dragonfly.git
17 years agoUse a new pkgsrc bootstrap to work with recent tar changes.
Simon Schubert [Wed, 7 Feb 2007 16:35:25 +0000 (16:35 +0000)]
Use a new pkgsrc bootstrap to work with recent tar changes.

17 years agoDo not create /var/db/pkg, pkgsrc bootstrap doesn't like it.
Simon Schubert [Wed, 7 Feb 2007 16:34:32 +0000 (16:34 +0000)]
Do not create /var/db/pkg, pkgsrc bootstrap doesn't like it.

17 years agoConvert RX desc's rssi field to relative RX signal strength (relative to
Sepherosa Ziehau [Wed, 7 Feb 2007 14:52:42 +0000 (14:52 +0000)]
Convert RX desc's rssi field to relative RX signal strength (relative to
noise floor).

The conversion for Maxim, Philips and GCT RF are adapted from Realtek's
818x Linux driver.

17 years agoFor rum(4) and rt2661 part of ral(4):
Sepherosa Ziehau [Wed, 7 Feb 2007 12:34:26 +0000 (12:34 +0000)]
For rum(4) and rt2661 part of ral(4):
Return -1 in {rum,rt2661}_get_rssi(), if rssi can't be converted to relative
RX signal strength.  Caller of these two functions will detect the conversion
error and use old rssi to feed ieee80211_input(), so when rssi smoothing is
in place, rssi conversion error will have no impacts on average rssi.

17 years agoMisc cleanups for NATA ataraid, and write ata_raid_dump() to look like
Thomas E. Spanjaard [Tue, 6 Feb 2007 15:17:44 +0000 (15:17 +0000)]
Misc cleanups for NATA ataraid, and write ata_raid_dump() to look like
something that will actually work. I did not take the same approach as with
ad_dump() here, because that takes just too much code duplication between
ata_raid_strategy() and ata_raid_dump().

17 years agoAbsolute RX siganl strength correction value from EEPROM should be within
Sepherosa Ziehau [Tue, 6 Feb 2007 14:40:32 +0000 (14:40 +0000)]
Absolute RX siganl strength correction value from EEPROM should be within
[-10, 10], if it is out of range, reset it to zero.

This range is referenced from Linux RT61 driver.

17 years agoCorrectly calculate absolute RX signal strength and use relative RX signal
Sepherosa Ziehau [Tue, 6 Feb 2007 14:33:39 +0000 (14:33 +0000)]
Correctly calculate absolute RX signal strength and use relative RX signal
strength (relative to noise floor) as RSSI.  This makes the signal of node
reported by ifconfig(8) consistent with reality, i.e. weaker signal, lower
RSSI.

Calculation of absolute RX signal strength is adapted from Linux RT73 driver.

17 years agoUse relative RX signal strength (relative to noise floor) as RSSI.
Sepherosa Ziehau [Tue, 6 Feb 2007 13:50:25 +0000 (13:50 +0000)]
Use relative RX signal strength (relative to noise floor) as RSSI.
RSSI correction value is obtained from Linux RT25USB driver.

17 years agoFor rt2661 part:
Sepherosa Ziehau [Tue, 6 Feb 2007 12:46:09 +0000 (12:46 +0000)]
For rt2661 part:
Use relative RX signal strength (relative to noise floor) as RSSI.

17 years agoUse relative RX signal strength (relative to noise floor) as RSSI.
Sepherosa Ziehau [Tue, 6 Feb 2007 12:38:30 +0000 (12:38 +0000)]
Use relative RX signal strength (relative to noise floor) as RSSI.

17 years agoleftover from 1:1 Userland threading stage 2.11/4
YONETANI Tomokazu [Tue, 6 Feb 2007 05:56:03 +0000 (05:56 +0000)]
leftover from 1:1 Userland threading stage 2.11/4

17 years agoAdd flag to optionally disable stripping of old kernel and modules.
Simon Schubert [Mon, 5 Feb 2007 22:19:28 +0000 (22:19 +0000)]
Add flag to optionally disable stripping of old kernel and modules.

Submitted-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk>
17 years agoMake natadisk's ad_dump() actually work. Ditch the fake bio/buf scheme to
Thomas E. Spanjaard [Mon, 5 Feb 2007 18:10:14 +0000 (18:10 +0000)]
Make natadisk's ad_dump() actually work. Ditch the fake bio/buf scheme to
pass information to ad_strategy() in favour of calling ata_queue_request()
directly. This way, we avoid hairy bio stuff and request callbacks.

17 years agoConvert RSSI to receive signal strength (dbm). The real convertion is not yet
Sepherosa Ziehau [Mon, 5 Feb 2007 15:19:04 +0000 (15:19 +0000)]
Convert RSSI to receive signal strength (dbm).  The real convertion is not yet
hooked, but it will be used when roaming/background scan is imported.

EEPROM offset of RSSI mapping value, default RSSI mapping value and noise floor
are obtained from Linux RT2500 driver.

17 years agoNuke unused header files
Sepherosa Ziehau [Mon, 5 Feb 2007 10:53:18 +0000 (10:53 +0000)]
Nuke unused header files

17 years agoDon't hand-roll an own version on selrecord. This unbreaks LINT build.
Simon Schubert [Mon, 5 Feb 2007 09:38:19 +0000 (09:38 +0000)]
Don't hand-roll an own version on selrecord.  This unbreaks LINT build.

Reported-by: swildner@
17 years agoRemove unnecessary CFLAGS and SRCS lines.
Peter Avalos [Sun, 4 Feb 2007 21:17:34 +0000 (21:17 +0000)]
Remove unnecessary CFLAGS and SRCS lines.

Obtained-from:  FreeBSD

17 years agoSync with FreeBSD.
Peter Avalos [Sun, 4 Feb 2007 21:08:28 +0000 (21:08 +0000)]
Sync with FreeBSD.

17 years agoRemove bogus "-" operand from usage message. There is no restriction on
Peter Avalos [Sun, 4 Feb 2007 21:06:34 +0000 (21:06 +0000)]
Remove bogus "-" operand from usage message. There is no restriction on
where it may be placed, it is treated the same as the file arguments.

Obtained-from:  FreeBSD

17 years agoUse the saved ucontext pointer and not the return value of get_mcontext.
Simon Schubert [Sun, 4 Feb 2007 20:28:21 +0000 (20:28 +0000)]
Use the saved ucontext pointer and not the return value of get_mcontext.

Reported-by: pavalos@
17 years agoDon't rely on sys/resource.h including all necessary headers.
Simon Schubert [Sun, 4 Feb 2007 19:45:24 +0000 (19:45 +0000)]
Don't rely on sys/resource.h including all necessary headers.

This fixes compilation on 1.4-RELEASE.

17 years agoAvoid truncating the pathname to UNIX Domain Sockets with snprintf(),
Peter Avalos [Sun, 4 Feb 2007 19:27:58 +0000 (19:27 +0000)]
Avoid truncating the pathname to UNIX Domain Sockets with snprintf(),
giving a more sensible warning when the (relatively meagre) sun_path
limit is exceeded.

Obtained-from:  FreeBSD

17 years agoPrint the "$" symbol on blank lines when the -s and -e options are used
Peter Avalos [Sun, 4 Feb 2007 18:43:39 +0000 (18:43 +0000)]
Print the "$" symbol on blank lines when the -s and -e options are used
together.

Obtained-from:  FreeBSD

17 years agoAdd note on using 'handle SIGSEGV noprint' when gdb'ing a virtual kernel.
Matthew Dillon [Sun, 4 Feb 2007 17:24:56 +0000 (17:24 +0000)]
Add note on using 'handle SIGSEGV noprint' when gdb'ing a virtual kernel.

Suggested-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
17 years agoregen
Sepherosa Ziehau [Sun, 4 Feb 2007 13:43:53 +0000 (13:43 +0000)]
regen

17 years agoAdd PCI ids for Intel WL-3945ABG wireless minipci adapters.
Sepherosa Ziehau [Sun, 4 Feb 2007 13:42:08 +0000 (13:42 +0000)]
Add PCI ids for Intel WL-3945ABG wireless minipci adapters.

17 years agoAdd firewire dependency.
Sascha Wildner [Sun, 4 Feb 2007 05:26:05 +0000 (05:26 +0000)]
Add firewire dependency.

17 years agoFix typo.
Sascha Wildner [Sun, 4 Feb 2007 04:31:13 +0000 (04:31 +0000)]
Fix typo.

17 years agoRestore {SYM,M}LINKS.
Sascha Wildner [Sun, 4 Feb 2007 03:48:07 +0000 (03:48 +0000)]
Restore {SYM,M}LINKS.

17 years agoFix typo: firmare -> firmware
Sascha Wildner [Sat, 3 Feb 2007 23:13:19 +0000 (23:13 +0000)]
Fix typo: firmare -> firmware

17 years ago1:1 Userland threading stage 2.11/4:
Simon Schubert [Sat, 3 Feb 2007 17:05:59 +0000 (17:05 +0000)]
1:1 Userland threading stage 2.11/4:

Move signals into lwps, take p_lwp out of proc.

Originally-Submitted-by: David Xu <davidxu@freebsd.org>
Reviewed-by: Thomas E. Spanjaard <tgen@netphreax.net>
17 years agoFor acd(4), create the 'root device' (e.g. acd0) and link a and c to it.
Thomas E. Spanjaard [Sat, 3 Feb 2007 17:00:12 +0000 (17:00 +0000)]
For acd(4), create the 'root device' (e.g. acd0) and link a and c to it.
This allows for specifying 'acd0' as device instead of the semi-deprecated
acd0a and acd0c.

17 years agoBump date for r1.6.
Sascha Wildner [Sat, 3 Feb 2007 11:38:25 +0000 (11:38 +0000)]
Bump date for r1.6.

17 years agoUse .Pq for parentheses.
Sascha Wildner [Sat, 3 Feb 2007 11:35:41 +0000 (11:35 +0000)]
Use .Pq for parentheses.

17 years agoAdd some words about signals and mention that CTRL-\ will drop the vkernel
Sascha Wildner [Sat, 3 Feb 2007 11:27:07 +0000 (11:27 +0000)]
Add some words about signals and mention that CTRL-\ will drop the vkernel
into ddb.

Suggested-by: dillon, Steve Mynott <steve.mynott@gmail.com>
17 years agoRemove now unused cpu_coredump.
Simon Schubert [Sat, 3 Feb 2007 10:30:12 +0000 (10:30 +0000)]
Remove now unused cpu_coredump.

17 years agoClean up stale code.
Simon Schubert [Sat, 3 Feb 2007 10:02:01 +0000 (10:02 +0000)]
Clean up stale code.

17 years agoleftover from aout_coredump() removal:
YONETANI Tomokazu [Sat, 3 Feb 2007 09:56:04 +0000 (09:56 +0000)]
leftover from aout_coredump() removal:
- remove stale prototype for aout_coredump()
- remove the last reference to aout_coredump()

17 years agoadd prototype for linprocfs_init()
YONETANI Tomokazu [Sat, 3 Feb 2007 09:50:49 +0000 (09:50 +0000)]
add prototype for linprocfs_init()

17 years agoFix ps %CPU output in vkernel by exporting hw.availpages.
Simon Schubert [Sat, 3 Feb 2007 09:25:10 +0000 (09:25 +0000)]
Fix ps %CPU output in vkernel by exporting hw.availpages.

Ps tries to read hw.availpages as part of its kernel variable setup and aborts
calculations for any value depending on any kernel variable.

17 years agoDo not generate a trap signal for a spurious SIGTRAP caused by single-
Thomas E. Spanjaard [Fri, 2 Feb 2007 20:00:22 +0000 (20:00 +0000)]
Do not generate a trap signal for a spurious SIGTRAP caused by single-
stepping in syscalls. Otherwise, a signal 0 is sent, which leads to a
panic, which leads to discovering NULL dereferences in DDB machdep code.

17 years agoFix indention.
Simon Schubert [Fri, 2 Feb 2007 19:57:31 +0000 (19:57 +0000)]
Fix indention.

17 years agoRemove some leftover gcc40 info files.
Peter Avalos [Fri, 2 Feb 2007 16:06:13 +0000 (16:06 +0000)]
Remove some leftover gcc40 info files.

Reviewed-by: corecode
17 years agoDon't segfault the kernel if it can't resolve a name on a backtrace.
Simon Schubert [Fri, 2 Feb 2007 15:57:51 +0000 (15:57 +0000)]
Don't segfault the kernel if it can't resolve a name on a backtrace.

17 years agoFix example: Put /dev/vkd0a in /etc/fstab.
Sascha Wildner [Fri, 2 Feb 2007 15:22:08 +0000 (15:22 +0000)]
Fix example: Put /dev/vkd0a in /etc/fstab.

17 years agoRemove GNU tar.
Peter Avalos [Fri, 2 Feb 2007 06:33:17 +0000 (06:33 +0000)]
Remove GNU tar.

Reviewed-by: corecode
17 years agoImplement window size passing between real tty and virtual console.
Simon Schubert [Thu, 1 Feb 2007 20:53:19 +0000 (20:53 +0000)]
Implement window size passing between real tty and virtual console.

Now less, vi, etc.  will use the existing real estate on the tty and
won't limit themselves to 24x80.  TTY window size changes will be passed
through as well.

17 years agoRestore exporting of kernel threads as 'fake' processes. Different from
Thomas E. Spanjaard [Thu, 1 Feb 2007 20:27:05 +0000 (20:27 +0000)]
Restore exporting of kernel threads as 'fake' processes. Different from
before is the fact that we don't take initproc as blueprint anymore, but
rather fake up or set according to the thread the things we care about
explicitly. This also makes use of the newly added fill_kinfo_proc_thread()
in libkvm for operation on dumps feasible.

Also some minor cleanups to related code.

In-collaboration-with: Simon 'corecode' Schubert <corecode@fs.ei.tum.de>

17 years agoTeach the vkernel to behave and not fsck with my tty.
Simon Schubert [Thu, 1 Feb 2007 17:31:31 +0000 (17:31 +0000)]
Teach the vkernel to behave and not fsck with my tty.

This makes the vkernel console set proper termios state when the console
goes out of control, for instance when suspending, interrupting or shutting
down the vkernel.  As a result the console isn't in raw mode afterwards
and can thus still be used.

17 years agoReport non-existing procs as pid -1, not NO_PID.
Simon Schubert [Thu, 1 Feb 2007 12:15:13 +0000 (12:15 +0000)]
Report non-existing procs as pid -1, not NO_PID.

17 years agoChange kinfo_proc interface between kernel and userland.
Simon Schubert [Thu, 1 Feb 2007 10:33:26 +0000 (10:33 +0000)]
Change kinfo_proc interface between kernel and userland.

Before, we were embedding a struct proc (among others) into struct
kinfo_proc.  Every time we change implementation details in the kernel,
userland has to be adapted (recompiled).  In preparation for the coming
LWP changes this interface has been reworked.  Now kinfo_proc is a
structure which does not depend on other structures on the kernel which
are subject to change.  Instead, the routines fill_kinfo_proc and
fill_kinfo_lwp copy all values which are of interest between the kernel
structure and the stable kinfo_proc structure.

Furthermore, this change adds infrastructure to export LWP-specific data.
If userland requests LWP data, it sets the flag KERN_PROC_FLAG_LWP in the
sysctl oid.  This leads to multiple kinfo_procs being exported.  If not
set, the first LWP will used.  This is like FreeBSD do it, and it seems
easy and simple.  Note that userland was not yet adjusted to actually
request LWPs and aggregate this information if necessary.  Besides, the
kernel does not yet have more than one LWP per process anyways.

This introduces a new file, kern/kern_kinfo.c, which is shared between
kernel and libkvm.  This was done to avoid and remove code duplication.
Now kvm_getprocs constructs a complete struct proc, including pointers,
and then calls fill_kinfo_proc to do its job.

In-collaboration-with: Thomas E. Spanjaard <tgen@netphreax.net>

17 years agoDo not build bus modules for vkernels.
Simon Schubert [Thu, 1 Feb 2007 08:42:18 +0000 (08:42 +0000)]
Do not build bus modules for vkernels.

17 years agoSync with FreeBSD (adding DragonFly 1.8.0).
Sascha Wildner [Wed, 31 Jan 2007 08:28:23 +0000 (08:28 +0000)]
Sync with FreeBSD (adding DragonFly 1.8.0).

17 years agoFix typos.
Sascha Wildner [Tue, 30 Jan 2007 16:03:04 +0000 (16:03 +0000)]
Fix typos.

Taken-from: FreeBSD

17 years agoRemove unused VIRTUAL_KERNEL option.
Sascha Wildner [Tue, 30 Jan 2007 15:53:30 +0000 (15:53 +0000)]
Remove unused VIRTUAL_KERNEL option.

OK-by: dillon
17 years agoEnhance markup.
Sascha Wildner [Tue, 30 Jan 2007 15:23:34 +0000 (15:23 +0000)]
Enhance markup.

17 years agoRework DEV_SUPPORT infrastructure.
Simon Schubert [Tue, 30 Jan 2007 14:50:11 +0000 (14:50 +0000)]
Rework DEV_SUPPORT infrastructure.

Reviewed-by: tgen@
17 years agoFix values for CCVER, they are gcc*, not cc*.
Simon Schubert [Tue, 30 Jan 2007 14:47:56 +0000 (14:47 +0000)]
Fix values for CCVER, they are gcc*, not cc*.

Noticed-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk> (via unrelated patch)
17 years agoFix iovec checks to match properly. iovec == MAXIOV and iovec == SMALLIOV
Thomas E. Spanjaard [Mon, 29 Jan 2007 20:44:02 +0000 (20:44 +0000)]
Fix iovec checks to match properly. iovec == MAXIOV and iovec == SMALLIOV
are both legal in those respective statements.

On-behalf-of: Joerg Sonnenberger <joerg@britannica.bec.de>

17 years agosmbfs was not guarenteeing a NULL return vnode on error. This is required
Matthew Dillon [Mon, 29 Jan 2007 06:10:01 +0000 (06:10 +0000)]
smbfs was not guarenteeing a NULL return vnode on error.  This is required
by DragonFly.

Reported-by: Gergo Szakal <bastyaelvtars@gmail.com>
17 years agoo Add modified tables based on the reports/patches from the following people:
YONETANI Tomokazu [Sun, 28 Jan 2007 17:32:35 +0000 (17:32 +0000)]
o Add modified tables based on the reports/patches from the following people:
  Pentium M 750 1.86GHz (Vlad Galu)
  Pentium M 760 2.00GHz (Victor Balada Diaz)
  Pentium M 715 1.50GHz (Ryan O'Connor)

o Bring in some fixes from NetBSD
  rev 1.27: Fix Pentium M 770 table.
  rev 1.25: Added 'Low Voltage Intel Pentium M processor 1.30 GHz' entry.

17 years agoDocument the recently added {STAILQ,TAILQ}_CONCAT macros.
Sascha Wildner [Sun, 28 Jan 2007 14:52:02 +0000 (14:52 +0000)]
Document the recently added {STAILQ,TAILQ}_CONCAT macros.

17 years agoEnable setting debug.acpi.{layer,level} via sysctl command.
YONETANI Tomokazu [Sun, 28 Jan 2007 06:33:49 +0000 (06:33 +0000)]
Enable setting debug.acpi.{layer,level} via sysctl command.

17 years agokern_recvmsg() may not return a valid(non-NULL) pointer to `sa'
YONETANI Tomokazu [Sun, 28 Jan 2007 06:31:00 +0000 (06:31 +0000)]
kern_recvmsg() may not return a valid(non-NULL) pointer to `sa'
even if its return value is 0.  Only sys_recvfrom() knew this.
Fix other callers to deal with this.

17 years agoFix argument lists of locking primitives (thereby fixing some compilation
Sascha Wildner [Sat, 27 Jan 2007 15:03:25 +0000 (15:03 +0000)]
Fix argument lists of locking primitives (thereby fixing some compilation
warnings).

While I'm here, remove several empty functions.

17 years agoi should be pre-incremented in the loop maintenance code or else
YONETANI Tomokazu [Sat, 27 Jan 2007 14:14:13 +0000 (14:14 +0000)]
i should be pre-incremented in the loop maintenance code or else
cp is assigned one element earlier than expected.  Convert the loop
to a simpler form.

17 years agoFix redefinition of KEX_DEFAULT_PK_ALG.
Sascha Wildner [Sat, 27 Jan 2007 08:54:05 +0000 (08:54 +0000)]
Fix redefinition of KEX_DEFAULT_PK_ALG.

17 years agoAdd missing '\n'
Sepherosa Ziehau [Sat, 27 Jan 2007 03:36:10 +0000 (03:36 +0000)]
Add missing '\n'

Obtained-from: FreeBSD (sam@freebsd.org)

17 years agoImplement -D
Matthew Dillon [Fri, 26 Jan 2007 18:58:55 +0000 (18:58 +0000)]
Implement -D

Submitted-by: "Steve O'Hara-Smith" <steve@sohara.org>
17 years agoRemove gobsd from the list.
Matthew Dillon [Fri, 26 Jan 2007 18:52:58 +0000 (18:52 +0000)]
Remove gobsd from the list.

Submitted-by: "Steve Mynott" <steve.mynott@gmail.com>
17 years agoFix generation of the mount path for "/" when a process is chrooted into
Matthew Dillon [Fri, 26 Jan 2007 18:05:23 +0000 (18:05 +0000)]
Fix generation of the mount path for "/" when a process is chrooted into
a subdirectory of a mount point.

Reported-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
17 years agoSetup for 1.8 release - Adjust HEAD to 1.9.0
Matthew Dillon [Thu, 25 Jan 2007 18:57:48 +0000 (18:57 +0000)]
Setup for 1.8 release - Adjust HEAD to 1.9.0

17 years agoSetup for 1.8 release - add new targets to Makefile to udpate the slip tag
Matthew Dillon [Thu, 25 Jan 2007 18:46:02 +0000 (18:46 +0000)]
Setup for 1.8 release - add new targets to Makefile to udpate the slip tag
for 1.8, and add a new cvsup example file for 1.8.

17 years agoSetup for 1.8 release - bump HEAD to 1.7.1 and synchronize preview tag.
Matthew Dillon [Thu, 25 Jan 2007 18:42:41 +0000 (18:42 +0000)]
Setup for 1.8 release - bump HEAD to 1.7.1 and synchronize preview tag.

17 years agoFix the incorrect addition of a leading '/' in file paths in the journaling
Matthew Dillon [Thu, 25 Jan 2007 18:19:31 +0000 (18:19 +0000)]
Fix the incorrect addition of a leading '/' in file paths in the journaling
stream.  This addition was a side effect of changes made to the
namecache topology.

Submitted-by: "Steve O'Hara-Smith" <steve@sohara.org>
17 years agoSilence harmless warnings (for module build).
YONETANI Tomokazu [Thu, 25 Jan 2007 15:12:06 +0000 (15:12 +0000)]
Silence harmless warnings (for module build).

17 years agoStop init before ripping its filesystem references out in order to
Matthew Dillon [Thu, 25 Jan 2007 09:18:13 +0000 (09:18 +0000)]
Stop init before ripping its filesystem references out in order to
prevent a panic from occuring during the shutdown sequence.  This is
a bit of a hack and a better solution is needed.

Reported-by: Gergo Szakal <bastyaelvtars@gmail.com>
17 years agoInitialize ACPI task thread before AcpiEnableSubsystem() installs
YONETANI Tomokazu [Thu, 25 Jan 2007 02:43:35 +0000 (02:43 +0000)]
Initialize ACPI task thread before AcpiEnableSubsystem() installs
interrupt handlers.

17 years agoPoor attempt to tracking the stack frame through a trap. Adjust for
Matthew Dillon [Wed, 24 Jan 2007 18:56:30 +0000 (18:56 +0000)]
Poor attempt to tracking the stack frame through a trap.  Adjust for
the additional elements pushed onto the stack to pass a trapframe pointer
instead of the trapframe itself.

17 years agoDeclare dumpmag as used, even though it is an unused static const.
Simon Schubert [Wed, 24 Jan 2007 16:09:29 +0000 (16:09 +0000)]
Declare dumpmag as used, even though it is an unused static const.

This prevents gcc from optimizing it away and makes savecore happy.

Dragonfly-Bug: <http://bugs.dragonflybsd.org/issue536>

17 years agoMove acpica-unix-20050309 into the Attic
YONETANI Tomokazu [Wed, 24 Jan 2007 15:55:50 +0000 (15:55 +0000)]
Move acpica-unix-20050309 into the Attic

17 years agocheckdirs() was being passed the wrong mount point, resulting in a panic
Matthew Dillon [Wed, 24 Jan 2007 01:25:47 +0000 (01:25 +0000)]
checkdirs() was being passed the wrong mount point, resulting in a panic
when mounting over an already existing mountpoint and improperly adjusting
the current or root directory for processes when finding a matching ncp
whether they were relative to the old mount or not.

Rewrite and document checkdirs() to fix the problems.

Reported-by: "Vincent Labrecque" <vnc@hush.ai>
17 years agoCheck EEPROM twice in em_hardware_init(), because for some PCIe em(4) first
Sepherosa Ziehau [Tue, 23 Jan 2007 11:09:25 +0000 (11:09 +0000)]
Check EEPROM twice in em_hardware_init(), because for some PCIe em(4) first
checking will fail if link is down.  Add comment about it.

This should solve the problem for Lenovo/IBM laptop users: if link is down,
then em(4) will not be initialized properly.

Submitted-and-Tested-by: Michal Belczyk <belczyk@bsd.krakow.pl>
Obtained-from: FreeBSD current list, Jack Vogel <jfvogel@gmail.com>

17 years agoDon't use readline anymore. kadmin/ktutil should be updated to use
Sascha Wildner [Tue, 23 Jan 2007 10:22:55 +0000 (10:22 +0000)]
Don't use readline anymore. kadmin/ktutil should be updated to use
libedit.

In-discussion-with: Matt Emmerton, dillon, corecode

17 years agoFix a crash related to the NPX (floating point) interrupt. The interrupt
Matthew Dillon [Tue, 23 Jan 2007 08:43:02 +0000 (08:43 +0000)]
Fix a crash related to the NPX (floating point) interrupt.  The interrupt
frame was not being properly translated to a trapframe due to recent
trapframe changes.

Reported-by: Sascha Wildner <saw@online.de>
17 years agoPass structs by reference if you expect the callee to modify them.
Simon Schubert [Mon, 22 Jan 2007 19:37:05 +0000 (19:37 +0000)]
Pass structs by reference if you expect the callee to modify them.

This fixes kernel boot with gcc41.  The gpfault people were seeing comes from
vm86_bioscall() in init386().  The cause is that the assembler code passes the
struct vm86frame by value, i.e. simply creating it on the stack.  This worked
up to gcc34, but gcc41 now optimizes stores to unused memory locations away,
whis is allowed per the standards.  This led to an uninitialized stack frame
which in turn panicked the box.

Oooohh...-please-commit-by: dillon@

17 years agoIncorrect calculation of a virtual block address within
YONETANI Tomokazu [Mon, 22 Jan 2007 15:31:37 +0000 (15:31 +0000)]
Incorrect calculation of a virtual block address within
a single indirect block causes corruption of large file.

Submitted-by: Yuichiro Goto <y7goto at gmail.com>
17 years agoIf there is no commitid, we need to initialize Delta->commitid to NULL to
Peter Avalos [Sun, 21 Jan 2007 17:58:42 +0000 (17:58 +0000)]
If there is no commitid, we need to initialize Delta->commitid to NULL to
prevent segmentation faults later.

17 years agoRemove GCC 4.0's manpages, not those of 4.1.
Sascha Wildner [Sun, 21 Jan 2007 12:24:43 +0000 (12:24 +0000)]
Remove GCC 4.0's manpages, not those of 4.1.

17 years agoAdd a workaround for some Lenovo/IBM laptops' onboard em(4).
Sepherosa Ziehau [Sun, 21 Jan 2007 10:44:45 +0000 (10:44 +0000)]
Add a workaround for some Lenovo/IBM laptops' onboard em(4).

Without this workaround, em(4) is extremely slow when TXing packets, though
polling(4) seems to make the situation better, the speed is still far from
expectation for 100BaseTX-FDX network.

Reported-by: Michal Belczyk <belczyk@bsd.krakow.pl>
Obtained-from: FreeBSD-current mail list, Jack Vogel <jfvogel@gmail.com>

# According to Jack Vogel:
# "saying its known to be a problem doesnt mean its cause is known"

17 years agoRemove unused argument.
Sascha Wildner [Sun, 21 Jan 2007 10:40:39 +0000 (10:40 +0000)]
Remove unused argument.

17 years agopbn * vn->sc_secsize may wraparound, because both pbn and vn->sc_secsize
YONETANI Tomokazu [Sun, 21 Jan 2007 10:37:29 +0000 (10:37 +0000)]
pbn * vn->sc_secsize may wraparound, because both pbn and vn->sc_secsize
are of type int.  Cast to off_t before multiplication.

17 years agoSync the network and sound modules sections with what we actually have.
Sascha Wildner [Sun, 21 Jan 2007 10:10:37 +0000 (10:10 +0000)]
Sync the network and sound modules sections with what we actually have.

Submitted-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk>
17 years agoRemove BUGS section and note that Matt fixed nullfs.
Sascha Wildner [Sun, 21 Jan 2007 10:07:50 +0000 (10:07 +0000)]
Remove BUGS section and note that Matt fixed nullfs.

Submitted-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk>
17 years agoDragonFly supports up to 16 partitions, not 8.
Sascha Wildner [Sun, 21 Jan 2007 10:06:10 +0000 (10:06 +0000)]
DragonFly supports up to 16 partitions, not 8.

Submitted-by: Thomas Nikolajsen <thomas.nikolajsen@mail.dk>
17 years agoAssume 0.0 if no float value is passed.
Sascha Wildner [Sun, 21 Jan 2007 00:24:55 +0000 (00:24 +0000)]
Assume 0.0 if no float value is passed.

17 years agoRemove wrong declarations (fix gcc41 build).
Sascha Wildner [Sat, 20 Jan 2007 21:32:36 +0000 (21:32 +0000)]
Remove wrong declarations (fix gcc41 build).

Found-by: pavalos, walt
17 years agoAdd CC.1 MLINK.
Sascha Wildner [Sat, 20 Jan 2007 20:26:37 +0000 (20:26 +0000)]
Add CC.1 MLINK.

17 years agoRemove the advertising clause from vinum with permission from Greg Lehey,
Matthew Dillon [Sat, 20 Jan 2007 19:20:39 +0000 (19:20 +0000)]
Remove the advertising clause from vinum with permission from Greg Lehey,
as encapsulated in his response to a query by Matt Emmerton:

---
On Friday, 19 January 2007 at 15:23:36 -0500, Matt Emmerton wrote:
> Greg,
>
> I'm working on DragonFlyBSD, which branched from FreeBSD-4.x.  As such, it
> has an "old" version of vinum which uses a BSD licence which includes an
> advertising clause.  (Interestingly, not requiring Berkeley to be mention=
ed,
> but some other organization.)
>
> I know that vinum has undergone a GEOM-ification in newer FreeBSD release=
s,
> and it appears that all the new code in FreeBSD does not have any adverti=
sing
> clauses.  However, this is not useful to us since we don't have GEOM.

Understood.  I've already sent a message to the FreeBSD people saying
that they can remove the advertising clause; of course you may do so
as well.

Greg
---

Submitted-by: Matt Emmerton <matt@gsicomp.on.ca>
17 years agoFix MLINK.
Sascha Wildner [Sat, 20 Jan 2007 19:18:02 +0000 (19:18 +0000)]
Fix MLINK.