]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoRemove dublicated device entry from the synopsis.
brueffer [Thu, 17 Feb 2005 16:01:20 +0000 (16:01 +0000)]
Remove dublicated device entry from the synopsis.

Submitted by: Ulrich Spoerlein <q@uni.de>
MFC after: 3 days

19 years agoMark netatm and netnatm explicitly as requiring Giant, as they still do.
rwatson [Thu, 17 Feb 2005 14:21:22 +0000 (14:21 +0000)]
Mark netatm and netnatm explicitly as requiring Giant, as they still do.

MFC after: 3 days

19 years agoIn accept1(), extend coverage of the socket lock from just covering
rwatson [Thu, 17 Feb 2005 13:00:23 +0000 (13:00 +0000)]
In accept1(), extend coverage of the socket lock from just covering
soref() to also covering the update of so_state.  While no other user
threads can update the socket state here as it's not yet hooked up to
the file descriptor array yet, the protocol could also frob the
socket state here, leading to a lost update to the so_state field.
No reported instances of this bug (as yet).

MFC after:      3 days

19 years agoIn sonewconn(), set the new socket's state to show the protocol-provided
rwatson [Thu, 17 Feb 2005 12:53:45 +0000 (12:53 +0000)]
In sonewconn(), set the new socket's state to show the protocol-provided
connection status before inserting the new socket into the listen
socket's accept queue, or there might be a race in which another thread
wakes up when the accept lock is released, and sees the socket before its
state is set correctly.  The wakeup still occurs after the accept lock is
released.  There have been no diagnoses of this bug in real-world systems
(as yet).

MFC after: 3 days

19 years agoWhitespace cleanup: substitute mixed tabs and spaces by canonical
harti [Thu, 17 Feb 2005 12:35:32 +0000 (12:35 +0000)]
Whitespace cleanup: substitute mixed tabs and spaces by canonical
whitespace and line up some variable definitions.

19 years agoMove error case to begin of if-else chain. Do not needless initialize
harti [Thu, 17 Feb 2005 12:31:53 +0000 (12:31 +0000)]
Move error case to begin of if-else chain. Do not needless initialize
startc, but only at the place where the initialisation is needed. Remove
a needless else.

Submitted by: Max Okumoto <okumoto@ucsd.edu> (partly)

19 years agoDocument the new default init_path.
des [Thu, 17 Feb 2005 11:14:45 +0000 (11:14 +0000)]
Document the new default init_path.

Reminded by: ru
MFC after: 2 weeks

19 years agoIntroduce vx_wait{l}() and use it instead of home-rolled versions.
phk [Thu, 17 Feb 2005 10:49:51 +0000 (10:49 +0000)]
Introduce vx_wait{l}() and use it instead of home-rolled versions.

19 years agoConvert KASSERTS to VNASSERTS
phk [Thu, 17 Feb 2005 10:28:58 +0000 (10:28 +0000)]
Convert KASSERTS to VNASSERTS

19 years agoAdd /rescue/init to the default init_path, before /stand/sysinstall.
des [Thu, 17 Feb 2005 10:00:10 +0000 (10:00 +0000)]
Add /rescue/init to the default init_path, before /stand/sysinstall.

MFC after: 2 weeks

19 years agoFix two typos in comments.
harti [Thu, 17 Feb 2005 09:09:34 +0000 (09:09 +0000)]
Fix two typos in comments.

Submitted by: ru & Max Okumoto <okumoto@ucsd.edu>

19 years agoSort out the error case that a single '$' was passed a little bit
harti [Thu, 17 Feb 2005 09:01:19 +0000 (09:01 +0000)]
Sort out the error case that a single '$' was passed a little bit
earlier instead of mixing its handling with other cases.

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoFix inteface clear time. pf printed "Thu Jan 1 09:00:01 1970"
yongari [Thu, 17 Feb 2005 03:36:31 +0000 (03:36 +0000)]
Fix inteface clear time. pf printed "Thu Jan  1 09:00:01 1970"
in "pfctl -vvsI" output when pf was statically linked to kernel.

Discussed with: mlaier

19 years agoThe correct error value for not having enough storage is E2BIG, not
njl [Thu, 17 Feb 2005 01:02:58 +0000 (01:02 +0000)]
The correct error value for not having enough storage is E2BIG, not
ENOMEM.  The manpage and ichss(4) are correct.

19 years agoFix the check for acpi_perf(4) so that we verify if it is fully attached
njl [Thu, 17 Feb 2005 01:01:40 +0000 (01:01 +0000)]
Fix the check for acpi_perf(4) so that we verify if it is fully attached
or just offering info.  In the former case, we don't probe/attach to allow
the ACPI driver precedence.  A refinement of this would be to actually
use the info provided by acpi_perf(4) to get the real CPU clock rates
instead of estimating them but since all systems that support both
acpi_perf(4) and ichss(4) export the control registers to acpi_perf(4),
it can just handle the registers on its own.

19 years agoUltraSparc II[e,i] based systems come up with the tick compare register
marius [Thu, 17 Feb 2005 00:13:49 +0000 (00:13 +0000)]
UltraSparc II[e,i] based systems come up with the tick compare register
loaded, the tick interrupt enabled and a handler that resets the tick
counter on every tick interrupt. While this isn't documented this can
cause DELAY() to wait for a value the tick counter will not reach when
used in early boot, i.e. before cpu_initclocks() is called, depending
on when in the cycle DELAY() is called, the delay value and the value
the tick compare register is set to. The excessive use of DELAY() in
uart(4) when probing Sun keyboards seems to always manage to trigger
this, resulting in a hang during boot.
Disable the tick interrupt in tick_init(), which is called early in
sparc64_init(), until the interrupt is enabled again in tick_start(),
called by cpu_initclocks(), with our own handler. This fixes the hang
during probing Sun keyboards on AXi boards and Ultra 10, with other
machines like Ultra 5 probably being affected but not tested.

Additional testing by: Matthias Muthmann
MFC after: 1 week

19 years agoaic79xx.c:
gibbs [Wed, 16 Feb 2005 23:13:38 +0000 (23:13 +0000)]
aic79xx.c:
aic7xxx.c:
Allow print_reg() to be called with a NULL column.

aic79xx.c:
Correct new usage of SCB_GET_TAG().

aic7xxx.c:
Fix stray ahd that snuck in here.

19 years agoFix year in copyrights.
pjd [Wed, 16 Feb 2005 22:26:34 +0000 (22:26 +0000)]
Fix year in copyrights.

19 years agoCopyright year update.
pjd [Wed, 16 Feb 2005 22:24:15 +0000 (22:24 +0000)]
Copyright year update.

19 years agoFix year in copyrights.
pjd [Wed, 16 Feb 2005 22:19:13 +0000 (22:19 +0000)]
Fix year in copyrights.

19 years agoUpdate copyright in files changed this year.
pjd [Wed, 16 Feb 2005 22:14:52 +0000 (22:14 +0000)]
Update copyright in files changed this year.

19 years agoFix year in copyrights.
pjd [Wed, 16 Feb 2005 22:13:22 +0000 (22:13 +0000)]
Fix year in copyrights.

19 years agoWell, it seems that I pre-maturely removed the "All rights reserved"
bmilekic [Wed, 16 Feb 2005 21:45:59 +0000 (21:45 +0000)]
Well, it seems that I pre-maturely removed the "All rights reserved"
statement from some files, so re-add it for the moment, until the
related legalese is sorted out.  This change affects:

sys/kern/kern_mbuf.c
sys/vm/memguard.c
sys/vm/memguard.h
sys/vm/uma.h
sys/vm/uma_core.c
sys/vm/uma_dbg.c
sys/vm/uma_dbg.h
sys/vm/uma_int.h

19 years agoRemove redundant label.
wpaul [Wed, 16 Feb 2005 21:24:04 +0000 (21:24 +0000)]
Remove redundant label.

19 years agoGenerate locale-agnostic configuration date.
ru [Wed, 16 Feb 2005 21:02:50 +0000 (21:02 +0000)]
Generate locale-agnostic configuration date.

19 years agoDefine PLATFORM correctly when cross-building.
ru [Wed, 16 Feb 2005 20:55:47 +0000 (20:55 +0000)]
Define PLATFORM correctly when cross-building.

19 years agoMake UMA set the overloaded page->object back to kmem_object for
bmilekic [Wed, 16 Feb 2005 20:06:11 +0000 (20:06 +0000)]
Make UMA set the overloaded page->object back to kmem_object for
UMA_ZONE_REFCNT and UMA_ZONE_MALLOC zones, as the page(s) undoubtedly
came from kmem_map for those two.  Previously it would set it back
to NULL for UMA_ZONE_REFCNT zones and although this was probably not
fatal, it added MORE code for no reason.

19 years agoFix freeing of custom driver extensions. (ExFreePool() was being
wpaul [Wed, 16 Feb 2005 19:21:07 +0000 (19:21 +0000)]
Fix freeing of custom driver extensions. (ExFreePool() was being
called with the wrong pointer.)

19 years agoRe-staticize a few functions I un-staticized for debugging purposes
wpaul [Wed, 16 Feb 2005 18:37:14 +0000 (18:37 +0000)]
Re-staticize a few functions I un-staticized for debugging purposes
on amd64 and accidentally forgot to put back. (Have I mentioned that
gdb on amd64 needs work? It does. Boy howdy.)

19 years agoKeAcquireSpinLockRaiseToDpc() and KeReleaseSpinLock() are (at least
wpaul [Wed, 16 Feb 2005 18:18:30 +0000 (18:18 +0000)]
KeAcquireSpinLockRaiseToDpc() and KeReleaseSpinLock() are (at least
for now) exactly the same as KfAcquireSpinLock() and KfReleaseSpinLock().
I implemented the former as small routines in subr_ntoskrnl.c that just
turned around and invoked the latter. But I don't really need the wrapper
routines: I can just create an entries in the ntoskrnl func table that
map KeAcquireSpinLockRaiseToDpc() and KeReleaseSpinLock() to
KfAcquireSpinLock() and KfReleaseSpinLock() directly. This means
the stubs can go away.

19 years agoOn Rev. B silicon, we disabled the enhanced busfree detection logic to
gibbs [Wed, 16 Feb 2005 18:16:35 +0000 (18:16 +0000)]
On Rev. B silicon, we disabled the enhanced busfree detection logic to
close holes in detecting busfrees that occur after a packetized target
transitions to a non-packetized phase.  The most common case where this
occurs is when a target is externally reset so the controller believes
a packetzied negotiation agreement is still in effect.  Unfortunately,
disabling this feature seems to cause problems for the 7901B.  Re-enable
ehanced busfree detection for this part until I can get my hands on a
samble to figure out if the old workaround is necessary and, if so, how
to make it work correctly.

19 years agoMF5S: Explicitly initialize timedout_scb lists, use SCB_TAG for all access
gibbs [Wed, 16 Feb 2005 18:09:41 +0000 (18:09 +0000)]
MF5S: Explicitly initialize timedout_scb lists, use SCB_TAG for all access
      to the hardware_scb->tag field, limit max lun reported to CAM to 63,
      return after a panic to silence a warning.

19 years agoMostly stylistic issues: move a variable into local scope, make
harti [Wed, 16 Feb 2005 17:20:09 +0000 (17:20 +0000)]
Mostly stylistic issues: move a variable into local scope, make
condition positive and fix long lines.

Submitted by: Max Okumoto <okumoto@ucsd.edu>

19 years agoRemove mutex asserion from g_gate_find(). We don't want g_gate_list_mtx
pjd [Wed, 16 Feb 2005 16:13:56 +0000 (16:13 +0000)]
Remove mutex asserion from g_gate_find(). We don't want g_gate_list_mtx
mutex to be held here, because we want speed here.

19 years agoRemove TDP_GEOM flag from thread after ggate device creation.
pjd [Wed, 16 Feb 2005 16:12:28 +0000 (16:12 +0000)]
Remove TDP_GEOM flag from thread after ggate device creation.
This flag means "wait for all pending requests before returning to userland".
There are pending events for sure, because we just created new provider and
other classes want to taste it, but we cannot answer on I/O requests until
we're here.

19 years agoRemove a recursion protection, which we inherited from splnet() netgraph times.
glebius [Wed, 16 Feb 2005 16:00:35 +0000 (16:00 +0000)]
Remove a recursion protection, which we inherited from splnet() netgraph times.
Now several threads may write data to ng_ksocket. Locking of socket is done in
sosend().

Reviewed by: archie, julian, rwatson
MFC after: 2 weeks

19 years agoFix a memory leak: when freeing the connection structure, don't forget to
des [Wed, 16 Feb 2005 12:46:46 +0000 (12:46 +0000)]
Fix a memory leak: when freeing the connection structure, don't forget to
free the connection buffer as well.

PR: bin/76153
MFC after: 1 week

19 years agoBetter version of the patch in 1.117: bring a variable into local scope
harti [Wed, 16 Feb 2005 12:39:32 +0000 (12:39 +0000)]
Better version of the patch in 1.117: bring a variable into local scope
to prepare for function splitting and slightly reorganise the code
in anticipation of Var_Subst returning a Buffer.

Submitted by: Max Okumoto <okumoto@ucsd.edu> (with slight changes)

19 years agoAdd some consistency checks to the signal-related code.
yar [Wed, 16 Feb 2005 11:35:51 +0000 (11:35 +0000)]
Add some consistency checks to the signal-related code.

MFC: along with rev. 1.202

19 years agoA call to maskurg() makes sense only when a transfer is under way,
yar [Wed, 16 Feb 2005 11:22:20 +0000 (11:22 +0000)]
A call to maskurg() makes sense only when a transfer is under way,
the function will emit an annoying log message otherwise.

Reported by: kris
MFC: along with rev. 1.202

19 years agova_list style tweaks
obrien [Wed, 16 Feb 2005 06:48:35 +0000 (06:48 +0000)]
va_list style tweaks

19 years agoOn second though, print the OUI, model and revision. This is the same
imp [Wed, 16 Feb 2005 05:56:39 +0000 (05:56 +0000)]
On second though, print the OUI, model and revision.  This is the same
information that's in the id1 and id2 fields we were using, but is in
a form that the drivers will be using in their matching routines.

19 years agoAdd support for Windows/x86-64 binaries to Project Evil.
wpaul [Wed, 16 Feb 2005 05:41:18 +0000 (05:41 +0000)]
Add support for Windows/x86-64 binaries to Project Evil.
Ville-Pertti Keinonen (will at exomi dot comohmygodnospampleasekthx)
deserves a big thanks for submitting initial patches to make it
work. I have mangled his contributions appropriately.

The main gotcha with Windows/x86-64 is that Microsoft uses a different
calling convention than everyone else. The standard ABI requires using
6 registers for argument passing, with other arguments on the stack.
Microsoft uses only 4 registers, and requires the caller to leave room
on the stack for the register arguments incase the callee needs to
spill them. Unlike x86, where Microsoft uses a mix of _cdecl, _stdcall
and _fastcall, all routines on Windows/x86-64 uses the same convention.
This unfortunately means that all the functions we export to the
driver require an intermediate translation wrapper. Similarly, we have
to wrap all calls back into the driver binary itself.

The original patches provided macros to wrap every single routine at
compile time, providing a secondary jump table with a customized
wrapper for each exported routine. I decided to use a different approach:
the call wrapper for each function is created from a template at
runtime, and the routine to jump to is patched into the wrapper as
it is created. The subr_pe module has been modified to patch in the
wrapped function instead of the original. (On x86, the wrapping
routine is a no-op.)

There are some minor API differences that had to be accounted for:

- KeAcquireSpinLock() is a real function on amd64, not a macro wrapper
  around KfAcquireSpinLock()
- NdisFreeBuffer() is actually IoFreeMdl(). I had to change the whole
  NDIS_BUFFER API a bit to accomodate this.

Bugs fixed along the way:
- IoAllocateMdl() always returned NULL
- kern_windrv.c:windrv_unload() wasn't releasing private driver object
  extensions correctly (found thanks to memguard)

This has only been tested with the driver for the Broadcom 802.11g
chipset, which was the only Windows/x86-64 driver I could find.

19 years agoChange /bin/sh so *it* implements the processing needed for scripts to
gad [Wed, 16 Feb 2005 05:17:58 +0000 (05:17 +0000)]
Change /bin/sh so *it* implements the processing needed for scripts to
work as expected when they have a "shebang line" of:

     #!/bin/sh -- # -*- perl -*- -p

This specific line is recommended in some perl documentation, and I think
I've seen similar lines in documentation for ruby and python.  Those
write-ups expect `sh' to ignore everything after the '--' if the first
thing after the '--' is a '#'.  See chapter 19, "The Command-Line Interface"
in 3rd edition of "Programming Perl", for some discussion of why perl
recommends using this line in some circumstances.

The above line does work on solaris, irix and aix (as three data points),
and it used to work on FreeBSD by means of a similar patch to execve().
However, that change to execve() effected *all* shells (which caused
other problems), and that processing was recently removed.

PR: 16393  (the original request to fix the same issue)
Reviewed by: freebsd-current (looking at a slightly different patch)
MFC after: 1 week

19 years agoFix for a SACK (receiver) bug where incorrect SACK blocks are
ps [Wed, 16 Feb 2005 01:46:17 +0000 (01:46 +0000)]
Fix for a SACK (receiver) bug where incorrect SACK blocks are
reported to the sender - in the case where the sender sends data
outside the window (as WinXP does :().

Reported by: Sam Jensen <sam at wand dot net dot nz>
Submitted by: Mohan Srinivasan

19 years agoAdd location and PNP info to the mii bus
imp [Wed, 16 Feb 2005 01:08:43 +0000 (01:08 +0000)]
Add location and PNP info to the mii bus

19 years agoAdd an XXX comment about string quoting.
imp [Wed, 16 Feb 2005 01:03:30 +0000 (01:03 +0000)]
Add an XXX comment about string quoting.

19 years agoSet TCP_NOPUSH on HTTP requests, reducing the number of round-trips
kbyanc [Wed, 16 Feb 2005 00:22:20 +0000 (00:22 +0000)]
Set TCP_NOPUSH on HTTP requests, reducing the number of round-trips
necessary to establish each connection.

MFC after: 2 weeks

19 years agoDon't say that mtx_lock() will "sleep" if another kernel thread is
ru [Tue, 15 Feb 2005 23:52:35 +0000 (23:52 +0000)]
Don't say that mtx_lock() will "sleep" if another kernel thread is
holding the mutex, say it will "block".  Later in this manual page
we say that sleeping while holding a mutex isn't allowed, and this
can be confusing.

Submitted by: jhb

19 years agoFix grammar error.
obrien [Tue, 15 Feb 2005 22:31:05 +0000 (22:31 +0000)]
Fix grammar error.

19 years agoRather than overloading the page->object field like UMA does, use instead
bmilekic [Tue, 15 Feb 2005 22:17:07 +0000 (22:17 +0000)]
Rather than overloading the page->object field like UMA does, use instead
an unused pageq queue reference in the page structure to stash a pointer
to the MemGuard FIFO.  Using the page->object field caused problems
because when vm_map_protect() was called the second time to set
VM_PROT_DEFAULT back onto a set of pages in memguard_map, the protection
in the VM would be changed but the PMAP code would lazily not restore
the PG_RW bit on the underlying pages right away (see pmap_protect()).
So when a page fault finally occured and the VM noticed the faulting
address corresponds to a page that _does_ have write access now, it
would then call into PMAP to set back PG_RW (i386 case being discussed
here).  However, before it got to do that, an assertion on the object
lock not being owned would get triggered, as the object of the faulting
page would need to be locked but was overloaded by MemGuard.  This is
precisely why MemGuard cannot overload page->object.

Submitted by: Alan Cox (alc@)

19 years agoInitialize Netgraph type at a correct time, before device probing.
ru [Tue, 15 Feb 2005 18:41:21 +0000 (18:41 +0000)]
Initialize Netgraph type at a correct time, before device probing.

19 years agoRemove an outdated comment about ifnet not being locked.
ru [Tue, 15 Feb 2005 17:47:57 +0000 (17:47 +0000)]
Remove an outdated comment about ifnet not being locked.

OK'ed by: njl, rwatson, sam

19 years agoSet the default guardsize and stacksize in the default thread
deischen [Tue, 15 Feb 2005 15:02:11 +0000 (15:02 +0000)]
Set the default guardsize and stacksize in the default thread
attribute when the library is initialized.

19 years agoBe concerned about huge callback numbers by truncating them rather than
brian [Tue, 15 Feb 2005 10:59:54 +0000 (10:59 +0000)]
Be concerned about huge callback numbers by truncating them rather than
scribbling past the end of our buffer.

Problem spotted by: Damien COUDERC couderc at openbsd dot org

19 years agoBe more careful when doing el_parse() - only do it when el is
delphij [Tue, 15 Feb 2005 10:23:01 +0000 (10:23 +0000)]
Be more careful when doing el_parse() - only do it when el is
properly initialized, that happens when lpc is called from a tty.
Without this change, it's possible to get SIGSEGV simply doing:
echo "..:" | lpc

Reported by: Wojciech A. Koszek <dunstan at freebsd czest pl>
PR: 77462 (patch rewritten by myself)
MFC After: 1 week

19 years agoExpand contractions.
ru [Tue, 15 Feb 2005 09:27:00 +0000 (09:27 +0000)]
Expand contractions.

19 years agoWhen dealing with systems with no absolute drivers attached, only calibrate
njl [Tue, 15 Feb 2005 07:43:48 +0000 (07:43 +0000)]
When dealing with systems with no absolute drivers attached, only calibrate
the rate for the 100% state once.  Afterwards, use that value for deriving
states.  This should fix the problem where the calibrated frequency was
different once a switch was done, giving a different set of levels each
time.  Also, properly search for the right cpufreqX device when detaching.

19 years agoMFi386 rev 1.61: Fix a few bugs in the legacy cpu attachment ivars.
njl [Tue, 15 Feb 2005 07:26:28 +0000 (07:26 +0000)]
MFi386 rev 1.61: Fix a few bugs in the legacy cpu attachment ivars.

19 years agoBind to the driver's parent cpu before switching, for both absolute and
njl [Tue, 15 Feb 2005 07:22:42 +0000 (07:22 +0000)]
Bind to the driver's parent cpu before switching, for both absolute and
relative drivers.  Remove some extraneous KASSERTs since NULL pointers
will be found when they're used right afterwards.

19 years agoCorrect a few bugs in the legacy cpu attachment. Get the unit from the
njl [Tue, 15 Feb 2005 07:21:20 +0000 (07:21 +0000)]
Correct a few bugs in the legacy cpu attachment.  Get the unit from the
parent cpu device before passing it to pcpu_find().  Get the ivars from the
child, not parent cpu device.  These bugs would cause a panic when
dereferencing the pcpu ivar, but weren't present in the acpi attachment
which it seems most people are using.

19 years agoRemove mention of the -k and -wcore options because they don't
marcel [Tue, 15 Feb 2005 07:13:51 +0000 (07:13 +0000)]
Remove mention of the -k and -wcore options because they don't
exist anymore.

PR: doc/70943
Submitted by: Jun <junsu at delphij dot net>
Reviewed by: delphij

19 years agoBump __FreeBSD_version for increased size for default thread stacks.
njl [Tue, 15 Feb 2005 06:33:28 +0000 (06:33 +0000)]
Bump __FreeBSD_version for increased size for default thread stacks.

19 years agoNew release notes: tzdata2004g (+MFC), netcat (+MFC), OpenPAM
bmah [Tue, 15 Feb 2005 06:16:05 +0000 (06:16 +0000)]
New release notes:  tzdata2004g (+MFC), netcat (+MFC), OpenPAM
Feterita, OpenSSH 3.9p1, sendmail 8.13.3.

These were all culled from CVS import log messages.

19 years agoUse ANSI function definitions, in preference to the K&R definitions.
imp [Tue, 15 Feb 2005 06:02:34 +0000 (06:02 +0000)]
Use ANSI function definitions, in preference to the K&R definitions.

19 years agoRemove more deadwood that never got implemented in NEWCARD, since NEWCARD
imp [Tue, 15 Feb 2005 02:54:53 +0000 (02:54 +0000)]
Remove more deadwood that never got implemented in NEWCARD, since NEWCARD
went a different direction than was anticipated when these compatibility
shims were added.

19 years agoMove the harvesting of the MAC address out of the generic novell probe
imp [Mon, 14 Feb 2005 23:00:41 +0000 (23:00 +0000)]
Move the harvesting of the MAC address out of the generic novell probe
and into the bus front ends.  For ISA and C-BUS cards, we always need
to grab it.  For PC Card, already committed, we need to do some sanity
checking on the data that's in the ROMs before we decide that they are
OK to use.  The PC Card code has already been committed and is
independent of this code (which also has to work on NE-1000 cards,
assuming that those cards still work :-).

19 years agoMove the #defines from edreg to edvar which don't have anything to do
imp [Mon, 14 Feb 2005 22:28:51 +0000 (22:28 +0000)]
Move the #defines from edreg to edvar which don't have anything to do
with talking to the hardware.

19 years agoo It turns out that most of the ne-2000 cards that I have got real unhappy
imp [Mon, 14 Feb 2005 22:27:03 +0000 (22:27 +0000)]
o It turns out that most of the ne-2000 cards that I have got real unhappy
  with the latest changes.  They actually have valid ROM data at location
  0 of memory, just like a real NE-2000 ISA card.  Use this data, if
  the ROM passes a few basic tests, as an additional source for the MAC
  address.  Prefer the CIS over this source, but have it take precidence
  over falling back to reading the attribtue memory.
o Minor cleanup of a few devices that we match on based on CIS string.

19 years agoAdapt for new KDB world order.
brueffer [Mon, 14 Feb 2005 21:14:00 +0000 (21:14 +0000)]
Adapt for new KDB world order.

PR: 77528
Submitted by: Jamin Brown <alec@gwi.net>
MFC after: 3 days

19 years ago- Retransmit just one segment on initiation of SACK recovery.
ps [Mon, 14 Feb 2005 21:01:08 +0000 (21:01 +0000)]
- Retransmit just one segment on initiation of SACK recovery.
  Remove the SACK "initburst" sysctl.
- Fix bugs in SACK dupack and partialack handling that can cause
  large bursts while in SACK recovery.

Submitted by: Mohan Srinivasan

19 years ago - Remove the unused and unsafe ufs_ihashlookup. This function returned a
jeff [Mon, 14 Feb 2005 20:51:39 +0000 (20:51 +0000)]
 - Remove the unused and unsafe ufs_ihashlookup.  This function returned a
   vnode pointer that could not be used since no locks were held.

Sponsored by: Isilon Systems, Inc.

19 years agoFix English grammar.
glebius [Mon, 14 Feb 2005 18:49:19 +0000 (18:49 +0000)]
Fix English grammar.

Submitted by: ru

19 years agoFix typo.
stefanf [Mon, 14 Feb 2005 18:40:31 +0000 (18:40 +0000)]
Fix typo.

Submitted by: Antoine Brodin

19 years agoUpdate information now that support for priorities has been added.
njl [Mon, 14 Feb 2005 18:17:47 +0000 (18:17 +0000)]
Update information now that support for priorities has been added.

19 years agoImplement priorities. This allows a driver (say, for cooling purposes) to
njl [Mon, 14 Feb 2005 18:16:35 +0000 (18:16 +0000)]
Implement priorities.  This allows a driver (say, for cooling purposes) to
override the current freq level temporarily and restore it when the
higher priority condition is past.  Note that only the first overridden
value is saved.  Callers pass NULL to CPUFREQ_SET to restore the saved
level.  Priorities are not yet used so this commit should have no effect.

19 years ago- Use socklen_t.
stefanf [Mon, 14 Feb 2005 17:59:52 +0000 (17:59 +0000)]
- Use socklen_t.
- No need for 'fromlen' to have file scope.
- Remove an unused variable.

19 years agoUse socklen_t.
stefanf [Mon, 14 Feb 2005 17:55:33 +0000 (17:55 +0000)]
Use socklen_t.

19 years ago- Use socklen_t.
stefanf [Mon, 14 Feb 2005 17:51:45 +0000 (17:51 +0000)]
- Use socklen_t.
- No need for two instances of 'num'.

19 years agoDocument NGM_ETHER_DETACH functionality.
glebius [Mon, 14 Feb 2005 17:43:42 +0000 (17:43 +0000)]
Document NGM_ETHER_DETACH functionality.

19 years agoFix most cases where the address of an int is passed to a function expecting a
stefanf [Mon, 14 Feb 2005 17:42:58 +0000 (17:42 +0000)]
Fix most cases where the address of an int is passed to a function expecting a
socklen_t * argument.

19 years agoSeveral improvements to ps.1:
delphij [Mon, 14 Feb 2005 16:56:15 +0000 (16:56 +0000)]
Several improvements to ps.1:
- Document the fact that empty heading text suppresses the
  heading line (e.g. 'ps -o pid='), as this is very useful
  in scripts.
- Describe logname keyword more completely.
- Describe the printing of arguments more completely.
- Put lockname in the correct alphabetical order in the list
  of all keywords.
- Correct sentence in standards section.

Submitted by: Jilles Tjoelker <jilles at stack nl>
PR: docs/73618
MFC After: 1 week

19 years agoNote addition of MCFG support.
njl [Mon, 14 Feb 2005 16:32:32 +0000 (16:32 +0000)]
Note addition of MCFG support.

19 years agoFix parsing of '0' and non-alphanumerics in steps. Previously, an
delphij [Mon, 14 Feb 2005 14:09:21 +0000 (14:09 +0000)]
Fix parsing of '0' and non-alphanumerics in steps.  Previously, an
entry having stepping value of zero can cause crontab to hang there,
and if the main crontab is being changed in this way, then cron(8)
will keep spining.

Obtained from: OpenBSD [src/usr.sbin/cron/entry.c,v 1.17]
PR: 68683 (my own, but forgot to commit it...)
MFC After: 1 week

19 years agoMake WITNESS happier:
glebius [Mon, 14 Feb 2005 13:47:06 +0000 (13:47 +0000)]
Make WITNESS happier:
- refactor ngd_constructor, so that make_dev() is called without
  any locks held, since it mallocs memory with M_WAITOK flag.
- rename global mtx, to have name different to per-node mtx

MFC after: 2 weeks

19 years agoAdd a SUBDIR_TARGETS variable which can be set to a list of
ru [Mon, 14 Feb 2005 12:57:51 +0000 (12:57 +0000)]
Add a SUBDIR_TARGETS variable which can be set to a list of
additional targets that will cause descending into subdirs.
Example:

cd /sys/modules; make load SUBDIR_TARGETS=load

(But don't try it with your pet.)

Submitted by: Alexey Klimov
PR: 47601

19 years agoUse the system gnuregex library vs. building GNU regex bits into libiberty
obrien [Mon, 14 Feb 2005 12:10:14 +0000 (12:10 +0000)]
Use the system gnuregex library vs. building GNU regex bits into libiberty
and using them.

Reviewed by: marcel,imp
Desired by: ache

19 years agoAdd new netgraph control message NGM_ETHER_DETACH, which actually
glebius [Mon, 14 Feb 2005 12:01:09 +0000 (12:01 +0000)]
Add new netgraph control message NGM_ETHER_DETACH, which actually
removes netgraph node and unwraps Ethernet interface.

This gives us ability to unload ng_ether.ko, when all interfaces
are detached, making ng_ether(4) developers happy.

Reviewed by: ru

19 years agoCheck for non-NULL ac_netgraph field in interface arpcom, instead of
glebius [Mon, 14 Feb 2005 11:58:54 +0000 (11:58 +0000)]
Check for non-NULL ac_netgraph field in interface arpcom, instead of
checking global presence of ng_ether(4).

Reviewed by: ru

19 years ago. Convert return type of gai_strerror() to 'const char *' as POSIX requires.
phantom [Mon, 14 Feb 2005 11:33:12 +0000 (11:33 +0000)]
. Convert return type of gai_strerror() to 'const char *' as POSIX requires.
. Convert ai_errlist[] to simple 'char *' array, and appropriately
  optimize gai_strerror()

19 years agoEAI_ADDRFAMILY and EAI_NODATA are obsoleted, and not definined anymore,
phantom [Mon, 14 Feb 2005 11:24:58 +0000 (11:24 +0000)]
EAI_ADDRFAMILY and EAI_NODATA are obsoleted, and not definined anymore,
so do not export these definitions via manual page

19 years agoAdd support for parsing MCFG tables.
scottl [Mon, 14 Feb 2005 11:21:48 +0000 (11:21 +0000)]
Add support for parsing MCFG tables.

19 years agoIf no vlan(4) interfaces are configured for the interface, and the
ru [Mon, 14 Feb 2005 08:29:42 +0000 (08:29 +0000)]
If no vlan(4) interfaces are configured for the interface, and the
driver did VLAN decapsulation in hardware, we were passing a frame
as if it came for the parent (non-VLAN) interface.  Stop this from
happening.

Reminded by: glebius
Security: This could pose a security risk in some setups

19 years agoMerge mci.c change to add mci_close() from the vendor branch.
gshapiro [Mon, 14 Feb 2005 08:04:08 +0000 (08:04 +0000)]
Merge mci.c change to add mci_close() from the vendor branch.

Problem noted by: marcus

19 years agoo Add handling of an IPv4-mapped IPv6 address.
maxim [Mon, 14 Feb 2005 07:37:51 +0000 (07:37 +0000)]
o Add handling of an IPv4-mapped IPv6 address.
o Use SYSCTL_IN() macro instead of direct call of copyin(9).

Submitted by: ume

o Move sysctl_drop() implementation to sys/netinet/tcp_subr.c where
most of tcp sysctls live.
o There are net.inet[6].tcp[6].getcred sysctls already, no needs in
a separate struct tcp_ident_mapping.

Suggested by: ume

19 years agoCorrect the SiS 755 PCI ID. Confirmed against Linux code.
anholt [Mon, 14 Feb 2005 07:30:04 +0000 (07:30 +0000)]
Correct the SiS 755 PCI ID.  Confirmed against Linux code.

PR: kern/76411
Submitted by: Jonathan Fosburgh, jonathan at fosburgh dot org
Obtained from: Jung-uk Kim, jkim at niksun.com

19 years agoNo use for this AMD64 special-case "return NULL;" in probe now that we don't
anholt [Mon, 14 Feb 2005 07:16:25 +0000 (07:16 +0000)]
No use for this AMD64 special-case "return NULL;" in probe now that we don't
do fake "generic" support.

19 years agoRemove card_get_function. It looks like it was intended to be a
imp [Mon, 14 Feb 2005 07:00:39 +0000 (07:00 +0000)]
Remove card_get_function.  It looks like it was intended to be a
bridge between OLDCARD and NEWCARD for drivers to inquire after the
function number (eg, 0, 1, 2).  Nobody ever used it, so retire it
with honors.  NEWCARD never implemented it, and the same information
can be obtained by the pccard_get_function_number().

MFC After: 3 days

19 years agoRework DL10019/DL10022 support. This tries to reset things in a more
imp [Mon, 14 Feb 2005 06:54:06 +0000 (06:54 +0000)]
Rework DL10019/DL10022 support.  This tries to reset things in a more
proper way, or at least the same way that NetBSD and Linux do things
(I've been unable to obtain datasheets for these parts to know for
sure).  This has some marginal improvement in the DL10022 and DL10019
cards that I have.  Also, report which type, exactly.

# There's one or two ed cards that I have which still don't work, but I think
# that's due to MII losage on the card that's not presently compensated
# for in the MII drivers.

19 years agoMFp4: Merge in AX88790 support from my p4 tree. I've had this in my
imp [Mon, 14 Feb 2005 06:47:22 +0000 (06:47 +0000)]
MFp4: Merge in AX88790 support from my p4 tree.  I've had this in my
tree since 2003/02/20, and I recently cleaned it up.  I'd even closed
the PR that I obtained this from Fri Jul 18 23:25:08 MDT 2003 since
I looked at my p4 tree.

PR: 46889
Submitted by: HASEGAWA Tomoki