 (08:56 +0000)]
bleh

14 years ago*** empty log message ***
Matthew Dillon [Wed, 18 Aug 2004 08:47:16 +0000 (08:47 +0000)]
*** empty log message ***

14 years ago*** empty log message ***
Matthew Dillon [Wed, 18 Aug 2004 08:45:58 +0000 (08:45 +0000)]
*** empty log message ***

14 years agomore testing
Matthew Dillon [Wed, 18 Aug 2004 08:43:23 +0000 (08:43 +0000)]
more testing

14 years agoyet more testing
Matthew Dillon [Wed, 18 Aug 2004 08:31:50 +0000 (08:31 +0000)]
yet more testing

14 years agotest
Matthew Dillon [Wed, 18 Aug 2004 08:24:44 +0000 (08:24 +0000)]
test

14 years ago*** empty log message ***
Matthew Dillon [Wed, 18 Aug 2004 08:00:38 +0000 (08:00 +0000)]
*** empty log message ***

14 years agofeh2
Matthew Dillon [Wed, 18 Aug 2004 07:24:30 +0000 (07:24 +0000)]
feh2

14 years agofeh
Matthew Dillon [Wed, 18 Aug 2004 07:23:54 +0000 (07:23 +0000)]
feh

14 years agosigh
Matthew Dillon [Wed, 18 Aug 2004 07:18:52 +0000 (07:18 +0000)]
sigh

14 years agomore testing 3
Matthew Dillon [Wed, 18 Aug 2004 07:00:25 +0000 (07:00 +0000)]
more testing 3

14 years agomore testing
Matthew Dillon [Wed, 18 Aug 2004 06:37:03 +0000 (06:37 +0000)]
more testing

14 years agoTest commits after machine upgrade.
Matthew Dillon [Wed, 18 Aug 2004 06:23:55 +0000 (06:23 +0000)]
Test commits after machine upgrade.

14 years agoAdd some robustness to the error-requeue code. FreeBSD-5's (new) ata driver
Matthew Dillon [Tue, 17 Aug 2004 20:59:39 +0000 (20:59 +0000)]
Add some robustness to the error-requeue code.  FreeBSD-5's (new) ata driver
had an issue with the donecount not being properly reset.  This issue is not
believed to occur with the old code but add sanity checks to be sure.

14 years agoProperly record and print 64 bit file sizes, do not truncate the file size
Matthew Dillon [Tue, 17 Aug 2004 20:34:38 +0000 (20:34 +0000)]
Properly record and print 64 bit file sizes, do not truncate the file size
to 32 bits.

14 years agoUpdate list of installer packages to match installer version 1.1.
Chris Pressey [Tue, 17 Aug 2004 19:23:52 +0000 (19:23 +0000)]
Update list of installer packages to match installer version 1.1.

14 years agoVFS messaging/interfacing work stage 2/99. This stage retools the vnode ops
Matthew Dillon [Tue, 17 Aug 2004 18:57:36 +0000 (18:57 +0000)]
VFS messaging/interfacing work stage 2/99.  This stage retools the vnode ops
vector dispatch, making the vop_ops a per-mount structure rather then a
per-filesystem structure.  Filesystem mount code, typically in blah_vfsops.c,
must now register various vop_ops pointers in the struct mount to compile
its VOP operations set.

This change will allow us to begin adding per-mount hooks to VFSes to support
things like kernel-level journaling, various forms of cache coherency
management, and so forth.

In addition, the vop_*() calls now require a struct vop_ops pointer as the
first argument instead of a vnode pointer (note: in this commit the VOP_*()
macros currently just pull the vop_ops pointer from the vnode in order to
call the vop_*() procedures).  This change is intended to allow us to divorce
ourselves from the requirement that a vnode pointer always be part of a VOP
call.  In particular, this will allow namespace based routines such as
remove(), mkdir(), stat(), and so forth to pass namecache pointers rather then
locked vnodes and is a very important precursor to the goal of using the
namecache for namespace locking.

14 years agoAdd an installer-fetchpkgs target and other related stuff to reduce the
Matthew Dillon [Tue, 17 Aug 2004 18:05:41 +0000 (18:05 +0000)]
Add an installer-fetchpkgs target and other related stuff to reduce the
work required to bring in the installer packages when building an
installer-enabled ISO.

Submitted-by: Chris Pressey <cpressey@catseye.mine.nu>