]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoremove EAI_NODATA aliased to EAI_NONAME.
Hajimu UMEMOTO [Thu, 15 Jan 2004 15:10:02 +0000 (15:10 +0000)]
remove EAI_NODATA aliased to EAI_NONAME.

PR: bin/61369

20 years agoCleanup obsolete text:
Atsushi Onoe [Thu, 15 Jan 2004 14:06:51 +0000 (14:06 +0000)]
Cleanup obsolete text:
- hostap and ibss creation now works.
- delete messages for common net80211 layer from DIAGNOSTICS.

20 years agoCorrect beacon (src/bssid) for hostap and adhoc mode.
Atsushi Onoe [Thu, 15 Jan 2004 13:30:06 +0000 (13:30 +0000)]
Correct beacon (src/bssid) for hostap and adhoc mode.

20 years agoo Document -s flag.
Maxim Konovalov [Thu, 15 Jan 2004 13:23:21 +0000 (13:23 +0000)]
o Document -s flag.

No objections from: ume
MFC after: 2 weeks

20 years agoo -c (compact) flag is ipfw2 feature.
Maxim Konovalov [Thu, 15 Jan 2004 12:59:44 +0000 (12:59 +0000)]
o -c (compact) flag is ipfw2 feature.

PR: bin/56328
MFC after: 3 days

20 years agoo -f (force) in conjunction with -p (preprocessor) is ipfw2 feature.
Maxim Konovalov [Thu, 15 Jan 2004 12:57:04 +0000 (12:57 +0000)]
o -f (force) in conjunction with -p (preprocessor) is ipfw2 feature.

MFC after: 3 days

20 years agoAdd RSH to the list of enviroment variables.
Alexander Langer [Thu, 15 Jan 2004 12:13:54 +0000 (12:13 +0000)]
Add RSH to the list of enviroment variables.

20 years agoAssorted markup fixes.
Ruslan Ermilov [Thu, 15 Jan 2004 11:15:27 +0000 (11:15 +0000)]
Assorted markup fixes.

20 years agoAdd missing '\' in previous change.
Yoshihiro Takahashi [Thu, 15 Jan 2004 10:17:43 +0000 (10:17 +0000)]
Add missing '\' in previous change.

20 years agoNew file descriptor allocation code, derived from similar code introduced
Dag-Erling Smørgrav [Thu, 15 Jan 2004 10:15:04 +0000 (10:15 +0000)]
New file descriptor allocation code, derived from similar code introduced
in OpenBSD by Niels Provos.  The patch introduces a bitmap of allocated
file descriptors which is used to locate available descriptors when a new
one is needed.  It also moves the task of growing the file descriptor table
out of fdalloc(), reducing complexity in both fdalloc() and do_dup().

Debts of gratitude are owed to tjr@ (who provided the original patch on
which this work is based), grog@ (for the gdb(4) man page) and rwatson@
(for assistance with pxeboot(8)).

20 years agoFixed "make clean".
Ruslan Ermilov [Thu, 15 Jan 2004 10:07:59 +0000 (10:07 +0000)]
Fixed "make clean".

20 years agoUse generic net80211 framework for awi driver.
Atsushi Onoe [Thu, 15 Jan 2004 10:04:21 +0000 (10:04 +0000)]
Use generic net80211 framework for awi driver.
Restore awi to be workable again; it was broken..
XXX: The initialization is still unreliable yet, it sometimes fails on
     some card.

20 years agoFixed "make clean".
Ruslan Ermilov [Thu, 15 Jan 2004 10:02:34 +0000 (10:02 +0000)]
Fixed "make clean".

20 years agoAlongside revision 1.382, don't build profiled libraries so early
Ruslan Ermilov [Thu, 15 Jan 2004 09:50:54 +0000 (09:50 +0000)]
Alongside revision 1.382, don't build profiled libraries so early
in the build -- it's OK to keep this for the "make all" phase.

20 years agoAdd support for FH phy, which will be used by awi driver.
Atsushi Onoe [Thu, 15 Jan 2004 08:44:27 +0000 (08:44 +0000)]
Add support for FH phy, which will be used by awi driver.
Also some if_media constants to indicate operational mode are changed
to bitmasks to reduce diffs from NetBSD.

20 years ago- add openpic macio and psim/iobus attachments
Peter Grehan [Thu, 15 Jan 2004 08:43:46 +0000 (08:43 +0000)]
- add openpic macio and psim/iobus attachments
- alpha sort powermac files

20 years agoImplement NdisCopyFromPacketToPacket() and NdisCopyFromPacketToPacketSafe().
Bill Paul [Thu, 15 Jan 2004 07:42:26 +0000 (07:42 +0000)]
Implement NdisCopyFromPacketToPacket() and NdisCopyFromPacketToPacketSafe().
I only have one driver that references this routine (for the 3Com 3cR990)
and it never gets called, but just in case, here it is.

20 years agoAdd a per controller IOCTL interface.
Matthew N. Dodd [Thu, 15 Jan 2004 06:37:52 +0000 (06:37 +0000)]
Add a per controller IOCTL interface.

20 years ago- Add comments.
Matthew N. Dodd [Thu, 15 Jan 2004 04:05:47 +0000 (04:05 +0000)]
- Add comments.
- Add more command defines and data structures.
- Re-organize struct ida_drive_info to factor out struct ida_drive_info
  which will be used elsewhere.

20 years agoInsure values of adapter structure members are in correct byte order.
Matthew N. Dodd [Thu, 15 Jan 2004 02:42:20 +0000 (02:42 +0000)]
Insure values of adapter structure members are in correct byte order.

20 years agoVOP_GETATTR() wants the vnode passed to it to be locked. Instead
Don Lewis [Wed, 14 Jan 2004 22:38:03 +0000 (22:38 +0000)]
VOP_GETATTR() wants the vnode passed to it to be locked.  Instead
of adding the code to lock and unlock the vnodes and taking care
to avoid deadlock, simplify linux_emul_convpath() by comparing the
vnode pointers directly instead of comparing their va_fsid and
va_fileid attributes.  This allows the removal of the calls to
VOP_GETATTR().

20 years agoSpell POSIX threads library as "libpthread", mdoc(7) is smart
Ruslan Ermilov [Wed, 14 Jan 2004 21:53:19 +0000 (21:53 +0000)]
Spell POSIX threads library as "libpthread", mdoc(7) is smart
enough to render it as "-lkse" these days.

20 years agoHave a single set of POSIX threads man pages. The LIBRARY section
Ruslan Ermilov [Wed, 14 Jan 2004 21:44:26 +0000 (21:44 +0000)]
Have a single set of POSIX threads man pages.  The LIBRARY section
of each manpage lists libraries that have corresponding interfaces
implemented.

Prodded by: threads
Reviewed by: deischen

20 years agoUse UMA instead of plain malloc for getting ATA request storage.
Søren Schmidt [Wed, 14 Jan 2004 21:26:35 +0000 (21:26 +0000)]
Use UMA instead of plain malloc for getting ATA request storage.
This gives +10% performance on simple tests, so definitly worth it.
A few percent more could be had by not using M_ZERO'd alloc's, but
we then need to clear fields all over the place to be safe, and
that was deemed not worth the trouble (and it makes life dangerous).

20 years agoThe libc_r/man/sigwait.3 manpage has been repocopied to libc/sys/sigwait.2.
Ruslan Ermilov [Wed, 14 Jan 2004 21:22:10 +0000 (21:22 +0000)]
The libc_r/man/sigwait.3 manpage has been repocopied to libc/sys/sigwait.2.

Reviewed by: deischen
Repocopy by: markm

20 years agoThis interface is not provided by libc_r. It is, on the other
Ruslan Ermilov [Wed, 14 Jan 2004 21:12:39 +0000 (21:12 +0000)]
This interface is not provided by libc_r.  It is, on the other
hand, is also provided by libthr, but this will be addressed
later.

20 years agoBack out previous commit, which as bde@ pointed out is a no-op.
Dag-Erling Smørgrav [Wed, 14 Jan 2004 21:02:06 +0000 (21:02 +0000)]
Back out previous commit, which as bde@ pointed out is a no-op.

20 years ago- libc/sys/sem.c was repocopied to libc/gen/sem.c.
Ruslan Ermilov [Wed, 14 Jan 2004 20:54:16 +0000 (20:54 +0000)]
- libc/sys/sem.c was repocopied to libc/gen/sem.c.
- sem_*(3) manpages were repocopied from libc_r.

Reviewed by: deischen
Repocopy by: markm

20 years agoThis commit was generated by cvs2svn to compensate for changes in r124528,
Nate Lawson [Wed, 14 Jan 2004 18:54:22 +0000 (18:54 +0000)]
This commit was generated by cvs2svn to compensate for changes in r124528,
which included commits to RCS files with non-trunk default branches.

20 years agoWhen creating a package element via "Store(0xXXX, Index(ArgX, 0xXXX))",
Nate Lawson [Wed, 14 Jan 2004 18:54:22 +0000 (18:54 +0000)]
When creating a package element via "Store(0xXXX, Index(ArgX, 0xXXX))",
be sure to increment the refcount of the argument so it is not
prematurely deleted.  This is a workaround and may appear in a different
form in ACPI-CA.  This fixes battery evaluation on Thinkpads that was
broken by fixing the Dell battery state.

Submitted by: Luming Yu <luming.yu@intel.com>

20 years agoUpdate firmware sets for the 12160, 2200 and 2300 cards to the (more or
Matt Jacob [Wed, 14 Jan 2004 18:38:03 +0000 (18:38 +0000)]
Update firmware sets for the 12160, 2200 and 2300 cards to the (more or
less) latest from QLogic.

20 years ago-DSMALL will remove the need of dump.c (for rtsol)
Hajimu UMEMOTO [Wed, 14 Jan 2004 17:59:38 +0000 (17:59 +0000)]
-DSMALL will remove the need of dump.c (for rtsol)

Obtained from: KAME

20 years agoadd -F flag, which configures sysctl(8) setting by rtsold
Hajimu UMEMOTO [Wed, 14 Jan 2004 17:42:03 +0000 (17:42 +0000)]
add -F flag, which configures sysctl(8) setting by rtsold
itself (rather than warn about the current setting).

Obtained from: KAME

20 years agoansify.
Hajimu UMEMOTO [Wed, 14 Jan 2004 17:16:19 +0000 (17:16 +0000)]
ansify.

Obtained from: KAME

20 years agouse timeradd/sub/cmp.
Hajimu UMEMOTO [Wed, 14 Jan 2004 16:55:08 +0000 (16:55 +0000)]
use timeradd/sub/cmp.

Obtained from: KAME

20 years agoMFi386: revision 1.462
Yoshihiro Takahashi [Wed, 14 Jan 2004 15:11:08 +0000 (15:11 +0000)]
MFi386: revision 1.462

20 years agoFix potential cable detection problem on older CMD chips.
Søren Schmidt [Wed, 14 Jan 2004 14:24:36 +0000 (14:24 +0000)]
Fix potential cable detection problem on older CMD chips.

20 years agoAdd MLINK:
Christian Brueffer [Wed, 14 Jan 2004 13:35:15 +0000 (13:35 +0000)]
Add MLINK:

rue.4 -> if_rue.4

20 years agoRe-add libkern/ffs.c. I thought sparc64 had an inline version, but
Dag-Erling Smørgrav [Wed, 14 Jan 2004 08:38:13 +0000 (08:38 +0000)]
Re-add libkern/ffs.c.  I thought sparc64 had an inline version, but
failed to notice that it's #if 0'ed out.

20 years ago#include <machine/cpufunc.h>, which may define inline versions of some
Dag-Erling Smørgrav [Wed, 14 Jan 2004 08:36:36 +0000 (08:36 +0000)]
#include <machine/cpufunc.h>, which may define inline versions of some
of the functions in libkern.  Without this, parts of the kernel would
reference a non-existent (undeclared and undefined) ffs() function; the
only reason this didn't break the kernel build is that gcc happens to
have a built-in ffs() and incorrectly fails to warn about the lack of
prototypes for built-in functions.

20 years agoTranslate from GNU C to ISO C.
Dag-Erling Smørgrav [Wed, 14 Jan 2004 07:47:10 +0000 (07:47 +0000)]
Translate from GNU C to ISO C.

20 years agoRemove vm_page_alloc_contig(). It's now unused.
Alan Cox [Wed, 14 Jan 2004 06:21:38 +0000 (06:21 +0000)]
Remove vm_page_alloc_contig().  It's now unused.

20 years agoUse contigmalloc() instead of vm_page_alloc_contig(). Pass M_ZERO to
Alan Cox [Wed, 14 Jan 2004 06:14:35 +0000 (06:14 +0000)]
Use contigmalloc() instead of vm_page_alloc_contig().  Pass M_ZERO to
contigmalloc() instead of calling bzero().

20 years agoinvalidate secpolicy pcb cache on key_timehandler. part of
Hajimu UMEMOTO [Wed, 14 Jan 2004 04:39:40 +0000 (04:39 +0000)]
invalidate secpolicy pcb cache on key_timehandler.  part of
http://sources.zabbadoz.net/freebsd/patchset/110-ipsec-netkey-key.diff

Submitted by: "Bjoern A. Zeeb" <bzeeb+freebsd@zabbadoz.net>
Reviewed by: itojun

20 years agoadd missing key_freesp()s. part of
Hajimu UMEMOTO [Wed, 14 Jan 2004 04:10:28 +0000 (04:10 +0000)]
add missing key_freesp()s.  part of
http://sources.zabbadoz.net/freebsd/patchset/110-ipsec-netkey-key.diff
with some modification.

Submitted by: "Bjoern A. Zeeb" <bzeeb+freebsd@zabbadoz.net>
Reviewed by: itojun
Obtained from: KAME

20 years agomp_ncpus is always defined now, so no need to do an #ifdef SMP in
Bill Paul [Wed, 14 Jan 2004 01:16:14 +0000 (01:16 +0000)]
mp_ncpus is always defined now, so no need to do an #ifdef SMP in
ndis_cpu_cnt().

20 years agoCatch up with the English version:
Hiroki Sato [Wed, 14 Jan 2004 01:08:05 +0000 (01:08 +0000)]
Catch up with the English version:

1.57 -> 1.59 errata/article.sgml

20 years agoUse contigmalloc() instead of vm_page_alloc_contig().
Alan Cox [Wed, 14 Jan 2004 00:52:53 +0000 (00:52 +0000)]
Use contigmalloc() instead of vm_page_alloc_contig().

20 years agoAdd an "ethernet" hook to the rfc1490 netgraph module. It will send
Brian Feldman [Wed, 14 Jan 2004 00:39:28 +0000 (00:39 +0000)]
Add an "ethernet" hook to the rfc1490 netgraph module.  It will send
and receive FCS-less RFC1490-"bridged" Ethernet packets that are
currently just ignored.

20 years agoIf a device attach routine fails during boot and calls bus_teardown_intr(),
Don Lewis [Tue, 13 Jan 2004 22:55:46 +0000 (22:55 +0000)]
If a device attach routine fails during boot and calls bus_teardown_intr(),
ithread_remove_handler() may fail to remove the interrupt handler if
it decides to let the ithread do the removal.  The problem is that during
boot "cold" is set, which causes msleep() to return immediately.  This
will cause ithread_remove_handler() to fail to wait for the ithread
to do the removal from the handler TAILQ before freeing the handler
back to the heap.  Bad things will happen when some other user of the
TAILQ, such as ithread_add_handler() or the actual ithread attempts to use
the freed handler.  Fix the problem by forcing ithread_remove_handler()
to do the actual removal itself if the "cold" flag is set.

Reviewed by: jhb

20 years agoAMD64 has a single MS-Win calling convention, so provide an empty __stdcall.
David E. O'Brien [Tue, 13 Jan 2004 22:49:45 +0000 (22:49 +0000)]
AMD64 has a single MS-Win calling convention, so provide an empty __stdcall.
Centralize the definition to make it easier to change.

20 years agoClarify revision 1.49. Specifically, using ${MAKE} target instead of
Brian Feldman [Tue, 13 Jan 2004 22:30:20 +0000 (22:30 +0000)]
Clarify revision 1.49.  Specifically, using ${MAKE} target instead of
depending on a target will cause ${MAKE} to have been "reset" to
the initial value by make(1).  This breaks, for example,
make MAKE="make -f subMakefile" clean.  The supported way to pass
these sorts of make flags is actually the .MAKEFLAGS variable.

20 years agoUse 'vm_offset_t' rather than 'u_int32_t'.
David E. O'Brien [Tue, 13 Jan 2004 22:26:37 +0000 (22:26 +0000)]
Use 'vm_offset_t' rather than 'u_int32_t'.

Tested on: AMD64
Reviewed by: wpaul

20 years agoAMD64 has a single MS-Win calling convention, so provide an empty __stdcall.
David E. O'Brien [Tue, 13 Jan 2004 22:23:47 +0000 (22:23 +0000)]
AMD64 has a single MS-Win calling convention, so provide an empty __stdcall.

20 years agoInitialize drive device_t in softc.
Matthew N. Dodd [Tue, 13 Jan 2004 21:44:03 +0000 (21:44 +0000)]
Initialize drive device_t in softc.

20 years agoAdd the Qualcomm CDMA Technologies MSM phone (umodem) device.
Brian Feldman [Tue, 13 Jan 2004 21:42:32 +0000 (21:42 +0000)]
Add the Qualcomm CDMA Technologies MSM phone (umodem) device.

Submitted by: Sean Welch <welchsm@earthlink.net>
MFC after: 1 week

20 years agoFix ata_getparam to accept the fact that some crappy devices can pose as
Søren Schmidt [Tue, 13 Jan 2004 21:35:39 +0000 (21:35 +0000)]
Fix ata_getparam to accept the fact that some crappy devices can pose as
both master and slave at the same time confusing the probe code.

20 years agoAdd a component constant for ACPI_TOSHIBA to fix compilation of
John Baldwin [Tue, 13 Jan 2004 21:27:50 +0000 (21:27 +0000)]
Add a component constant for ACPI_TOSHIBA to fix compilation of
acpi_toshiba(4) driver with ACPI_DEBUG and thus fix LINT on i386.

20 years ago- Use the standard .y and .l transformation rules -- this fixes
Ruslan Ermilov [Tue, 13 Jan 2004 20:53:56 +0000 (20:53 +0000)]
- Use the standard .y and .l transformation rules -- this fixes
  the usual parallel make race in custom .y rules.

- Fixed some style bugs.

- Removed -v from YFLAGS.

Prodded by:     bde
Reviewed by:    bde, njl

20 years agoUse contigmalloc() and contigfree() instead of vm_page_alloc_contig() and
Alan Cox [Tue, 13 Jan 2004 20:36:03 +0000 (20:36 +0000)]
Use contigmalloc() and contigfree() instead of vm_page_alloc_contig() and
kmem_free().  Note: The FreeBSD-specific code in this file has been
subsumed by the FreeBSD-specific header file, pdq_freebsd.h.  That header
file already specifies the use of contigmalloc() and contigfree().  Thus,
the purpose of this change is to avoid having nonsensical examples of
FreeBSD-specific memory allocation in our source tree.

20 years ago* if-like-constructs now use <TAB>body consistently.
David E. O'Brien [Tue, 13 Jan 2004 18:46:09 +0000 (18:46 +0000)]
* if-like-constructs now use <TAB>body consistently.
* Long statements with '\' continuation now use <4 spaces>continuation
  consistently.
* wraps some long lines
* More consistent on how very simular repeated commands are wrapped.

20 years agoForgot ffsl() and flsl() on alpha.
Dag-Erling Smørgrav [Tue, 13 Jan 2004 18:05:49 +0000 (18:05 +0000)]
Forgot ffsl() and flsl() on alpha.

20 years agobsd.dep.mk,v 1.43 allows us to replace a hack with a solution.
Ruslan Ermilov [Tue, 13 Jan 2004 17:38:42 +0000 (17:38 +0000)]
bsd.dep.mk,v 1.43 allows us to replace a hack with a solution.

20 years agoIntroduce DPSRCS, that holds the list of source files which are
Ruslan Ermilov [Tue, 13 Jan 2004 17:37:45 +0000 (17:37 +0000)]
Introduce DPSRCS, that holds the list of source files which are
needed for generating dependencies.  SRCS are always part of it,
and normally only they.

This can be useful in some random cases where it's necessary to
have something in .depend that isn't part of SRCS.  This will be
used to replace a hack in lib/libpam/libpam/Makefile.

20 years agoThis commit was generated by cvs2svn to compensate for changes in r124488,
Bill Fenner [Tue, 13 Jan 2004 17:28:59 +0000 (17:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r124488,
which included commits to RCS files with non-trunk default branches.

20 years agoCommit vendor update for vulnerabilities found by
Bill Fenner [Tue, 13 Jan 2004 17:28:59 +0000 (17:28 +0000)]
Commit vendor update for vulnerabilities found by
Jonathan Heusser <jonny@drugphish.ch>

20 years agoThis commit was generated by cvs2svn to compensate for changes in r124486,
Bill Fenner [Tue, 13 Jan 2004 17:28:06 +0000 (17:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r124486,
which included commits to RCS files with non-trunk default branches.

20 years agoCommit vendor fix for l2tp vulnerability reported by
Bill Fenner [Tue, 13 Jan 2004 17:28:06 +0000 (17:28 +0000)]
Commit vendor fix for l2tp vulnerability reported by
Przemyslaw Frasunek on bugtraq list.

20 years agoNew errata: KUser can corrupt passwd file.
Bruce A. Mah [Tue, 13 Jan 2004 17:15:58 +0000 (17:15 +0000)]
New errata:  KUser can corrupt passwd file.

Requested by: scottl, "Eugenia Loli-Queru" <eugenia at osnews dot com>

20 years agoNew errata: NSS problems and workaround [1], GCC typo in release
Bruce A. Mah [Tue, 13 Jan 2004 16:46:35 +0000 (16:46 +0000)]
New errata:  NSS problems and workaround [1], GCC typo in release
notes [2].

Submitted by: nectar [1], gerald [2]

20 years agoAdd and document ffsl(), fls() and flsl().
Dag-Erling Smørgrav [Tue, 13 Jan 2004 16:05:47 +0000 (16:05 +0000)]
Add and document ffsl(), fls() and flsl().

20 years agoCast to unsigned to avoid sign-extension problems.
Dag-Erling Smørgrav [Tue, 13 Jan 2004 16:02:20 +0000 (16:02 +0000)]
Cast to unsigned to avoid sign-extension problems.

20 years agoAdd ffsl(), fls() flsl() to platforms that don't already have them.
Dag-Erling Smørgrav [Tue, 13 Jan 2004 15:37:23 +0000 (15:37 +0000)]
Add ffsl(), fls() flsl() to platforms that don't already have them.

20 years agoAdd C implementations of ffsl(), fls() and flsl().
Dag-Erling Smørgrav [Tue, 13 Jan 2004 15:36:34 +0000 (15:36 +0000)]
Add C implementations of ffsl(), fls() and flsl().

20 years agoANSIfy, de-register, replace VAX reference with actual description.
Dag-Erling Smørgrav [Tue, 13 Jan 2004 15:35:22 +0000 (15:35 +0000)]
ANSIfy, de-register, replace VAX reference with actual description.

20 years agoWhitespace nit.
Dag-Erling Smørgrav [Tue, 13 Jan 2004 15:30:36 +0000 (15:30 +0000)]
Whitespace nit.

20 years agoInclude "../Makefile.inc" to propagate the configuration to subdirectories.
Yoshihiro Takahashi [Tue, 13 Jan 2004 13:08:27 +0000 (13:08 +0000)]
Include "../Makefile.inc" to propagate the configuration to subdirectories.

20 years agoFix typo in a comment
Christian Brueffer [Tue, 13 Jan 2004 11:31:09 +0000 (11:31 +0000)]
Fix typo in a comment

20 years agoMODULE_DEPEND is a C macro, not a make(1).
Ruslan Ermilov [Tue, 13 Jan 2004 11:30:37 +0000 (11:30 +0000)]
MODULE_DEPEND is a C macro, not a make(1).

20 years agobsd.kmod.mk does not deal with manpages anymore.
Ruslan Ermilov [Tue, 13 Jan 2004 11:28:51 +0000 (11:28 +0000)]
bsd.kmod.mk does not deal with manpages anymore.

20 years agostyle(9): single tab after #define.
Matthew N. Dodd [Tue, 13 Jan 2004 11:28:21 +0000 (11:28 +0000)]
style(9): single tab after #define.

20 years agoUse a device identify entry point to attach to nexus, since the
Peter Grehan [Tue, 13 Jan 2004 11:25:40 +0000 (11:25 +0000)]
Use a device identify entry point to attach to nexus, since the
nexus code no longer searches for interrupt controllers.

20 years agoMake the OpenPic driver bus-independent, with attachments for
Peter Grehan [Tue, 13 Jan 2004 11:24:36 +0000 (11:24 +0000)]
Make the OpenPic driver bus-independent, with attachments for
the MacIO chip and PSIM's IOBus. Bus-specific drivers should
use the identify method to attach themselves to nexus so
interrupt can be allocated before the h/w is probed. The
'early attach' routine in openpic is used for this stage
of boot. When h/w is probed, the openpic can be attached
properly. It will enable interrupts allocated prior to
this.

20 years agoRemove hard-coded knowledge of specific OFW devices. Use bus_generic_probe
Peter Grehan [Tue, 13 Jan 2004 11:20:52 +0000 (11:20 +0000)]
Remove hard-coded knowledge of specific OFW devices. Use bus_generic_probe
and add_child entry point to allow devices to use the identify
method to add themselves if need be (e.g. openpic, syscons).
Export interrupt-controller-add routine for extern int cntlr drivers.
Eliminate recursive OFW device-tree walk and only iterate the
top-level ala sparc64. Allow child devices to set the device
type with write_ivars.

Step 1 of many in removing the hard-dependency on OpenFirmware.

20 years agoDon't put .depend in CLEANFILES, it's cleaned by cleandepend.
Ruslan Ermilov [Tue, 13 Jan 2004 11:13:00 +0000 (11:13 +0000)]
Don't put .depend in CLEANFILES, it's cleaned by cleandepend.

20 years agoCatch up with ATA changes by including <sys/sema.h>
Peter Grehan [Tue, 13 Jan 2004 10:54:35 +0000 (10:54 +0000)]
Catch up with ATA changes by including <sys/sema.h>

20 years agocall ipsec_pcbconn()/ipsec_pcbdisconn() from in6_pcbconnect().
Hajimu UMEMOTO [Tue, 13 Jan 2004 10:45:02 +0000 (10:45 +0000)]
call ipsec_pcbconn()/ipsec_pcbdisconn() from in6_pcbconnect().

Obtained from: KAME

20 years agodo not deref freed pointer
Hajimu UMEMOTO [Tue, 13 Jan 2004 09:51:47 +0000 (09:51 +0000)]
do not deref freed pointer

Submitted by: "Bjoern A. Zeeb" <bzeeb+freebsd@zabbadoz.net>
Reviewed by: itojun

20 years agoImplement some more unicode handling routines. This will hopefully bring
Bill Paul [Tue, 13 Jan 2004 09:12:47 +0000 (09:12 +0000)]
Implement some more unicode handling routines. This will hopefully bring
us closer to being able to run the Intel PRO/Wireless 5000 driver.

20 years agoAdded FreeBSD 5.2-RELEASE.
Maxim Konovalov [Tue, 13 Jan 2004 08:35:00 +0000 (08:35 +0000)]
Added FreeBSD 5.2-RELEASE.

MFC after: 3 days

20 years agotypo, duh!
Poul-Henning Kamp [Tue, 13 Jan 2004 08:00:55 +0000 (08:00 +0000)]
typo, duh!

20 years agoReplace calls to vm_page_alloc_contig() by calls to contigmalloc().
Alan Cox [Tue, 13 Jan 2004 07:45:10 +0000 (07:45 +0000)]
Replace calls to vm_page_alloc_contig() by calls to contigmalloc().
vm_page_alloc_contig() will be removed after the three remaining drivers
that use it are also converted to contigmalloc().

20 years agoThis is simply too true to not become a fortune.
Poul-Henning Kamp [Tue, 13 Jan 2004 07:36:17 +0000 (07:36 +0000)]
This is simply too true to not become a fortune.

Wisdom by: wes

20 years agoHandle SIOCSIFMTU ioctl directly so we can apply 802.11-specific bounds.
Sam Leffler [Tue, 13 Jan 2004 06:22:55 +0000 (06:22 +0000)]
Handle SIOCSIFMTU ioctl directly so we can apply 802.11-specific bounds.
Note that the min is actually constrained to IF_MINMTU by the if layer.

20 years agocorrect spelling
Hajimu UMEMOTO [Tue, 13 Jan 2004 05:39:07 +0000 (05:39 +0000)]
correct spelling

Submitted by: "Bjoern A. Zeeb" <bzeeb+freebsd@zabbadoz.net>
Reviewed by: itojun

20 years agofix potential 'cannot-happen' memory leak
Hajimu UMEMOTO [Tue, 13 Jan 2004 05:32:12 +0000 (05:32 +0000)]
fix potential 'cannot-happen' memory leak

Submitted by: "Bjoern A. Zeeb" <bzeeb+freebsd@zabbadoz.net>
Reviewed by: itojun

20 years agoLoosen up the range test in ndis_register_ioport(). Allow drivers to
Bill Paul [Tue, 13 Jan 2004 01:12:28 +0000 (01:12 +0000)]
Loosen up the range test in ndis_register_ioport(). Allow drivers to
map ranges that are smaller than what our resource manager code knows
is available, rather than requiring that they match exactly. This
fixes a problem with the Intel PRO/1000 gigE driver: it wants to map
a range of 32 I/O ports, even though some chips appear set up to
decode a range of 64. With this fix, it loads and runs correctly.

20 years agoPrevent a panic when mounting a v2/v3 only server with mount_nfs4.
Alfred Perlstein [Tue, 13 Jan 2004 01:04:36 +0000 (01:04 +0000)]
Prevent a panic when mounting a v2/v3 only server with mount_nfs4.

Submitted by: Jim Rees <rees@umich.edu>
Reported/testing: Florian C. Smeets <flo@kasimir.com>

20 years agoYet another inf file handling tweak: when checking for Ndi\Params\foo
Bill Paul [Tue, 13 Jan 2004 00:29:17 +0000 (00:29 +0000)]
Yet another inf file handling tweak: when checking for Ndi\Params\foo
keys, don't just do a substring match on "Ndi\Params". Instead, check
explicitly for strings that begin with "Ndi\Params". Why? Because it's
possible to create your own keys with different paths, like
"PROSetNdi\NdiExt\Params" which is what Intel does in their PRO/1000
driver's .INF file.

20 years agoLast revision also added duplicate SRCS. Fixed this so that
Ruslan Ermilov [Mon, 12 Jan 2004 22:02:31 +0000 (22:02 +0000)]
Last revision also added duplicate SRCS.  Fixed this so that
the resulting .a library has only unique .o files.

Noticed by: bde