dragonfly.git
14 years agoMerge from vendor branch OPENSSH:
Simon Schubert [Thu, 28 Sep 2006 18:38:21 +0000 (18:38 +0000)]
Merge from vendor branch OPENSSH:
Import stripped down sources of OpenSSH-4.4p1

14 years agoImport stripped down sources of OpenSSH-4.4p1
Simon Schubert [Thu, 28 Sep 2006 18:38:21 +0000 (18:38 +0000)]
Import stripped down sources of OpenSSH-4.4p1

14 years agoUpdate strtonum(3) to the latest OpenBSD version, which uses long long's
Simon Schubert [Thu, 28 Sep 2006 17:20:45 +0000 (17:20 +0000)]
Update strtonum(3) to the latest OpenBSD version, which uses long long's
instead of unsigned long long's.

Noticed-by: OpenSSH openbsd-portable regression test suite
14 years agoThe "is_eof" commit made this WARNS 3.
Peter Avalos [Thu, 28 Sep 2006 04:24:05 +0000 (04:24 +0000)]
The "is_eof" commit made this WARNS 3.

14 years agoFix compiler warnings by creating a new macro "is_eof" that does the PEOF
Peter Avalos [Thu, 28 Sep 2006 04:19:40 +0000 (04:19 +0000)]
Fix compiler warnings by creating a new macro "is_eof" that does the PEOF
comparison, and don't use it if the character comes from a string.  Taken
from FreeBSD.

14 years agoRemove WARNS since it's in ../Makefile.inc.
Peter Avalos [Thu, 28 Sep 2006 03:46:03 +0000 (03:46 +0000)]
Remove WARNS since it's in ../Makefile.inc.

14 years agoGrammer fixes taken from OpenBSD.
Peter Avalos [Wed, 27 Sep 2006 21:58:08 +0000 (21:58 +0000)]
Grammer fixes taken from OpenBSD.

14 years agoRemove crufty build options that are not relevant.
Peter Avalos [Wed, 27 Sep 2006 19:18:00 +0000 (19:18 +0000)]
Remove crufty build options that are not relevant.

14 years agoWelcome the build infrastructure for gcc41
Simon Schubert [Wed, 27 Sep 2006 12:10:35 +0000 (12:10 +0000)]
Welcome the build infrastructure for gcc41

14 years agoAdd our READMEs
Simon Schubert [Wed, 27 Sep 2006 10:27:33 +0000 (10:27 +0000)]
Add our READMEs

14 years agoMerge from vendor branch GCC:
Simon Schubert [Wed, 27 Sep 2006 10:25:53 +0000 (10:25 +0000)]
Merge from vendor branch GCC:
Import a stripped down version of gcc-4.1.1

14 years agoImport a stripped down version of gcc-4.1.1
Simon Schubert [Wed, 27 Sep 2006 10:25:53 +0000 (10:25 +0000)]
Import a stripped down version of gcc-4.1.1

14 years agoRun trimdir in a subshell so that our cwd doesn't change under us
Simon Schubert [Wed, 27 Sep 2006 10:24:15 +0000 (10:24 +0000)]
Run trimdir in a subshell so that our cwd doesn't change under us

14 years agoFollow up to kern_conf.c 1.16. We can't just ignore the ops comparison, it
Matthew Dillon [Tue, 26 Sep 2006 18:57:14 +0000 (18:57 +0000)]
Follow up to kern_conf.c 1.16.  We can't just ignore the ops comparison, it
is needed to keep user-invisible devices user-invisible.  Add a flag so
hashdev() knows when it can ignore the comparison and when it can't ignore
the comparison.

14 years agoFix a bug in the device intercept code used by /dev/console. The
Matthew Dillon [Tue, 26 Sep 2006 18:15:28 +0000 (18:15 +0000)]
Fix a bug in the device intercept code used by /dev/console.  The
bug caused device opens on the underlying device to create a new
device instead of using the console intercept, causing make_dev
warnings and preventing getty from operating properly on the console.

Reported-by: "Frank W. Josellis" <frank@dynamical-systems.org>
14 years agoFix an off-by-one error. Track #1 is index 0 in the TOC buffer.
Matthew Dillon [Tue, 26 Sep 2006 17:32:09 +0000 (17:32 +0000)]
Fix an off-by-one error.  Track #1 is index 0 in the TOC buffer.

Submitted-by: "Ben Woolley" <tautolog@gmail.com>
14 years agoAlways warn about identical target and source. Nevertheless, don't treat
Simon Schubert [Mon, 25 Sep 2006 09:27:21 +0000 (09:27 +0000)]
Always warn about identical target and source.  Nevertheless, don't treat
it as a failure when -f is specified.

14 years agoDon't unlink the target if the source and target are the same file. Instead,
Simon Schubert [Sun, 24 Sep 2006 21:28:13 +0000 (21:28 +0000)]
Don't unlink the target if the source and target are the same file.  Instead,
complain unless -f is given.

Add a comment about the inverted nomenclature within the source
wrt target and source.

14 years agoTry to clean up any remaining filesystem references when rebooting. Clean
Matthew Dillon [Sun, 24 Sep 2006 19:43:55 +0000 (19:43 +0000)]
Try to clean up any remaining filesystem references when rebooting.  Clean
up references made by the process calling reboot(), the proc0 template,
the init process, and by the VFS subsystem.

14 years agoMake bsdtar our default tar. GNU tar is still installed as gtar, and can
Peter Avalos [Sat, 23 Sep 2006 20:31:45 +0000 (20:31 +0000)]
Make bsdtar our default tar.  GNU tar is still installed as gtar, and can
be "tar" by using WITH_GTAR in make.conf.

Discussed on: users

14 years agoWhitespace adjustment.
Sepherosa Ziehau [Sat, 23 Sep 2006 01:59:16 +0000 (01:59 +0000)]
Whitespace adjustment.

Add two device ids:
- Intel Pro/100 VE Network Connection card
- NIC found on ICH7 based systems

Submitted-by: Rauf Kuliyev <rauf@kuliyev.com>
Obtained-from: FreeBSD

14 years agoFix a compile error when DDB is not defined. db_print_backtrace() is
Matthew Dillon [Fri, 22 Sep 2006 16:47:09 +0000 (16:47 +0000)]
Fix a compile error when DDB is not defined.  db_print_backtrace() is
only available if DDB is defined.

Reported-by: "Rauf Kuliyev" <rauf@kuliyev.com>, Petr Janda <elekktretterr@exemail.com.au>
14 years agoRemove some unnecessary -r options and use csh for man page removal.
Sascha Wildner [Fri, 22 Sep 2006 09:46:48 +0000 (09:46 +0000)]
Remove some unnecessary -r options and use csh for man page removal.

14 years agoM_WRITABLE() is not applicable to DFLY since if*mp is from the driver
Scott Ullrich [Fri, 22 Sep 2006 01:18:14 +0000 (01:18 +0000)]
M_WRITABLE() is not applicable to DFLY since if*mp is from the driver
which supports jumbo frames.

Pointed-out-by: Sepherosa Ziehau
14 years agoUpgrade to bind-9.3.2-P1. Delete unused cat files for dnssec-makekeyset
Victor Balada Diaz [Thu, 21 Sep 2006 20:56:50 +0000 (20:56 +0000)]
Upgrade to bind-9.3.2-P1. Delete unused cat files for dnssec-makekeyset
and dnssec-signkey.

Suggested-By: Sascha Wildner
14 years agoRecent dev_t work confused sysctl. Adjust the reported type to udev_t
Matthew Dillon [Thu, 21 Sep 2006 16:16:09 +0000 (16:16 +0000)]
Recent dev_t work confused sysctl.  Adjust the reported type to udev_t
and make sysctl recognize it.

Reported-by: "Frank W. Josellis" <frank@dynamical-systems.org>
14 years agoRemove unused function.
Sascha Wildner [Thu, 21 Sep 2006 11:49:22 +0000 (11:49 +0000)]
Remove unused function.

Submitted-by: Trevor Kendall <trevorjkendall@gmail.com>
14 years agoTypo in commented out line.
YONETANI Tomokazu [Thu, 21 Sep 2006 05:18:24 +0000 (05:18 +0000)]
Typo in commented out line.

Revealed-by: Haidut <haidut at gmail.com>
14 years agoAdd an option that causes cpdup to skip CHR or BLK devices. This option
Matthew Dillon [Thu, 21 Sep 2006 04:09:28 +0000 (04:09 +0000)]
Add an option that causes cpdup to skip CHR or BLK devices.  This option
would typically be needed if using cpdup across a network to a non-BSD box
which does not support the creation of device nodes.

14 years agoFix a bug in the script softlink code. The softlinks were not being
Matthew Dillon [Thu, 21 Sep 2006 00:18:13 +0000 (00:18 +0000)]
Fix a bug in the script softlink code.  The softlinks were not being
properly updated because 'ln -sf' creates a softlink in the directory pointed
to by the existing softlink rather then replacing the existing softlink.

14 years agoMerge from vendor branch BIND:
Victor Balada Diaz [Wed, 20 Sep 2006 22:16:11 +0000 (22:16 +0000)]
Merge from vendor branch BIND:
Add updated README files based on our previous version

14 years agoAdd updated README files based on our previous version
Victor Balada Diaz [Wed, 20 Sep 2006 22:16:11 +0000 (22:16 +0000)]
Add updated README files based on our previous version

14 years agoUpgrade bind to bind-9.3.2-P1. This tools are no longer used.
Victor Balada Diaz [Wed, 20 Sep 2006 21:56:00 +0000 (21:56 +0000)]
Upgrade bind to bind-9.3.2-P1. This tools are no longer used.

14 years agoUpgrade bind to bind-9.3.2-P1
Victor Balada Diaz [Wed, 20 Sep 2006 21:48:39 +0000 (21:48 +0000)]
Upgrade bind to bind-9.3.2-P1

14 years agoImport of bind-9.3.2-P1
Victor Balada Diaz [Wed, 20 Sep 2006 21:28:53 +0000 (21:28 +0000)]
Import of bind-9.3.2-P1

14 years agoMerge from vendor branch BIND:
Victor Balada Diaz [Wed, 20 Sep 2006 21:28:53 +0000 (21:28 +0000)]
Merge from vendor branch BIND:
Import of bind-9.3.2-P1

14 years agoAs of the recent libarchive upgrade, archive_read_open_file() and
Sascha Wildner [Wed, 20 Sep 2006 16:51:33 +0000 (16:51 +0000)]
As of the recent libarchive upgrade, archive_read_open_file() and
archive_write_open_file() have become deprecated. Therefore, remove
the old MLINKS.

14 years agoCheck that namecache references to the mount point are no longer present
Matthew Dillon [Tue, 19 Sep 2006 18:17:46 +0000 (18:17 +0000)]
Check that namecache references to the mount point are no longer present
before unmounting a filesystem.  Forced unmounts ignore the check but will
print a warning.

This patch is primarily designed to prevent nullfs partitions from being
unmounted while processes are still present within them.  The normal vnode
check does not work for nullfs mounts since nullfs mounts do not hold any
vnodes of their own.

Note that this will cause a warning to be reported for the root filesystem
when rebooting.

14 years agoRemove the last bits of code that stored mount point linkages in vnodes.
Matthew Dillon [Tue, 19 Sep 2006 16:06:17 +0000 (16:06 +0000)]
Remove the last bits of code that stored mount point linkages in vnodes.
Mount point linkages are now ENTIRELY a function of the namecache topology,
made possible by DragonFly's advanced namecache.

This fixes a number of problems with NULLFS and adds two major features to
our NULLFS mounting capabilities.

NULLFS mounting paths NO LONGER NEED TO BE DISTINCT.  For example, you
can now safely do things like 'mount_null -o ro / /fubar/jail1' without
creating a recursion and you can now create SUB-MOUNTS within nullfs
mounts, such as 'mount_null -o ro /usr /fubar/jail1/usr', without creating
problems in the original master partitions.

The result is that NULLFS can now be used to glue arbitrary pieces of
filesystems together using a mixture of read-only and read-write NULLFS
mounts for situations where localhost NFS mounts had to be used before.
Jail or chroot construction is now utterly trivial.

With-input-from: Joerg Sonnenberger <joerg@britannica.bec.de>

14 years ago1:1 Userland threading stage 2.9/4:
Simon Schubert [Tue, 19 Sep 2006 11:47:36 +0000 (11:47 +0000)]
1:1 Userland threading stage 2.9/4:

Push out p_thread a little bit more

14 years agoSet f_ncp in the struct file unconditionally. Previously we only set it
Matthew Dillon [Mon, 18 Sep 2006 18:19:35 +0000 (18:19 +0000)]
Set f_ncp in the struct file unconditionally.  Previously we only set it
when opening directories.  This allows the f*() system calls such as
fchmod() to check the actual mount point instead of the aliased mount
point (in the case of a NULLFS mount).  Also, the fstat program will
properly report the path for descriptors opened via nullfs mounts.

Add code to all f*() system calls such as fchmod() to check f_ncp
in order to detect read-only nullfs mounts.

14 years agoDisallow writes to filesystems mounted read-only via NULLFS. In this case
Matthew Dillon [Mon, 18 Sep 2006 17:42:28 +0000 (17:42 +0000)]
Disallow writes to filesystems mounted read-only via NULLFS.  In this case
the ncp->nc_mount in the namecache must be checked since the vnode's
mount point is the actual filesystem and not the NULLFS mount.

Reported-by: Joerg Sonnenberger <joerg@britannica.bec.de>
14 years agoRemove old version of bsdtar and libarchive.
Peter Avalos [Mon, 18 Sep 2006 03:09:39 +0000 (03:09 +0000)]
Remove old version of bsdtar and libarchive.

14 years agoUpgrade libarchive and bsdtar to 1.3.1.
Peter Avalos [Mon, 18 Sep 2006 01:47:10 +0000 (01:47 +0000)]
Upgrade libarchive and bsdtar to 1.3.1.

libarchive Makefile changes inspired by FreeBSD.

14 years agoAdd comment saying this is used by both libarchive and bsdtar.
Peter Avalos [Mon, 18 Sep 2006 01:29:38 +0000 (01:29 +0000)]
Add comment saying this is used by both libarchive and bsdtar.

14 years agoAdd our READMEs to the import.
Peter Avalos [Mon, 18 Sep 2006 01:22:12 +0000 (01:22 +0000)]
Add our READMEs to the import.

14 years agoMerge from vendor branch LIBARCHIVE:
Peter Avalos [Mon, 18 Sep 2006 01:05:12 +0000 (01:05 +0000)]
Merge from vendor branch LIBARCHIVE:
Import of libarchive and bsdtar 1.3.1

14 years agoImport of libarchive and bsdtar 1.3.1
Peter Avalos [Mon, 18 Sep 2006 01:05:12 +0000 (01:05 +0000)]
Import of libarchive and bsdtar 1.3.1

14 years agoMove an assertion in the bridge code so it only gets hit if the bridge
Matthew Dillon [Sun, 17 Sep 2006 21:10:18 +0000 (21:10 +0000)]
Move an assertion in the bridge code so it only gets hit if the bridge
code is actually enabled for an interface.

14 years agoMake some adjustments to low level madvise/mcontrol/mmap support code to
Matthew Dillon [Sun, 17 Sep 2006 21:09:40 +0000 (21:09 +0000)]
Make some adjustments to low level madvise/mcontrol/mmap support code to
accomodate vmspace_*() calls.

Reformulate the new vmspace_*() calls so they operate similarly to the
MAP_VPAGETABLE and mcontrol() calls.  This also makes vmspace's more
'programmable' in the sense that it will be possible to mix virtual
pagetable mmap()ings with other mmap()ing in a vmspace.

Fill in the code for all the new vmspace_*() calls except for
vmspace_ctl().  NOTE: vmspace calls are effectively disabled unless
vm.vkernel_enable is turned on, just like MAP_VPAGETABLE.

Renumber the new mcontrol() and vmspace_*() calls and regenerate.

14 years agoMake some adjustments to low level madvise/mcontrol/mmap support code to
Matthew Dillon [Sun, 17 Sep 2006 21:07:32 +0000 (21:07 +0000)]
Make some adjustments to low level madvise/mcontrol/mmap support code to
accomodate vmspace_*() calls.

Reformulate the new vmspace_*() calls so they operate similarly to the
MAP_VPAGETABLE and mcontrol() calls.  This also makes vmspace's more
'programmable' in the sense that it will be possible to mix virtual
pagetable mmap()ings with other mmap()ing in a vmspace.

Fill in the code for all the new vmspace_*() calls except for
vmspace_ctl().  NOTE: vmspace calls are effectively disabled unless
vm.vkernel_enable is turned on, just like MAP_VPAGETABLE.

Renumber the new mcontrol() and vmspace_*() calls and regenerate.

14 years agonullfs_subr doesn't exist anymore.
Joerg Sonnenberger [Sun, 17 Sep 2006 15:37:48 +0000 (15:37 +0000)]
nullfs_subr doesn't exist anymore.

14 years agoCommit a comprehensive file describing how to do incremental backups along
Matthew Dillon [Sat, 16 Sep 2006 21:57:10 +0000 (21:57 +0000)]
Commit a comprehensive file describing how to do incremental backups along
with some example scripts.

14 years agoSync with FreeBSD.
Peter Avalos [Sat, 16 Sep 2006 19:18:06 +0000 (19:18 +0000)]
Sync with FreeBSD.

Submitted By:  Trevor Kendall <trevorjkendall@gmail.com>

14 years agoFix ng_mppc module building.
Sascha Wildner [Sat, 16 Sep 2006 18:39:13 +0000 (18:39 +0000)]
Fix ng_mppc module building.

14 years agoWARNS6 cleanup
Peter Avalos [Sat, 16 Sep 2006 18:38:00 +0000 (18:38 +0000)]
WARNS6 cleanup

style(9) cleanups

Use sysexits

Sync calendar files with FreeBSD

Submitted by:  Trevor Kendall <trevorjkendall@gmail.com>

14 years agoAdd a README file with some helpful porting hints. I'd rather do this in
Matthew Dillon [Sat, 16 Sep 2006 18:24:37 +0000 (18:24 +0000)]
Add a README file with some helpful porting hints.  I'd rather do this in
the master repository then maintain it out of band.

14 years agoBump the version number reported by cpdup from 1.06 to 1.07
Matthew Dillon [Sat, 16 Sep 2006 18:18:05 +0000 (18:18 +0000)]
Bump the version number reported by cpdup from 1.06 to 1.07

14 years agoFix a bug when '-f -H' is used and the target already exists. cpdup was
Matthew Dillon [Sat, 16 Sep 2006 16:47:29 +0000 (16:47 +0000)]
Fix a bug when '-f -H' is used and the target already exists.  cpdup was
copying the file in that case instead of hardlinking.

14 years agonfe_jref() wrongly uses atomic_subtract_int() instead of atomic_add_int(),
Sepherosa Ziehau [Sat, 16 Sep 2006 06:37:11 +0000 (06:37 +0000)]
nfe_jref() wrongly uses atomic_subtract_int() instead of atomic_add_int(),
because of I "copy & paste"ed it from nfe_jfree()

14 years agoFix a bug in sendmsg() and two compatibility versions of sendmsg().
Matthew Dillon [Sat, 16 Sep 2006 03:37:15 +0000 (03:37 +0000)]
Fix a bug in sendmsg() and two compatibility versions of sendmsg().
iovec_free() should not be called when iovec_copyin() fails.

Reported-by: Ilja van Sprundel
Found-using: A trivial system call fuzzer

14 years agoFix a bug in sysctl()'s handling of user data. You can't wire 0 bytes
Matthew Dillon [Sat, 16 Sep 2006 03:32:44 +0000 (03:32 +0000)]
Fix a bug in sysctl()'s handling of user data.  You can't wire 0 bytes
without blowing up vm_map_wire().

Reported-by: Ilja van Sprundel
Found-using: A trivial system call fuzzer

14 years agoFirst commit
Victor Balada Diaz [Sat, 16 Sep 2006 00:28:16 +0000 (00:28 +0000)]
First commit

14 years agoCollapse some bits of repetitive code into their own procedures and
Matthew Dillon [Wed, 13 Sep 2006 22:25:00 +0000 (22:25 +0000)]
Collapse some bits of repetitive code into their own procedures and
allocate a maximally sized default object to back MAP_VPAGETABLE
mappings, allowing us to access logical memory beyond the size of the
original mmap() call by programming the page table to point at it.

This gives us an abstraction and capability similar to a real kernel's
ability to map e.g. 2GB of physical memory into its 1GB address space.

14 years agoClean up some #include's that shouldn't have been in there. Unbreak
Matthew Dillon [Wed, 13 Sep 2006 21:05:22 +0000 (21:05 +0000)]
Clean up some #include's that shouldn't have been in there.  Unbreak
buildworld.

14 years ago4BSD is history. Long live DragonFly!
Sascha Wildner [Wed, 13 Sep 2006 19:41:41 +0000 (19:41 +0000)]
4BSD is history. Long live DragonFly!

14 years agoAdd cvs annotate -b which allows to annotate "backwards":
Simon Schubert [Wed, 13 Sep 2006 19:34:24 +0000 (19:34 +0000)]
Add cvs annotate -b which allows to annotate "backwards":

When specifying a revision to start with, it will show for each line
at which point (later) this line was removed.

14 years agoFactor patching out into a macro, so that it can be easily used in other
Simon Schubert [Wed, 13 Sep 2006 19:33:14 +0000 (19:33 +0000)]
Factor patching out into a macro, so that it can be easily used in other
Makefiles as well.

14 years agoAnsify the rest of the K&R-style function declarations in sys/i386.
Sascha Wildner [Wed, 13 Sep 2006 18:45:12 +0000 (18:45 +0000)]
Ansify the rest of the K&R-style function declarations in sys/i386.

Those were somehow[tm] forgotten last time.

Noticed-by: corecode
While I'm here, perform some stylistic cleanup in math_emulate.c.

14 years agoMore cleanups + fix a bug when taking a write fault on a mapping that uses
Matthew Dillon [Wed, 13 Sep 2006 18:12:18 +0000 (18:12 +0000)]
More cleanups + fix a bug when taking a write fault on a mapping that uses
a virtual page table.  The page was not being pmap'd with the correct
permissions.

14 years agoFix typo.
Sascha Wildner [Wed, 13 Sep 2006 17:18:18 +0000 (17:18 +0000)]
Fix typo.

Noticed-by: Erik Wikström <erik-wikstrom@telia.com>
14 years agoMAP_VPAGETABLE support part 3/3.
Matthew Dillon [Wed, 13 Sep 2006 17:10:42 +0000 (17:10 +0000)]
MAP_VPAGETABLE support part 3/3.

Implement a new system call called mcontrol() which is an extension of
madvise(), adding an additional 64 bit argument.  Add two new advisories,
MADV_INVAL and MADV_SETMAP.

MADV_INVAL will invalidate the pmap for the specified virtual address
range.  You need to do this for the virtual addresses effected by changes
made in a virtual page table.

MADV_SETMAP sets the top-level page table entry for the virtual page table
governing the mapped range.  It only works for memory governed by a virtual
page table and strange things will happen if you only set the root
page table entry for part of the virtual range.

Further refine the virtual page table format.  Keep with 32 bit VPTE's for
the moment, but properly implement VPTE_PS and VPTE_V.  VPTE_PS can be
used to suport 4MB linear maps in the top level page table and it can also
be used when specifying the 'root' VPTE to disable the page table entirely
and just linear map the backing store.  VPTE_V is the 'valid' bit (before
it was inverted, now it is normal).

14 years agoAdjust README
Simon Schubert [Wed, 13 Sep 2006 16:14:33 +0000 (16:14 +0000)]
Adjust README

14 years agoMerge from vendor branch CVS:
Simon Schubert [Wed, 13 Sep 2006 16:13:30 +0000 (16:13 +0000)]
Merge from vendor branch CVS:
Import useful files

14 years agoImport useful files
Simon Schubert [Wed, 13 Sep 2006 16:13:30 +0000 (16:13 +0000)]
Import useful files

14 years agoNote that Netgear WG311 v3 does *not* use an Atheros chip.
Sascha Wildner [Wed, 13 Sep 2006 09:13:26 +0000 (09:13 +0000)]
Note that Netgear WG311 v3 does *not* use an Atheros chip.

Taken-from: NetBSD
Noticed-by: Trevor Kendall <trevorjkendall@gmail.com>
14 years agoFix small mistake.
Sascha Wildner [Wed, 13 Sep 2006 09:08:50 +0000 (09:08 +0000)]
Fix small mistake.

14 years agoChange one forgotten malloc() to kmalloc().
Sascha Wildner [Wed, 13 Sep 2006 06:59:24 +0000 (06:59 +0000)]
Change one forgotten malloc() to kmalloc().

Reported-by: Ferruccio Zamuner <nonsolosoft@diff.org>
14 years agoMAP_VPAGETABLE support part 2/3.
Matthew Dillon [Tue, 12 Sep 2006 22:03:12 +0000 (22:03 +0000)]
MAP_VPAGETABLE support part 2/3.

Implement preliminary virtual page table handling code in vm_fault.  This
code is strictly temporary so subsystem and userland interactions can be
tested, but the real code will be very similar.

14 years agoMAP_VPAGETABLE support part 1/3.
Matthew Dillon [Tue, 12 Sep 2006 18:41:32 +0000 (18:41 +0000)]
MAP_VPAGETABLE support part 1/3.

Reorganize vm_fault() to get more direct access to the VM page resolved by
a VM fault.  Move vm_fault()'s core shadow object traversal and fault I/O
code to a new procedure called vm_fault_object().

Begin adding support for memory mappings which are backed by a virtualized
page table under userland control.

14 years agoMove flag(s) representing the type of vm_map_entry into its own vm_maptype_t
Matthew Dillon [Mon, 11 Sep 2006 20:25:31 +0000 (20:25 +0000)]
Move flag(s) representing the type of vm_map_entry into its own vm_maptype_t
type.  This is a precursor to adding a new VM mapping type for virtualized
page tables.

14 years agoSync with FreeBSD.
Sascha Wildner [Sun, 10 Sep 2006 22:04:38 +0000 (22:04 +0000)]
Sync with FreeBSD.

14 years agoRemove obsolete pre-Citrus file.
Sascha Wildner [Sun, 10 Sep 2006 21:42:45 +0000 (21:42 +0000)]
Remove obsolete pre-Citrus file.

14 years agoReserve upcall IDs 0-31 for system use.
Matthew Dillon [Sun, 10 Sep 2006 21:35:11 +0000 (21:35 +0000)]
Reserve upcall IDs 0-31 for system use.

14 years ago* Remove support for VAX-style and IBM mainframe-style floating-point
Sascha Wildner [Sun, 10 Sep 2006 21:22:32 +0000 (21:22 +0000)]
* Remove support for VAX-style and IBM mainframe-style floating-point
  arithmetic.

* Remove K&R function header support.

* Remove Bad_float_h support.

* While I'm here, perform some minor style cleanup.

14 years agoUpdate default make.conf and its manpage:
Sascha Wildner [Sun, 10 Sep 2006 20:53:21 +0000 (20:53 +0000)]
Update default make.conf and its manpage:

* Sync CPU types with share/mk/bsd.cpu.gcc{34,40}.mk.

* Add some generic words of warning about building with the
  experimental compiler.

* Add NO_GDB and WANT_LUKEMFTPD.

* Remove duplicate NO_SHARE.

* Remove ports specific FETCH_ENV.

Based-on-patch-by: Trevor Kendall <trevorjkendall@gmail.com>
Thanks to pavalos for noting that WANT_LUKEMFTPD is missing.

14 years agoRemove DEC Alpha support.
Sascha Wildner [Sun, 10 Sep 2006 10:09:25 +0000 (10:09 +0000)]
Remove DEC Alpha support.

14 years agoacpiioctl() can be left without ACPI_UNLOCK(), leading to
YONETANI Tomokazu [Sun, 10 Sep 2006 06:05:35 +0000 (06:05 +0000)]
acpiioctl() can be left without ACPI_UNLOCK(), leading to
critical section count mismatch panic.

14 years agounbreak LINT building
Sepherosa Ziehau [Sun, 10 Sep 2006 05:02:14 +0000 (05:02 +0000)]
unbreak LINT building

14 years agoAdd an ifmedia(4) manual page.
Sascha Wildner [Sun, 10 Sep 2006 04:14:01 +0000 (04:14 +0000)]
Add an ifmedia(4) manual page.

Taken-from: NetBSD (modified to reflect what's in our <net/if_media.h>).

Thanks to sephe for assistance.

14 years agoFix comment.
Sepherosa Ziehau [Sun, 10 Sep 2006 03:40:11 +0000 (03:40 +0000)]
Fix comment.

Noticed-by: swildner
14 years agoAdd "Turbo mode" and "Monitor mode"
Sepherosa Ziehau [Sun, 10 Sep 2006 02:57:28 +0000 (02:57 +0000)]
Add "Turbo mode" and "Monitor mode"

Noticed-by: swildner
14 years agoChange the kernel dev_t, representing a pointer to a specinfo structure,
Matthew Dillon [Sun, 10 Sep 2006 01:26:41 +0000 (01:26 +0000)]
Change the kernel dev_t, representing a pointer to a specinfo structure,
to cdev_t.  Change struct specinfo to struct cdev.  The name 'cdev' was taken
from FreeBSD.  Remove the dev_t shim for the kernel.

This commit generally removes the overloading of 'dev_t' between userland and
the kernel.

Also fix a bug in libkvm where a kernel dev_t (now cdev_t) was not being
properly converted to a userland dev_t.

14 years agoSilence warnings regarding yyunput being defined and never used.
Peter Avalos [Sat, 9 Sep 2006 23:24:43 +0000 (23:24 +0000)]
Silence warnings regarding yyunput being defined and never used.

14 years agoRemove (forgotten?) files from last year's upgrade.
Peter Avalos [Sat, 9 Sep 2006 21:03:16 +0000 (21:03 +0000)]
Remove (forgotten?) files from last year's upgrade.

14 years agoRename the kernel NODEV to NOCDEV to avoid conflicts with the userland NODEV.
Matthew Dillon [Sat, 9 Sep 2006 19:34:47 +0000 (19:34 +0000)]
Rename the kernel NODEV to NOCDEV to avoid conflicts with the userland NODEV.

14 years agoRename struct specinfo into struct cdev. Add a new typedef 'cdev_t' for cdev
Matthew Dillon [Sat, 9 Sep 2006 19:07:29 +0000 (19:07 +0000)]
Rename struct specinfo into struct cdev.  Add a new typedef 'cdev_t' for cdev
pointers.  Temporarily retain dev_t for cdev pointers until the kernel can
be converted over to cdev_t.

14 years agoGet rid of a struct device naming conflict.
Matthew Dillon [Sat, 9 Sep 2006 18:54:36 +0000 (18:54 +0000)]
Get rid of a struct device naming conflict.

14 years agoAttempt to fix a vnode<->namecache deadlock in NFS's handling of stale
Matthew Dillon [Sat, 9 Sep 2006 18:29:09 +0000 (18:29 +0000)]
Attempt to fix a vnode<->namecache deadlock in NFS's handling of stale
file handles.  This is a fairly bad hack.

Reported-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>,
     Joerg Sonnenberger <joerg@britannica.bec.de>

14 years agoRevert last commits and make man treating the argument
Simon Schubert [Sat, 9 Sep 2006 17:34:46 +0000 (17:34 +0000)]
Revert last commits and make man treating the argument
as file name, as a last resort.