]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoPut back the undocumented change from rev. 1.334 too: no
Ruslan Ermilov [Mon, 7 Apr 2003 00:01:33 +0000 (00:01 +0000)]
Put back the undocumented change from rev. 1.334 too: no
need to create ${WORLDTMP}/legacy/usr/include explicitly.

21 years agoPut back parts of 1.335 and 1.336 that 1.337 accidentally backed out.
Warner Losh [Sun, 6 Apr 2003 23:46:02 +0000 (23:46 +0000)]
Put back parts of 1.335 and 1.336 that 1.337 accidentally backed out.

Submitted by: ru

21 years agoMove a bus_dmamap_sync() to the correct place.
Maxime Henrion [Sun, 6 Apr 2003 23:16:00 +0000 (23:16 +0000)]
Move a bus_dmamap_sync() to the correct place.

21 years agoBecause alpha can't access memory in 16-bit granularity,
Maxime Henrion [Sun, 6 Apr 2003 23:09:57 +0000 (23:09 +0000)]
Because alpha can't access memory in 16-bit granularity,
we're using an atomic operation to clear the suspend flag
in fxp_start().  Since other architectures may need the
same thing, we want to do it all the time and not only
in the __alpha__ case.  However, we don't want to use
atomic operations on 16-bit integers, because those may
not be available on any architecture.  We're thus faking
a 32-bit atomic operation here.  This patch also deals
with endianness here.

21 years agoSufficient access checks are performed by vmapbuf() that calling useracc()
Alan Cox [Sun, 6 Apr 2003 22:21:03 +0000 (22:21 +0000)]
Sufficient access checks are performed by vmapbuf() that calling useracc()
is pointless.  Remove the calls to useracc().

21 years ago-legacy and /.../legacy/... looks better than build or bootstrap in
Warner Losh [Sun, 6 Apr 2003 21:46:44 +0000 (21:46 +0000)]
-legacy and /.../legacy/... looks better than build or bootstrap in
the logs, so use that instead.

Submitted by: obrien.

21 years agoAlways remove ${WORLDTMP}/build/usr/include, even in the NOCLEAN
Ruslan Ermilov [Sun, 6 Apr 2003 21:39:20 +0000 (21:39 +0000)]
Always remove ${WORLDTMP}/build/usr/include, even in the NOCLEAN
case.  This way, we won't have stale compatibility headers there.

21 years agoRevert the s/u_int/u_int8_t/ changes, we can't really use other
Maxime Henrion [Sun, 6 Apr 2003 21:35:45 +0000 (21:35 +0000)]
Revert the s/u_int/u_int8_t/ changes, we can't really use other
integer types than int with bit-fields in a portable way.

Prodded by: bde

21 years agoMoved libbuild target to where it belongs. Added a comment.
Ruslan Ermilov [Sun, 6 Apr 2003 21:33:49 +0000 (21:33 +0000)]
Moved libbuild target to where it belongs.  Added a comment.

21 years agoRemove the 32KB VHPT section from the kernel image. We don't really
Marcel Moolenaar [Sun, 6 Apr 2003 21:31:26 +0000 (21:31 +0000)]
Remove the 32KB VHPT section from the kernel image. We don't really
use it because we allocate a VHPT based on the size of the physical
memory and even if the allocated VHPT is 32KB, we don't use the in-
image section for it. Since the VHPT must be naturally aligned, we
save 48K on average (due to alignment).
Consequently, we start off with the VHPT disabled (it is assumed
the VHPT is disabled because the EFI loader runs without memory
address translation and thus has no need to setup the VHPT). It's
probably a good idea to explicitly disable the VHPT if we make the
use of the VHPT optional.

21 years agoFixed buildworld stages names in comments.
Ruslan Ermilov [Sun, 6 Apr 2003 21:23:02 +0000 (21:23 +0000)]
Fixed buildworld stages names in comments.

21 years agoImprove, and slightly soften, the deprecated flags. Call them deprecated
Warner Losh [Sun, 6 Apr 2003 19:27:49 +0000 (19:27 +0000)]
Improve, and slightly soften, the deprecated flags.  Call them deprecated
and not obsolete, but add a warning about their disappearance.

Add additional notes that explain the lameness of WEP.

21 years agoSufficient access checks are performed by vmapbuf() that calling useracc()
Alan Cox [Sun, 6 Apr 2003 19:26:30 +0000 (19:26 +0000)]
Sufficient access checks are performed by vmapbuf() that calling useracc()
is pointless.  Remove the call to useracc().

Don't reinitialize fields that are already initialized by getpbuf().

Reviewed by: tegge

21 years agoo Tone down the obsolete messages since we actually do something, for
Warner Losh [Sun, 6 Apr 2003 19:11:33 +0000 (19:11 +0000)]
o Tone down the obsolete messages since we actually do something, for
  the moment, with the deprecated flags.
o Better error reporting on getting values from the driver.  When we can't
  get one for the default dumpinfo output.  The old driver would succeed
  for invalid RIDs, while the new driver reports errors.  Since the info
  we're getting from the card/driver doesn't exist for all cards, we just
  don't report them.  Improve error reporting elsewhere now that wi_getval
  doesn't exit.  Also fix a file descriptor leak as a side effect.

Reported by: scottl

21 years agoInitialize the PIIX timecounter in piix_attach(), which is called only
Dag-Erling Smørgrav [Sun, 6 Apr 2003 18:42:22 +0000 (18:42 +0000)]
Initialize the PIIX timecounter in piix_attach(), which is called only
once, instead of doing it in piix_probe(), which is called every time
the PCI bus is rescanned.

21 years agoRemove a largely useless statistic (its kept elsewhere too).
Jake Burkholder [Sun, 6 Apr 2003 18:18:17 +0000 (18:18 +0000)]
Remove a largely useless statistic (its kept elsewhere too).

21 years agoMake the pmap stats writeable. It can be useful to clear them.
Jake Burkholder [Sun, 6 Apr 2003 18:17:31 +0000 (18:17 +0000)]
Make the pmap stats writeable.  It can be useful to clear them.

21 years agoAdd support for the CLIE 5.0 series of PDAs
Scott Long [Sun, 6 Apr 2003 17:34:50 +0000 (17:34 +0000)]
Add support for the CLIE 5.0 series of PDAs

21 years agoRegen from usbdevs v.116
Scott Long [Sun, 6 Apr 2003 17:34:08 +0000 (17:34 +0000)]
Regen from usbdevs v.116

21 years agoRegen
Scott Long [Sun, 6 Apr 2003 17:32:39 +0000 (17:32 +0000)]
Regen

21 years agoAdd ID's for the CLIE 5.0 series and the Belkin F5U109 usb->serial adapter.
Scott Long [Sun, 6 Apr 2003 17:32:12 +0000 (17:32 +0000)]
Add ID's for the CLIE 5.0 series and the Belkin F5U109 usb->serial adapter.

21 years agoUse the vis block copy/zero functions for pmap_copy_page and pmap_zero_page.
Jake Burkholder [Sun, 6 Apr 2003 17:05:26 +0000 (17:05 +0000)]
Use the vis block copy/zero functions for pmap_copy_page and pmap_zero_page.
These are called through function pointers so that different implementations
can be provided for cheetah, where the block load instructions may or may
not be a win, and so they can be disabled with the machdep.use_vis tunable.
In terms of raw bandwidth the integer versions are faster, but not allocating
lines in the L2 cache for useless data gives a measurable improvement in user
time for the benchmarks I tested (mostly buildworld with -j8).

As far as I can tell the instructions used are implemented on everything
back to UltraSPARC I, so there should not be a problem with different cpu
types.

21 years agoIgnore attempts to pmap_kremove or pmap_qremove pages which do not have
Jake Burkholder [Sun, 6 Apr 2003 15:14:24 +0000 (15:14 +0000)]
Ignore attempts to pmap_kremove or pmap_qremove pages which do not have
a valid mapping.  This is bug for bug compatible with other platforms.

21 years agoThese entries (for multilingual sysinstall) are not for Japanese term only.
Tatsumi Hosokawa [Sun, 6 Apr 2003 11:47:24 +0000 (11:47 +0000)]
These entries (for multilingual sysinstall) are not for Japanese term only.
Changed to more generic names.

21 years agomake prototpyes match functions, and declare things static as needed
Warner Losh [Sun, 6 Apr 2003 08:30:25 +0000 (08:30 +0000)]
make prototpyes match functions, and declare things static as needed

21 years agoTest for bug in revision 1.54 of vfprintf.c.
Tim J. Robbins [Sun, 6 Apr 2003 08:02:20 +0000 (08:02 +0000)]
Test for bug in revision 1.54 of vfprintf.c.

21 years agoZap some a.out leftovers
Peter Wemm [Sun, 6 Apr 2003 06:28:08 +0000 (06:28 +0000)]
Zap some a.out leftovers

21 years agoAlso set the access bit in the PTE when we get a data dirty bit fault.
Marcel Moolenaar [Sun, 6 Apr 2003 05:55:36 +0000 (05:55 +0000)]
Also set the access bit in the PTE when we get a data dirty bit fault.
This avoids an immediate access bit fault when we serviced the dirty
bit fault in case the access bit is unset. This typically happens for
newly allocated memory that's being zeroed and thus very common.

21 years agoSync up with kern/subr_prf.c. This adds %ll, %j, %q, %z, etc.
Peter Wemm [Sun, 6 Apr 2003 05:25:48 +0000 (05:25 +0000)]
Sync up with kern/subr_prf.c.  This adds %ll, %j, %q, %z, etc.

21 years agoSearch for "elf32 kernel" (and elf64) and "elf32 module" (and elf64)
Peter Wemm [Sun, 6 Apr 2003 05:20:00 +0000 (05:20 +0000)]
Search for "elf32 kernel" (and elf64) and "elf32 module" (and elf64)
as well as "elf kernel" and "elf module".  This is a precursor to
x86-64 support in the i386 loader so it can load an elf64 x86-64 kernel.

21 years agoWe can't use ${WORLDTMP}/usr/{include,lib} for the compat layer. This
Warner Losh [Sun, 6 Apr 2003 03:50:28 +0000 (03:50 +0000)]
We can't use ${WORLDTMP}/usr/{include,lib} for the compat layer.  This
is because we populate these directories later, and a subsequent
-DNOCLEAN build may fail.  So, we put them in
${WORLDTMP}/build/usr/{include,lib} instead and adjust Makefile.boot.

Again, this works on -stable and -current, but might break older
versions.

Submitted by: ru@

21 years agoDon't need the gccism include_next since we won't create this file
Warner Losh [Sun, 6 Apr 2003 03:30:44 +0000 (03:30 +0000)]
Don't need the gccism include_next since we won't create this file
when a real sys/endian.h exists.

Submitted by: ru@

21 years agoUse __FBSDID.
Maxime Henrion [Sun, 6 Apr 2003 01:27:12 +0000 (01:27 +0000)]
Use __FBSDID.

21 years agoCorrect the definition of the link_addr and rbd_addr fields
Maxime Henrion [Sun, 6 Apr 2003 01:11:38 +0000 (01:11 +0000)]
Correct the definition of the link_addr and rbd_addr fields
in struct fxp_rfa.  This should have been committed with my
last endianness fixes.

21 years ago- Instead of rolling our own alignment-safe function, use le32enc()
Maxime Henrion [Sun, 6 Apr 2003 01:04:17 +0000 (01:04 +0000)]
- Instead of rolling our own alignment-safe function, use le32enc()
  which deals with both endianness and alignment issues.
- Collect low-hanging fruits for endianness safety.
- Use 0xffffffff instead of -1 where appropriate.

21 years agoRemove an unnecessary trunc_page() from vmapbuf().
Alan Cox [Sun, 6 Apr 2003 00:40:54 +0000 (00:40 +0000)]
Remove an unnecessary trunc_page() from vmapbuf().

Reviewed by: tegge

21 years ago- Use __FXP_BITFIELDX macros to make the configuration bitfield
Maxime Henrion [Sat, 5 Apr 2003 23:46:58 +0000 (23:46 +0000)]
- Use __FXP_BITFIELDX macros to make the configuration bitfield
  endian safe.
- Change some u_int to u_int8_t which make more sense here since
  we're really defining bytes.  That produces the same code due to
  how bitfields work.
- Add the definition of the vlan_drop_en bit (not used yet).
- Add some useful comments.

Obtained from: NetBSD

21 years agoUse bus_dmamap_load_mbuf() instead of bus_dmamap_load() for the
Maxime Henrion [Sat, 5 Apr 2003 23:24:23 +0000 (23:24 +0000)]
Use bus_dmamap_load_mbuf() instead of bus_dmamap_load() for the
RX part of this driver too.  It's better since the code wasn't
dealing with bus_dmamap_load() returning EINPROGRESS, and this
can't happen with bus_dmamap_load_mbuf().

Submitted by: jake

21 years agoDon't reinitialize fields that are already initialized by getpbuf().
Alan Cox [Sat, 5 Apr 2003 23:02:58 +0000 (23:02 +0000)]
Don't reinitialize fields that are already initialized by getpbuf().

21 years agoUnbreak the !LAZY_SWITCH case. I #ifdef'ed too much when I added
Peter Wemm [Sat, 5 Apr 2003 22:18:14 +0000 (22:18 +0000)]
Unbreak the !LAZY_SWITCH case.  I #ifdef'ed too much when I added
the ifdefs prior to commit and killed the same-address-space test.

Submitted by: bde

21 years agoRegression tests for printf()'s floating point formats.
David Schultz [Sat, 5 Apr 2003 22:13:01 +0000 (22:13 +0000)]
Regression tests for printf()'s floating point formats.
Tests for correct printing of subnormals are lacking.

21 years agoRework the floating point code in printf(). Significant changes:
David Schultz [Sat, 5 Apr 2003 22:11:42 +0000 (22:11 +0000)]
Rework the floating point code in printf().  Significant changes:

- We used to round long double arguments to double.  Now we print
  them properly.

- Bugs involving '%F', corner cases of '#' and 'g' format
  specifiers, and the '.*' precision specifier have been
  fixed.

- Added support for the "'" specifier to print thousands' grouping
  characters in a locale-dependent manner.

- Implement the __vfprintf() side of hexadecimal floating point
  support.  All that is still needed is a routine to convert the
  mantissa to hex digits one nibble at a time in the style of ultoa().

Reviewed by: silence on standards@

21 years agoAdd __ldtoa(), a wrapper around gdtoa() to make it look like dtoa().
David Schultz [Sat, 5 Apr 2003 22:10:13 +0000 (22:10 +0000)]
Add __ldtoa(), a wrapper around gdtoa() to make it look like dtoa().
In support of this, add some MD macros to assist in converting long
doubles to the format expected by gdtoa().

Reviewed by: silence on standards@

21 years agoAdd missing prototype for gdtoa().
David Schultz [Sat, 5 Apr 2003 22:09:26 +0000 (22:09 +0000)]
Add missing prototype for gdtoa().

21 years agoAdd missing #include to unbreak previous commit.
David Schultz [Sat, 5 Apr 2003 22:08:53 +0000 (22:08 +0000)]
Add missing #include to unbreak previous commit.

21 years agoCorrect some buffer sizes.
David Schultz [Sat, 5 Apr 2003 22:03:43 +0000 (22:03 +0000)]
Correct some buffer sizes.

- __vfprintf()'s 'buf' has never been used for floating point, so
  don't define it in terms of (incorrect) constants describing
  floating point numbers.  The actual size needed depends on
  sizeof(uintmax_t) and locale details, so I slightly overestimated.

- We don't need a 308-character buffer to store the string "308".
  With long doubles and %a we need more than three characters, though.

21 years agoSufficient access checks are performed by vmapbuf() that calling
Alan Cox [Sat, 5 Apr 2003 21:19:58 +0000 (21:19 +0000)]
Sufficient access checks are performed by vmapbuf() that calling
useracc() is pointless.  Remove the call to useracc() from physio().

Reviewed by: tegge

21 years agoInclude <geom/geom_disk.h> and stop including <sys/disk.h>. The
Marcel Moolenaar [Sat, 5 Apr 2003 21:14:05 +0000 (21:14 +0000)]
Include <geom/geom_disk.h> and stop including <sys/disk.h>. The
former gives us 'struct disk'.

21 years agoupdate man page to reflect change in default syslog level ipmon is now compiled with
Darren Reed [Sat, 5 Apr 2003 21:12:58 +0000 (21:12 +0000)]
update man page to reflect change in default syslog level ipmon is now compiled with

21 years agoRemove GIANT_REQUIRED from getpbuf(). Reviewed by: tegge
Alan Cox [Sat, 5 Apr 2003 21:01:16 +0000 (21:01 +0000)]
Remove GIANT_REQUIRED from getpbuf().  Reviewed by: tegge

Reduce pbuf_mtx's scope in relpbuf().  Submitted by: tegge

21 years agoUse assignment, not equivalence test.
Nate Lawson [Sat, 5 Apr 2003 20:54:54 +0000 (20:54 +0000)]
Use assignment, not equivalence test.

No response from: suz

21 years agoMigrate to a new way of dealing with building from old revisions of
Warner Losh [Sat, 5 Apr 2003 20:30:30 +0000 (20:30 +0000)]
Migrate to a new way of dealing with building from old revisions of
FreeBSD.  This method attempts to centralize all the necessary hacks
or work arounds in one of two places in the tree (src/Makefile.inc1
and src/tools/build).  We build a small compatibility library
(libbuild.a) as well as selectively installing necessary include
files.  We then include this directory when building host binaries.

This removes all the past release compatibilty hacks from various
places in the tree.  We still build on tip of stable and current.  I
will work with those that want to support more, although I anticipate
it will just work.

Many thanks to ru@, obrien@ and jhb@ for providing valuable input at
various stage of implementation, as well as for working together to
positively effect a change for the better.

21 years agoAdd a basic manual page for wlan(4).
Tom Rhodes [Sat, 5 Apr 2003 19:25:44 +0000 (19:25 +0000)]
Add a basic manual page for wlan(4).

Reviewed by: imp, ru

21 years agoAdd a manual page for the smapi module.
Tom Rhodes [Sat, 5 Apr 2003 19:12:51 +0000 (19:12 +0000)]
Add a manual page for the smapi module.

Reviewed by: ru, mdodd

21 years agoFix ed compilation with PAE by using %jx instead of %x.
Olivier Houchard [Sat, 5 Apr 2003 18:12:36 +0000 (18:12 +0000)]
Fix ed compilation with PAE by using %jx instead of %x.

21 years agoUse vm_paddr_t instead of vm_offset_t for the paddr parameter of vesa_mmap.
Olivier Houchard [Sat, 5 Apr 2003 18:08:22 +0000 (18:08 +0000)]
Use vm_paddr_t instead of vm_offset_t for the paddr parameter of vesa_mmap.

21 years agoDon't forget to send the Content-length header after calculating it.
David Malone [Sat, 5 Apr 2003 17:15:38 +0000 (17:15 +0000)]
Don't forget to send the Content-length header after calculating it.

PR: 29725
Submitted by: Eugene Grosbein <eugen@svzserv.kemerovo.su>

21 years agoMore warns cleanups and misc cleanups:
David Malone [Sat, 5 Apr 2003 15:27:27 +0000 (15:27 +0000)]
More warns cleanups and misc cleanups:
1) Make fetch_mode a pointer 'cos we can just use the optarg.
2) Constness.
3) Add/complete prototypes.
4) Change an int to a socklen_t.
5) Don't use C++ style comments.
6) Check return values from read and fork a little more carefully.
7) Avoid closing an uninitialised int.
8) Get rid of unneeded extern declarations.

21 years agoFix/add missing accents on the weekday names.
Tom Rhodes [Sat, 5 Apr 2003 14:43:05 +0000 (14:43 +0000)]
Fix/add missing accents on the weekday names.

PR: 37219
Submitted by: Nicola Vitale <nivit@libero.it>
Reivewed by: Alex Dupre <sysadmin@alexdupre.com>

21 years agoThe Netgear MA311 is a Prism 2.5 card.
Warner Losh [Sat, 5 Apr 2003 13:07:44 +0000 (13:07 +0000)]
The Netgear MA311 is a Prism 2.5 card.

Submitted by: duncan@quantumlogic.net

21 years ago- Change a FXP_NRFABUFS which would have been a FXP_NTXCB.
Maxime Henrion [Sat, 5 Apr 2003 12:43:21 +0000 (12:43 +0000)]
- Change a FXP_NRFABUFS which would have been a FXP_NTXCB.
- Correct some bus_dmamap_sync() calls.

Submitted by: jake

21 years agoChange the default syslog facility from LOG_LOCAL0 to LOG_SECURITY after
Darren Reed [Sat, 5 Apr 2003 10:42:38 +0000 (10:42 +0000)]
Change the default syslog facility from LOG_LOCAL0 to LOG_SECURITY after
discussion on security@freebsd.org.

21 years agoWhen session is over, IPv6 default route to tun should be
Hajimu UMEMOTO [Sat, 5 Apr 2003 10:10:33 +0000 (10:10 +0000)]
When session is over, IPv6 default route to tun should be
removed, too.

MFC after: 1 week

21 years agoChange the default syslog facility from LOG_LOCAL0 to LOG_SECURITY after
Darren Reed [Sat, 5 Apr 2003 09:25:19 +0000 (09:25 +0000)]
Change the default syslog facility from LOG_LOCAL0 to LOG_SECURITY after
discussion on security@freebsd.org.

21 years agoDWL-520 is used to describe three or four different cards. only some
Warner Losh [Sat, 5 Apr 2003 07:41:52 +0000 (07:41 +0000)]
DWL-520 is used to describe three or four different cards.  only some
of them are prism based and are thus supported by wi.  Document this
fact.

21 years agoMFp4: Link strtof.3 and strtold.3 to strtod.3.
Tim J. Robbins [Sat, 5 Apr 2003 07:33:46 +0000 (07:33 +0000)]
MFp4: Link strtof.3 and strtold.3 to strtod.3.

21 years agoZero out the struct tm supplied by the caller. Otherwise,
Mike Makonnen [Sat, 5 Apr 2003 05:46:43 +0000 (05:46 +0000)]
Zero out the struct tm supplied by the caller. Otherwise,
strange things might happen when garbage values in the struct
get passed in to localtime_r() and family.

Noticed by: marcus
Approved by: markm (mentor)(implicit)

21 years agoMFp4: Pentium/Athlon-optimised implementation of wcschr().
Tim J. Robbins [Sat, 5 Apr 2003 04:17:05 +0000 (04:17 +0000)]
MFp4: Pentium/Athlon-optimised implementation of wcschr().

21 years agoSetup our interrupt only at the end of fxp_attach(), when all
Maxime Henrion [Sat, 5 Apr 2003 01:27:17 +0000 (01:27 +0000)]
Setup our interrupt only at the end of fxp_attach(), when all
other allocations/initializations have been successful.  I kinda
doubt it will fix the recent breakage that some people are seeing,
but this could have caused problems for sure.

21 years agoAdd SMP_TSC option, which can be used on SMP systems where the TSCs
Tor Egge [Fri, 4 Apr 2003 23:54:46 +0000 (23:54 +0000)]
Add SMP_TSC option, which can be used on SMP systems where the TSCs
are synchronized to reduce context switch cost.

21 years agoDiff reduction with my p4 changes:
Warner Losh [Fri, 4 Apr 2003 22:46:08 +0000 (22:46 +0000)]
Diff reduction with my p4 changes:

Add @ before ${ECHODIR} where appropriate because we don't need to echo
the echo command...  This gets rid of extra echo ===> in log files...

21 years agoAdd a rudimentary but working driver for the Adlink "NuDaq PCI-9812".
Poul-Henning Kamp [Fri, 4 Apr 2003 18:53:04 +0000 (18:53 +0000)]
Add a rudimentary but working driver for the Adlink "NuDaq PCI-9812".

This is a 4 channel 20 msps 12 bit ADC card.

Anyone wanting to play with GNUradio or similar can start here.

21 years agostyle.Makefile(5)
David E. O'Brien [Fri, 4 Apr 2003 17:49:21 +0000 (17:49 +0000)]
style.Makefile(5)

21 years agoDefine ovbcopy() as a macro which expands to the equivalent bcopy() call,
Dag-Erling Smørgrav [Fri, 4 Apr 2003 17:29:55 +0000 (17:29 +0000)]
Define ovbcopy() as a macro which expands to the equivalent bcopy() call,
to take care of the KAME IPv6 code which needs ovbcopy() because NetBSD's
bcopy() doesn't handle overlap like ours.

Remove all implementations of ovbcopy().

Previously, bzero was a function pointer on i386, to save a jmp to
bzero_vector.  Get rid of this microoptimization as it only confuses
things, adds machine-dependent code to an MD header, and doesn't really
save all that much.

This commit does not add my pagezero() / pagecopy() code.

21 years agoUse sized cast matching the sized pointer.
Poul-Henning Kamp [Fri, 4 Apr 2003 16:59:39 +0000 (16:59 +0000)]
Use sized cast matching the sized pointer.

21 years agoRetire <sys/diskslice.h>
Poul-Henning Kamp [Fri, 4 Apr 2003 16:57:51 +0000 (16:57 +0000)]
Retire <sys/diskslice.h>

21 years agoLibdisk does not need to include <sys/diskslice.h> any more.
Poul-Henning Kamp [Fri, 4 Apr 2003 16:35:16 +0000 (16:35 +0000)]
Libdisk does not need to include <sys/diskslice.h> any more.

Move the remaining bits of <sys/diskslice.h> to <i386/include/bootinfo.h>

Move i386/pc98 specific bits from <sys/reboot.h> to
<i386/include/bootinfo.h> as well.

Adjust includes in sys/boot accordingly.

21 years agoAdd NE2000DVF_AX88190 flag to Corega FEther PCC-TXD
MIHIRA Sanpei Yoshiro [Fri, 4 Apr 2003 14:46:50 +0000 (14:46 +0000)]
Add NE2000DVF_AX88190 flag to Corega FEther PCC-TXD
(see also /etc/defaults/pccard.conf rev.1.172)

21 years ago - Print out an error message instead of dereferencing a NULL pointer
Robert Drehmel [Fri, 4 Apr 2003 14:40:49 +0000 (14:40 +0000)]
 - Print out an error message instead of dereferencing a NULL pointer
   if matchinstalled() found no packages, which happens to be the
   case after fresh installations.
 - Instead of using strstr(3) to match the package name, depend on
   matchinstalled()'s MATCH_REGEX package matching.

PR: bin/50384
MFC after: 2 weeks

21 years agofix typo
MIHIRA Sanpei Yoshiro [Fri, 4 Apr 2003 14:40:01 +0000 (14:40 +0000)]
fix typo

21 years agoFix braino in definition of isfinite().
David Schultz [Fri, 4 Apr 2003 13:27:47 +0000 (13:27 +0000)]
Fix braino in definition of isfinite().

Noticed by: marcus
Pointy hat to: das

21 years agoovbcopy -> bcopy
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:56:20 +0000 (12:56 +0000)]
ovbcopy -> bcopy

21 years agoReplace ovbcopy() with bcopy().
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:15:20 +0000 (12:15 +0000)]
Replace ovbcopy() with bcopy().

21 years agoReplace memcpy() and ovbcopy() with bcopy(); ditch some caddr_t usage.
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:14:00 +0000 (12:14 +0000)]
Replace memcpy() and ovbcopy() with bcopy(); ditch some caddr_t usage.

21 years agoDon't use ovbcopy().
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:12:34 +0000 (12:12 +0000)]
Don't use ovbcopy().

21 years agoDon't use ovbcopy(); use void * instead of char *.
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:11:46 +0000 (12:11 +0000)]
Don't use ovbcopy(); use void * instead of char *.

21 years agoThe kernel bcopy() is safe for overlapping regions (and always has), so
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:10:04 +0000 (12:10 +0000)]
The kernel bcopy() is safe for overlapping regions (and always has), so
there is no use for a separate ovbcopy().

21 years agoRename a static variable to avoid future conflicts.
Dag-Erling Smørgrav [Fri, 4 Apr 2003 12:08:42 +0000 (12:08 +0000)]
Rename a static variable to avoid future conflicts.

21 years agoSet link-local address of tun interface with prefixlen = 64
Hajimu UMEMOTO [Fri, 4 Apr 2003 11:09:08 +0000 (11:09 +0000)]
Set link-local address of tun interface with prefixlen = 64
instead of 128.  It makes RA happy.

Reported by: rafa@dif.um.es,
SHIRASAKI Yasuhiro <yasuhiro@nttv6.jp>
Reviewed by: SHIRASAKI Yasuhiro <yasuhiro@nttv6.jp>
MFC after: 1 week

21 years ago-Wall implies -Wuninitialized if -O is also in effect.
Ruslan Ermilov [Fri, 4 Apr 2003 10:47:06 +0000 (10:47 +0000)]
-Wall implies -Wuninitialized if -O is also in effect.
-Wuninitialized does not work without -O.

This fixes the ${WARNS} > 4 compilations with -O0.

Spotted by: marcel

21 years agoAdd a manpage for the 'vpd' driver.
Matthew N. Dodd [Fri, 4 Apr 2003 10:12:09 +0000 (10:12 +0000)]
Add a manpage for the 'vpd' driver.

21 years agoBandaid fix for previous commit while I figure out why it broke. This
Jake Burkholder [Fri, 4 Apr 2003 10:09:44 +0000 (10:09 +0000)]
Bandaid fix for previous commit while I figure out why it broke.  This
caused crashes early in boot on i386 UP machines.

Reported by: phk
Pointy hat to: jake

21 years agoAdditional codec ids.
Orion Hodson [Fri, 4 Apr 2003 07:37:25 +0000 (07:37 +0000)]
Additional codec ids.

21 years agoFix mismatch between bus address stored for buffer descriptors and
Orion Hodson [Fri, 4 Apr 2003 07:15:19 +0000 (07:15 +0000)]
Fix mismatch between bus address stored for buffer descriptors and
actual address of buffer descriptor.  This should fix the reported
calibration failures and subsequent speed problems with ich chipsets.

Minor calibration comment updates.

21 years agoAdd release of FreeBSD 4.8.
Murray Stokely [Fri, 4 Apr 2003 06:59:27 +0000 (06:59 +0000)]
Add release of FreeBSD 4.8.

MFC after:  3 days

21 years ago o Remove useracc() calls from aio_qphysio(); they are redundant
Alan Cox [Fri, 4 Apr 2003 06:26:28 +0000 (06:26 +0000)]
 o Remove useracc() calls from aio_qphysio(); they are redundant
   given the checks performed by vmapbuf().

Reviewed by: tegge

21 years ago o Check the b_bufsize passed to vmapbuf() returning an error
Alan Cox [Fri, 4 Apr 2003 06:14:54 +0000 (06:14 +0000)]
 o Check the b_bufsize passed to vmapbuf() returning an error
   if it is invalid.
 o Remove a debugging printf() from vmapbuf().

Suggested by:   tegge

21 years agoSimplify compatibility ifdef.
Warner Losh [Fri, 4 Apr 2003 04:39:29 +0000 (04:39 +0000)]
Simplify compatibility ifdef.

21 years agoNo need to check to see if we're running a version of FreeBSD 3.0 current or
Warner Losh [Fri, 4 Apr 2003 04:17:14 +0000 (04:17 +0000)]
No need to check to see if we're running a version of FreeBSD 3.0 current or
newer anymore.

21 years agoNo need to have ifdef < FreeBSD 4.0-current in here. Remove it to
Warner Losh [Fri, 4 Apr 2003 04:16:06 +0000 (04:16 +0000)]
No need to have ifdef < FreeBSD 4.0-current in here.  Remove it to
avoid false positive while searching for __FreeBSD_version abuse.