Jan Lentfer [Sat, 19 Dec 2009 12:04:17 +0000 (13:04 +0100)]
top: Remove WCPU, implement CTIME.
* WCPU is a quite useless value for a long time,
thus it has been replaced with
* CTIME, which presents the same value as 'ps -S'
does for the CPU value. This is the process's own
cpu time plus the cpu time of it's exited child
processes
Sascha Wildner [Fri, 18 Dec 2009 21:10:57 +0000 (22:10 +0100)]
Fix buildworld with -DNDEBUG in CFLAGS.
ANSI C requires that if NDEBUG is defined, assert(3) shall be disabled,
that is, defined "((void)0)".
The usage of NDEBUG in libevent is not related to assert(3), so simply
remove the #ifdef's (as was done upstream in later versions of
libevent).
Joe Talbott [Fri, 18 Dec 2009 15:56:46 +0000 (10:56 -0500)]
examples/kld - Fix for devfs.
Remove dev_ops_{add,remove} and remove mknod from Makefile.
Alex Hornung [Fri, 18 Dec 2009 10:22:33 +0000 (10:22 +0000)]
CONFIG - untangle whitespace mess
* fix whitespace mess introduced by previous commit.
Alex Hornung [Fri, 18 Dec 2009 09:39:37 +0000 (09:39 +0000)]
CONFIG - add NULLFS
* add option NULLFS to all kernel config files, as nullfs is more or
less a requirement for a hammer system due to pfs.
* Incidentally this solves problems when mounting /usr and other PFS
when the modules are out of sync with the kernel; for example after a
bad buildkernel and using an old kernel.
Discussed-With: Matt Dillon
Alex Hornung [Fri, 18 Dec 2009 08:54:21 +0000 (08:54 +0000)]
dump - rename md dumpsys to md_dumpsys
* Rename the machine dependent dumpsys() to md_dumpsys().
* Add dumpsys() which takes no arguments, giving backward compatibility
and the possibility to use 'call dumpsys' from the ddb> prompt.
dumpsys() then calls md_dumpsys(), except for vkernels.
Alex Hornung [Thu, 17 Dec 2009 15:29:15 +0000 (15:29 +0000)]
devfs - Populate getattr va_size for disks properly
* As getattr() va_size is used for lseek SEEK_END, we need to set
va_size for disk devices to the underlying device's media size.
* Beware that *stat() will not be affected, and st_size will remain
the size of the devfs_node structure, as it has been so far.
Dragonfly-bug: http://bugs.dragonflybsd.org/issue1632
Reported-by: Steve O'Hara-Smith
Jan Lentfer [Thu, 17 Dec 2009 14:57:20 +0000 (15:57 +0100)]
ncurses: Update master to work with v5.7 in contrib.
Jan Lentfer [Thu, 17 Dec 2009 14:05:32 +0000 (15:05 +0100)]
Merge branch 'vendor/NCURSES'
Jan Lentfer [Thu, 17 Dec 2009 14:03:46 +0000 (15:03 +0100)]
vendor/NCURSES: Import v5.7 sources
Jan Lentfer [Thu, 17 Dec 2009 13:56:13 +0000 (14:56 +0100)]
ncurses: Update to work without version tag.
Jan Lentfer [Thu, 17 Dec 2009 13:17:25 +0000 (14:17 +0100)]
Merge branch 'vendor/NCURSES'
Jan Lentfer [Thu, 17 Dec 2009 13:10:54 +0000 (14:10 +0100)]
vendor/NCURSES: Remove version tag.
Jan Lentfer [Thu, 17 Dec 2009 08:18:50 +0000 (09:18 +0100)]
kern/kern_kinfo.c: Fix incorrect population of kinfo_proc.kp_cru
Sascha Wildner [Wed, 16 Dec 2009 08:36:59 +0000 (09:36 +0100)]
Fix some warnings that creep up when compiling without INVARIANTS.
* Add a __debugvar define to <sys/systm.h> which is empty when INVARIANTS
is set (default) and __attribute__((unused)) when INVARIANTS is not set.
* Use __debugvar in the declaration of variables which are only used in a
KASSERT or KKASSERT to avoid 'unused variable' warnings.
Suggested-by: aggelos
Reported-by: Joel K. Petterson <joelkpettersson@gmail.com>
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1629>
Sascha Wildner [Wed, 16 Dec 2009 08:22:35 +0000 (09:22 +0100)]
Fix the SOEKRIS build (kernel without I686_CPU).
Sascha Wildner [Wed, 16 Dec 2009 08:21:48 +0000 (09:21 +0100)]
Fix the LINT build.
Michael Neumann [Wed, 16 Dec 2009 00:10:40 +0000 (01:10 +0100)]
Unbreak HAMMER root mounts
Absolute device names in vfs.root.mountfrom where handled incorrectly.
Reported-by: aggelos
Matthew Dillon [Tue, 15 Dec 2009 20:31:02 +0000 (12:31 -0800)]
kernel - adjust falloc and arguments to dupfdopen, fsetfd, fdcheckstd
* Make changes to the pointer type passed (proc, lwp, filedesc) to
numerous routines.
* falloc() needs access to td_ucred (it was previously using p_ucred which
is not MPSAFE).
* Adjust fsetfd() to make it conform to the other fsetfd*() procedures.
* Related changes to fdcheckstd() and dupfdopen().
Matthew Dillon [Tue, 15 Dec 2009 18:43:48 +0000 (10:43 -0800)]
kernel - use new td_ucred in numerous places
* Use curthread->td_ucred in numerous places, primarily system calls,
where curproc->p_ucred was used before.
* Clean up local variable use related to the above.
* Adjust several places where p_ucred is replaced to properly deal
with lwp threading races to avoid accessing and freeing a potentially
stale ucred.
* Adjust static procedures in the ktrace code to generally take lwp
pointers instead of proc pointers.
Matthew Dillon [Tue, 15 Dec 2009 18:41:50 +0000 (10:41 -0800)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Sascha Wildner [Tue, 15 Dec 2009 14:52:07 +0000 (15:52 +0100)]
Add -Werror to the kernel build.
* Add a new 'nowerror' keyword to config(8) for parsing the 'files' files.
When specified, it will prevent the file from being compiled with
-Werror.
Note that files that have their own compile-with rules need "${WERROR}"
added to get -Werror.
* Adjust all 'files' files accordingly: Set nowerror where we still warn
(i386 or x86_64) and add "${WERROR}" to compile-with rules where it
applies.
* Remove explicit -Werror-implicit-function-declaration setting as it is
now useless (-Wimplicit-function-declaration is already part of -Wall).
You will need an updated config(8) (i.e., one that understands 'nowerror')
for this to work so don't forget to quick/buildworld before recompiling
the kernel. nativekernel users will have to update their host config(8).
NO_WERROR in /etc/make.conf applies here as well and turns off the setting
of -Werror for the kernel, too.
In-discussion-with: corecode, aggelos, and others
Sascha Wildner [Tue, 15 Dec 2009 14:10:40 +0000 (15:10 +0100)]
mp_machdep.c: Fix some warnings of the SMP without APIC_IO build.
Alex Hornung [Tue, 15 Dec 2009 12:42:14 +0000 (12:42 +0000)]
LINT - clear a bunch of warnings
* Clear a bunch of warnings that appear when building LINT kernel.
Sascha Wildner [Tue, 15 Dec 2009 11:14:30 +0000 (12:14 +0100)]
Fix warning in the UP build.
Matthew Dillon [Tue, 15 Dec 2009 08:24:00 +0000 (00:24 -0800)]
kernel - Third time's the charm - move lwp_ucred to td_ucred
* Easier access to the cached ucred via the thread structure instead of
the lwp.
Alex Hornung [Tue, 15 Dec 2009 08:00:53 +0000 (08:00 +0000)]
savecore(8) - Update manpage
* Update savecore man page to reflect changes to the savecore utility itself.
Obtained-from: FreeBSD
Matthew Dillon [Tue, 15 Dec 2009 04:40:54 +0000 (20:40 -0800)]
kernel - Make adjustments to the per-lwp ucred
* Rename lwp_syscall_ucred to lwp_ucred and move ucred synchronization
from syscall2() to userenter(). It now runs on any user entry into
the kernel, not just for system calls.
NOTE: A VM fault from the kernel itself does not call userenter() and
does not trigger another synchronization. This is the behavior
we desire, to ensure that lwp_ucred is stable.
There are numerous non-syscall cases (e.g. signal, fault, etc) where
we may need to access a safe ucred.
* Remove the SMP conditional. lwp_ucred is now tracked for both UP and SMP.
There is no point optimizing this for UP since there is basically no cost
for tracking lwp_ucred.
Matthew Dillon [Tue, 15 Dec 2009 03:29:43 +0000 (19:29 -0800)]
kernel - move proc->p_lock, add lwp ucred caching
* Move p_lock out from the struct proc copy area, change it from a
char to an int, and make PHOLD/PRELE MPSAFE.
* System calls now cache a read-only copy of p->p_ucred in
lp->lwp_syscall_ucred. This will be used by system call code
to access the ucred in a MPSAFE manner.
Sascha Wildner [Mon, 14 Dec 2009 20:16:50 +0000 (21:16 +0100)]
vga(4): Simplify conditions for mode rejection somewhat.
When mapping the video parameter table to our mode_map, only reject modes
which are neither 40x25, 80x25 nor 80x30, instead of rejecting or accepting
the whole table based on comparing the initial mode's BIOS and power-up
register settings.
Add a VGA_KEEP_POWERON_MODE option which (when set) preserves the register
settings which were found for the initial mode upon power-on for switching
back to it (or some mode derived from it) from another mode. The power-on
settings are also taken for the initial mode in case the video parameter
table has no valid settings for it.
This is kinda similar to what's proposed in FreeBSD's kern/64114.
Joe Talbott [Mon, 14 Dec 2009 17:25:39 +0000 (12:25 -0500)]
periodic - Don't include timestamps in e-mail output by default.
Matthew Dillon [Mon, 14 Dec 2009 16:30:08 +0000 (08:30 -0800)]
HAMMER Utility - fix unfreeable pointer in 'hammer info'
* Fix the construction of the mountedon variable to ensure the
variable can be free()'d.
Reported-by: "Steve O'Hara-Smith" <steve@sohara.org>
Also-suggested-by: Antonio Huete Jimenez <ahuete.devel@gmail.com>
Matthew Dillon [Mon, 14 Dec 2009 16:01:57 +0000 (08:01 -0800)]
HAMMER Utility - Fix seg-fault in hammer cleanup if blank lines in config
* If the HAMMER config contained blank lines the hammer cleanup
directive would seg-fault trying to parse it.
* Ignore blank lines. Also ignore lines beginning with a '#' when
parsing the config.
Reported-by: elekktretterr@exemail.com.au
Sascha Wildner [Mon, 14 Dec 2009 04:37:36 +0000 (05:37 +0100)]
setlocale.3: Correct function name.
Sascha Wildner [Mon, 14 Dec 2009 04:24:37 +0000 (05:24 +0100)]
pnp.4: Comment out some docs (code either gone or #if 0'd).
Sascha Wildner [Mon, 14 Dec 2009 03:38:51 +0000 (04:38 +0100)]
rman.9: Comment out documentation for rman_await_resource().
The code is #if 0'd too.
Sascha Wildner [Mon, 14 Dec 2009 03:27:35 +0000 (04:27 +0100)]
vcount.9: count_udev() no longer exists.
See
aec8eea47f8acbdec20843e7d8e7bc6d4255cce3
Sascha Wildner [Mon, 14 Dec 2009 03:06:04 +0000 (04:06 +0100)]
pthread_rwlock_timedwrlock.3: Correct function name.
Sascha Wildner [Mon, 14 Dec 2009 02:55:41 +0000 (03:55 +0100)]
sleep.9: Adjust for serialize_sleep() -> zsleep() rename.
See
ed3f6624e9093425dd733be90efc696df86c9e0e
Also, sync the MLINKS with the manual page.
Matthew Dillon [Sun, 13 Dec 2009 20:19:30 +0000 (12:19 -0800)]
kernel - Move MP lock inward, plus misc other stuff
* Remove the MPSAFE flag from the syscalls.master file. All system calls
are now called without the MP lock held and will acquire the MP lock if
necessary.
* Shift the MP lock inward. Try to leave most copyin/copyout operations
outside the MP lock. Reorder some of the copyouts in the linux emulation
code to suit.
Kernel resource operations are MP safe.
Process ucred access is now outside the MP lock but not quite MP safe
yet (will be fixed in a followup).
* Remove unnecessary KKASSERT(p) calls left over from the time before
system calls where prefixed with sys_*
* Fix a bunch of cases in the linux emulation code when setting groups
where the ngrp range check is incorrect.
Michael Neumann [Sun, 13 Dec 2009 20:15:09 +0000 (21:15 +0100)]
Remove misleading comments
Michael Neumann [Sun, 13 Dec 2009 18:56:55 +0000 (19:56 +0100)]
HAMMER - Stabilize and refactor volume removal
* Correctly handle already freed layer1 entries.
* Correclty avoid reallocation on the volume to be removed
during reblocking.
* Erase volume header after removal so that trying to mount it
afterwards will fail (otherwise mount will silently use it).
Sascha Wildner [Sun, 13 Dec 2009 18:26:08 +0000 (19:26 +0100)]
libc: Some warning fixes.
Sascha Wildner [Sun, 13 Dec 2009 16:19:57 +0000 (17:19 +0100)]
Sync libc's un-namespace.h with namespace.h.
Alexander Polakov [Sat, 12 Dec 2009 10:10:39 +0000 (13:10 +0300)]
i386: sync k6_mem.c with FreeBSD
* style(9)
* replace strcmp() with cpu_vendor_id
Alexander Polakov [Sat, 12 Dec 2009 09:33:21 +0000 (12:33 +0300)]
i386: sync i686_mem with FreeBSD
* style cleanup
* replace cpu_vendor with cpu_vendor_id
* fix performance problem with some cpus
(see http://svn.freebsd.org/viewvc/base?view=revision&revision=177125)
* add constants for the various fields in MTRR registers
* hide MTRR under bootverbose
Alexander Polakov [Sat, 12 Dec 2009 07:37:33 +0000 (10:37 +0300)]
identcpu.c: add PLCMULDQ, AES and VMM feature strings
Alexander Polakov [Sat, 12 Dec 2009 21:47:27 +0000 (00:47 +0300)]
i386: sync identcpu with FreeBSD
* more feature strings
* core/package detection (from x86_64)
* replace cpu_vendor with cpu_vendor_id
* move cpu identification variables declaration
to initcpu.c
Michael Neumann [Sat, 12 Dec 2009 20:58:46 +0000 (21:58 +0100)]
HAMMER - Fix reblocking operation of volume removal
* Reblock everything
* Interruptable
* Sync before free L1/L2 entries (bigblocks might be reserved).
* Fix a bug (layer2->zone vs. layer2_zone)
Matthew Dillon [Sat, 12 Dec 2009 18:50:22 +0000 (10:50 -0800)]
HAMMER VFS - Fix incorrect hammer_cursor_removed_node() call in btree_remove()
* hammer_cursor_removed_node() was being called on the wrong node. This
fixes a parent != NULL assertion later on.
* There is still at least one known issue where btree_iterate can panic
due to a cursor tracking issue that has not yet been located.
Michael Neumann [Sat, 12 Dec 2009 13:02:03 +0000 (14:02 +0100)]
HAMMER - Fix panic when removing volume
* unload buffers before unloading volume
* fix missing call to hammer_rel_volume() which lead to
kernel panic upon umount
Sascha Wildner [Sat, 12 Dec 2009 12:05:15 +0000 (13:05 +0100)]
Remove inclusion of <sys/cdefs.h> from kernel .c files.
They are not needed and mostly leftovers from former __FBSDID and the
likes.
Also add some missing CVS ids to the comments.
Michael Neumann [Sat, 12 Dec 2009 01:25:41 +0000 (02:25 +0100)]
HAMMER - Stabilize volume removal
* assert that L1 entries point to the volume
* don't leave the volume in an inconsistent state
if it is not completely empty.
Michael Neumann [Sat, 12 Dec 2009 00:02:39 +0000 (01:02 +0100)]
Fix typo
Michael Neumann [Sat, 12 Dec 2009 00:00:22 +0000 (01:00 +0100)]
HAMMER - Fix umount flushing issue for volume-add/del
Only changes to the header of the root volume are automatically
flushed to disk, changes to headers of other volumes not. Do it
manually.
Sascha Wildner [Fri, 11 Dec 2009 23:39:31 +0000 (00:39 +0100)]
Add hotplugd options to the default rc.conf and rc.conf.5.
Michael Neumann [Fri, 11 Dec 2009 21:59:21 +0000 (22:59 +0100)]
HAMMER - Implement experimental volume removal
A volume other than the root volume can be removed with:
hammer volume-del device filesystem
WARNING: Experimental!
Aggelos Economopoulos [Thu, 10 Dec 2009 19:36:55 +0000 (21:36 +0200)]
Kill T/TCP
T/TCP has serious security issues by design and has not
found widespread usage. Keeping it around is also a
maintenance burden.
Jan Lentfer [Thu, 10 Dec 2009 14:02:12 +0000 (15:02 +0100)]
top - bring in "pid" sort order from freebsd module (N-KEY and option -o "pid").
- rework pointer casts in compare_* functions after netbsd module's pattern.
Jan Lentfer [Thu, 10 Dec 2009 11:51:01 +0000 (12:51 +0100)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Jan Lentfer [Thu, 10 Dec 2009 11:49:00 +0000 (12:49 +0100)]
make.conf and make.conf.5: Remove obsolete TOP_TABLE_SIZE option.
Sascha Wildner [Thu, 10 Dec 2009 08:58:15 +0000 (09:58 +0100)]
hotplug.4 & hotplugd.8: Some mdoc cleanup.
Michael Neumann [Wed, 9 Dec 2009 22:34:18 +0000 (23:34 +0100)]
Rename "hammer expand" to "hammer volume-add"
Also rename the ioctl interface function accordingly.
NOTE:
hammer expand fs device
changed to:
hammer volume-add device fs
Sascha Wildner [Wed, 9 Dec 2009 11:36:44 +0000 (12:36 +0100)]
Fix argument order for netmsg_init().
Sascha Wildner [Wed, 9 Dec 2009 09:58:19 +0000 (10:58 +0100)]
Actually hook hotplug.4 into the build.
Jan Lentfer [Tue, 8 Dec 2009 15:30:41 +0000 (16:30 +0100)]
top - enable full command line display.
Sascha Wildner [Tue, 8 Dec 2009 07:55:53 +0000 (08:55 +0100)]
Fix LINT kernel build.
Specifically, building with IPDIVERT & SMP was broken.
Thomas Nikolajsen [Mon, 7 Dec 2009 22:39:09 +0000 (23:39 +0100)]
hammer: update Version 4 description
Matthew Dillon [Mon, 7 Dec 2009 19:59:10 +0000 (11:59 -0800)]
HAMMER VFS - Fix WIP
* Change WIP to 5 so hammer version-upgrade doesn't complain when upgrading
to version 4 (which is no longer WIP).
Reported-by: Johannes Hofmann
Michael Neumann [Mon, 7 Dec 2009 17:51:54 +0000 (18:51 +0100)]
HAMMER Utility - Allow expanding root filesystems again
This reverts commit
e3102897c486b844e6a59e419e7e98a0d1ca3246, because
as of commit
104cb849cad1c21c8929f73d7654e16c4fb8a268, multi-volume
HAMMER root-mounts should work.
Please note that "hammer expand" is still considered EXPERIMENTAL.
Michael Neumann [Mon, 7 Dec 2009 17:18:45 +0000 (18:18 +0100)]
HAMMER - Implement multi-volume root mounts
Alex Hornung [Mon, 7 Dec 2009 10:48:29 +0000 (10:48 +0000)]
dump - Unbreak VKERNELs
* VKERNELs don't support dumping, so add a dummy dumpsys() to indicate this
on VKERNELs.
* hence also don't allocate vm_page_dump.
Reported-by: Simon "corecode" Schubert
Sascha Wildner [Mon, 7 Dec 2009 10:16:18 +0000 (11:16 +0100)]
top(1): Remove top.x from CLEANFILES, since it's not generated.
Matthew Dillon [Mon, 7 Dec 2009 06:56:30 +0000 (22:56 -0800)]
HAMMER Utility - Refuse to expand the root filesystem
* Refuse to expand a HAMMER filesystem mounted as the root filesystem.
The DragonFly kernel is currently not able to mount a multi-volume
HAMMER root.
Reported-by: Jan Lentfer <Jan.Lentfer@web.de>
Matthew Dillon [Mon, 7 Dec 2009 04:28:36 +0000 (20:28 -0800)]
HAMMER Utility - Add -p <ssh-port> option
* Allows the ssh port for remote specification to be specified. This
option is passed as '-p <ssh-port>' to ssh invocations.
Sascha Wildner [Mon, 7 Dec 2009 02:26:30 +0000 (03:26 +0100)]
loader.8: Bump .Dd and fix a manlint warning.
Sascha Wildner [Mon, 7 Dec 2009 02:09:43 +0000 (03:09 +0100)]
Clean up BIND upgrade.
Remove some auto-generated files which were accidentally added.
Alex Hornung [Sun, 6 Dec 2009 19:57:03 +0000 (19:57 +0000)]
libkvm - Update for new dumps & minidumps
* Update libkvm so it will work with the new-style dumps and minidumps.
* If the ELF header of a core dump is not valid, assume it's an old style
dump. This way we can still read old-style dumps after this update.
Obtained-from: FreeBSD
Alex Hornung [Sun, 6 Dec 2009 19:55:51 +0000 (19:55 +0000)]
savecore - Update savecore to work with new dumps/minidumps
* Update savecore so it can save the new ELF dumps and minidumps.
Obtained-from: FreeBSD
Alex Hornung [Sun, 6 Dec 2009 19:52:58 +0000 (19:52 +0000)]
dump - Make use of the new dumping
* Adapt our dumping infrastructure to the new dump type.
* Update all disk/raid dump methods (except aac) to work with the new
dumps. These now don't take matters into their own hands and just write
what they are told to.
Alex Hornung [Sun, 6 Dec 2009 19:48:53 +0000 (19:48 +0000)]
dump - Bring in FreeBSD's dumping (new dumps & minidumps)
* Bring in FreeBSD's dumps and minidumps, which use an ELF header instead of
a raw dump.
* Adapt to our needs by, for example, saving the dumppcb and dumpthread.
Obtained-from: FreeBSD
Alex Hornung [Sun, 6 Dec 2009 19:44:26 +0000 (19:44 +0000)]
pmap - Change pmap_map syntaxis
* Change pmap_map syntaxis to take a pointer to vm_offset_t, so the vaddr is
modified when memory is mapped.
* Hence also don't return the modified vaddr, but rather return the mapped
memory.
* This unifies the syntax for the different platforms, as on amd64, pmap_map
returned the mapped area, instead of the changed vaddr.
* Modify vm_page.c accordingly, and also set aside some memory that minidumps
need.
Jan Lentfer [Sun, 6 Dec 2009 09:33:44 +0000 (10:33 +0100)]
lentferj - commit test.
Sascha Wildner [Sun, 6 Dec 2009 07:55:00 +0000 (08:55 +0100)]
network - Fix SMP build
Reported-by: lentferj
Matthew Dillon [Sun, 6 Dec 2009 05:54:39 +0000 (21:54 -0800)]
network - Fix UP build
Reported-by: Sascha Wildner <saw@online.de>
Matthew Dillon [Sat, 5 Dec 2009 19:45:34 +0000 (11:45 -0800)]
network - Move socket from netmsg ext to netmsg header, add port to socket
These changes should make it easier to debug mbufs routed to the wrong
protocol threads. Note that IPV6 is routed to netisr0, even for tcp and udp
packets, so for now we do not KKASSERT that the port matches directly.
The TCP code still KKASSERTs that the timers are running on the correct cpu,
since the cpu is still correct for IPV6 TCP/UDP packets even when the
protocol thread is wrong (netisr_cpu 0 instead of tcp_thread 0 or
udp_thread 0).
* Instead of recalculating the port based on the inp or mbuf all the time,
add a so_port field to the socket structure directly.
* The socket pointer is now part of the netmsg header, even though some
subsystems do not need it. This allows us to validate the message port
more easily.
Sascha Wildner [Sat, 5 Dec 2009 19:37:22 +0000 (20:37 +0100)]
development.7: Add style(9) xref and a pointer to dragonfly.el.
Sascha Wildner [Sat, 5 Dec 2009 19:30:33 +0000 (20:30 +0100)]
Move tools/tools/editing/dragonfly.el to share/misc.
Sascha Wildner [Sat, 5 Dec 2009 18:51:28 +0000 (19:51 +0100)]
Add a profile for indent(1) which matches our style quite well.
Adapted-from: NetBSD
Sascha Wildner [Sat, 5 Dec 2009 18:14:19 +0000 (19:14 +0100)]
Clean up some Makefiles (let the user decide which options they want).
Aggelos Economopoulos [Sat, 5 Dec 2009 17:57:01 +0000 (19:57 +0200)]
Merge commit 'lentferj/top_update'
Jan Lentfer [Sat, 5 Dec 2009 17:41:19 +0000 (18:41 +0100)]
top - correctly prototype compare_* functions. Remove debug option from Makefile.
Matthew Dillon [Sat, 5 Dec 2009 17:20:27 +0000 (09:20 -0800)]
HAMMER VFS - Change default from version 3 to version 4
* Version 4 is now the default for newfs_hammer and is no longer
considered a work in progress.
* Filesystems may be upgraded to version 4 in-place. The UNDO FIFO
(typically 1G) is reformatted so the upgrade might take a minute or
two depending.
* Version 4 allows the UNDO FIFO to be flushed without also having
to flush the volume header, removing 2 of the 4 disk syncs typically
required for an fsync() and removing 1 of the 2 disk syncs typically
required for a flush sequence.
This is accomplished by adding a sequence number of the UNDO entries
in the FIFO and preventing them from crossing a 512-byte boundary,
so each 512-byte block starts with an entry. HAMMER's mount code
is then able to scan the UNDO FIFO to determine the start and end
for recovery purposes.
* Version 4 implements the infrastructure needed for REDO, but does not
implement any REDO operations.
Jan Lentfer [Sat, 5 Dec 2009 16:58:45 +0000 (17:58 +0100)]
top - style cleanup and removal of unnecessary code
Alexander Polakov [Fri, 4 Dec 2009 22:32:26 +0000 (01:32 +0300)]
Import hotplugd(8) from OpenBSD.
Alexander Polakov [Fri, 4 Dec 2009 21:43:14 +0000 (00:43 +0300)]
Import hotplug(4) pseudo-device.
Obtained-from: OpenBSD.
Matthew Dillon [Fri, 4 Dec 2009 22:32:48 +0000 (14:32 -0800)]
kernel - Fix recursive lockmgr lock panic when creating a hardlink
* HAMMER uses non-recursive vnode locks. Fix a case where attempting
to hardlink A to B where both already exist and point to the same
inode can result in a panic if B is not in the namecache.
HAMMER must temporarily lock the vnode it returns in the resolver
and kern_link() was already holding A locked. This all occurs
before kern_link() is able to check whether A and B are the same
so it can't be avoided.
Solve the problem by unlocking A while resolving B.
Reported-by: Goetz Isenmann <info@goetz-isenmann.de>
Aggelos Economopoulos [Fri, 4 Dec 2009 22:27:49 +0000 (00:27 +0200)]
Merge commit 'lentferj/top_update'
Jan Lentfer [Fri, 4 Dec 2009 21:21:59 +0000 (22:21 +0100)]
top - Import DragonFly specific patches for 3.8beta1.
* fix commands.c renice function to prevent segfault on wrong number of parameters.
Matthew Dillon [Thu, 3 Dec 2009 23:13:02 +0000 (15:13 -0800)]
HAMMER VFS - Fix bug in statvfs()
* statvfs() had a coding error which caused it to return 0 free space
until the first statfs() call was made.
Reported-by: Johannes Hofmann <johannes.hofmann@gmx.de>