]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoFix static array overrun.
Gleb Smirnoff [Wed, 17 May 2006 07:38:58 +0000 (07:38 +0000)]
Fix static array overrun.

(This will be also fixed in next vendor release.)

Coverity ID: 916
Reviewed by: Jack Vogel

18 years agoMake the printfs relating to purging threads from a device less intrusive.
Poul-Henning Kamp [Wed, 17 May 2006 06:37:14 +0000 (06:37 +0000)]
Make the printfs relating to purging threads from a device less intrusive.

18 years agoUpdate to current DRM CVS, bringing in a fix for a NULL pointer chase when
Eric Anholt [Wed, 17 May 2006 06:36:28 +0000 (06:36 +0000)]
Update to current DRM CVS, bringing in a fix for a NULL pointer chase when
running 3d apps on i915.

18 years agoMerge from DRM CVS:
Eric Anholt [Wed, 17 May 2006 06:29:36 +0000 (06:29 +0000)]
Merge from DRM CVS:
Set entry->virtual for sg maps, fixing ATI PCI/PCIE GART support.

PR: kern/97056
Submitted by: Stanislav Sedov <ssedov@mbsd.msk.ru>

18 years agoFix return value.
David Xu [Wed, 17 May 2006 05:14:45 +0000 (05:14 +0000)]
Fix return value.

18 years agoget thread signal info.
David Xu [Wed, 17 May 2006 05:12:42 +0000 (05:12 +0000)]
get thread signal info.

18 years agoExport offsets of thread signal pending set and signal mask for debugger.
David Xu [Wed, 17 May 2006 04:59:59 +0000 (04:59 +0000)]
Export offsets of thread signal pending set and signal mask for debugger.

18 years agoAdd new SIOC_HCI_RAW_NODE_LIST_NAMES ioctl. User-space applications can
Maksim Yevmenkin [Wed, 17 May 2006 00:13:07 +0000 (00:13 +0000)]
Add new SIOC_HCI_RAW_NODE_LIST_NAMES ioctl. User-space applications can
use this ioctl to obtain the list of HCI nodes. User-space application
is expected to preallocate 'ng_btsocket_hci_raw_node_list_names' structure
and set limit in 'num_nodes' field. The 'nodes' field should be allocated
as well and it should have space for at least 'num_nodes' elements.

The SIOC_HCI_RAW_NODE_LIST_NAMES should be issued on bound raw HCI socket.
It does not really really matter what HCI name the socket is bound to, as
long as it is not empty.

MFC after: 1 week

18 years agoConsistently use `superuser privileges' instead of its singular
Giorgos Keramidas [Tue, 16 May 2006 22:58:43 +0000 (22:58 +0000)]
Consistently use `superuser privileges' instead of its singular
form here and its plural there.

Submitted by: Rostislav Krasny <rosti.bsd@gmail.com>
MFC after:  3 days

18 years agoFix style(9) nits, whitespace and parentheses.
Andrew Thompson [Tue, 16 May 2006 22:50:41 +0000 (22:50 +0000)]
Fix style(9) nits, whitespace and parentheses.

18 years agoDisconnect mount_reiserfs from build.
Craig Rodrigues [Tue, 16 May 2006 22:12:55 +0000 (22:12 +0000)]
Disconnect mount_reiserfs from build.
It is no longer needed, because "mount -t reiserfs" calls nmount(2)
directly to mount a ReiserFS file system without the use of
this external mount program.

Approved by: dumbbell

18 years agoAdd reiserfs.5 man page to build.
Craig Rodrigues [Tue, 16 May 2006 22:08:38 +0000 (22:08 +0000)]
Add reiserfs.5 man page to build.

18 years agoAdd man page for ReiserFS.
Craig Rodrigues [Tue, 16 May 2006 22:06:47 +0000 (22:06 +0000)]
Add man page for ReiserFS.

Reviewed by: dumbbell

18 years ago- Revert if_le_pci.c rev. 1.2; although lnc(4) is now gone, le_pci_probe()
Marius Strobl [Tue, 16 May 2006 21:04:01 +0000 (21:04 +0000)]
- Revert if_le_pci.c rev. 1.2; although lnc(4) is now gone, le_pci_probe()
  still should return BUS_PROBE_LOW_PRIORITY instead of BUS_PROBE_DEFAULT
  in order to give pcn(4) a chance to attach in case it probes after le(4).
- Rearrange the code related to RX interrupt handling so that ownership of
  RX descriptors is immediately returned to the NIC after we have copied
  the data of the hardware, allowing the NIC to already reuse the descriptor
  while we are processing the data in ifp->if_input(). This results in a
  small but measurable increase in RX throughput.
  As a side-effect, this moves the workaround for the LANCE revision C bug
  to am7900.c (still off by default as I doubt we will actually encounter
  such an old chip in a machine running FreeBSD) and the workaround for the
  bug in the VMware PCnet-PCI emulation to am79000.c, which is now also
  only compiled on i386 (resulting in a small increase in RX throughput on
  the other platforms).
- Change the RX interrupt handlers so that the descriptor error bits are
  only check once in case there was no error instead of twice (inspired
  by the NetBSD pcn(4), which additionally predicts the error branch as
  false).
- Fix the debugging output of the RX and TX interrupt handlers; while
  looping through the descriptors print info about the currently processed
  one instead of always the previously last used one; remove pointless
  printing of info about the RX descriptor bits after their values were
  reset.
- Create the DMA tags used to allocate the memory for the init block,
  descriptors and packet buffers with the alignment the respective NIC
  actually requires rather than using PAGE_SIZE unconditionally. This might
  as well fix the alignment of the memory as it seems we do not inherit
  the alignment constraint from the parent DMA tag.
- For the PCI variants double the number of RX descriptors and buffers
  from 8 to 16 as this minimizes the number of RX overflows im seeing with
  one NIC-mainboard combination. Nevertheless move reporting of overflows
  under debugging as they seem unavoidable with some crappy hardware.
- Set the software style of the PCI variants to ILACC rather than PCnet-PCI
  as the former is was am79000.c actually implements. Should not make a
  difference for this driver though.
- Fix the driver name part in the MODULE_DEPEND of the PCI front-end for
  ether.
- Use different device descriptions for PCnet-Home and PCnet-PCI.
- Fix some 0/NULL confusion in lance_get().
- Use bus_addr_t for sc_addr and bus_size_t for sc_memsize as these are
  more appropriate than u_long for these.
- Remove the unused LE_DRIVER_NAME macro.
- Add a comment describing why we are taking the LE_HTOLE* etc approach
  instead of using byteorder(9) functions directly.
- Improve some comments and fix some wording.

MFC after: 2 weeks

18 years ago* Document the fact that non-superusers cannot change file flags for
Giorgos Keramidas [Tue, 16 May 2006 20:24:41 +0000 (20:24 +0000)]
* Document the fact that non-superusers cannot change file flags for
  objects with SF_IMMUTABLE, SF_APPEND, or SF_NOUNLINK.
* Document that non-superusers cannot set or clear any SF_* flag
  (setting fails with EPERM, clearing is silently ignored).
* Document that superusers cannot change any flag if one of
  SF_IMMUTABLE, SF_APPEND, SF_NOUNLINK is set and securelevel is
  greater than 0.
* Document SF_SNAPSHOT and note that it is maintained by the
  system and is, for this reason, impossible to set to clear by
  any user.

PR:             docs/33877
Submitted by:   harti
Help by:        George Marsellis <gam9478@njit.edu>
MFC after:      1 week

18 years agoThe current routing code allows insertion of indirect routes that have
Qing Li [Tue, 16 May 2006 19:11:11 +0000 (19:11 +0000)]
The current routing code allows insertion of indirect routes that have
gateways which are unreachable except through the default router. For
example, assuming there is a default route configured, and inserting
a route

"route add 64.102.54.0/24 60.80.1.1"

is currently allowed even when 60.80.1.1 is only reachable through
the default route. However, an error is thrown when this route is
utilized, say,

"ping 64.102.54.1"  will return an error

This type of route insertion should be disallowed becasue:

1) Let's say that somehow our code allowed this packet to flow to
   the default router, and the default router knows the next hop is
   60.80.1.1, then the question is why bother inserting this route in
   the 1st place, just simply use the default route.

2) Since we're not talking about source routing here, the default
   router could very well choose a different path than using 60.80.1.1
   for the next hop, again it defeats the purpose of adding this route.

Reviewed by: ru, gnn, bz
Approved by: andre

18 years agoSome minor wording/description changes.
Tom Rhodes [Tue, 16 May 2006 18:59:00 +0000 (18:59 +0000)]
Some minor wording/description changes.

Reviewed by: ru (two months ago)

18 years agoProvide a less cryptic panic message in place of just "found inode."
Tom Rhodes [Tue, 16 May 2006 18:51:22 +0000 (18:51 +0000)]
Provide a less cryptic panic message in place of just "found inode."

18 years agoFix file leaking in translate_path_major_minor.
Doug Ambrisko [Tue, 16 May 2006 17:57:00 +0000 (17:57 +0000)]
Fix file leaking in translate_path_major_minor.

18 years agoClose a "(" I left open.
Matteo Riondato [Tue, 16 May 2006 17:10:16 +0000 (17:10 +0000)]
Close a "(" I left open.

18 years agoMove a define depending on __FreeBSD_versoin to after where it
Matt Jacob [Tue, 16 May 2006 16:31:58 +0000 (16:31 +0000)]
Move a define depending on __FreeBSD_versoin to after where it
would be defined.

Submitted by:   Ruslan Ermilov

18 years agoAdd support for allocating one larger than page-sized contiguous block of memory
Eric Anholt [Tue, 16 May 2006 16:19:48 +0000 (16:19 +0000)]
Add support for allocating one larger than page-sized contiguous block of memory
with a physical address.  This is used for hardware ARGB cursor support on newer
chipsets.

18 years agoDon't try to upload alpha-specific docs anymore.
Bruce A. Mah [Tue, 16 May 2006 16:12:45 +0000 (16:12 +0000)]
Don't try to upload alpha-specific docs anymore.

18 years agoUpdate the description of the `STATE' column for SMP systems,
Giorgos Keramidas [Tue, 16 May 2006 15:27:43 +0000 (15:27 +0000)]
Update the description of the `STATE' column for SMP systems,
and add the missing `C' SMP-only column.

PR: docs/29245
Submitted by: ada@bsd.org
Patch by: Pawel Worach <pawel.worach@telia.com>
MFC after: 3 days

18 years agoMake this compile with -Wundef.
Ruslan Ermilov [Tue, 16 May 2006 15:23:27 +0000 (15:23 +0000)]
Make this compile with -Wundef.

18 years agoSince DELAY() was moved, most <machine/clock.h> #includes have been
Poul-Henning Kamp [Tue, 16 May 2006 14:37:58 +0000 (14:37 +0000)]
Since DELAY() was moved,  most <machine/clock.h> #includes have been
unnecessary.

18 years agoMake a bootable CD on pc98.
Yoshihiro Takahashi [Tue, 16 May 2006 14:22:37 +0000 (14:22 +0000)]
Make a bootable CD on pc98.

18 years ago1) Fix mii_phy_tick():
Oleg Bulyzhin [Tue, 16 May 2006 12:26:00 +0000 (12:26 +0000)]
1) Fix mii_phy_tick():
   Current code does not report link loss correctly - when link goes down,
   mii_phy_tick() will notice that with up to mii_anegticks delay.
   If link goes up during this delay then link flapping will be unnoticed
   by driver.

2) mii_phy_add_media(): initialize sc->mii_anegticks for 10/100 media

3) Use MII_ANEGTICKS/MII_ANEGTICKS_GIGE defines instead of hardcoded values.

Approved by: glebius (mentor)
MFC after: 1 month

18 years agoIntroduce MII_ANEGTICKS and MII_ANEGTICKS_GIGE defines.
Oleg Bulyzhin [Tue, 16 May 2006 12:23:57 +0000 (12:23 +0000)]
Introduce MII_ANEGTICKS and MII_ANEGTICKS_GIGE defines.
(How many ticks should we wait before retrying autonegotiation process).

Approved by: glebius (mentor)
MFC after: 1 month

18 years agoKill more references to lnc(4).
Ruslan Ermilov [Tue, 16 May 2006 12:15:39 +0000 (12:15 +0000)]
Kill more references to lnc(4).

Submitted by: grep(1)

18 years agoRemove unneeded check.
Gleb Smirnoff [Tue, 16 May 2006 11:49:26 +0000 (11:49 +0000)]
Remove unneeded check.

Coverity ID: 445

18 years agoSince m_pullup() can return a new mbuf, change gre_input2() to
Gleb Smirnoff [Tue, 16 May 2006 11:15:22 +0000 (11:15 +0000)]
Since m_pullup() can return a new mbuf, change gre_input2() to
return mbuf back to gre_input(). If the former returns mbuf back
to the latter, then pass it to raw_input().

Coverity ID: 829

18 years ago- Backout one line from 1.78. The tp can be freed by tcp_drop().
Gleb Smirnoff [Tue, 16 May 2006 10:51:26 +0000 (10:51 +0000)]
- Backout one line from 1.78. The tp can be freed by tcp_drop().
- Style next line.

Coverity ID: 912

18 years agoDo not leak kernel memory in case if userland has been compiled
Gleb Smirnoff [Tue, 16 May 2006 09:32:58 +0000 (09:32 +0000)]
Do not leak kernel memory in case if userland has been compiled
against older NG_VERSION.

Coverity ID: 1131

18 years agoAllow concurrent read(2)/readv(2) access to a file.
Paul Saab [Tue, 16 May 2006 07:50:54 +0000 (07:50 +0000)]
Allow concurrent read(2)/readv(2) access to a file.
Lock file offset against multiple read calls.

Submitted by: ups
Obtained from: Yahoo!
MFC after: 2 weeks

18 years agoRead block hints list from last snapshot on the active snapshot list.
Tor Egge [Tue, 16 May 2006 00:14:20 +0000 (00:14 +0000)]
Read block hints list from last snapshot on the active snapshot list.

18 years agoCopy last block on file system again after file system has been suspended.
Tor Egge [Mon, 15 May 2006 23:18:49 +0000 (23:18 +0000)]
Copy last block on file system again after file system has been suspended.

Obtained from: NetBSD

18 years agoDon't leak a locked buffer if last block on file system cannot be read.
Tor Egge [Mon, 15 May 2006 22:59:23 +0000 (22:59 +0000)]
Don't leak a locked buffer if last block on file system cannot be read.

18 years agoErrors detected while file system is suspended should not trigger an
Tor Egge [Mon, 15 May 2006 22:52:22 +0000 (22:52 +0000)]
Errors detected while file system is suspended should not trigger an
assertion failure.

18 years ago- Fix warning when compiling with -DDEBUG
Pav Lucistnik [Mon, 15 May 2006 22:09:45 +0000 (22:09 +0000)]
- Fix warning when compiling with -DDEBUG

Approved by: alfred

18 years ago- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage,
Pav Lucistnik [Mon, 15 May 2006 21:18:28 +0000 (21:18 +0000)]
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage,
  rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock,
  pathconf, truncate, ftruncate, kill

- Decode more arguments of open, mprot, *stat, and fcntl.

- Convert all constant-macro and bitfield decoding to lookup tables; much
  cleaner than previous code.

- Print the timestamp of process exit and signal reception when -d or -D are in
  use

- Try six times with 1/2 second delay to debug the child

PR: bin/52190 (updated)
Submitted by: Dan Nelson <dnelson@allantgroup.com>
Approved by: alfred

18 years agoShoot my own dog :-(
Wilko Bulte [Mon, 15 May 2006 21:13:55 +0000 (21:13 +0000)]
Shoot my own dog :-(

18 years agoPrepare to shoot my own dog, decouple the alpha docs from the build.
Wilko Bulte [Mon, 15 May 2006 21:12:12 +0000 (21:12 +0000)]
Prepare to shoot my own dog, decouple the alpha docs from the build.

18 years agos/alpha/sparc64/ in examples.
Wilko Bulte [Mon, 15 May 2006 21:10:34 +0000 (21:10 +0000)]
s/alpha/sparc64/ in examples.

18 years ago- Fix crash when hitting unknown syscall. Copied from i386-fbsd.c
Pav Lucistnik [Mon, 15 May 2006 21:03:02 +0000 (21:03 +0000)]
- Fix crash when hitting unknown syscall.  Copied from i386-fbsd.c

Approved by: alfred

18 years agoRemove reference to Alpha architecture.
Wilko Bulte [Mon, 15 May 2006 20:58:05 +0000 (20:58 +0000)]
Remove reference to Alpha architecture.

18 years agoRemove reference to Alpha
Wilko Bulte [Mon, 15 May 2006 20:55:36 +0000 (20:55 +0000)]
Remove reference to Alpha

18 years agoo In the cross-build example replace alpha by sparc64.
Maxim Konovalov [Mon, 15 May 2006 20:53:00 +0000 (20:53 +0000)]
o In the cross-build example replace alpha by sparc64.

18 years agoRemove reference to the Alpha architecture.
Wilko Bulte [Mon, 15 May 2006 20:51:53 +0000 (20:51 +0000)]
Remove reference to the Alpha architecture.

18 years agoRemove reference to Alpha.
Wilko Bulte [Mon, 15 May 2006 20:48:18 +0000 (20:48 +0000)]
Remove reference to Alpha.

18 years agoRemove reference to Alpha machine
Wilko Bulte [Mon, 15 May 2006 20:42:42 +0000 (20:42 +0000)]
Remove reference to Alpha machine

18 years agoo Join "lnc(4) killers" gang and add lnc.4 man page to a list
Maxim Konovalov [Mon, 15 May 2006 20:41:11 +0000 (20:41 +0000)]
o Join "lnc(4) killers" gang and add lnc.4 man page to a list
of obsoleted files.

o Fix a timestamp in the previous revision: 20061205 -> 20060512.

18 years agoNow that lnc(4) is removed from the source tree, replace lnc0 in an example
Marius Strobl [Mon, 15 May 2006 20:28:18 +0000 (20:28 +0000)]
Now that lnc(4) is removed from the source tree, replace lnc0 in an example
with fxp0 (fxp(4) was chosen as it's also used in a couple of examples in
other man pages and as it has good grep(1)'ability).

18 years agoNow that lnc(4) is removed from the source tree, remove it from the list
Marius Strobl [Mon, 15 May 2006 20:16:42 +0000 (20:16 +0000)]
Now that lnc(4) is removed from the source tree, remove it from the list
of drivers supporting ALTQ.

18 years agoReplace references to lnc(4) with references to le(4) (so far the notes
Marius Strobl [Mon, 15 May 2006 20:07:24 +0000 (20:07 +0000)]
Replace references to lnc(4) with references to le(4) (so far the notes
still apply to le(4)) now that lnc(4) is removed and le(4) is going to
replace it.

18 years agoUpdate the description of pcn(4) from pcn(4) vs. lnc(4) to pcn(4) vs. le(4)
Marius Strobl [Mon, 15 May 2006 20:03:10 +0000 (20:03 +0000)]
Update the description of pcn(4) from pcn(4) vs. lnc(4) to pcn(4) vs. le(4)
now that lnc(4) is removed and le(4) is going to replace it.

18 years agoRestore the ability to mount procfs and fdescfs filesystems via the
Kelly Yancey [Mon, 15 May 2006 19:42:10 +0000 (19:42 +0000)]
Restore the ability to mount procfs and fdescfs filesystems via the
mount(2) system call:

  * Add cmount hook to fdescfs and pseudofs (and, by extension, procfs and
    linprocfs).  This (mostly) restores the ability to mount these
    filesystems using the old mount(2) system call (see below for the
    rest of the fix).

  * Remove not-NULL check for the data argument from the mount(2) entry
    point.  Per the mount(2) man page, it is up to the individual
    filesystem being mounted to verify data.  Or, in the case of procfs,
    etc. the filesystem is free to ignore the data parameter if it does
    not use it.  Enforcing data to be not-NULL in the mount(2) system call
    entry point prevented passing NULL to filesystems which ignored the
    data pointer value.  Apparently, passing NULL was common practice
    in such cases, as even our own mount_std(8) used to do it in the
    pre-nmount(2) world.

All userland programs in the tree were converted to nmount(2) long ago,
but I've found at least one external program which broke due to this
(presumably unintentional) mount(2) API change.  One could argue that
external programs should also be converted to nmount(2), but then there
isn't much point in keeping the mount(2) interface for backward
compatibility if it isn't backward compatible.

18 years agoUpdate README.examples according to the current state of the examples/etc directory
Matteo Riondato [Mon, 15 May 2006 17:43:51 +0000 (17:43 +0000)]
Update README.examples according to the current state of the examples/etc directory

PR: conf/97232

18 years agoNote that an old adjkerntz is not really usable after the recent
Giorgos Keramidas [Mon, 15 May 2006 15:47:45 +0000 (15:47 +0000)]
Note that an old adjkerntz is not really usable after the recent
sysctlbyname() changes, so the new adjkerntz binary should be
used while in single-user mode.

Reviewed by: ceri, maxim

18 years agoRefer to csup in the make.conf(5) manpage.
Maxime Henrion [Mon, 15 May 2006 14:08:41 +0000 (14:08 +0000)]
Refer to csup in the make.conf(5) manpage.

Insta-MFC to follow.

Reminded by: ru (some time ago)

18 years agoAdd definitions for atomic_subtract_rel_32, atomic_add_rel_32 and
Olivier Houchard [Mon, 15 May 2006 13:08:12 +0000 (13:08 +0000)]
Add definitions for atomic_subtract_rel_32, atomic_add_rel_32 and
atomic_load_acq_32, needed for hwpmc.

18 years agoRecalculate IP checksum after running pfil hooks.
Daniel Hartmeier [Mon, 15 May 2006 11:49:01 +0000 (11:49 +0000)]
Recalculate IP checksum after running pfil hooks.

Reviewed by: thompsa
Tested by: Adam McDougall <mcdouga9@egr.msu.edu>

18 years agoSwitch from the lnc driver to the le driver. But C-NET(98)S support is
Yoshihiro Takahashi [Mon, 15 May 2006 11:15:34 +0000 (11:15 +0000)]
Switch from the lnc driver to the le driver.  But C-NET(98)S support is
dropped.

18 years agoDisplay real/avail memory as per other platforms.
Benno Rice [Mon, 15 May 2006 10:40:37 +0000 (10:40 +0000)]
Display real/avail memory as per other platforms.

Approved by: cognet

18 years agoo In rip_disconnect() do not call rip_abort(), just mark a socket
Maxim Konovalov [Mon, 15 May 2006 09:28:57 +0000 (09:28 +0000)]
o In rip_disconnect() do not call rip_abort(), just mark a socket
as not connected.  In soclose() case rip_detach() will kill inpcb for
us later.

It makes rawconnect regression test do not panic a system.

Reviewed by: rwatson
X-MFC after: with all 1th April inpcb changes

18 years agoo Fix typo in a comment.
Maxim Konovalov [Mon, 15 May 2006 08:20:38 +0000 (08:20 +0000)]
o Fix typo in a comment.

PR: bin/97292
Submitted by: clsung
MFC after: 3 days

18 years agoRe-wrok PHY setup, media handling and dual-port detection.
Pyun YongHyeon [Mon, 15 May 2006 04:50:33 +0000 (04:50 +0000)]
Re-wrok PHY setup, media handling and dual-port detection.
With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0
adapter now works.

Obtained from: OpenBSD
Reported by: Ganbold ganbold ! micom ( mng $ net
Tested by: Ganbold ganbold ! micom ( mng $ net

18 years agoSwitch to a 64bit time_t, while it's not a big problem to do so.
Olivier Houchard [Mon, 15 May 2006 00:17:27 +0000 (00:17 +0000)]
Switch to a 64bit time_t, while it's not a big problem to do so.

Suggested by: imp

18 years agoUse only lower 64bit of src/dest (and src/dest port) for hashing of IPv6
Max Laier [Sun, 14 May 2006 23:42:24 +0000 (23:42 +0000)]
Use only lower 64bit of src/dest (and src/dest port) for hashing of IPv6
connections and get rid of the flow_id as it is not guaranteed to be stable
some (most?) current implementations seem to just zero it out.

PR: kern/88664
Reported by: jylefort
Submitted by: Joost Bekkers (w/ changes)
Tested by "regisr" <regisrApoboxDcom>

18 years agoFix formatting. Add missing break;
Maksim Yevmenkin [Sun, 14 May 2006 22:19:38 +0000 (22:19 +0000)]
Fix formatting. Add missing break;

Submitted by: Iain Hibbert
MFC after: 3 days

18 years agoWith the demise of Alpha, remove Alpha specific diagnostic.
Wilko Bulte [Sun, 14 May 2006 22:19:11 +0000 (22:19 +0000)]
With the demise of Alpha, remove Alpha specific diagnostic.

18 years agoo A tiny regression test exposes bug in rip_disconnect().
Maxim Konovalov [Sun, 14 May 2006 22:03:00 +0000 (22:03 +0000)]
o A tiny regression test exposes bug in rip_disconnect().

18 years agoNew release notes:
Christian Brueffer [Sun, 14 May 2006 21:02:25 +0000 (21:02 +0000)]
New release notes:

- cpufreq(4) support for VIA C7-M processors
- lnc(4) removed
- ip6fw(8) removed

18 years agoFix typo.
Christian Brueffer [Sun, 14 May 2006 20:54:45 +0000 (20:54 +0000)]
Fix typo.

18 years agoRemove autogeneration of the lnc(4) hardware list to fix the build.
Christian Brueffer [Sun, 14 May 2006 20:31:32 +0000 (20:31 +0000)]
Remove autogeneration of the lnc(4) hardware list to fix the build.

18 years agoThe last execution of -exec {} + is not done if the -exec primary is
Kirill Ponomarev [Sun, 14 May 2006 20:23:01 +0000 (20:23 +0000)]
The last execution of -exec {} + is not done if the -exec primary is
not on the top-level -and sequence, e.g. inside of ! or -or.

Create a separate linked list of all active -exec {} + primaries and
do the last execution for all at termination.

PR: bin/79263
Submitted by: Jilles Tjoelker <jilles@stack.nl>
MFC after: 7 days

18 years ago- Replace the entry for the no longer existing lnc(4) module with an
Marius Strobl [Sun, 14 May 2006 19:04:12 +0000 (19:04 +0000)]
- Replace the entry for the no longer existing lnc(4) module with an
  entry for the replacement le(4) module.
- Add an entry for the gem(4) module.
- Remove gratuitous whitespace in the description of the hme(4) entry.

18 years agoRemove some remnants of lnc(4).
Marius Strobl [Sun, 14 May 2006 18:49:25 +0000 (18:49 +0000)]
Remove some remnants of lnc(4).

18 years agoRemove the lnc module Makefile.
Scott Long [Sun, 14 May 2006 18:37:03 +0000 (18:37 +0000)]
Remove the lnc module Makefile.

18 years agoFix build (unhook lnc).
Xin LI [Sun, 14 May 2006 18:31:34 +0000 (18:31 +0000)]
Fix build (unhook lnc).

Submitted by: David H. Wolfskill <david catwhisker org>

18 years agoAdd if_* MLINKS for bce(4) and le(4).
Christian Brueffer [Sun, 14 May 2006 15:45:44 +0000 (15:45 +0000)]
Add if_* MLINKS for bce(4) and le(4).

MFC after: 3 days

18 years agoFix a long-standing limitation in IPv4 multicast group membership.
Bruce M Simpson [Sun, 14 May 2006 14:22:49 +0000 (14:22 +0000)]
Fix a long-standing limitation in IPv4 multicast group membership.

By making the imo_membership array a dynamically allocated vector,
this minimizes disruption to existing IPv4 multicast code. This
change breaks the ABI for the kernel module ip_mroute.ko, and may
cause a small amount of churn for folks working on the IGMPv3 merge.

Previously, sockets were subject to a compile-time limitation on
the number of IPv4 group memberships, which was hard-coded to 20.
The imo_membership relationship, however, is 1:1 with regards to
a tuple of multicast group address and interface address. Users who
ran routing protocols such as OSPF ran into this limitation on machines
with a large system interface tree.

18 years agoDon't quote IP_MAX_MEMBERSHIPS as 20 any more.
Bruce M Simpson [Sun, 14 May 2006 14:13:30 +0000 (14:13 +0000)]
Don't quote IP_MAX_MEMBERSHIPS as 20 any more.

18 years agoAdd multicast group join/leave functionality to IPv4 regression
Bruce M Simpson [Sun, 14 May 2006 14:11:54 +0000 (14:11 +0000)]
Add multicast group join/leave functionality to IPv4 regression
test suite.

18 years agoHandle SIGINFO in time(1) by printing the time the given command is running.
Pawel Jakub Dawidek [Sun, 14 May 2006 13:16:19 +0000 (13:16 +0000)]
Handle SIGINFO in time(1) by printing the time the given command is running.
I find it quite handy to just hit ^T and see the time.

18 years agoAdd a bootable CD support.
Yoshihiro Takahashi [Sun, 14 May 2006 07:26:02 +0000 (07:26 +0000)]
Add a bootable CD support.

18 years agoThe VERBOSE_SYSINIT stuff sees the DDB define a lot better if we include
Benno Rice [Sun, 14 May 2006 07:11:28 +0000 (07:11 +0000)]
The VERBOSE_SYSINIT stuff sees the DDB define a lot better if we include
opt_ddb.h.

Spotted by: benno
Pointy hat to: benno

18 years agoFor src/dest parsing take off the netmask before checking for AF with
Max Laier [Sun, 14 May 2006 03:53:04 +0000 (03:53 +0000)]
For src/dest parsing take off the netmask before checking for AF with
inet_pton.  This fixes cases like "fe02::/16".

PR: bin/91245
Reported by: Fredrik Lindberge

18 years agoRemove the ipfw6 config from NOTES
Mathieu Arnold [Sun, 14 May 2006 02:37:56 +0000 (02:37 +0000)]
Remove the ipfw6 config from NOTES

Forgotten by: mlaier
Approved by: mlaier
Pointy hat to:  mlaier :-)

18 years agoRemove the documentation from the Makefile as well.
George V. Neville-Neil [Sun, 14 May 2006 02:01:59 +0000 (02:01 +0000)]
Remove the documentation from the Makefile as well.

18 years agoRemoved the deprecated lance driver, lnc, from files.
George V. Neville-Neil [Sun, 14 May 2006 01:59:12 +0000 (01:59 +0000)]
Removed the deprecated lance driver, lnc, from files.

18 years agoRemove documentation for the deprecated lnc driver.
George V. Neville-Neil [Sun, 14 May 2006 01:55:20 +0000 (01:55 +0000)]
Remove documentation for the deprecated lnc driver.

Reviewed By: simon

18 years agoRemove autogeneration of lnc(4) hardware notes before gnn blows it away.
Simon L. B. Nielsen [Sun, 14 May 2006 01:55:06 +0000 (01:55 +0000)]
Remove autogeneration of lnc(4) hardware notes before gnn blows it away.

18 years agoFor nmount(), if "rw" is specified as a mount option,
Craig Rodrigues [Sun, 14 May 2006 01:51:38 +0000 (01:51 +0000)]
For nmount(), if "rw" is specified as a mount option,
add "noro" to the list of mount options.  This allows
a read-only mount to be converted to read-write via:
mount -u -o rw

Requested by: kris

18 years agoRemove the old, and now deprecated lnc driver. The Lance style hardware
George V. Neville-Neil [Sun, 14 May 2006 01:47:51 +0000 (01:47 +0000)]
Remove the old, and now deprecated lnc driver.  The Lance style hardware
is supported by the le and pnc drivers.

Reviewed by: jmg

18 years agoPrefer the le device driver for Lance (AMD7990 et al) hardware over the
George V. Neville-Neil [Sun, 14 May 2006 01:40:41 +0000 (01:40 +0000)]
Prefer the le device driver for Lance (AMD7990 et al) hardware over the
older, and less capable lnc driver.

Reviewed by: imp

18 years agoResurrect Skyeye support :
Olivier Houchard [Sat, 13 May 2006 23:41:16 +0000 (23:41 +0000)]
Resurrect Skyeye support :
Add a new option, SKYEYE_WORKAROUNDS, which as the name suggests adds
workarounds for things skyeye doesn't simulate. Specifically :
- Use USART0 instead of DBGU as the console, make it not use DMA, and           manually provoke an interrupt when we're done in the transmit function.
- Skyeye maintains an internal counter for clock, but apparently there's
no way to access it, so hack the timecounter code to return a value which
is increased at every clock interrupts. This is gross, but I didn't find a
better way to implement timecounters without hacking Skyeye to get the
counter value.
- Force the write-back of PTEs once we're done writing them, even if they
are supposed to be write-through. I don't know why I have to do that.

18 years agoAdd a manual dependency on ssh_namespace.h.
Dag-Erling Smørgrav [Sat, 13 May 2006 21:38:16 +0000 (21:38 +0000)]
Add a manual dependency on ssh_namespace.h.

Discussed with: ru

18 years agoExpunge traces of unlinked snapshot files when making a new snapshot.
Tor Egge [Sat, 13 May 2006 20:41:37 +0000 (20:41 +0000)]
Expunge traces of unlinked snapshot files when making a new snapshot.

18 years agoo Add an option allowing to start search from some offset instead of offset 0;
Maxim Sobolev [Sat, 13 May 2006 20:23:06 +0000 (20:23 +0000)]
o Add an option allowing to start search from some offset instead of offset 0;

o mesure distance between UFS1 and UFS2 superblocks independently.

MFC after: 1 week