]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoRemove the clause 3, per berkeley and mail from rwatson representing
Warner Losh [Mon, 10 Jan 2005 08:21:45 +0000 (08:21 +0000)]
Remove the clause 3, per berkeley and mail from rwatson representing
that nai was OK with all third clause removals.

19 years agoNE200 -> NE200T
Warner Losh [Mon, 10 Jan 2005 08:08:30 +0000 (08:08 +0000)]
NE200 -> NE200T

19 years agoTweaks to the name of this card, based on input from Sean Shapira.
Warner Losh [Mon, 10 Jan 2005 08:07:41 +0000 (08:07 +0000)]
Tweaks to the name of this card, based on input from Sean Shapira.

19 years agoRemove matcd. It isn't in the build, has issues and can be retired.
Warner Losh [Mon, 10 Jan 2005 08:00:14 +0000 (08:00 +0000)]
Remove matcd.  It isn't in the build, has issues and can be retired.

Submitted by: trhodes (the real one)
Approved by: core, mdodd, uhclem

19 years agosort more things alphabetically
Warner Losh [Mon, 10 Jan 2005 05:11:40 +0000 (05:11 +0000)]
sort more things alphabetically

19 years agolibarchive does much more than just tar ;-)
Tim Kientzle [Mon, 10 Jan 2005 05:11:15 +0000 (05:11 +0000)]
libarchive does much more than just tar ;-)

19 years agoSort entries.
Warner Losh [Mon, 10 Jan 2005 04:40:23 +0000 (04:40 +0000)]
Sort entries.
Remove a couple of 'card' lines that were somehow missed when OLDCARD was
desupported.

19 years agoSort entires better.
Warner Losh [Mon, 10 Jan 2005 04:38:58 +0000 (04:38 +0000)]
Sort entires better.

19 years agoAdd support for:
Warner Losh [Mon, 10 Jan 2005 03:48:17 +0000 (03:48 +0000)]
Add support for:
fe1: <EAGLE Technology NE200 ETHERNET LAN MBH10302  04>

As reported by Sean Shapira.  This appears to be working.  Eagle used
Fujitsu's vendor number, with a product number of 4 (which is the same
as the vendor number, which is a little suspect).  Since there's no
apparent conflict, go ahead and use it.

Submitted by: Sean Shapira

19 years agoAdd Eagle NE200 Ethernet LAN MBH10302 card. This appears to be a fe
Warner Losh [Mon, 10 Jan 2005 03:45:21 +0000 (03:45 +0000)]
Add Eagle NE200 Ethernet LAN MBH10302 card.  This appears to be a fe
card, and works with that driver.  However, Eagle is using Fujitsu's
vendor number and a product code of 4, which seems a little odd.
Still, there's no conflicts...

19 years agoFree the shared devq last since CAM expects it to be there if
Doug White [Mon, 10 Jan 2005 02:34:26 +0000 (02:34 +0000)]
Free the shared devq last since CAM expects it to be there if
xpt_alloc_device() gets called, which can happen during detach in
certain situations. Fixes module unload.

MFC after: 3 days

19 years agoWording nit.
Tom Rhodes [Mon, 10 Jan 2005 00:35:54 +0000 (00:35 +0000)]
Wording nit.

19 years agofix a "little-endian-big-endian confusion that luckily:
Julian Elischer [Sun, 9 Jan 2005 23:49:45 +0000 (23:49 +0000)]
fix a "little-endian-big-endian confusion that luckily:
1/ doesn't matter on most of our architectures
2/ will never happen unless we start queueing multiple trasactions
to a single endpoint at one time (which we do not allow yet).
If anyone has a big_endian machine with EHCI they might check this
if they are having problems with EHCI but it's unlikely even there..

Submitted by: Hans Petter Selasky <hselasky@c2i.net>
MFC after: 3 days

19 years agoCosmetic typo in check_pidfile()
Giorgos Keramidas [Sun, 9 Jan 2005 23:46:37 +0000 (23:46 +0000)]
Cosmetic typo in check_pidfile()

PR: bin/75946
Submitted by: zero@gddn.org (Finn)

19 years agoOnly update the boot sector when there is a valid drive number provided.
Peter Edwards [Sun, 9 Jan 2005 23:30:35 +0000 (23:30 +0000)]
Only update the boot sector when there is a valid drive number provided.
(After squeezing a few more bytes out of boot0)

Discussed With: jhb, julian
PR: 66248
Submitted By: Hans Petter Selasky
MFC After: 1 week

19 years agoFix typos and add .Pp after the end of a display to separate the
Giorgos Keramidas [Sun, 9 Jan 2005 22:49:24 +0000 (22:49 +0000)]
Fix typos and add .Pp after the end of a display to separate the
displayed text from the paragraph right after it.

19 years agoIn my last commit, I'd assumed that LINE30 was always defined. It
Warner Losh [Sun, 9 Jan 2005 22:17:39 +0000 (22:17 +0000)]
In my last commit, I'd assumed that LINE30 was always defined.  It
turns out that LINE30_ROW was always defined, not LINE30.  I confused
this for LINE30 and did the unifdef -DLINE30 using that mistaken
belief.  This corrects that problem.

Submitted by: nyan-san

19 years agoRemove the meteor helper files, imp removed the driver and manual page
Tom Rhodes [Sun, 9 Jan 2005 21:21:25 +0000 (21:21 +0000)]
Remove the meteor helper files, imp removed the driver and manual page
over a year ago.

19 years agocorrect direction for bus_dma sync of rx buffer
Sam Leffler [Sun, 9 Jan 2005 19:57:55 +0000 (19:57 +0000)]
correct direction for bus_dma sync of rx buffer

Submitted by: Tai-hwa Liang

19 years agoScheduled mdoc(7) sweep.
Ruslan Ermilov [Sun, 9 Jan 2005 19:54:23 +0000 (19:54 +0000)]
Scheduled mdoc(7) sweep.

19 years agoDon't panic when we're asked to allocate a resource type that we know
Marcel Moolenaar [Sun, 9 Jan 2005 18:58:08 +0000 (18:58 +0000)]
Don't panic when we're asked to allocate a resource type that we know
won't exist for EBus. Just fail the allocation by returning NULL.
Now drivers that are MI can try resources that the driver knows may
be used by the device.

19 years agoUse uint32_t instead of u_long when appropriate.
Olivier Houchard [Sun, 9 Jan 2005 17:42:03 +0000 (17:42 +0000)]
Use uint32_t instead of u_long when appropriate.
That's enough to make my bt848 work on amd64.

MFC after: 1 week

19 years agoIncrease default synchronization speed.
Pawel Jakub Dawidek [Sun, 9 Jan 2005 14:43:39 +0000 (14:43 +0000)]
Increase default synchronization speed.

MFC after: 3 days

19 years agoRevert previous revision. Including bsd.compat.mk is necessary
Ruslan Ermilov [Sun, 9 Jan 2005 12:55:47 +0000 (12:55 +0000)]
Revert previous revision.  Including bsd.compat.mk is necessary
both from sys.mk (to handle old knobs set in /etc/make.conf),
and from here (to handle old knobs set in individual makefiles).

19 years agoo Make telnet[d] -S (IP TOS) flag really work. We do not have
Maxim Konovalov [Sun, 9 Jan 2005 10:24:46 +0000 (10:24 +0000)]
o Make telnet[d] -S (IP TOS) flag really work.  We do not have
/etc/iptos implementation so only numeric values supported.

o telnetd.8: steal the -S flag description from telnet.1, bump
the date of the document.

MFC after: 6 weeks

19 years agoFix build on all (hopefully) 64 bit architectures.
Tom Rhodes [Sun, 9 Jan 2005 09:10:24 +0000 (09:10 +0000)]
Fix build on all (hopefully) 64 bit architectures.

Noticed by: tinderbox
Tested on: panther

19 years agoMark the IPX netisr as MPSAFE so that inbound IPX traffic is processed
Robert Watson [Sun, 9 Jan 2005 07:34:55 +0000 (07:34 +0000)]
Mark the IPX netisr as MPSAFE so that inbound IPX traffic is processed
without Giant, and can be directly dispatched in the ithread when
net.isr.enable is turned on.

MFC after: 4 weeks

19 years agoRecent changes have locked down most of the highly dynamic data
Robert Watson [Sun, 9 Jan 2005 05:34:37 +0000 (05:34 +0000)]
Recent changes have locked down most of the highly dynamic data
structures in IPX/SPX -- primarily, sequence numbering, PCB lists,
and PCBs for IPX raw sockets, IPX datagram sockets, and IPX/SPX.
As such, remove remove NET_NEEDS_GIANT() for IPX, and remove the
assertion of Giant in the ipxintr() IPX input path.

Note that IPX/SPX is not fully MPSAFE, and that there are some
problems with IPX/SPX locking that will require some further work.
However, it is now safe enough to run in general without the Giant
lock.

MFC after: 4 weeks

19 years agoUse the IPX PCB list mutex and IPX PCB mutexes to lock down the SPX
Robert Watson [Sun, 9 Jan 2005 05:31:16 +0000 (05:31 +0000)]
Use the IPX PCB list mutex and IPX PCB mutexes to lock down the SPX
portion of IPX/SPX:

- Protect IPX PCB lists with the IPX PCB list mutex, in particular
  when calling PCB and PCB list manipulation routines in ipx_pcb.c.
- Protect both IPX PCB state and SPX PCB state using the IPX PCB
  mutex.
- Generally annotate locking, as well as adding liberal use of lock
  assertions to document locking requirements.
- Where possible, use unlocked reads when reading integer or smaller
  sized socket options on SPX sockets.
- De-spl throughout.

Notes:

- spx_input() expects both the list mutex and PCB mutex to be held
  on entry, but will release both on return.  Because sonewconn() is
  called from spx_input(), it may actually drop one PCB lock and
  acquire another during generation of a new connection, meaning the
  caller is not in a position to unlock the PCB mutex.

MFC after: 3 weeks

19 years agoClean up return handling for a number of SPX-related routines that
Robert Watson [Sun, 9 Jan 2005 05:25:02 +0000 (05:25 +0000)]
Clean up return handling for a number of SPX-related routines that
were derived from more complex TCP versions of the same:

- spx_close(), spx_disconnect(), spx_drop(), and spx_usrclosed() all
  always free's the spxpcb invalidating the argument, so a return
  value is not required to indicate if it has.
- Annotate that the cb arguments to each of these functions is
  invalidated via a comment.
- When tearing down a pcb due to sonewconn() having failed, mark the
  cb as NULL; later, when deciding whether to store trace information
  due to SO_DEBUG, check that cb is not NULL before dereferencing or
  a NULL pointer dereference may occur.

MFC after: 3 weeks

19 years agoProtect ipx_pexseq with the IPX PCB list mutex.
Robert Watson [Sun, 9 Jan 2005 05:15:59 +0000 (05:15 +0000)]
Protect ipx_pexseq with the IPX PCB list mutex.

When processing socket options against IPX PCBs, generally protect
PCB fields using the IPX PCB mutex.  Where possible, use unlocked
reads on integer values to avoid locking overhead.

MFC after: 3 weeks

19 years agoAcquire or assert the IPX PCB list lock or IPX PCB lock during various
Robert Watson [Sun, 9 Jan 2005 05:13:14 +0000 (05:13 +0000)]
Acquire or assert the IPX PCB list lock or IPX PCB lock during various
protocol methods relating to IPX.  Conditionally acquire the PCB list
lock in the send operation only if the socket requires binding in order
to use the requested address.

Remove spl's generally no longer required during these accesses.

MFC after: 3 weeks

19 years agoAssert or acquire the IPX PCB list lock or IPX PCB locks throughout
Robert Watson [Sun, 9 Jan 2005 05:10:43 +0000 (05:10 +0000)]
Assert or acquire the IPX PCB list lock or IPX PCB locks throughout
the IPX-related PCB routines.  In general, the list lock is required
to iterate the PCB list, either for read or write; the PCB lock is
required to access or modify a PCB.  To change the binding of a PCB,
both locks must be held.

MFC after: 3 weeks

19 years agoHold the IPX PCB mutex around calls to ipx_input() in the IPX input
Robert Watson [Sun, 9 Jan 2005 05:08:47 +0000 (05:08 +0000)]
Hold the IPX PCB mutex around calls to ipx_input() in the IPX input
path.

MFC after: 3 weeks

19 years agoHold the global IPX PCB list mutex in the IPX input path when walking
Robert Watson [Sun, 9 Jan 2005 05:06:19 +0000 (05:06 +0000)]
Hold the global IPX PCB list mutex in the IPX input path when walking
the IPX PCB list.

MFC after: 3 weeks

19 years agoIntroduce a global mutex, ipxpcb_list_mtx, to protect the global
Robert Watson [Sun, 9 Jan 2005 05:00:41 +0000 (05:00 +0000)]
Introduce a global mutex, ipxpcb_list_mtx, to protect the global
IPX PCB lists.  Add macros to initialize, destroy, lock, unlock,
and assert the mutex.  Initialize the mutex when IPX is started.

Add per-IPX PCB mutexes, ipxp_mtx in struct ipxpcb, to protect
per-PCB IPX/SPX state.  Add macros to initialize, destroy, lock,
unlock, and assert the mutex.  Initialize the mutex when a new
PCB is allocated; destroy it when the PCB is free'd.

MFC after: 2 weeks

19 years agoIn ipx_setsockaddr(), use M_WAITOK instead of M_NOWAIT so that the
Robert Watson [Sun, 9 Jan 2005 04:47:42 +0000 (04:47 +0000)]
In ipx_setsockaddr(), use M_WAITOK instead of M_NOWAIT so that the
call always succeeds, avoiding causing the caller to return success
even though the returned *sockaddr is NULL.

MFC after: 2 weeks

19 years agoEliminate jump to 'bad' label in order to clean up the ipx_input()
Robert Watson [Sun, 9 Jan 2005 04:39:16 +0000 (04:39 +0000)]
Eliminate jump to 'bad' label in order to clean up the ipx_input()
return/unwind path for locking work.

MFC after: 2 weeks

19 years agoImplement rpmatch(), a semi-standard interface (as found on AIX, Tru64,
Tim J. Robbins [Sun, 9 Jan 2005 03:55:13 +0000 (03:55 +0000)]
Implement rpmatch(), a semi-standard interface (as found on AIX, Tru64,
GNU) for determining whether a string is an affirmative or negative
response to a question according to the current locale. This is done
by matching the response against nl_langinfo(3) items YESEXPR and NOEXPR.

19 years agoMove the acquisition and release of the page queues lock outside of a loop
Alan Cox [Sat, 8 Jan 2005 23:41:11 +0000 (23:41 +0000)]
Move the acquisition and release of the page queues lock outside of a loop
in vm_object_split() to avoid repeated acquisition and release.

19 years agoUse static const char and bump WARNS.
Tom Rhodes [Sat, 8 Jan 2005 22:56:08 +0000 (22:56 +0000)]
Use static const char and bump WARNS.

19 years agoRemove support for FreeBSD < 4.recent from this driver.
Warner Losh [Sat, 8 Jan 2005 22:52:00 +0000 (22:52 +0000)]
Remove support for FreeBSD < 4.recent from this driver.

19 years agoRemove unused variable.
Tom Rhodes [Sat, 8 Jan 2005 22:47:10 +0000 (22:47 +0000)]
Remove unused variable.

19 years agoRemove vnode.h and adjust includes to compensate for pollution.
Warner Losh [Sat, 8 Jan 2005 22:43:33 +0000 (22:43 +0000)]
Remove vnode.h and adjust includes to compensate for pollution.

19 years agoLINE30 is always defined now, so unifdef -DLINE30 for clarity.
Warner Losh [Sat, 8 Jan 2005 22:29:53 +0000 (22:29 +0000)]
LINE30 is always defined now, so unifdef -DLINE30 for clarity.

19 years agoMerge module.h into 30line.h and remove it. It lacked a
Warner Losh [Sat, 8 Jan 2005 22:29:23 +0000 (22:29 +0000)]
Merge module.h into 30line.h and remove it. It lacked a
copyright/license header and was only used by 30line.h.  It appears
that the copyright/license in 30line.h covers the old contents
module.h anyway, so this simplifies things a little while cleaning up
one obscure potential license confusion...

Revired by: nyan-san

19 years agoRecognize and parse symlinks in ISO9660 CDROM images with Rockridge extensions.
Tim Kientzle [Sat, 8 Jan 2005 19:56:07 +0000 (19:56 +0000)]
Recognize and parse symlinks in ISO9660 CDROM images with Rockridge extensions.

19 years agoDocumentation improvements. In particular,
Tim Kientzle [Sat, 8 Jan 2005 19:51:04 +0000 (19:51 +0000)]
Documentation improvements.  In particular,
expand and clarify the description of the client
callback functions and how they should handle errors.

Thanks to: Antony Dovgal

19 years agoCorrectly document the order of arguments
Tim Kientzle [Sat, 8 Jan 2005 18:32:49 +0000 (18:32 +0000)]
Correctly document the order of arguments
to archive_read_open and archive_write_open.

Thanks to: Antony Dovgal

19 years agoClear the error buffer on entry
Tim Kientzle [Sat, 8 Jan 2005 18:28:11 +0000 (18:28 +0000)]
Clear the error buffer on entry
to archive_read_next_header so the
next error doesn't just get appended.

MFC after: 7 days

19 years agoCast to intmax_t when using %jd format.
Pawel Jakub Dawidek [Sat, 8 Jan 2005 17:19:56 +0000 (17:19 +0000)]
Cast to intmax_t when using %jd format.

MFC after: 3 days

19 years agoMore fixes to the copyright notice.
Jeffrey Hsu [Sat, 8 Jan 2005 17:16:43 +0000 (17:16 +0000)]
More fixes to the copyright notice.

Submitted by: Ken Smith <kensmith@cse.Buffalo.EDU>

19 years agoCleanup usr.sbin/fd* so they can compile under WARNS=6.
Xin LI [Sat, 8 Jan 2005 15:46:06 +0000 (15:46 +0000)]
Cleanup usr.sbin/fd* so they can compile under WARNS=6.

fdcontrol/fdcontrol.c:
- Add const constraint to an intermediate value
  which is not supposed to be changed elsewhere.
fdread/fdread.c:
- Use _devname in favor of devname to avoid name
  conflicit.
- -1 is less than any positive number so in order
  to get the block to function, we should get the
  block a little earlier.
- Cast to remove signed when we are sure that a
  return value is positive, or is compared with
  an positive number (tracknumber of a floppy
  disk is not likely to have UINT_MAX/2 anyway)
fdread/fdutil.c:
- Use more specific initializer
fdwrite/fdwrite.c:
- Use static on format_track since it's not
  referenced in other places.
- Use const char* to represent string constant.

Bump WARNS accordingly.

19 years agoFollow style.Makefile(5):
Xin LI [Sat, 8 Jan 2005 15:15:42 +0000 (15:15 +0000)]
Follow style.Makefile(5):
WARNS comes before CFLAGS

This reduces diff against my local branch.

19 years agoThis change adds reliability for Ethernet trunks built with ng_one2many:
Gleb Smirnoff [Sat, 8 Jan 2005 12:42:03 +0000 (12:42 +0000)]
This change adds reliability for Ethernet trunks built with ng_one2many:

- Introduce another ng_ether(4) callback ng_ether_link_state_p, which
  is called from if_link_state_change(), every time link is changed.
- In ng_ether_link_state() send netgraph control message notifying
  of link state change to a node connected to "lower" hook.

Reviewed by: sam
MFC after: 2 weeks

19 years agoFix typo, function name pthread_rwlock_timedrwlock should be
David Xu [Sat, 8 Jan 2005 11:07:13 +0000 (11:07 +0000)]
Fix typo, function name pthread_rwlock_timedrwlock should be
pthread_rwlock_timedwrlock.

19 years ago[NO]MAN -> NO_[MAN].
Tom Rhodes [Sat, 8 Jan 2005 09:21:35 +0000 (09:21 +0000)]
[NO]MAN -> NO_[MAN].
Bump doc date.

19 years agoIn total violation of at least 4 sections in the ACPI spec, some systems
Nate Lawson [Sat, 8 Jan 2005 09:10:20 +0000 (09:10 +0000)]
In total violation of at least 4 sections in the ACPI spec, some systems
place device objects in \ (in this case, PCI links.)  Work around this by
starting our probe from \.  To avoid attaching system scope objects,
explicitly skip them.  (I think it's an ACPI-CA bug that \_SB and \_TZ have
device and thermal object types.)  Thanks to pjd@ for testing.

MFC after: 2 weeks

19 years agoBreak out of loop earlier if it is not timeout.
David Xu [Sat, 8 Jan 2005 06:57:46 +0000 (06:57 +0000)]
Break out of loop earlier if it is not timeout.

19 years agoChange the USB keyboard example to use /dev/ukbd0 instead of /dev/kbd1.
Brooks Davis [Sat, 8 Jan 2005 06:02:56 +0000 (06:02 +0000)]
Change the USB keyboard example to use /dev/ukbd0 instead of /dev/kbd1.

This example caused me to incorrectly believe that you must use the
generic device nodes when you can in fact use either.  It's often better
to use the driver specific node.

19 years agoWhen ukbd0 arrives, attach to /dev/ukbd0 rather then /dev/kbd1 since
Brooks Davis [Sat, 8 Jan 2005 06:00:24 +0000 (06:00 +0000)]
When ukbd0 arrives, attach to /dev/ukbd0 rather then /dev/kbd1 since
kbd1 might be something else.

19 years agoAnnotate that pfs_exit() always acquires and releases two mutexes for
Robert Watson [Sat, 8 Jan 2005 04:56:38 +0000 (04:56 +0000)]
Annotate that pfs_exit() always acquires and releases two mutexes for
every process exist, even if procfs isn't mounted.  And one of those
mutexes is Giant.  No immediate thoughts on fixing this.

19 years agoIn acct_process(), do a lockless read of acctvp to see if it's NULL
Robert Watson [Sat, 8 Jan 2005 04:45:57 +0000 (04:45 +0000)]
In acct_process(), do a lockless read of acctvp to see if it's NULL
before deciding to do more expensive locking to account for process
exit.  This acceptable minor race avoids two mutex operations in
that highly common case of accounting not being enabled.

MFC after: 2 weeks

19 years agoElminate 1 LOR (actually a recursive mutex grab) involving ipfilter where
Darren Reed [Sat, 8 Jan 2005 04:32:02 +0000 (04:32 +0000)]
Elminate 1 LOR (actually a recursive mutex grab) involving ipfilter where
we loop through all the list of NICs (struct ifnet), holding the lock on
it and then do a name lookup with ifunit() whilst holding it.

19 years agoIn kern_wait(), let the compiler copy the rusage structure rather than
Robert Watson [Sat, 8 Jan 2005 04:17:48 +0000 (04:17 +0000)]
In kern_wait(), let the compiler copy the rusage structure rather than
an explicit bcopy() -- it probably does a better job.

19 years agoRemove the LSI MegaRAID SATA 150-2, this is a SiL 3112 based software RAID
Christian Brueffer [Sat, 8 Jan 2005 03:07:49 +0000 (03:07 +0000)]
Remove the LSI MegaRAID SATA 150-2, this is a SiL 3112 based software RAID
device.

Submitted by: Martin Nilsson <martin@gneto.com>
MFC after: 3 days

19 years agoFix copyright notice.
Jeffrey Hsu [Sat, 8 Jan 2005 01:59:36 +0000 (01:59 +0000)]
Fix copyright notice.

19 years agoRevert local experiment which leaked into commit.
Poul-Henning Kamp [Fri, 7 Jan 2005 23:54:16 +0000 (23:54 +0000)]
Revert local experiment which leaked into commit.

19 years agoFix compilation of DEVICE_POLLING code.
Poul-Henning Kamp [Fri, 7 Jan 2005 23:52:41 +0000 (23:52 +0000)]
Fix compilation of DEVICE_POLLING code.

19 years agoDocument slot specific location for base I/O board on AS600.
Wilko Bulte [Fri, 7 Jan 2005 21:05:45 +0000 (21:05 +0000)]
Document slot specific location for base I/O board on AS600.

Thanks to: Charles Smeijer for getting me my AS600.

19 years agoReorganize the DESCRIPTION section, using .Bl lists to separate the
Giorgos Keramidas [Fri, 7 Jan 2005 19:41:00 +0000 (19:41 +0000)]
Reorganize the DESCRIPTION section, using .Bl lists to separate the
descriptions of items from each other and have related things appear
in the same nesting 'level'.

The .Fn function/macro, and bump document date.

Reviewed by: jkoshy

19 years agoNote that fdisk is not used on Alpha.
Wilko Bulte [Fri, 7 Jan 2005 19:32:32 +0000 (19:32 +0000)]
Note that fdisk is not used on Alpha.
Remove outdated info on install media.

Submitted by: "William H. Magill" <magill@mcgillsociety.org>

19 years agoComment typo.
Giorgos Keramidas [Fri, 7 Jan 2005 19:31:43 +0000 (19:31 +0000)]
Comment typo.

PR: kern/75923
Submitted by: Matthew D. Fuller <fullermd@over-yonder.net>

19 years agoFix support for machines with default MP Table configurations:
John Baldwin [Fri, 7 Jan 2005 18:42:59 +0000 (18:42 +0000)]
Fix support for machines with default MP Table configurations:
- Fix the MP Table pci bridge drivers to not probe the configuration table
  unless we actually have one.  Machines using a default configuration do
  not have such a table.
- Only allow default configuration types of 5 (ISA + PCI) and 6 (EISA +
  PCI) as the others are not likely to work.  Types 1 through 4 use an
  external APIC (probably with 80486 processors) which we certainly do not
  support, and type 7 uses an MCA bus which has not been tested with the
  new MP Table code.
- Correct the fact that the single I/O APIC in a default configuration has
  an ID of 2, not 0.
- Fix off by one errors in setting the bus types from the default_data[]
  arrays for default configurations.
- Explicitly configure each of the 16 interrupt pins on the sole I/O APIC
  when using a default configuration.  This is especially helpful for type
  6 (EISA + PCI) since the EISA interrupts need to have their polarity
  programmed based on the values in the ELCR.

Much thanks to the submitter and tester who endured several rounds of
testing to get this fixed.

MFC after: 1 week
Tested by: Georg Schwarz georg dot schwarz at freenet dot de

19 years agoFinish src part VFS_INIT.9 removal. Need to check www/.
Maxim Konovalov [Fri, 7 Jan 2005 15:38:21 +0000 (15:38 +0000)]
Finish src part VFS_INIT.9 removal.  Need to check www/.

19 years agoThe `timep' argument of utime() is not an array of structs.
Giorgos Keramidas [Fri, 7 Jan 2005 14:38:29 +0000 (14:38 +0000)]
The `timep' argument of utime() is not an array of structs.
Make sure we don't confuse the reader by claiming it is.

PR: docs/75615
Submitted by: Nobuyuki Koganemaru <n-kogane@syd.odn.ne.jp>

19 years agoThe .Fn macro.
Giorgos Keramidas [Fri, 7 Jan 2005 14:10:59 +0000 (14:10 +0000)]
The .Fn macro.

19 years agoThe .Fn macro and addition of .Fa in the tag width of the
Giorgos Keramidas [Fri, 7 Jan 2005 14:01:08 +0000 (14:01 +0000)]
The .Fn macro and addition of .Fa in the tag width of the
argument description list.

19 years ago- The .Fn macro and description of the `struct statfs' type of `sbp'.
Giorgos Keramidas [Fri, 7 Jan 2005 12:56:04 +0000 (12:56 +0000)]
- The .Fn macro and description of the `struct statfs' type of `sbp'.
- List of struct statfs fields that are related to file systems, much
  like the list in stat.2 for the userlevel `struct stat'.
- Bump document date.

19 years agoFix the derivation of the GEOM name from the specified device name by
Ralf S. Engelschall [Fri, 7 Jan 2005 12:19:57 +0000 (12:19 +0000)]
Fix the derivation of the GEOM name from the specified device name by
complementing the existing special case of a not existing /dev prefix
with the recognition of an already existing /dev prefix.

This implicitly solves the following two issues related to working on
GEOM devices /dev/foo/bar (which have the GEOM provider name "foo/bar")
with the expected commands like "bsdlabel /dev/foo/bar":

1. the error "Geom not found" when trying to write or edit the BSD
   label (because previously the incorrect GEOM name "bar" instead of
   "foo/bar" was derived from "/dev/foo/bar").

2. the multiple times reported "magically introduced" partition offset
   of 63 blocks and the resulting errors like "partition extends past
   end of unit" and "partition c doesn't start at 0!".

   This implicitly resulted because bsdlabel(8) determines the "MBR
   offset" via GEOM and (intentionally) silently falls back to an offset
   of 0 if it could not be queried (which is the case if the name was
   incorrectly derived).

   Usually (at least on PCs) the offset for the first slice is 63 blocks
   and bsdlabel(8) automatically subtracts them from the absolute
   offsets in the read on-disk BSD label, resulting in the display of an
   effective offset of 0. If the GEOM query fails, the assumed offset of
   0 is subtracted and an incorrect effective offset of 63 is displayed
   and tried to be worked upon.

Reviewed by: pjd
MFC after: 1 week

19 years agoThe .Fn macro, document `flags' argument & bump document date.
Giorgos Keramidas [Fri, 7 Jan 2005 12:14:43 +0000 (12:14 +0000)]
The .Fn macro, document `flags' argument & bump document date.

19 years agoWARNS=6 cleanup. This includes:
Xin LI [Fri, 7 Jan 2005 12:06:30 +0000 (12:06 +0000)]
WARNS=6 cleanup.  This includes:
- Apply __unused on unused parameters
- Use const where suitable
- Use PRIu64 instead of the deprecated %q
- Bump WARNS to 6

19 years agoTrim excessively verbose sentence.
Giorgos Keramidas [Fri, 7 Jan 2005 12:01:11 +0000 (12:01 +0000)]
Trim excessively verbose sentence.

19 years agoThe .Fn macro
Giorgos Keramidas [Fri, 7 Jan 2005 11:45:59 +0000 (11:45 +0000)]
The .Fn macro

19 years agoMove the description of MNT_FORCE closer to mntflags and use a .Bl
Giorgos Keramidas [Fri, 7 Jan 2005 11:45:11 +0000 (11:45 +0000)]
Move the description of MNT_FORCE closer to mntflags and use a .Bl
list for the valid flag values.  This way, if VFS_UNMOUNT(9) supports
more flags in the future, adding a single list item is going to be
easy and all the flags are going to be in one place.

19 years agoThe .Fn macro.
Giorgos Keramidas [Fri, 7 Jan 2005 11:39:36 +0000 (11:39 +0000)]
The .Fn macro.

19 years agoThere is no VFS_INIT(9).
Giorgos Keramidas [Fri, 7 Jan 2005 11:34:05 +0000 (11:34 +0000)]
There is no VFS_INIT(9).

Approved by: phk

19 years agoThe VFS_MOUNT() macro expects only 2 arguments in the nmount world
Giorgos Keramidas [Fri, 7 Jan 2005 10:57:33 +0000 (10:57 +0000)]
The VFS_MOUNT() macro expects only 2 arguments in the nmount world
(since revision 1.173 of src/sys/sys/mount.h).

Update its prototype, the argument list description and bump the date.

19 years agoLink a couple of missing manpages
Guido van Rooij [Fri, 7 Jan 2005 09:22:02 +0000 (09:22 +0000)]
Link a couple of missing manpages

Submitted by: Hideyuki KURASHINA <rushani@FreeBSD.org>
MFC after: 1 week

19 years agoUse a .Bl -tag list instead of plain .Pp paragraphs for describing the
Giorgos Keramidas [Fri, 7 Jan 2005 09:02:40 +0000 (09:02 +0000)]
Use a .Bl -tag list instead of plain .Pp paragraphs for describing the
macro arguments expected by SYSCALL_MODULE().

MFC after: 3 days

19 years agoFix time-travel bug.
Giorgos Keramidas [Fri, 7 Jan 2005 09:00:19 +0000 (09:00 +0000)]
Fix time-travel bug.

Noticed by: krion

19 years agoAvoid overusing "these" by breaking a long sentence in 2 smaller ones.
Giorgos Keramidas [Fri, 7 Jan 2005 08:51:10 +0000 (08:51 +0000)]
Avoid overusing "these" by breaking a long sentence in 2 smaller ones.

MFC after: 3 days

19 years agoConvert the descriptions of module versions to an explicit list and
Giorgos Keramidas [Fri, 7 Jan 2005 08:44:31 +0000 (08:44 +0000)]
Convert the descriptions of module versions to an explicit list and
note that these versions numbers refer to moddepend, not the current
module.

MFC after: 3 days

19 years agoSyntax nit.
Giorgos Keramidas [Fri, 7 Jan 2005 08:31:09 +0000 (08:31 +0000)]
Syntax nit.

MFC after: 3 days

19 years agoIntroduce bus_dmamap_load_mbuf_sg(). Instead of taking a callback arg, this
Scott Long [Fri, 7 Jan 2005 07:57:18 +0000 (07:57 +0000)]
Introduce bus_dmamap_load_mbuf_sg().  Instead of taking a callback arg, this
cuts to the chase and fills in a provided s/g list.  This is meant to optimize
out the cost of the callback since the callback doesn't serve much purpose for
mbufs since mbuf loads will never be deferred.  This is just for amd64 and
i386 at the moment, other arches will be coming shortly.

19 years agoFix typo from previous commit.
Scott Long [Fri, 7 Jan 2005 07:08:33 +0000 (07:08 +0000)]
Fix typo from previous commit.

19 years agoUpdate manual page after adding 'fts_bignum' field.
Pawel Jakub Dawidek [Fri, 7 Jan 2005 07:03:19 +0000 (07:03 +0000)]
Update manual page after adding 'fts_bignum' field.

Submitted by: Peter Jeremy <PeterJeremy@optushome.com.au>
MFC after: 5 days

19 years agoAdd a default driver to attach to the "hidden" scsi channels of the Dell
Scott Long [Fri, 7 Jan 2005 05:59:45 +0000 (05:59 +0000)]
Add a default driver to attach to the "hidden" scsi channels of the Dell
PERC 3 controllers.  This is needed to keep the PM code from powering them
down.

19 years agoTransfer responsibility for freeing the page taken from the cache
Alan Cox [Fri, 7 Jan 2005 05:02:19 +0000 (05:02 +0000)]
Transfer responsibility for freeing the page taken from the cache
queue and (possibly) unlocking the containing object from
vm_page_alloc() to vm_page_select_cache().  Recent optimizations to
vm_map_pmap_enter() (see vm_map.c revisions 1.362 and 1.363) and
pmap_enter_quick() have resulted in panic()s because vm_page_alloc()
mistakenly unlocked objects that had not been locked by
vm_page_select_cache().

Reported by: Peter Holm and Kris Kennaway

19 years agoFix typos from previous commit.
Scott Long [Fri, 7 Jan 2005 05:01:24 +0000 (05:01 +0000)]
Fix typos from previous commit.