dragonfly.git
19 years agoClean up some of the sockbuf append code.
Jeffrey Hsu [Wed, 20 Apr 2005 21:37:06 +0000 (21:37 +0000)]
Clean up some of the sockbuf append code.

19 years agoRIP compat libraries. Use misc/compatXX from ports instead.
Joerg Sonnenberger [Wed, 20 Apr 2005 21:06:06 +0000 (21:06 +0000)]
RIP compat libraries. Use misc/compatXX from ports instead.

19 years agoWARNS=6. The (int) case should be reevaluated when this code is converted
Joerg Sonnenberger [Wed, 20 Apr 2005 20:45:57 +0000 (20:45 +0000)]
WARNS=6. The (int) case should be reevaluated when this code is converted
to (s)size_t usage.

19 years agoUn-disable stack protector. I take responsibility for all problems this
Joerg Sonnenberger [Wed, 20 Apr 2005 20:27:04 +0000 (20:27 +0000)]
Un-disable stack protector. I take responsibility for all problems this
adds, which are not actual bugs.

19 years agoRemove old beforeinstall hack. Should be not used for ages.
Joerg Sonnenberger [Wed, 20 Apr 2005 20:07:19 +0000 (20:07 +0000)]
Remove old beforeinstall hack. Should be not used for ages.

19 years agoRemove some commented out warnings.
Joerg Sonnenberger [Wed, 20 Apr 2005 20:01:09 +0000 (20:01 +0000)]
Remove some commented out warnings.

19 years agoCosmetic changes, mostly changing zeros to NULLs.
Jeffrey Hsu [Wed, 20 Apr 2005 19:43:53 +0000 (19:43 +0000)]
Cosmetic changes, mostly changing zeros to NULLs.

19 years agoStrip away a layer of indirection. Now that we properly declare
Jeffrey Hsu [Wed, 20 Apr 2005 19:38:22 +0000 (19:38 +0000)]
Strip away a layer of indirection.  Now that we properly declare
so_pcb as a "void *", we don't need the sotounpcb() macro any more.
Directly reference so->so_pcb to get the unp control block.

19 years agoAdd NetBSD 3.0.
Sascha Wildner [Wed, 20 Apr 2005 18:21:16 +0000 (18:21 +0000)]
Add NetBSD 3.0.

19 years agobasetime should be static, all access done via PKI.
Joerg Sonnenberger [Wed, 20 Apr 2005 17:57:16 +0000 (17:57 +0000)]
basetime should be static, all access done via PKI.

19 years agoAdd entry for libarchive.
Sascha Wildner [Wed, 20 Apr 2005 17:26:15 +0000 (17:26 +0000)]
Add entry for libarchive.

19 years agoDocument the rwlock routines as being MP safe.
Matthew Dillon [Wed, 20 Apr 2005 17:03:35 +0000 (17:03 +0000)]
Document the rwlock routines as being MP safe.

19 years agoMake the kernel malloc() and free() MP safe by pushing the BGL past the
Matthew Dillon [Wed, 20 Apr 2005 17:03:08 +0000 (17:03 +0000)]
Make the kernel malloc() and free() MP safe by pushing the BGL past the
per-cpu frontend and into the VM access layer.

19 years agoGet rid of VPLACEMARKER and retool vmntvnodescan() to use a secondary
Matthew Dillon [Wed, 20 Apr 2005 17:01:56 +0000 (17:01 +0000)]
Get rid of VPLACEMARKER and retool vmntvnodescan() to use a secondary
structure to track ripouts that occur during the slowfunc() callback,
similar to how we do it for the mountlist and clean/dirty buffer trees.

19 years agoStyle(9) cleanup: use ANSI format for function definitions.
Chris Pressey [Wed, 20 Apr 2005 16:37:09 +0000 (16:37 +0000)]
Style(9) cleanup: use ANSI format for function definitions.

19 years agoRemove unnecessary .El (fixes groff warning).
Sascha Wildner [Wed, 20 Apr 2005 16:24:13 +0000 (16:24 +0000)]
Remove unnecessary .El (fixes groff warning).

19 years agoMake comments about bpf_validate()'s behaviour reflect reality. The
Chris Pressey [Wed, 20 Apr 2005 16:05:51 +0000 (16:05 +0000)]
Make comments about bpf_validate()'s behaviour reflect reality.  The
code does not check that jumps are forward; jumps are specified by
unsigned values, and as such, are inevitably forward jumps.

19 years agoFix grammar (associate -> associated) and start sentence on new line.
Chris Pressey [Wed, 20 Apr 2005 15:44:32 +0000 (15:44 +0000)]
Fix grammar (associate -> associated) and start sentence on new line.

19 years agoUse get_approximate_time_t().
Joerg Sonnenberger [Wed, 20 Apr 2005 14:27:29 +0000 (14:27 +0000)]
Use get_approximate_time_t().

19 years agoAdd support for ServerWorks chipsets.
Joerg Sonnenberger [Wed, 20 Apr 2005 10:51:24 +0000 (10:51 +0000)]
Add support for ServerWorks chipsets.

Obtained-from: FreeBSD

19 years agoFix whitespace in copyright dates.
Hiten Pandya [Wed, 20 Apr 2005 10:24:48 +0000 (10:24 +0000)]
Fix whitespace in copyright dates.

19 years agoAdd bpf_mtap_family(9) which allows the client to specify an address
Hiten Pandya [Wed, 20 Apr 2005 10:22:44 +0000 (10:22 +0000)]
Add bpf_mtap_family(9) which allows the client to specify an address
family.  The logic is quite repetetive in various psuedo drivers and
other networking subsystems, like netgraph.

In-discussion-with:  Joerg Sonnenberger.

19 years agosbappend() is called by stream-oriented as well as record-oriented
Jeffrey Hsu [Wed, 20 Apr 2005 09:28:29 +0000 (09:28 +0000)]
sbappend() is called by stream-oriented as well as record-oriented
protocols, so we can't use the sb_lastrecord hint, which is only
for record-oriented protocols, as an optimization.  (However,
stream-oriented protocols only have one record anyways, so a last
record hint provides zero speedup here anyhow.)

Original bug with X client occasionally not working reported by Simon Schubert.

19 years agoCorrect a bug in the positive 32bit overflow handling of ntp_tick_acc.
Joerg Sonnenberger [Wed, 20 Apr 2005 08:04:32 +0000 (08:04 +0000)]
Correct a bug in the positive 32bit overflow handling of ntp_tick_acc.
Simplify the logic by not playing wiered bit masking, just substract
the large part. Depending on whether 64bit math is native, this can be
even faster.

19 years agoNULL-out two stack-declared variables that were incorrectly assumed to
Matthew Dillon [Tue, 19 Apr 2005 20:38:57 +0000 (20:38 +0000)]
NULL-out two stack-declared variables that were incorrectly assumed to
be NULL on entry.

Reported-by: Jimmy Lim <jimmyblim@gmail.com>
Submitted-by: "Skoda Tomas" <tomas.skoda@zoznam.sk>
19 years agoChange NOACU to NULL.
Sascha Wildner [Tue, 19 Apr 2005 18:55:24 +0000 (18:55 +0000)]
Change NOACU to NULL.

19 years agoFix a buildkernel warning by properly initializing dom_next in struct domain.
Sascha Wildner [Tue, 19 Apr 2005 18:51:57 +0000 (18:51 +0000)]
Fix a buildkernel warning by properly initializing dom_next in struct domain.

Thanks for telling: joerg

19 years agoChange herz to hertz.
Sascha Wildner [Tue, 19 Apr 2005 18:05:02 +0000 (18:05 +0000)]
Change herz to hertz.

19 years agoAbstract out the routines which manipulate the mountlist.
Matthew Dillon [Tue, 19 Apr 2005 17:54:52 +0000 (17:54 +0000)]
Abstract out the routines which manipulate the mountlist.

Introduce an MP-safe mountlist scanning function.  This function keeps track
of scans which are in-progress and properly handles ripouts that occur during
the callback by advancing the matching pointers being tracked.  The callback
can safely block without confusing the scan.

This algorithm has already been successfully used for the buffer cache and
will soon be used for the vnode lists hanging off the mount point.

19 years agoProvide correct PLIST_SUB value for batch scripts.
Joerg Sonnenberger [Tue, 19 Apr 2005 17:17:03 +0000 (17:17 +0000)]
Provide correct PLIST_SUB value for batch scripts.

19 years agoMerge from vendor branch NTPD:
Joerg Sonnenberger [Tue, 19 Apr 2005 13:09:30 +0000 (13:09 +0000)]
Merge from vendor branch NTPD:
Print "reply from ..." before "adjust local clock ..." in debug mode.

19 years agoPrint "reply from ..." before "adjust local clock ..." in debug mode. vendor/NTPD
Joerg Sonnenberger [Tue, 19 Apr 2005 13:09:30 +0000 (13:09 +0000)]
Print "reply from ..." before "adjust local clock ..." in debug mode.

19 years agoMerge from vendor branch NTPD:
Joerg Sonnenberger [Tue, 19 Apr 2005 07:37:43 +0000 (07:37 +0000)]
Merge from vendor branch NTPD:
Sync in my paranoia check.

19 years agoSync in my paranoia check.
Joerg Sonnenberger [Tue, 19 Apr 2005 07:37:43 +0000 (07:37 +0000)]
Sync in my paranoia check.

19 years agoStyle cleanup: instead of seperate #defines for NULL cast to different
Chris Pressey [Tue, 19 Apr 2005 05:32:02 +0000 (05:32 +0000)]
Style cleanup: instead of seperate #defines for NULL cast to different
types (NOSTR, NOFILE, NOPWD, NOACU, NOVAL,) just spell NULL as NULL.

Inspired-by: NetBSD
19 years agoReduce warnings that appear under higher WARNS levels:
Chris Pressey [Tue, 19 Apr 2005 05:18:19 +0000 (05:18 +0000)]
Reduce warnings that appear under higher WARNS levels:
- Include a missing header;
- Spell out initializers in full;
- De-constify a char * that isn't really constant.

19 years agoWhen installing packages into an ISO-image-to-be file tree:
Chris Pressey [Tue, 19 Apr 2005 04:20:27 +0000 (04:20 +0000)]
When installing packages into an ISO-image-to-be file tree:
- Don't try to copy packages that already exist (with identical version
  numbers) on the image; and
- Bail out early if an error (e.g. missing dependency) occurs, instead
  of blindly carrying on with the next package.

19 years agoPrefer TAILQ_EMPTY() to null-check on the result of TAILQ_FIRST().
Jeffrey Hsu [Mon, 18 Apr 2005 23:50:30 +0000 (23:50 +0000)]
Prefer TAILQ_EMPTY() to null-check on the result of TAILQ_FIRST().

19 years agoDefer assigning to the forwarding route variable until the forwarding
Jeffrey Hsu [Mon, 18 Apr 2005 23:43:04 +0000 (23:43 +0000)]
Defer assigning to the forwarding route variable until the forwarding
route has been determined.

19 years agoMinor cosmetic cleanups. Use the canonical name "ro" for a variable
Jeffrey Hsu [Mon, 18 Apr 2005 22:57:52 +0000 (22:57 +0000)]
Minor cosmetic cleanups.  Use the canonical name "ro" for a variable
of type struct route.  Clean up ip_forward().

19 years agoUse a larger initial window size when restarting after a long idle period.
Jeffrey Hsu [Mon, 18 Apr 2005 22:41:23 +0000 (22:41 +0000)]
Use a larger initial window size when restarting after a long idle period.

19 years agoCVS was seg-faulting on systems with hostnames greater then 34 characters
Matthew Dillon [Mon, 18 Apr 2005 20:22:42 +0000 (20:22 +0000)]
CVS was seg-faulting on systems with hostnames greater then 34 characters
long due to assuming that gethostname() would return ENAMETOOLONG.  In
FreeBSD/DragonFly it returns ENOMEM.  Use our auto-patching mechanism to
patch in the additional error code.

Reported-by: Jeroen Ruigrok/asmodai <asmodai@wxs.nl>
19 years agoWARNS=6 + minor style issues while here.
Joerg Sonnenberger [Mon, 18 Apr 2005 20:17:58 +0000 (20:17 +0000)]
WARNS=6 + minor style issues while here.

Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
19 years agoSync with OpenBSD. The ntpd -s is included now.
Joerg Sonnenberger [Mon, 18 Apr 2005 18:17:20 +0000 (18:17 +0000)]
Sync with OpenBSD. The ntpd -s is included now.

19 years agoMerge from vendor branch NTPD:
Joerg Sonnenberger [Mon, 18 Apr 2005 18:17:20 +0000 (18:17 +0000)]
Merge from vendor branch NTPD:
Sync with OpenBSD. The ntpd -s is included now.

19 years agoSync with OpenBSD. The ntpd -s is included now.
Joerg Sonnenberger [Mon, 18 Apr 2005 18:17:20 +0000 (18:17 +0000)]
Sync with OpenBSD. The ntpd -s is included now.

19 years agoUse the local runetype.h here too.
Joerg Sonnenberger [Mon, 18 Apr 2005 17:42:15 +0000 (17:42 +0000)]
Use the local runetype.h here too.

19 years agoSwitch IP divert from mbuf based tagging to mbuf tags.
Joerg Sonnenberger [Mon, 18 Apr 2005 14:26:57 +0000 (14:26 +0000)]
Switch IP divert from mbuf based tagging to mbuf tags.
This patch is meant to keep the old semantic, the divert
handling can be pushed down in most places later.

19 years agoMake kern.ntp.permanent specify the frequency correction per second,
Joerg Sonnenberger [Mon, 18 Apr 2005 13:27:44 +0000 (13:27 +0000)]
Make kern.ntp.permanent specify the frequency correction per second,
not per tick. This makes a bit more sense by isolating implementation
detail of per-tick correction.

19 years agoAdd snprintf and vsnprintf.
Joerg Sonnenberger [Mon, 18 Apr 2005 07:55:09 +0000 (07:55 +0000)]
Add snprintf and vsnprintf.

19 years agoUse the headers from src/include for ctype.h and runetype.h instead
Joerg Sonnenberger [Mon, 18 Apr 2005 07:43:45 +0000 (07:43 +0000)]
Use the headers from src/include for ctype.h and runetype.h instead
of the installed versions. The latter might not exist or define different
conditionals.

19 years agostaticize lwkt_reqtoken_remote().
Matthew Dillon [Mon, 18 Apr 2005 01:03:33 +0000 (01:03 +0000)]
staticize lwkt_reqtoken_remote().

19 years agoFix a bug in the last commit. A FIFO calculation was incorrect (3/2 instead
Matthew Dillon [Mon, 18 Apr 2005 01:02:58 +0000 (01:02 +0000)]
Fix a bug in the last commit.  A FIFO calculation was incorrect (3/2 instead
of 2/3), resulting in overflows during heavy token handling activity.

Reported-by: =?ISO-8859-1?Q?Stefan_Kr=FCger?= <skrueger@meinberlikomm.de>
19 years agopatch-7.189
Max Okumoto [Sat, 16 Apr 2005 10:38:37 +0000 (10:38 +0000)]
patch-7.189
Remove unsed functions Var_GetTail() and Var_GetHead()

19 years agotch-7.188
Max Okumoto [Sat, 16 Apr 2005 10:36:19 +0000 (10:36 +0000)]
tch-7.188
  Var_SubstOnly(), match_var()
o Split out test to check if variable name should not be expanded.

19 years agopatch-7.187
Max Okumoto [Sat, 16 Apr 2005 10:35:54 +0000 (10:35 +0000)]
patch-7.187
  Var_Subst() Var_SubstOnly()
o unify syntax for accessing bytes in the input string.
o Simplify skipping bytes.

19 years agopatch-7.186
Max Okumoto [Sat, 16 Apr 2005 10:35:29 +0000 (10:35 +0000)]
patch-7.186
 Var_SubstOnly()
o remove continue statement.

19 years agopatch-7.185
Max Okumoto [Sat, 16 Apr 2005 10:35:02 +0000 (10:35 +0000)]
patch-7.185
  Var_Subst(), Var_SubstOnly()
o remove dead code.

19 years agopatch-7.184
Max Okumoto [Sat, 16 Apr 2005 10:34:26 +0000 (10:34 +0000)]
patch-7.184
o Copy Var_Subst() into Var_SubstOnly(), the new function
  is never passed NULL in the 'var' input parameter.
o Modified Var_Subst() to always expect var==NULL as input paramter.
o Update calling code by removing unused parameter.

19 years agoAdd a debug knob debug.acpi.avoid.paths, a variant of debug.acpi.avoid
YONETANI Tomokazu [Sat, 16 Apr 2005 08:17:02 +0000 (08:17 +0000)]
Add a debug knob debug.acpi.avoid.paths, a variant of debug.acpi.avoid
but without totally disabling execution of _INI and _STA.  Maybe useful
to narrow down the portion of namespace causing problems without messing
with editing and compiling ASL.  Disablement of _INI and _STA when
debug.acpi.avoid is specified has been added years before to workaround
interpreter crash on some systems, and I'm not sure whether the recent
ACPI-CA code still crashes on those systems, but I'll leave it as is
anyway for compatibility with FreeBSD.

Suggested-by: Sepherosa Ziehau
19 years agoForgot to add these files to previous commit. directive_hash.c
Max Okumoto [Fri, 15 Apr 2005 21:13:05 +0000 (21:13 +0000)]
Forgot to add these files to previous commit.  directive_hash.c
must be regenerated if new directives are added/removed/modified.

19 years agoAdd back .makenv compat until everyone's source tree is updated.
Max Okumoto [Fri, 15 Apr 2005 21:09:15 +0000 (21:09 +0000)]
Add back .makenv compat until everyone's source tree is updated.

19 years agoCreate VAR_ENV context, and fill it with everything in the environment.
Max Okumoto [Fri, 15 Apr 2005 21:06:35 +0000 (21:06 +0000)]
Create VAR_ENV context, and fill it with everything in the environment.

19 years agoadd .EXPORT: special tag and add ST_EXPORT variable.
Max Okumoto [Fri, 15 Apr 2005 21:05:52 +0000 (21:05 +0000)]
add .EXPORT: special tag and add ST_EXPORT variable.

19 years agoFreeBSD-Date: 2005/04/11 08:20:09
Max Okumoto [Fri, 15 Apr 2005 21:01:27 +0000 (21:01 +0000)]
FreeBSD-Date: 2005/04/11 08:20:09
FreeBSD-Date: 2005/04/11 08:26:21
FreeBSD-Date: 2005/04/11 08:40:54

Author: harti

Rework the directive parsing code. Instead of using a lot of strcmp()s
on every line that starts with a dot use a minimal perfect hash
function and a single strcmp() on the first word after the dot
to find out whether it is really a directive call and, if yes, which
one. Then directly dispatch to a handler function for that directive
(or fall through to the dependency handling code). This makes the
directive parse a little bit more strict about the syntax: the directive
word must be followed by a character that is not alphanumerical and not
an underline (making .undefFOO illegal); .endif and .else can only be
followed by comments.

19 years agoInline stripvarname() since it will be removed in the next few patches.
Max Okumoto [Fri, 15 Apr 2005 20:50:43 +0000 (20:50 +0000)]
Inline stripvarname() since it will be removed in the next few patches.

19 years agoSplit VarFind() into a series of functions tailored for the different sets
Max Okumoto [Fri, 15 Apr 2005 20:45:48 +0000 (20:45 +0000)]
Split VarFind() into a series of functions tailored for the different sets
of flags originally passed to VarFind(). This eliminates the code by
removing a bunch of tests.

patch-7.173
patch-7.174
patch-7.176

19 years agoImplement Red-Black trees for the vnode clean/dirty buffer lists.
Matthew Dillon [Fri, 15 Apr 2005 19:08:32 +0000 (19:08 +0000)]
Implement Red-Black trees for the vnode clean/dirty buffer lists.

Implement ranged fsyncs and adjust the syncer to use the new capability.
This capability will also soon be used to replace the write_behind
heuristic.  Rewrite the fsync code for all VFSs to use the new APIs
(generally simplifying them).

Get rid of B_WRITEINPROG, it is no longer useful or needed.
Get rid of B_SCANNED, it is no longer useful or needed.

Rewrite the NFS 2-phase commit protocol to take advantage of the new
Red-Black tree topology.

Add RB_SCAN() for callback-scanning of Red-Black trees.  Give RB_SCAN
the ability to track the 'next' scan node and automatically fix it up
if the callback directly or indirectly or through blocking indirectly
deletes nodes in the tree while the scan is in progress.

Remove most related loop restart conditions, they are no longer necessary.

Disable filesystem background bitmap writes.  This really needs to be
solved a different way and the concept does not work well with red-black
trees.

19 years agoRemove pccard RCNG script, the daemon config, the depency on pccard
Joerg Sonnenberger [Fri, 15 Apr 2005 18:40:52 +0000 (18:40 +0000)]
Remove pccard RCNG script, the daemon config, the depency on pccard
for the other RCNG scripts, the actual daemon and the control program.

pccard_ether is used by usbd, keep it for now.

19 years agoZero the whole structure, not just the name, so we don't trip up on this
Matthew Dillon [Fri, 15 Apr 2005 18:39:13 +0000 (18:39 +0000)]
Zero the whole structure, not just the name, so we don't trip up on this
in the future.

19 years agoGC OLDCARD userland tools.
Joerg Sonnenberger [Fri, 15 Apr 2005 18:26:58 +0000 (18:26 +0000)]
GC OLDCARD userland tools.

19 years agoFix the parameter order of __stack_smash_handler. Also print the
Joerg Sonnenberger [Fri, 15 Apr 2005 18:07:13 +0000 (18:07 +0000)]
Fix the parameter order of __stack_smash_handler. Also print the
return address on IA32 to aid debugging.

19 years agoExtend a local buffer to workarond a buffer overflow for now.
Joerg Sonnenberger [Fri, 15 Apr 2005 17:55:49 +0000 (17:55 +0000)]
Extend a local buffer to workarond a buffer overflow for now.

19 years agoRemove some uses of register keywords.
David Rhodus [Fri, 15 Apr 2005 17:55:29 +0000 (17:55 +0000)]
Remove some uses of register keywords.

19 years agoFix a small stack disclosure in ifconf().
Joerg Sonnenberger [Fri, 15 Apr 2005 07:08:27 +0000 (07:08 +0000)]
Fix a small stack disclosure in ifconf().

See FreeBSD-SA-05:04.ifconf for credits and details.

19 years agogetopt is WARNS 6 clean.
Eirik Nygaard [Thu, 14 Apr 2005 15:49:12 +0000 (15:49 +0000)]
getopt is WARNS 6 clean.

19 years agoA few more instances of leaf that should be leap.
Eirik Nygaard [Thu, 14 Apr 2005 15:05:29 +0000 (15:05 +0000)]
A few more instances of leaf that should be leap.

Noticed by: Bart Thate <dunker@freebsd.nl>

19 years agoIgnore replies with a negative delay. Correct time of next query and
Joerg Sonnenberger [Thu, 14 Apr 2005 11:22:16 +0000 (11:22 +0000)]
Ignore replies with a negative delay. Correct time of next query and
deadline for all peers in priv_settime is called.

This should fix the problems of ntpd -s with very large initial correction.

19 years agoleap second, not leaf second. fix totally stupid, but consistent spelling
Joerg Sonnenberger [Thu, 14 Apr 2005 11:15:52 +0000 (11:15 +0000)]
leap second, not leaf second. fix totally stupid, but consistent spelling
error.

Noticed-by: corecode
19 years agoSync with OpenBSD. This fixes the bug of -s not working and allows
Joerg Sonnenberger [Thu, 14 Apr 2005 10:41:29 +0000 (10:41 +0000)]
Sync with OpenBSD. This fixes the bug of -s not working and allows
defered DNS lookups, which might be necessary in certain network
setups.

19 years agoMerge from vendor branch NTPD:
Joerg Sonnenberger [Thu, 14 Apr 2005 10:41:29 +0000 (10:41 +0000)]
Merge from vendor branch NTPD:
Sync with OpenBSD. This fixes the bug of -s not working and allows
defered DNS lookups, which might be necessary in certain network
setups.

19 years agoUse corrected system time internally, adjust old offsets after
Joerg Sonnenberger [Thu, 14 Apr 2005 10:31:27 +0000 (10:31 +0000)]
Use corrected system time internally, adjust old offsets after
doing a relative adjust of kern.ntp.delta.

Change the default scaling interval to 0.005 and 0.010, which
means that offsets smaller than 5ms double to probe interval,
anything greater than 10ms uses the default. The value of
QSCALE_OFF_MIN can be decreases even more, once frequency correction
is implemented.

19 years agoUse corrected system time internally, adjust old offsets after
Joerg Sonnenberger [Thu, 14 Apr 2005 10:26:51 +0000 (10:26 +0000)]
Use corrected system time internally, adjust old offsets after
doing a relative adjust of kern.ntp.delta.

Change the default scaling interval to 0.005 and 0.010, which
means that offsets smaller than 5ms double to probe interval,
anything greater than 10ms uses the default. The value of
QSCALE_OFF_MIN can be decreases even more, once frequency correction
is implemented.

19 years agoResurrect some of functions shared by rrestore.
YONETANI Tomokazu [Thu, 14 Apr 2005 10:17:23 +0000 (10:17 +0000)]
Resurrect some of functions shared by rrestore.

19 years agowether ==> whether
Joerg Sonnenberger [Thu, 14 Apr 2005 08:14:31 +0000 (08:14 +0000)]
wether ==> whether

Noticed-by: asmodai
19 years agoSimplified NTP kernel interface:
Joerg Sonnenberger [Thu, 14 Apr 2005 07:55:36 +0000 (07:55 +0000)]
Simplified NTP kernel interface:
- kern.ntp.delta gives the delta to apply (ns)
- kern.ntp.tick_delta is the correction applied in each tick (ns)
- kern.ntp.default_tick_delta is the default correction for each tick (ns)
- kern.ntp.big_delta is the threshold for kern.ntp.delta to use
  10x kern.ntp.default_tick_delta, not the normal value
- kern.ntp.adjust can be used to change the current value of kern.ntp.delta
  relatively.
- kern.ntp.next_leaf_second specifies the time_t of the next leaf second change,
  kern.ntp.insert_leaf_second != 0 means a leaf second is inserted, otherwise
  it is removed

All ntp_* variables are manipulated on CPU #0 with the exception of
set_timeofday. It just sets ntp_delta to 0, which is fine for the moment.

19 years agoRemove another argument name in a prototype.
Joerg Sonnenberger [Wed, 13 Apr 2005 16:08:04 +0000 (16:08 +0000)]
Remove another argument name in a prototype.

19 years agoDecommon a lot of variables, makes some static, fix a small bug in rmtcall
Joerg Sonnenberger [Wed, 13 Apr 2005 16:07:15 +0000 (16:07 +0000)]
Decommon a lot of variables, makes some static, fix a small bug in rmtcall
introduced earlier.

19 years agoRemove a cast, correct another.
Joerg Sonnenberger [Wed, 13 Apr 2005 15:45:37 +0000 (15:45 +0000)]
Remove a cast, correct another.

19 years ago- Let receiver know if the sender accepts replies
Liam J. Foy [Wed, 13 Apr 2005 15:44:03 +0000 (15:44 +0000)]
- Let receiver know if the sender accepts replies
- Use correct headers
- Use err(). In some cases an incorrect error message can be shown
- Strlcpy
- warnx() instead of errx() if sender has messages disabled (OpenBSD)
- Remove non-POSIX S_IWRITE (OpenBSD)
- General cleanup

19 years agoDon't write to tape, strdup it first.
Joerg Sonnenberger [Wed, 13 Apr 2005 15:43:36 +0000 (15:43 +0000)]
Don't write to tape, strdup it first.

Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
19 years agoRemove argument names from prototypes. Adjust prototypes to deal
Joerg Sonnenberger [Wed, 13 Apr 2005 15:21:36 +0000 (15:21 +0000)]
Remove argument names from prototypes. Adjust prototypes to deal
with void * vs. char * and const vs. non-const behaviour.

19 years agoMove initialisation of wrote into the loop to prevent longjmp/setjmp
Joerg Sonnenberger [Wed, 13 Apr 2005 14:32:01 +0000 (14:32 +0000)]
Move initialisation of wrote into the loop to prevent longjmp/setjmp
from clobbering it.

19 years agoWarning fixes.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:29:20 +0000 (14:29 +0000)]
Warning fixes.

Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
19 years agoSplit atomic into two functions, atomic_read and atomic_write.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:21:06 +0000 (14:21 +0000)]
Split atomic into two functions, atomic_read and atomic_write.
This makes the code more readable and helps the const correctness.

Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
19 years agoCompute buffer length only once per rmtcall.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:12:35 +0000 (14:12 +0000)]
Compute buffer length only once per rmtcall.

19 years agoRemove unused functions.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:10:18 +0000 (14:10 +0000)]
Remove unused functions.

Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
19 years agoFirst bunch of dump(8) cleanups.
Joerg Sonnenberger [Wed, 13 Apr 2005 14:05:35 +0000 (14:05 +0000)]
First bunch of dump(8) cleanups.

Submitted-by: Sepherosa Ziehau <sepherosa@softhome.net>
19 years agoOnly bump the switch_count counter when lwkt_switch() actually switches
Matthew Dillon [Wed, 13 Apr 2005 04:02:08 +0000 (04:02 +0000)]
Only bump the switch_count counter when lwkt_switch() actually switches
to a different thread.  Otherwise the idle loop causes this counter
to go haywire.

19 years agoOptimize lwkt_send_ipiq() - the IPI based inter-cpu messaging routine.
Matthew Dillon [Wed, 13 Apr 2005 04:00:56 +0000 (04:00 +0000)]
Optimize lwkt_send_ipiq() - the IPI based inter-cpu messaging routine.

* Add a passive version which does not initiate any actual hardware IPI.
  The message will be handled the next time the target cpu polls the
  queue (on each tick typically).  Adjust the free() path to use this
  version when freeing memory owned by another cpu.

* Add an interlock to avoid reissuing and unnecessarily stalling on
  the hardware IPI if a prior hardware IPI to the target cpu has not
  yet completed processing.

  This feature theoretically means that two cpus can tightly couple a
  large number of pipelined messages with only a single actual IPI being
  sent.

* Reorganize the hystersis points in the IPIQ FIFOs.

* Change a token livelock warning into a panic if it occurs 10 times in
  a row.

* Add a call to lwkt_process_ipiq() just after the AP startup code enables
  a cpu, to process any messages that might have built up during startup.
  There shouldn't be any, but this may avoid surprises later.