]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoCatch up with libpthread/support/Makefile.inc,v 1.6 and revert
ru [Mon, 15 Sep 2003 21:18:30 +0000 (21:18 +0000)]
Catch up with libpthread/support/Makefile.inc,v 1.6 and revert
Makefile.inc1 revisions 1.365 and 1.367: libc_pic.a is no longer
necessary to build libpthread.so (ie: libkse.so).

21 years agoCorrectly reset ich[3-5] sound cards on resume. This fixes audio playback
njl [Mon, 15 Sep 2003 21:16:47 +0000 (21:16 +0000)]
Correctly reset ich[3-5] sound cards on resume.  This fixes audio playback
after suspend/resume for me.

PR:
Submitted by: iwasaki
Reviewed by: orion
Approved by: cg
Obtained from:
MFC after:

21 years agoo do not filter received frames based on type or length; pass 'em all up
sam [Mon, 15 Sep 2003 19:41:54 +0000 (19:41 +0000)]
o do not filter received frames based on type or length; pass 'em all up
  to the 802.11 layer if they are at least IEEE80211_MIN_LEN
o mask off interrupt status bits that we don't care about so we don't do
  the wrong thing; this fixes a problem where the beacon miss interrupt status
  bit is delivered together with other status bits when operating in monitor
  mode (we would post a beacon miss swi and then do the wrong thing)

21 years agoChange input handling to not expect drivers to filter short frames.
sam [Mon, 15 Sep 2003 19:38:29 +0000 (19:38 +0000)]
Change input handling to not expect drivers to filter short frames.
In particular, let drivers send up control frames so we can dispatch
them to bpf in monitor mode.

This is the first (small) step to adding more functionality such as
power save mode.

21 years agoadd definitions for various control frames and a minimum-sized frame that
sam [Mon, 15 Sep 2003 19:36:34 +0000 (19:36 +0000)]
add definitions for various control frames and a minimum-sized frame that
we use to define IEEE80211_MIN_LEN; the minimal length for frames drivers
may pass up into the 802.11 layer

21 years agoRelease a previous lease if there is any, instead of
mbr [Mon, 15 Sep 2003 18:27:38 +0000 (18:27 +0000)]
Release a previous lease if there is any, instead of
just killing dhclient. Needed my some broken dhcpd
implementations to be able to get a lease again.

21 years agoNote the addition of `-C 60' to inetd_flags in etc/defaults/rc.conf.
nectar [Mon, 15 Sep 2003 17:20:05 +0000 (17:20 +0000)]
Note the addition of `-C 60' to inetd_flags in etc/defaults/rc.conf.

21 years agoAdd `-C 60' to the default flags for inetd, so that it is less
nectar [Mon, 15 Sep 2003 16:44:24 +0000 (16:44 +0000)]
Add `-C 60' to the default flags for inetd, so that it is less
vulnerable to run-of-the-mill DoS attacks in the default installation.

21 years agoRemove nsswitch.conf from the list of examples, because there is no
nectar [Mon, 15 Sep 2003 16:42:52 +0000 (16:42 +0000)]
Remove nsswitch.conf from the list of examples, because there is no
such example.

Reported by: blackend

21 years agoBetter explain what INSTALL="install -C" does, and warn that this may
sheldonh [Mon, 15 Sep 2003 16:11:41 +0000 (16:11 +0000)]
Better explain what INSTALL="install -C" does, and warn that this may
be hardwired into makefiles, including those under /usr/share/mk.

The reporter submitted a patch, but I've watered it down.

Reported by: Ian Freislich <ianf@za.uu.net>
MFC after: 3 weeks

21 years agoThe previous fix (rev 1.25) is only needed in the second place.
mbr [Mon, 15 Sep 2003 15:48:12 +0000 (15:48 +0000)]
The previous fix (rev 1.25) is only needed in the second place.

21 years agoQuick fix for a pessimization in rev.1.194. An extra i/o instruction
bde [Mon, 15 Sep 2003 13:49:18 +0000 (13:49 +0000)]
Quick fix for a pessimization in rev.1.194.  An extra i/o instruction
was added to the fast path to support the COM_IIR_RXRDYBUG() case even
when that case is not configured.  This increased the relative overhead
of sio input by almost 25% in the worst case and by 2-3% in the usual
case (usually only about 0.2% absolute per port at 115200 bps).  The
quick fix is to significantly pessimize only the COM_IIR_RXRDYBUG()
case.

21 years agoAdd a module dependancy. Now CAM will autoload when you load this.
markm [Mon, 15 Sep 2003 06:41:33 +0000 (06:41 +0000)]
Add a module dependancy. Now CAM will autoload when you load this.

OK'ed by: mdodd

21 years agoOnly enable S4BIOS by default if the FACS says it is available. The
njl [Mon, 15 Sep 2003 06:29:31 +0000 (06:29 +0000)]
Only enable S4BIOS by default if the FACS says it is available.  The
user can override this with a sysctl.

Be sure to return the acpi_SetSleepState return value to userland.

21 years agoEliminate the use of Giant from vm_object_reference().
alc [Mon, 15 Sep 2003 05:58:27 +0000 (05:58 +0000)]
Eliminate the use of Giant from vm_object_reference().

21 years agoSort: build uart_bus_pccard.c before uart_bus_pci.c.
marcel [Mon, 15 Sep 2003 04:53:11 +0000 (04:53 +0000)]
Sort: build uart_bus_pccard.c before uart_bus_pci.c.

21 years agoOur getaddrinfo() and getnameinfo() are thread-safe but
ume [Mon, 15 Sep 2003 04:52:15 +0000 (04:52 +0000)]
Our getaddrinfo() and getnameinfo() are thread-safe but
some limitation.

Reported by: Marc G. Fournier <scrappy@hub.org>

21 years agoRemove inclusion of <sys/timepps.h>. It's included in "uart_bus.h"
marcel [Mon, 15 Sep 2003 04:49:22 +0000 (04:49 +0000)]
Remove inclusion of <sys/timepps.h>. It's included in "uart_bus.h"
to avoid having to include it in almost all other source files.

21 years agoOnly kill the default route in the release mode if the interface
mbr [Mon, 15 Sep 2003 04:47:41 +0000 (04:47 +0000)]
Only kill the default route in the release mode if the interface
where the default route is bound to is the same device. Keep it
working for the single user or sysinstall mode, where we just do
what we did before.

Suggested by:   dougb

21 years agoRemove useless #ifdef PC98.
takawata [Mon, 15 Sep 2003 03:12:27 +0000 (03:12 +0000)]
Remove useless #ifdef PC98.

Submitted by: nyan

21 years agoCorrect a typo in the previous revision.
alc [Mon, 15 Sep 2003 02:56:48 +0000 (02:56 +0000)]
Correct a typo in the previous revision.

21 years agoFix bogus comment and assign sigmask in critical region, use
davidxu [Mon, 15 Sep 2003 00:08:48 +0000 (00:08 +0000)]
Fix bogus comment and assign sigmask in critical region, use
SIG_CANTMASK to remove unmaskable signal masks.

21 years agoFix a bogus comment, sigmask must be maintained correctly,
davidxu [Mon, 15 Sep 2003 00:06:46 +0000 (00:06 +0000)]
Fix a bogus comment, sigmask must be maintained correctly,
it will be inherited in pthread_create.

21 years agoRemove invalid statement about the maximum number of swap drives.
hmp [Sun, 14 Sep 2003 23:44:55 +0000 (23:44 +0000)]
Remove invalid statement about the maximum number of swap drives.

PR: docs/55276
Submitted by: Roderick van Domburg <r.s.a.vandomburg@student.utwente.nl>

21 years agoUpdate supported Ethernet Controllers as per the Hardware Notes.
hmp [Sun, 14 Sep 2003 23:38:42 +0000 (23:38 +0000)]
Update supported Ethernet Controllers as per the Hardware Notes.

PR: docs/55559
Submitted by: Lukas Ertl <l.ertl@univie.ac.at>
MFC after: 3 days

21 years agomust also check for 5Ghz channels when marking short preamble capability in
sam [Sun, 14 Sep 2003 22:53:41 +0000 (22:53 +0000)]
must also check for 5Ghz channels when marking short preamble capability in
the beacon frames

Reminded by: Stephane Laroche <stephane.laroche@colubris.com>

21 years ago1. Allocating and freeing lock related resource in _thr_alloc and _thr_free
davidxu [Sun, 14 Sep 2003 22:52:16 +0000 (22:52 +0000)]
1. Allocating and freeing lock related resource in _thr_alloc and _thr_free
   to avoid potential memory leak, also fix a bug in pthread_create, contention
   scope should be inherited when PTHREAD_INHERIT_SCHED is set, and also check
   right field for PTHREAD_INHERIT_SCHED, scheduling inherit flag is in sched_inherit.
2. Execute hooks registered by atexit() on thread stack but not on scheduler
   stack.
3. Simplify some code in _kse_single_thread by calling xxx_destroy functions.

Reviewed by: deischen

21 years agoWhen invoking an old style signal handler, use true traditional BSD style to
davidxu [Sun, 14 Sep 2003 22:42:39 +0000 (22:42 +0000)]
When invoking an old style signal handler, use true traditional BSD style to
invoke signal handler.

Reviewed by: deischen

21 years agoRespect POSIX specification, a value return from pthread_attr_getguardsize
davidxu [Sun, 14 Sep 2003 22:39:44 +0000 (22:39 +0000)]
Respect POSIX specification, a value return from pthread_attr_getguardsize
should be a value past to pthread_attr_setguardsize, not a rounded up value.
Also fix a stack size matching bug in thr_stack.c, now stack matching code
uses number of pages but not bytes length to match stack size, so for example,
size 512 bytes and size 513 bytes should both match 1 page stack size.

Reviewed by: deischen

21 years agoo mark the device capable of short preamble (meaningless for the 5210 but
sam [Sun, 14 Sep 2003 22:39:19 +0000 (22:39 +0000)]
o mark the device capable of short preamble (meaningless for the 5210 but
  safe since the 802.11 layer does the right thing for 11a operation)
o select short preamble operation based on the negotiated capabilities; not
  just the local state/capability
o fillin the duration field in the 802.11 header as appropriate
o remove detection of 11g support; no longer needed

Obtained from: MADWIFI (with modifications)

21 years agoHonor the short preamble capability/state flag when calculating the
sam [Sun, 14 Sep 2003 22:34:24 +0000 (22:34 +0000)]
Honor the short preamble capability/state flag when calculating the
capabilities for outbound management frames.  But beware of sending
this when operating on 5GHz channels; some 11a AP's reject association
requests if this bit is set in the capabilities listed.

Obtained from: MADWIFI (with modifications)

21 years agoAvoid garbage bits in c_flags by direct assigning value.
davidxu [Sun, 14 Sep 2003 22:33:32 +0000 (22:33 +0000)]
Avoid garbage bits in c_flags by direct assigning value.

Reviewed by: deischen

21 years agoIf user is seting scope process flag, clear PTHREAD_SCOPE_SYSTEM bit
davidxu [Sun, 14 Sep 2003 22:32:28 +0000 (22:32 +0000)]
If user is seting scope process flag, clear PTHREAD_SCOPE_SYSTEM bit
accordingly.

Reviewed by: deischen

21 years agoshort preamble capability is not just for 11g; mark IEEE80211_F_SHPREAMBLE
sam [Sun, 14 Sep 2003 22:32:18 +0000 (22:32 +0000)]
short preamble capability is not just for 11g; mark IEEE80211_F_SHPREAMBLE
regardless of the operating mode

Obtained from: MADWIFI

21 years agoCheck invalid parameter and return EINVAL.
davidxu [Sun, 14 Sep 2003 22:28:13 +0000 (22:28 +0000)]
Check invalid parameter and return EINVAL.

Reviewed by: deischen

21 years agoRemove an unneccessary comma.
ceri [Sun, 14 Sep 2003 20:35:22 +0000 (20:35 +0000)]
Remove an unneccessary comma.

21 years agoTeach the PCI code to parse MSI extended capabilities. Re-arrange the
scottl [Sun, 14 Sep 2003 19:30:00 +0000 (19:30 +0000)]
Teach the PCI code to parse MSI extended capabilities.  Re-arrange the
pcicfg struct a bit to hold extcap structures instead of structure members.

21 years agoNew release notes: DRM update for 20030909, uart(4), ata(4) update
bmah [Sun, 14 Sep 2003 18:20:24 +0000 (18:20 +0000)]
New release notes:  DRM update for 20030909, uart(4), ata(4) update
(ATAng), rtld(1) does libmap by default, symorder gone, sysinstall(8)
alternate MTA selection, GCC -pthread gone.

21 years agoPrint notify values as hex.
njl [Sun, 14 Sep 2003 17:47:44 +0000 (17:47 +0000)]
Print notify values as hex.

21 years agoAdd uart pccard attachment decription on conf/files, too.
takawata [Sun, 14 Sep 2003 17:02:59 +0000 (17:02 +0000)]
Add uart pccard attachment decription on conf/files, too.

Pointed out by: ru

21 years agoDon't depend on IPv4-mapped IPv6 address to bind to both IPv4
ume [Sun, 14 Sep 2003 16:42:46 +0000 (16:42 +0000)]
Don't depend on IPv4-mapped IPv6 address to bind to both IPv4
and IPv6.

Wrote at: Hakone.
Powered by: Warner Losh's scotch whisky.
Requested by: nork

21 years ago- Avoid calling pci_get_device() more than once in a single function.
mdodd [Sun, 14 Sep 2003 16:33:48 +0000 (16:33 +0000)]
- Avoid calling pci_get_device() more than once in a single function.
- Provide a mechanism to prevent the use of MMIO.
- Prevent the use of MMIO for all 3c575 cardbus cards.

21 years agoAdd uart pccard bus attachment,based on sio_pccard.c .
takawata [Sun, 14 Sep 2003 16:21:06 +0000 (16:21 +0000)]
Add uart pccard bus attachment,based on sio_pccard.c .

Wrote at: Hakone.
Powered by: Warner Losh's scotch whisky.
Tested by: nork

21 years agoExpand the extended capabilities list and add definitions for MSI.
scottl [Sun, 14 Sep 2003 14:42:26 +0000 (14:42 +0000)]
Expand the extended capabilities list and add definitions for MSI.

21 years agoGet rid of duplicates.
ru [Sun, 14 Sep 2003 13:41:59 +0000 (13:41 +0000)]
Get rid of duplicates.

21 years ago- Cut out the code that caches the "." directory out of Dir_Init()
ru [Sun, 14 Sep 2003 12:31:33 +0000 (12:31 +0000)]
- Cut out the code that caches the "." directory out of Dir_Init()
  into a separate function, Dir_InitDot().

- Postpone the current and object directories detection (and caching
  of the "." directory) until after all command line arguments are
  parsed.  This makes the -C option DTRT.

PR: bin/47149

21 years agoAdd a new sysctl, security.bsd.conservative_signals, to disable
rwatson [Sun, 14 Sep 2003 07:22:38 +0000 (07:22 +0000)]
Add a new sysctl, security.bsd.conservative_signals, to disable
special signal-delivery protections for setugid processes.  In the
event that a system is relying on "unusual" signal delivery to
processes that change their credentials, this can be used to work
around application problems.

Also, add SIGALRM to the set of signals permitted to be delivered to
setugid processes by unprivileged subjects.

Reported by: Joe Greco <jgreco@ns.sol.net>

21 years agoRemove most of the magic constants from the extcap parsing code.
scottl [Sun, 14 Sep 2003 06:23:19 +0000 (06:23 +0000)]
Remove most of the magic constants from the extcap parsing code.

21 years agoCall vm_page_unmanage() on pages belonging to the kmem_object. This
alc [Sun, 14 Sep 2003 02:37:59 +0000 (02:37 +0000)]
Call vm_page_unmanage() on pages belonging to the kmem_object.  This
eliminates the unnecessary overhead of managing "PV" entries for these
pages.

21 years agoEnable IPv6 for Token Ring.
mdodd [Sun, 14 Sep 2003 02:32:31 +0000 (02:32 +0000)]
Enable IPv6 for Token Ring.

21 years agoCosmetic cleanups.
mdodd [Sun, 14 Sep 2003 02:26:45 +0000 (02:26 +0000)]
Cosmetic cleanups.

21 years agoCosmetic adjustment.
mdodd [Sun, 14 Sep 2003 02:18:54 +0000 (02:18 +0000)]
Cosmetic adjustment.

21 years agoCorrect the calculation of "a leap year" in parseDWM. The calculation
gad [Sun, 14 Sep 2003 00:56:50 +0000 (00:56 +0000)]
Correct the calculation of "a leap year" in parseDWM.  The calculation
would only match a leap year every 400 years.  The parseDWM code first
showed up in April 2000, so the first time this bug would cause any
confusion is in Feb 2004.

MFC after: 18 days

21 years agoCorrect typo in comment.
njl [Sun, 14 Sep 2003 00:40:24 +0000 (00:40 +0000)]
Correct typo in comment.

21 years agoFor dumping the DSDT (-o option), be sure to use the X_DSDT address if
njl [Sun, 14 Sep 2003 00:37:52 +0000 (00:37 +0000)]
For dumping the DSDT (-o option), be sure to use the X_DSDT address if
appropriate.  This should be the last change to make ia64 work.

21 years agoTeach the re(4) driver about the CFG2 register, which tells us whether
wpaul [Sat, 13 Sep 2003 23:51:35 +0000 (23:51 +0000)]
Teach the re(4) driver about the CFG2 register, which tells us whether
we're on a 32-bit/64-bit bus or not. Use this to decide if we should
set the PCI dual-address cycle enable bit in the C+ command register.
(Enabling DAC on a 32-bit bus seems to do bad things.)

Also, initialize the C+ command register early in the re_init() routine.
The documentation says this register should be configured first.

21 years agoMake the description of the connect, get, and put commands clearer.
simon [Sat, 13 Sep 2003 23:34:34 +0000 (23:34 +0000)]
Make the description of the connect, get, and put commands clearer.

Reported by: Gary W. Swearingen <underway@comcast.net>
Submitted by: Jim Brown <jpb@sixshooter.v6.thrupoint.net> (original version)
Reviewed by: ru
PR: docs/36459
MFC after: 3 weeks

21 years agoThe in6_ifattach() routine contains the following code:
wpaul [Sat, 13 Sep 2003 22:34:52 +0000 (22:34 +0000)]
The in6_ifattach() routine contains the following code:

        in6_pcbpurgeif0(LIST_FIRST(udbinfo.listhead), ifp);
        in6_pcbpurgeif0(LIST_FIRST(ripcbinfo.listhead), ifp);

The problem here is that udbinfo.listhead and ripcbinfo.listhead are
not initialized during the device probe/attach phase of the kernel
boot process. So if, for example, a network driver calls ether_ifattach()
in its foo_attach() routine and then decides that something is wrong
and calls ether_ifdetach() to reverse the process, we will panic trying
to dereference the uninitialized list head pointers. (Though the
same sequence of events performed after the kernel has come up works
file, i.e. doing kldload if_foo from multiuser.)

Change this to:

        if (udbinfo.listhead != NULL)
                in6_pcbpurgeif0(LIST_FIRST(udbinfo.listhead), ifp);
        if (ripcbinfo.listhead != NULL)
                in6_pcbpurgeif0(LIST_FIRST(ripcbinfo.listhead), ifp);

to avoid the NULL pointer dereferences.

21 years agoSimplify (and micro-optimize) pmap_unuse_pt(): Only one caller,
alc [Sat, 13 Sep 2003 21:57:38 +0000 (21:57 +0000)]
Simplify (and micro-optimize) pmap_unuse_pt(): Only one caller,
pmap_remove_pte(), passed NULL instead of the required page table
page to pmap_unuse_pt().  Compute the necessary page table page
in pmap_remove_pte().  Also, remove some unreachable code from
pmap_remove_pte().

21 years agoRetire the WITH_LIBMAP compile knob; libmap is now a standard feature.
mdodd [Sat, 13 Sep 2003 21:50:36 +0000 (21:50 +0000)]
Retire the WITH_LIBMAP compile knob; libmap is now a standard feature.

21 years agoChange libmap.c:lm_init() to return a status value; 0 for success
mdodd [Sat, 13 Sep 2003 21:43:08 +0000 (21:43 +0000)]
Change libmap.c:lm_init() to return a status value; 0 for success
(libmap available) and 1 for failure.  Assign this return to the
global 'libmap_disable' variable in rtld.c.

This totally prevents any libmap functions from being called after
lm_init() if no config file is present.

21 years agoBe a little bit more correct WRT counting numbers vs. integer numbers.
obrien [Sat, 13 Sep 2003 20:46:58 +0000 (20:46 +0000)]
Be a little bit more correct WRT counting numbers vs. integer numbers.

21 years agoAdd the -i flag to acpiconf(8) to retrieve battery information.
njl [Sat, 13 Sep 2003 20:13:01 +0000 (20:13 +0000)]
Add the -i flag to acpiconf(8) to retrieve battery information.
Rename a few structure elements.

21 years agoThere is no need for an atomic increment on the vm object's generation
alc [Sat, 13 Sep 2003 20:07:26 +0000 (20:07 +0000)]
There is no need for an atomic increment on the vm object's generation
count in _vm_object_allocate().  (Access to the generation count is
governed by the vm object's lock.)  Note: the introduction of the
atomic increment in revision 1.238 appears to be an accident.  The
purpose of that commit was to fix an Alpha-specific bug in UMA's
debugging code.

21 years agoC2/C3 latency is in microseconds (us), not ms.
njl [Sat, 13 Sep 2003 18:47:49 +0000 (18:47 +0000)]
C2/C3 latency is in microseconds (us), not ms.

21 years agosched_setscheduler: Return EINVAL when a invalid policy is specified,
nectar [Sat, 13 Sep 2003 18:46:24 +0000 (18:46 +0000)]
sched_setscheduler: Return EINVAL when a invalid policy is specified,
thus complying with POLA and the man page.  (Previously, no error was
returned for this case.)

21 years agoOnly print various optional fields if they are non-zero. Always print the
njl [Sat, 13 Sep 2003 18:44:05 +0000 (18:44 +0000)]
Only print various optional fields if they are non-zero.  Always print the
required fields, no matter what their value.

21 years agoApply Aleksander Fafula's crayons to the beastie. Add the this line to
scottl [Sat, 13 Sep 2003 18:35:01 +0000 (18:35 +0000)]
Apply Aleksander Fafula's crayons to the beastie.  Add the this line to
/boot/loader.conf to see the pretty colors =-)

loader_color="YES"

21 years agoReduce the default width for IO port GAS printing.
njl [Sat, 13 Sep 2003 17:32:45 +0000 (17:32 +0000)]
Reduce the default width for IO port GAS printing.

Courtesy of: rwatson's chart presentation

21 years agoCorrect mostly harmless off-by-one error in getdomainname().
nectar [Sat, 13 Sep 2003 17:12:22 +0000 (17:12 +0000)]
Correct mostly harmless off-by-one error in getdomainname().

Reviewed by: imp

21 years agoDocument /var/empty as generic empty directory, not sshd specific
eivind [Sat, 13 Sep 2003 16:23:49 +0000 (16:23 +0000)]
Document /var/empty as generic empty directory, not sshd specific

21 years agoasami is no longer our ports lead, so remove a statement saying
eivind [Sat, 13 Sep 2003 16:16:31 +0000 (16:16 +0000)]
asami is no longer our ports lead, so remove a statement saying
the ports collection is maintained by asami and the ports team.

21 years agoFix spelling/grammar errors and remove a duplicate
eivind [Sat, 13 Sep 2003 15:46:35 +0000 (15:46 +0000)]
Fix spelling/grammar errors and remove a duplicate

21 years agoMerged from sys/isa/fd.c revision 1.258.
nyan [Sat, 13 Sep 2003 12:56:05 +0000 (12:56 +0000)]
Merged from sys/isa/fd.c revision 1.258.

21 years agoThe re driver works on pc98 maybe.
nyan [Sat, 13 Sep 2003 12:51:12 +0000 (12:51 +0000)]
The re driver works on pc98 maybe.

21 years agoDo not assume there is only a space between #define and the macro name
schweikh [Sat, 13 Sep 2003 06:59:22 +0000 (06:59 +0000)]
Do not assume there is only a space between #define and the macro name
when grepping for JOBS. The recent style cleanup replaced the space with
a tab and broke job control detection. Little edits, disastrous consequences.

Submitted by: Peter Edwards <pmedwards@eircom.net>
X-MFC when:     in about 5 weeks with the other sh arithmetic fixes.

21 years agoAdd support for automatic hardware flow control for 16[679]50 UARTs.
marcel [Sat, 13 Sep 2003 06:25:04 +0000 (06:25 +0000)]
Add support for automatic hardware flow control for 16[679]50 UARTs.
We simply use the detected FIFO size to determine whether we have
a post 16550 UART or not. The support lacks proper serialization of
hardware access for now.

21 years agoConvert vmapbuf() from using pmap_extract() to using
alc [Sat, 13 Sep 2003 04:29:55 +0000 (04:29 +0000)]
Convert vmapbuf() from using pmap_extract() to using
pmap_extract_and_hold().  Note, however, that GIANT_REQUIRED should not be
removed until all platforms fully implement the "prot" parameter to
pmap_extract_and_hold().

Reviewed by: tegge

21 years agoI forgot whom I got this from- only set single initiator buffered mode
mjacob [Sat, 13 Sep 2003 02:01:56 +0000 (02:01 +0000)]
I forgot whom I got this from- only set single initiator buffered mode
if we've recorded in our softc that we should set it.

21 years agoNetBSD'er with time on their hands (dma->DMA)
mjacob [Sat, 13 Sep 2003 01:59:14 +0000 (01:59 +0000)]
NetBSD'er with time on their hands (dma->DMA)

21 years agoBland synchronization of common source with no effect on FreeBSD whatsomever.
mjacob [Sat, 13 Sep 2003 01:58:51 +0000 (01:58 +0000)]
Bland synchronization of common source with no effect on FreeBSD whatsomever.

21 years agoSome ridiculous target mode botches- like having the wrong definitions
mjacob [Sat, 13 Sep 2003 01:58:26 +0000 (01:58 +0000)]
Some ridiculous target mode botches- like having the wrong definitions
for messages. Some spelling fixes. Some target mode structure cleanups
to reflect reality.

21 years agoSome canonicalization function cleanups based upon some target mode
mjacob [Sat, 13 Sep 2003 01:57:31 +0000 (01:57 +0000)]
Some canonicalization function cleanups based upon some target mode
structure cleanups.

21 years agoNo time like the present to turn back on SMP locking.
mjacob [Sat, 13 Sep 2003 01:56:45 +0000 (01:56 +0000)]
No time like the present to turn back on SMP locking.

21 years agoReport correct active vs. nvram node/port WWNs in an ioctl.
mjacob [Sat, 13 Sep 2003 01:56:24 +0000 (01:56 +0000)]
Report correct active vs. nvram node/port WWNs in an ioctl.

21 years agoOn reset, make sure that we have some parameters set correctly. This
mjacob [Sat, 13 Sep 2003 01:55:44 +0000 (01:55 +0000)]
On reset, make sure that we have some parameters set correctly. This
fixes a longstanding issue WRT resetting the chip after startup- it
would fail if we were connected as an F-port to a switch. If we
were connected as an F-port, we got assigned a hard loop ID of 255,
which is really a bogus loop id. Then when we turned around to
reset ourselves, the firmware would reject the ICB_INIT request
because the loop id was bogus. *sputter*

Minor fixlet from somebody in NetBSD with too much time on their
hands (dma -> DMA).

21 years agoMove an overly verbose message under #ifdef CODA_VERBOSE.
tjr [Sat, 13 Sep 2003 01:13:56 +0000 (01:13 +0000)]
Move an overly verbose message under #ifdef CODA_VERBOSE.

21 years agomdoc(7): Fix common mistakes made in the SEE ALSO section.
ru [Fri, 12 Sep 2003 21:54:11 +0000 (21:54 +0000)]
mdoc(7): Fix common mistakes made in the SEE ALSO section.

21 years agoWhen determining the device class to use for the serial console, check
tmm [Fri, 12 Sep 2003 20:13:23 +0000 (20:13 +0000)]
When determining the device class to use for the serial console, check
the "compatible" property too in the ns8250 case. This gets the serial
console to work on Blade 100s, where the device name is just "serial".

Reviewed by: marcel

21 years agoHandle ISA devices in OF_decode_addr(), with the same code that is
tmm [Fri, 12 Sep 2003 20:04:29 +0000 (20:04 +0000)]
Handle ISA devices in OF_decode_addr(), with the same code that is
used in the EBus case.

21 years agoRemove double usage of performance, and combine it to one.
hmp [Fri, 12 Sep 2003 19:20:05 +0000 (19:20 +0000)]
Remove double usage of performance, and combine it to one.

21 years agoBump __FreeBSD_cc_version for (1) 5.1 (post-mortem) (2) -pthread changes.
obrien [Fri, 12 Sep 2003 19:19:10 +0000 (19:19 +0000)]
Bump __FreeBSD_cc_version for (1) 5.1 (post-mortem) (2) -pthread changes.

21 years agoUsing prefix=/usr will put these files in /usr/gnats, not /usr/ports.
ceri [Fri, 12 Sep 2003 19:02:34 +0000 (19:02 +0000)]
Using prefix=/usr will put these files in /usr/gnats, not /usr/ports.

21 years agoName the vnode method vectors consistently with the rest of the filesystems.
phk [Fri, 12 Sep 2003 16:44:40 +0000 (16:44 +0000)]
Name the vnode method vectors consistently with the rest of the filesystems.

This improves the output of src/tools/tools/vop_table

21 years agoAdd the ports/dns and ports/polish categories.
ceri [Fri, 12 Sep 2003 09:09:49 +0000 (09:09 +0000)]
Add the ports/dns and ports/polish categories.

21 years agoUnbreak "make release" by moving bfe and re driver into 2nd floppy.
kuriyama [Fri, 12 Sep 2003 08:20:47 +0000 (08:20 +0000)]
Unbreak "make release" by moving bfe and re driver into 2nd floppy.

21 years agopipe_build_write_buffer() only requires read access of the page that it
alc [Fri, 12 Sep 2003 07:13:15 +0000 (07:13 +0000)]
pipe_build_write_buffer() only requires read access of the page that it
obtains from pmap_extract_and_hold().

21 years agoAdd a new parameter to pmap_extract_and_hold() that is needed to eliminate
alc [Fri, 12 Sep 2003 07:07:49 +0000 (07:07 +0000)]
Add a new parameter to pmap_extract_and_hold() that is needed to eliminate
Giant from vmapbuf().

Idea from: tegge

21 years agoMove the parse8601 and parseDWM routines into a new ptime.c file. The
gad [Fri, 12 Sep 2003 01:33:37 +0000 (01:33 +0000)]
Move the parse8601 and parseDWM routines into a new ptime.c file.  The
only code-change is to add a "next_time" parameter to both routines (and
that is not used yet).  A later update will make "next_time" more useful.

MFC after: 20 days

21 years agoDelay most of the adapter initilization until after interrupts are
ps [Thu, 11 Sep 2003 23:30:28 +0000 (23:30 +0000)]
Delay most of the adapter initilization until after interrupts are
enabled.

Tested by: Richard Puga <puga@mauibuilt.com>