Sascha Wildner [Thu, 3 Apr 2014 07:21:04 +0000 (09:21 +0200)]
ls(1): Restore the traditional behavior of -f implying -a.
OpenBSD's commit msg:
------
restore the traditional behavior of -f implying -a; apparently Keith Bostic
forgot to restore it when the -f flag was put back on 2nd of September 1989,
after being removed on 16th of August as a consequence of issues getting it
working over NFS, so deviation from traditional UNIX behavior in all BSDs
looks like an historical accident; as a side effect, this change accommodates
behavior of this option to IEEE Std 1003.1-2008 (``POSIX.1'').
joint work with jmc@ (who found the inaccuracy in our implementation),
schwarze@ (who provided a detailed tracking of historical facts) and millert@
ok millert@, schwarze@
------
The same patch has been pushed to NetBSD and FreeBSD too.
Submitted-by: Igor Sobrado
Taken-from: OpenBSD
Sascha Wildner [Thu, 3 Apr 2014 07:10:23 +0000 (09:10 +0200)]
Move ieee80211_ratectl_node_init() from drivers into ieee80211.
This fixes the amrr_node_deinit() panics we were getting after my
recent ratectl commit broke it (it was trying to kfree a NULL
pointer).
It should also fix hostap mode for run(4) e.g.
This commit is essentially FreeBSD's r217511.
Pointed-out-by: Imre Vadasz <imre@vdsz.com>
Taken-from: FreeBSD
Sascha Wildner [Thu, 3 Apr 2014 07:00:42 +0000 (09:00 +0200)]
Revert "wlan - Allow NULL field in amrr_node_deinit()"
This reverts commit
d68ce29fbc7cb83d080f0abb8790151cd6fec598.
Matthew Dillon [Thu, 3 Apr 2014 02:09:55 +0000 (19:09 -0700)]
wlan - Allow NULL field in amrr_node_deinit()
* Allow a NULL field in amrr_node_deinit(), prevent kfree() from being
called with NULL, which DragonFly explicitly does not allow.
Matthew Dillon [Wed, 2 Apr 2014 21:41:00 +0000 (14:41 -0700)]
loader - Add slowprint() debugging helper
* Add a slowprint() function and command to enable it
to help trace out problems.
* Used for recent debugging but calls are not intended to be permanently
resident in the code. Leave the infrastructure in place for future
debugging.
Matthew Dillon [Wed, 2 Apr 2014 21:38:54 +0000 (14:38 -0700)]
libstand - Fix excessive memory use by hammer2 module
* Don't try to dynamically allocate 64KB for each open file. Just use
the same static media buffer that normal operations use.
This fixes some severe heap fragmentation and out-of-memory issues
when a lot (20+) modules are specified in /boot/loader.conf that
can cause the boot loader to fail.
* Add getheap() to retrieve heap stats.
Matthew Dillon [Wed, 2 Apr 2014 19:24:57 +0000 (12:24 -0700)]
kernel - Fix mid-kernel-boot lockups due to module order
* Sometimes the kernel would just lock up for no apparent reason. Changing
the preloaded module set (even randomly) could effect the behavior.
* The problem turned out to be an issue with kernel modules needing to
temporarily migrate to another cpu (such as when installing a SWI or
other interrupt handler). If the idle thread on cpu 0 had not yet
bootstrapped, lwkt_switch_return() would not be called properly and
the LWKT cpu migration code would never schedule on the target cpu.
* Fix the problem by handling the idle thread bootstrap case for cpu 0
properly in pc64.
Matthew Dillon [Wed, 2 Apr 2014 19:22:50 +0000 (12:22 -0700)]
boot - boot loader fixes
* Clarify some issues in comments.
* Don't trust the bios memory boundary, start 1K below it.
* Always bounce, not just for floppies. This should help with situations
when loading kernels and modules beyond the 16MB mark.
* Add some debugging memory commands
Sascha Wildner [Wed, 2 Apr 2014 17:56:03 +0000 (19:56 +0200)]
crunch: We use .Mt for mail addresses.
François Tigeot [Wed, 2 Apr 2014 08:04:47 +0000 (10:04 +0200)]
Revert "installer: Export /boot device name in loader.conf"
This reverts commit
8a01d9f7000ef8c274b595edf681b3ff0b17591e.
It turns out putting the boot device name in a sysctl variable for its
use by a rescue system is not such a good idea.
Matthew Dillon [Tue, 1 Apr 2014 23:56:43 +0000 (16:56 -0700)]
crunch - Remove FBSDID
* buildworld bootstrap doesn't seem to be using the world's
sys/cdefs.h, and the installed sys/cdefs.h might not have
the FBSDID hack.
* Remove the FBSDID stuff from crunch to fix buildworld. Wish I had
never invented FBSDID.
François Tigeot [Mon, 31 Mar 2014 20:36:03 +0000 (22:36 +0200)]
mkinitrd(8): Use cpdup to copy files
Preserving hard links
Sascha Wildner [Mon, 31 Mar 2014 17:16:31 +0000 (19:16 +0200)]
kernel/acpi: Fix a x86_64 warning in the ACPI_DEBUG_MEMMAP code.
Sascha Wildner [Mon, 31 Mar 2014 17:02:29 +0000 (19:02 +0200)]
kernel/acpi: Restore ACPI_DEBUG_{CACHE,LOCKS,MEMMAP} functionality.
Unfortunately we can't do this in acdragonfly.h alone since ACPICA
types such as ACPI_SIZE and so on are unavailable in acdragonfly.h.
This is because acpi.h includes acenv.h (and therefore acdragonfly.h)
before actypes.h.
François Tigeot [Sun, 30 Mar 2014 19:55:54 +0000 (21:55 +0200)]
Update crunched binaries tools from FreeBSD
* Fix amd64/x86_64 confusion
* Remove a.out support
François Tigeot [Sun, 30 Mar 2014 19:46:42 +0000 (21:46 +0200)]
Revert "Remove crunchgen(8)."
This reverts commit
97c14704c37c914dbd89b22465a421ffde7fe89d.
Conflicts:
Makefile.inc1
Makefile_upgrade.inc
sbin/ifconfig/Makefile
usr.bin/vi/Makefile
usr.sbin/ppp/bundle.c
usr.sbin/pppd/Makefile
usr.sbin/tcpdump/Makefile
Matthew Dillon [Mon, 31 Mar 2014 01:43:51 +0000 (18:43 -0700)]
hammer2 - multi-target mount part 4/many (stabilization)
* Fix spmp mirror_tid synchronization bug in the flush code. TIDs were
not being allocated properly for the super-root portion of the flush.
* Add in-memory chain dump ioctl.
* Fix another missing modsync.
Matthew Dillon [Mon, 31 Mar 2014 01:42:25 +0000 (18:42 -0700)]
hammer2 - Add debug ioctl
* Add a a debugging command to dump the in-memory chain topology.
Sascha Wildner [Sun, 30 Mar 2014 21:36:05 +0000 (23:36 +0200)]
Unbreak buildworld after recent hammer2 changes.
504565062f34ec55037ac0cf308fe3562f091460 renamed HAMMER2_MAX_KEY to
HAMMER2_KEY_MAX but libstand wasn't adjusted.
Reported-by: David Shao
Dragonfly-bug: <https://bugs.dragonflybsd.org/issues/2658>
Matthew Dillon [Sun, 30 Mar 2014 18:07:36 +0000 (11:07 -0700)]
hammer2 - multi-target mount part 3/many (stabilization)
* Fix cluster memory leaks, fix a malloc_type leak.
* Resync cached PFS (pmp) data in the root inode for the PFS on flush.
(inode allocator).
* Track total DIO memory.
* For now assert on logical mirror mismatches.
* Minor start on infrastructure for mirror synchronization.
* Refactor hammer2_cluster_copy() flags.
Sascha Wildner [Sun, 30 Mar 2014 10:27:07 +0000 (12:27 +0200)]
<sys/cdefs.h>: Simplify the CTASSERT definition better.
Matthew Dillon [Sun, 30 Mar 2014 00:09:08 +0000 (17:09 -0700)]
hammer2 - multi-target mount part 2/many
* Split media and in-memory tracking of TIDs. In-memory tracking is
effectively global, while on-media TIDs are now split into an independent
domain for each PFS plus one for the super-root and freemap topology.
Introduces hammer2_xid_t and reformulates some of the TID and XID
constants.
* Refactor the flush code to no longer depend on media modify_tid/mirror_tid
values during a flush. In-memory XIDs are used for all flush recursions.
* Refactor the transaction mechanisms in order to allow XIDs to cover multiple
TID domains. For the moment, the transaction management structure is
global but the code is designed to allow us to have multiple management
domains in a future commit.
* Associate a PMP structure with each storage topology's super-root and
freemap (still a bit rough).
* Each PFS now accounts for its own inode numbers and transaction ids,
independent of other PFSs. This also means that both inode numbers
and transaction ids are PFS-specific, so multiple storage medias which
are part of the same PFS will use the PFS-centric tracking rather than
per-media tracking.
* Cleanup a few memory leaks and remove some debugging from the chain code.
* Refactor mount and unmount code to handle the new PFS abstraction.
* Refactor the volume data lock/unlock/modify mechanics somewhat.
Sascha Wildner [Sat, 29 Mar 2014 09:07:03 +0000 (10:07 +0100)]
<sys/cdefs.h>: Use _Static_assert for CTASSERT.
Move CTASSERT from <sys/systm.h> to <sys/cdefs.h> and use _Static_assert
for its definition (which gives us a better error message in case it is
triggered).
But if there is no support for _Static_assert or static_assert in the
compiler, use the previous typedef method.
Sascha Wildner [Sat, 29 Mar 2014 08:06:44 +0000 (09:06 +0100)]
kernel/hptmv: Don't mess up the struct by copying too much.
VendorId is only 8 chars.
Sascha Wildner [Fri, 28 Mar 2014 22:18:02 +0000 (23:18 +0100)]
kernel/cpu_topology: Fix a casting issue in the topology tree printing.
compute_unit_id is uint8_t so it's actually 0xff we want to check for.
Before this commit, the "... != -1" checks were always true.
Sascha Wildner [Fri, 28 Mar 2014 21:09:55 +0000 (22:09 +0100)]
kernel/acpi_fujitsu: Doh, fix a copy/paste error I did in my last upgrade.
Sascha Wildner [Fri, 28 Mar 2014 20:54:07 +0000 (21:54 +0100)]
<sys/hash.h>: Fix wrong header guard.
Sascha Wildner [Thu, 27 Mar 2014 13:49:25 +0000 (14:49 +0100)]
kernel/acpi: Improve the errors which AcpiOsTableOverride() returns.
While here, use ACPI_COMPARE_NAME() and ACPI_SIG_DSDT.
Sascha Wildner [Wed, 26 Mar 2014 09:42:11 +0000 (10:42 +0100)]
Fix the usbd rc script a bit better.
Most importantly, take care about the case where usb isn't in the
kernel config.
Reported-by: lhmwzy on users@
Sascha Wildner [Tue, 25 Mar 2014 09:35:13 +0000 (10:35 +0100)]
Sync zoneinfo database with tzdata2014b from ftp://ftp.iana.org/tz/releases
* africa: Morocco's 2014 DST start will be as predicted. (Thanks to
Sebastien Willemijns.)
* australasia: Add comments about Johnston Island time in the 1960s.
(Thanks to Lyle McElhaney.)
* antarctica: New entry for Troll station, Antarctica. (Thanks to
Paul-Inge Flakstad and Bengt-Inge Larsson.) This is currently an
approximation; a better version will require zic and localtime
fixes. Also the NZAQ rules - which are never used - have been
removed.
* europe: Crimea switches to Moscow time on 2014-03-30 at 02:00 local
time. (Thanks to Alexander Krivenyshev.)
Move its zone.tab entry from UA to RU.
Sascha Wildner [Tue, 25 Mar 2014 03:32:16 +0000 (04:32 +0100)]
kernel/i386: Remove CPU_ENABLE_TCC traces.
Sepherosa Ziehau [Mon, 24 Mar 2014 13:22:27 +0000 (21:22 +0800)]
inpcb: Use a special version of in_pcbinsporthash() on TCP accept path
It holds the port token for in_pcbinsporthash(). Assert that the port
token is held in in_pcbinsport() now.
Sascha Wildner [Mon, 24 Mar 2014 08:59:54 +0000 (09:59 +0100)]
Fix kernel build.
Antonio Huete Jimenez [Thu, 20 Mar 2014 15:26:26 +0000 (08:26 -0700)]
netstat - Replace a couple malloc calls.
- Use calloc to ensure memory is zeroed.
Sepherosa Ziehau [Sun, 23 Mar 2014 13:30:27 +0000 (21:30 +0800)]
inpcb: Group port related fields into inpcbportinfo
Prepare for port token splitting.
Sascha Wildner [Sun, 23 Mar 2014 11:53:57 +0000 (12:53 +0100)]
libc: Remove a useless assignment.
Antonio Huete Jimenez [Sun, 23 Mar 2014 01:30:01 +0000 (02:30 +0100)]
ext2fs - Attempt to fix a panic on VOP_CLOSE
Follows-up commit
12cdc371fb859682eda7d1c26c5059920a7cf118
Sascha Wildner [Sat, 22 Mar 2014 20:37:55 +0000 (21:37 +0100)]
usb4bsd: Sync the run(4) driver with FreeBSD.
This adds support for a number of new chipsets/adapters (see the
manual page for a list).
Thanks to Markus Pfeiffer for testing with a RT5370 based adapter.
Taken-from: FreeBSD
Sascha Wildner [Sat, 22 Mar 2014 17:44:28 +0000 (18:44 +0100)]
kernel/802_11: Fix panic when the desired ratectl algorithm is unavailable.
Set the default to NONE. When a driver calls ieee80211_ratectl_init(), try
to use AMRR etc. if NONE is selected. Drivers may still select other
algorithms using ieee80211_ratectl_set() prior to calling _init().
Due to our broken ieee80211_load_module(), we do not currently autoload
the desired ratectl algorithm like FreeBSD does. Instead, a message about
manually loading the module will be printed (see ieee80211_dragonfly.c).
Reported-by: profmakx
Taken-from: FreeBSD
Sepherosa Ziehau [Sat, 22 Mar 2014 10:45:21 +0000 (18:45 +0800)]
in6_pcbbind: Rearrange local port bind/select code path.
Mainly group local port binding logic together. This paves way for the
upcoming port token splitting.
Tested-by: ftigeot@
Matthew Dillon [Sat, 22 Mar 2014 01:01:44 +0000 (18:01 -0700)]
mount_hammer2 - Check ERANGE for mount integration result
* For now the H2 vfs code returns ERANGE when mounting multiple
elements of a cluster, report it as an integration instead of
as an error.
Matthew Dillon [Sat, 22 Mar 2014 00:18:33 +0000 (17:18 -0700)]
newfs_hammer2 - add -R uuid and -I uuid
* Allow the super-root (-R) cluster id and the labeled PFS's cluster id
to be specified.
* Populate the cluster id in the inode data.
Matthew Dillon [Sat, 22 Mar 2014 00:15:22 +0000 (17:15 -0700)]
hammer2 - skeleton multi target mount detection
* Detect H2 filesystems with the same cluster id and add to an existing
mount's cluster instead of mounting separately.
* Very preliminary, not operational. Transactions, flush, and volume
management need to be rewritten.
Matthew Dillon [Thu, 20 Mar 2014 00:28:42 +0000 (17:28 -0700)]
hammer2 - cluster API, preliminary implementation of multi-chain updates
* Separate out read-only from read-write access to cluster data.
* Implement hammer2_cluster_modsync() and call in appropriate places.
* Clean up some media inode fields not currently in-use.
Sascha Wildner [Fri, 21 Mar 2014 20:39:12 +0000 (21:39 +0100)]
kernel/msdosfs: Merge deName and deExtension in the direntry struct.
* winChkSum() in msdosfs_conv.c was assuming that deExtension was
following deName in memory, which is meh.
* While here, add some include guards and put some reserved bytes
explicitly in the structs.
This is basically r203827 from FreeBSD.
Sascha Wildner [Thu, 20 Mar 2014 22:47:56 +0000 (23:47 +0100)]
Fix MLINKS for the last sbuf(9) update.
Joris Giovannangeli [Thu, 20 Mar 2014 17:23:59 +0000 (18:23 +0100)]
kernel: check that p_ucred is not NULL before calling p_trespass
Sascha Wildner [Thu, 20 Mar 2014 10:55:14 +0000 (11:55 +0100)]
crashinfo(8): Add -w and -R to the ps command.
Sascha Wildner [Thu, 20 Mar 2014 09:17:14 +0000 (10:17 +0100)]
mpt(4): Sync with FreeBSD.
Nothing big, as far as I can tell. Some fixes and removal of old
compatibility code.
While here, use callout_init_mp() and callout_stop_sync().
Sascha Wildner [Wed, 19 Mar 2014 22:11:54 +0000 (23:11 +0100)]
Add some missing MLINKS for lock.9 and pidfile.3.
Sascha Wildner [Wed, 19 Mar 2014 20:50:20 +0000 (21:50 +0100)]
tbridge.9: Adjust typedefs a bit.
Sascha Wildner [Wed, 19 Mar 2014 18:58:36 +0000 (19:58 +0100)]
aac(4): Sync with FreeBSD.
Some bug fixes, static/const, etc.
Also, add MSI support and use callout_init_mp().
Sascha Wildner [Wed, 19 Mar 2014 17:37:01 +0000 (18:37 +0100)]
usb4bsd/run: Fix compilation without INVARIANTS.
Since wlan_serializer is now held consistently, we can unconditionally
wlan_serialize_exit/enter around firmware_get().
This commit is identical to
11d169ef87e7b77e152510810bec814b36ac584d
for iwn(4).
Sepherosa Ziehau [Wed, 19 Mar 2014 12:36:12 +0000 (20:36 +0800)]
in6_pcblookup_local: Make in6_addr const
Sepherosa Ziehau [Wed, 19 Mar 2014 11:52:08 +0000 (19:52 +0800)]
in_pcbbind: Rearrange local port bind/select code path.
- Group local port binding logic together.
- Group local port auto-selecting logic together and factor out
in_pcbsetlport().
This paves way for the upcoming port token splitting.
Sascha Wildner [Wed, 19 Mar 2014 09:17:32 +0000 (10:17 +0100)]
Fix license in some USB related manual pages (taken from FreeBSD).
Matthew Dillon [Wed, 19 Mar 2014 05:17:08 +0000 (22:17 -0700)]
hammer2 - Cluster API stabilization
* Fix issues that come up with mount/umount/cpdup tests related to the
new cluster API. Ref-counting, locks, and many other (expected) items.
* Fleshing out cluster->focus was a bit more involved than I had hoped.
* Generally speaking, cluster refs and locks work the same way that chain
refs and locks do.
Sascha Wildner [Tue, 18 Mar 2014 18:21:31 +0000 (19:21 +0100)]
Try to start usbd(8) only if it actually exists.
It is specific to the old USB stack so we can remove the rc script
once we don't offer it anymore. For now, just check if it is there
before starting.
Matthew Dillon [Tue, 18 Mar 2014 16:35:21 +0000 (09:35 -0700)]
hammer2 - Start adding internal cluster API
Initial addition of the cluster API. H2 will be non-operational until this
is stabilized. Adding the cluster API will require a few stages. This first
stage is to add the API and make it work under degenerate (single-target)
conditions.
The hammer2_cluster structure collects and manages an array of up to
8 chains representing mirrors / cluster nodes / copies of the same point
in the topology.
* Add hammer2_cluster.c, and hammer2_cluster_t
* Replace nearly all high-level (vnops, vfsops) chain calls with
cluster calls.
Sascha Wildner [Tue, 18 Mar 2014 08:37:31 +0000 (09:37 +0100)]
usb4bsd: FWIW, port usb_template and hook it into the build.
Sascha Wildner [Tue, 18 Mar 2014 07:54:36 +0000 (08:54 +0100)]
Add umoscom(4) and usfs(4) manual pages (from FreeBSD).
Sascha Wildner [Tue, 18 Mar 2014 07:30:31 +0000 (08:30 +0100)]
tcplay: Fix a function name in tcplay.3 and .Dd in tcplay.8.
Sascha Wildner [Mon, 17 Mar 2014 18:32:37 +0000 (19:32 +0100)]
Remove old tcplay(3) MLINKS via 'make upgrade'.
Antonio Huete Jimenez [Mon, 17 Mar 2014 14:33:21 +0000 (07:33 -0700)]
test/debug - Tweak vnodeinfo a bit
- Remove trailing white spaces.
- Fix usage.
- Add '-p' option to show filesystem specific inode data. Currently
only dirfs is supported.
- Additional filesystems can be added easily to the dumplist struct.
Sascha Wildner [Sun, 16 Mar 2014 20:39:25 +0000 (21:39 +0100)]
libusb20.3: uint61_t -> uint16_t
Sascha Wildner [Sun, 16 Mar 2014 20:39:01 +0000 (21:39 +0100)]
usbhid.3: u_int -> unsigned int
Sascha Wildner [Sun, 16 Mar 2014 20:01:13 +0000 (21:01 +0100)]
devd: Regenerate usb.conf.
Sascha Wildner [Sun, 16 Mar 2014 17:57:13 +0000 (18:57 +0100)]
usb.4: Add ufm reference.
Sascha Wildner [Sun, 16 Mar 2014 17:43:38 +0000 (18:43 +0100)]
kernel/twa: Sync a bit with FreeBSD.
Mainly s/callout_init/callout_init_mp/ and using pci_enable_busmaster()
directly.
Sascha Wildner [Sun, 16 Mar 2014 17:49:38 +0000 (18:49 +0100)]
usb4bsd: Port ufm.
Alex Hornung [Sun, 16 Mar 2014 16:55:59 +0000 (16:55 +0000)]
tcplay - fix compile error on i386
Alex Hornung [Sun, 16 Mar 2014 13:45:21 +0000 (13:45 +0000)]
cryptdisks safe_mem - use weak symbols
* Use weak symbols so that the ones in libtcplay.a can override the
symbols in the local safe_mem.
* I'm not sure why this ever worked before, but it definitely doesn't
work anymore.
Alex Hornung [Sun, 16 Mar 2014 13:45:09 +0000 (13:45 +0000)]
cryptdisks - update to new tcplay API
Alex Hornung [Sun, 16 Mar 2014 07:37:01 +0000 (07:37 +0000)]
update tcplay to 2.0
Sepherosa Ziehau [Sun, 16 Mar 2014 13:18:02 +0000 (21:18 +0800)]
in6_pcbbind: Move stack variable declaration near its usage
Sepherosa Ziehau [Sun, 16 Mar 2014 13:09:11 +0000 (21:09 +0800)]
in_pcbbind: Move stack variable declaration near its usage
Sepherosa Ziehau [Sun, 16 Mar 2014 13:05:19 +0000 (21:05 +0800)]
inpcb: Break long line
Sepherosa Ziehau [Sun, 16 Mar 2014 12:45:55 +0000 (20:45 +0800)]
inpcb: in_pcbconn_bind -> in_pcbsetlport_remote; no functional change.
Sepherosa Ziehau [Sun, 16 Mar 2014 12:07:51 +0000 (20:07 +0800)]
in_pcbconn_bind: Change argument name from 'nam' to 'remote'
Increase readability.
Sepherosa Ziehau [Sun, 16 Mar 2014 11:48:05 +0000 (19:48 +0800)]
inpcb: in_pcblookup_addrport -> in_pcblookup_localremote
Compared w/ in_pcblookup_local, in_pcblookup_addrport checks remote
address and port (faddr,fport) too. Renaming it to
in_pcblookup_localremote to be consistent w/ in_pcblookup_local name.
No functional changes.
Sepherosa Ziehau [Sun, 16 Mar 2014 11:34:18 +0000 (19:34 +0800)]
inpcb/in6pcb: in_pcbinsporthash() never fails
Sepherosa Ziehau [Sun, 16 Mar 2014 07:31:50 +0000 (15:31 +0800)]
in6pcb: in6_pcblookup_local() must have been protected by port token
Sepherosa Ziehau [Sun, 16 Mar 2014 07:31:32 +0000 (15:31 +0800)]
inpcb: in_pcblookup_local() must have been protected by port token
Sepherosa Ziehau [Sun, 16 Mar 2014 07:24:08 +0000 (15:24 +0800)]
in6pcb: Properly hold port token for in6_pcbbind() and in6_pcbsetport()
Sascha Wildner [Sun, 16 Mar 2014 11:19:36 +0000 (12:19 +0100)]
man.1: Add missing section.
Antonio Huete Jimenez [Sun, 16 Mar 2014 01:12:16 +0000 (02:12 +0100)]
gdbinit - Fix previous commit.
Spotted-by: swildner
Antonio Huete Jimenez [Sun, 16 Mar 2014 00:34:53 +0000 (01:34 +0100)]
gdbinit - A couple fixes.
- Fix 'lstok' for proc_tokens and vmobj_tokens.
Sascha Wildner [Sat, 15 Mar 2014 21:07:09 +0000 (22:07 +0100)]
usb.4: Use .Nx and fix a wrong reference.
Antonio Huete Jimenez [Sat, 15 Mar 2014 15:05:44 +0000 (16:05 +0100)]
libkvm - Adjust for the proc_token changes
- allproc, zombproc are now gone. All processes are now in an array of
proclist structures.
- ps(1) did not work on coredumps because '_allproc' and '_zombproc'
were missing.
Antonio Huete Jimenez [Fri, 14 Mar 2014 09:04:19 +0000 (10:04 +0100)]
kernel - Add allproc_hsize global
- Used by kvm(3) to determine the proc hash size.
François Tigeot [Sat, 15 Mar 2014 16:37:33 +0000 (17:37 +0100)]
installer: Export /boot device name in loader.conf
This will eventually allow the kernel to mount /boot directly
Matthew Dillon [Sat, 15 Mar 2014 16:34:48 +0000 (09:34 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Sat, 15 Mar 2014 16:28:55 +0000 (09:28 -0700)]
dntpd - Improve offset correction precision
* Offset corrections are more accurate with more samples (whereas frequency
corrections are more accurate when the first and last sample are far apart
in the time domain).
* Even though dntpd uses a standard deviation test to conditionalize offset
corrections, it is still possible to get false positives with only 4
samples.
* Once dntpd has gone into maintainance mode increase the number of
required samples for an offset correction from 4 to 8. This makes
the standard deviation more accurate, reduces false positives, and
improves stability.
* Retain the original mechanic during aquisition so dntpd applies corrections
sooner rather than later when first started.
Sascha Wildner [Sat, 15 Mar 2014 16:28:19 +0000 (17:28 +0100)]
Update the pciconf(8) database.
March 14, 2014 snapshot from http://pciids.sourceforge.net/
Sepherosa Ziehau [Sat, 15 Mar 2014 13:30:55 +0000 (21:30 +0800)]
inpcb: in_pcblookup_addrport() must have been protected by port token
Sepherosa Ziehau [Sat, 15 Mar 2014 13:05:09 +0000 (21:05 +0800)]
in_pcbconn_bind: Reduce port token cover range
Sepherosa Ziehau [Sat, 15 Mar 2014 12:46:46 +0000 (20:46 +0800)]
in_pcbconn_bind: Remove duplicated prison_replace_wildcards()
Sepherosa Ziehau [Sat, 15 Mar 2014 12:43:58 +0000 (20:43 +0800)]
in_pcbbind: Merge duplicated prison_replace_wildcards()
Sascha Wildner [Sat, 15 Mar 2014 11:55:35 +0000 (12:55 +0100)]
Some usb4bsd manpage related shuffling/additions in Makefiles.
Sascha Wildner [Sat, 15 Mar 2014 09:20:58 +0000 (10:20 +0100)]
usb.4: Sync parts of the manual page with FreeBSD.