]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoDo not cast dval.
Maksim Yevmenkin [Wed, 11 Feb 2004 21:09:13 +0000 (21:09 +0000)]
Do not cast dval.
Make global _usage_page unsigned.

Reviewed by: imp (mentor)

20 years agoCreate symbolic link for g_print_bio(9) manual page.
Pawel Jakub Dawidek [Wed, 11 Feb 2004 18:43:20 +0000 (18:43 +0000)]
Create symbolic link for g_print_bio(9) manual page.

Approved by: scottl (mentor)

20 years agoDocument g_print_bio() function.
Pawel Jakub Dawidek [Wed, 11 Feb 2004 18:36:31 +0000 (18:36 +0000)]
Document g_print_bio() function.

Approved by: scottl (mentor)

20 years agoAdded g_print_bio() function to print informations about given bio.
Pawel Jakub Dawidek [Wed, 11 Feb 2004 18:21:32 +0000 (18:21 +0000)]
Added g_print_bio() function to print informations about given bio.

Approved by: phk, scottl (mentor)

20 years agoArgh! Fix a bogon. lim_cur() was returning the hard (max) limit rather
John Baldwin [Wed, 11 Feb 2004 18:04:13 +0000 (18:04 +0000)]
Argh!  Fix a bogon.  lim_cur() was returning the hard (max) limit rather
than the soft (cur) limit.

Submitted by: bde

20 years agoRemoved more vestiges of vfs_ioopt:
Bruce Evans [Wed, 11 Feb 2004 15:27:26 +0000 (15:27 +0000)]
Removed more vestiges of vfs_ioopt:
- rev.1.42 of ffs_readwrite.c added a special case in ffs_read() for reads
  that are initially at EOF, and rev.1.62 of ufs_readwrite.c fixed
  timestamp bugs in it.  Removal of most of vfs_ioopt made it just and
  optimization, and removal of the vm object reference calls made it less
  than an optimization.  It was cloned in rev.1.94 of ufs_readwrite.c as
  part of cloning ffs_extwrite() although it was always less than an
  optimization in ffs_extwrite().
- some comments, compound statements and vertical whitespace were vestiges
  of dead code.

20 years agoUse automatic major number allocation for the nfs4 device instead of
Tim J. Robbins [Wed, 11 Feb 2004 13:31:30 +0000 (13:31 +0000)]
Use automatic major number allocation for the nfs4 device instead of
stealing #29, which is already reserved for mcd.

20 years agoInterrupt statistics, vmstat -i now works.
Peter Grehan [Wed, 11 Feb 2004 13:18:31 +0000 (13:18 +0000)]
Interrupt statistics, vmstat -i now works.

Submitted by:  Suleiman Souhlal <refugee@segfaulted.com>
Slightly modified by: grehan
Derived from:  i386

20 years agoUn-reserve major 144 (nsmb) now that we are using automatic allocation.
Tim J. Robbins [Wed, 11 Feb 2004 12:58:06 +0000 (12:58 +0000)]
Un-reserve major 144 (nsmb) now that we are using automatic allocation.

20 years agoUse automatic major number allocation for nsmb devices.
Tim J. Robbins [Wed, 11 Feb 2004 12:49:49 +0000 (12:49 +0000)]
Use automatic major number allocation for nsmb devices.

20 years agoClean up header files, which fixes compile warning.
Peter Grehan [Wed, 11 Feb 2004 10:49:30 +0000 (10:49 +0000)]
Clean up header files, which fixes compile warning.

20 years ago- constify devinfo strings to eliminate compile warning
Peter Grehan [Wed, 11 Feb 2004 10:15:15 +0000 (10:15 +0000)]
- constify devinfo strings to eliminate compile warning
- remove trailing whitespace

20 years ago(atapi_cb): Reenable automatic retrieve of sense data on error, making
Thomas Quinot [Wed, 11 Feb 2004 10:14:08 +0000 (10:14 +0000)]
(atapi_cb): Reenable automatic retrieve of sense data on error, making
 it asynchronous to avoid incorrect use of ata_atapicmd within an
 ATAPI callback.

Tested by: harti

20 years agoAdded GEOM kernel API manuals to the build.
Pawel Jakub Dawidek [Wed, 11 Feb 2004 10:09:17 +0000 (10:09 +0000)]
Added GEOM kernel API manuals to the build.

Reviewed by: scottl
Approved by: phk, scottl (mentor)

20 years agoAdded first part of GEOM kernel API manuals pages.
Pawel Jakub Dawidek [Wed, 11 Feb 2004 10:06:18 +0000 (10:06 +0000)]
Added first part of GEOM kernel API manuals pages.
Documented function and macros are:
- DECLARE_GEOM_CLASS(),
- g_attach(),
- g_detach(),
- g_new_bio(),
- g_clone_bio(),
- g_destroy_bio(),
- g_new_consumer(),
- g_destroy_consumer(),
- g_read_data(),
- g_write_data(),
- g_post_event(),
- g_waitfor_event(),
- g_cancel_event(),
- g_new_geomf(),
- g_destroy_geom(),
- g_new_providerf(),
- g_destroy_provider(),
- g_error_provider(),
- g_provider_by_name(),
- g_wither_geom().
and more to come.

I want to thanks following people for help with those documents:
Slawek Zak <zaks@prioris.mini.pw.edu.pl>
Simon L. Nielsen <simon@FreeBSD.org>
Pieter de Boer <g.p.de.boer@st.hanze.nl>
and of course
Poul-Henning Kamp <phk@FreeBSD.org>

Reviewed by: phk, scottl
Approved by: phk, scottl (mentor)

20 years agoFix a typo; left out preprocessor conditional for sigoff variable, which
Bruce M Simpson [Wed, 11 Feb 2004 09:46:54 +0000 (09:46 +0000)]
Fix a typo; left out preprocessor conditional for sigoff variable, which
is only used by TCP_SIGNATURE code.

Noticed by: Roop Nanuwa

20 years agoSlight tweak to previous commit: don't forget to call _http_seterr() if
Dag-Erling Smørgrav [Wed, 11 Feb 2004 09:35:27 +0000 (09:35 +0000)]
Slight tweak to previous commit: don't forget to call _http_seterr() if
we get a genuine 416 reply.

20 years agoWhen restarting a transfer that has already completed, the server will
Dag-Erling Smørgrav [Wed, 11 Feb 2004 09:31:39 +0000 (09:31 +0000)]
When restarting a transfer that has already completed, the server will
reply with a 416 error code (requested range not satisfiable) because
we ask it to start at the end of the file.  Handle this gracefully by
considering a 416 reply a success if the requested offset exactly
matches the length of the file and the requested length is zero.

20 years agoRe-wrap some comments.
Dag-Erling Smørgrav [Wed, 11 Feb 2004 09:23:35 +0000 (09:23 +0000)]
Re-wrap some comments.

20 years agoGet rid of unnecessary use of m4(1) by using cpp(1) instead.
Ruslan Ermilov [Wed, 11 Feb 2004 08:42:38 +0000 (08:42 +0000)]
Get rid of unnecessary use of m4(1) by using cpp(1) instead.
(John tells me there were problems when trying this before,
but it appears to be safe these day.)

OK'ed by: jhb
Repocopied by: joe

20 years ago- fix compile warnings
Peter Grehan [Wed, 11 Feb 2004 08:07:19 +0000 (08:07 +0000)]
- fix compile warnings
- removed obsolete NetBSD-derived ADB conditionals

20 years ago- fixed trailing whitespace and indentation
Peter Grehan [Wed, 11 Feb 2004 07:58:43 +0000 (07:58 +0000)]
- fixed trailing whitespace and indentation
- removed unused variable to fix compile warning

20 years agoFix compile warning
Peter Grehan [Wed, 11 Feb 2004 07:48:19 +0000 (07:48 +0000)]
Fix compile warning

20 years ago- remove trailing whitespace
Peter Grehan [Wed, 11 Feb 2004 07:44:56 +0000 (07:44 +0000)]
- remove trailing whitespace
- fix compile warnings. badaddr() will go to a header file soon.

20 years agoCleaned up param.h:
Peter Grehan [Wed, 11 Feb 2004 07:27:34 +0000 (07:27 +0000)]
Cleaned up param.h:

 - culled long-dead #define's
 - segment register defs moved to sr.h
 - NPMAPS moved to pmap.h
 - KERNBASE moved to vmparam.h
 - removed include of <machine/cpu.h> and fixed src files that
   relied on this.

Modifying segment register code no longer causes gcc rebuilds :-)

20 years agoInvalide pcb's fpu cpu # by setting it to INT_MAX, not NULL.
Peter Grehan [Wed, 11 Feb 2004 07:19:15 +0000 (07:19 +0000)]
Invalide pcb's fpu cpu # by setting it to INT_MAX, not NULL.

20 years agoGNU tar in the base system is now called "gtar"
Tim Kientzle [Wed, 11 Feb 2004 05:44:20 +0000 (05:44 +0000)]
GNU tar in the base system is now called "gtar"

"tar" is now just a link to "gtar".

This lays the groundwork for an orderly migration
from GNU tar to some other tar.  (First, we introduce
the new tar program, then we migrate the 'tar' name,
then we remove gtar, with intervals of months between
these steps.)

Approved by: gordon

20 years agoAdd sysctl hw.uma_mdpages to track how many pages have been allocated
Peter Grehan [Wed, 11 Feb 2004 04:42:48 +0000 (04:42 +0000)]
Add sysctl hw.uma_mdpages to track how many pages have been allocated
by UMA_MD_SMALL_ALLOC

20 years agoInitial import of RFC 2385 (TCP-MD5) digest support.
Bruce M Simpson [Wed, 11 Feb 2004 04:34:34 +0000 (04:34 +0000)]
Initial import of RFC 2385 (TCP-MD5) digest support.

This is the second of two commits; bring in the userland support to finish.

Teach libipsec and setkey about the tcp-md5 class of security associations,
thus allowing administrators to add per-host keys to the SADB for use by
the tcpsignature_compute() function.

Document that a single SPI must be used until such time as the code which
adds support to the SPD to specify flows for tcp-md5 treatment is suitable
for production.

Sponsored by: sentex.net

20 years agoInitial import of RFC 2385 (TCP-MD5) digest support.
Bruce M Simpson [Wed, 11 Feb 2004 04:26:04 +0000 (04:26 +0000)]
Initial import of RFC 2385 (TCP-MD5) digest support.

This is the first of two commits; bringing in the kernel support first.
This can be enabled by compiling a kernel with options TCP_SIGNATURE
and FAST_IPSEC.

For the uninitiated, this is a TCP option which provides for a means of
authenticating TCP sessions which came into being before IPSEC. It is
still relevant today, however, as it is used by many commercial router
vendors, particularly with BGP, and as such has become a requirement for
interconnect at many major Internet points of presence.

Several parts of the TCP and IP headers, including the segment payload,
are digested with MD5, including a shared secret. The PF_KEY interface
is used to manage the secrets using security associations in the SADB.

There is a limitation here in that as there is no way to map a TCP flow
per-port back to an SPI without polluting tcpcb or using the SPD; the
code to do the latter is unstable at this time. Therefore this code only
supports per-host keying granularity.

Whilst FAST_IPSEC is mutually exclusive with KAME IPSEC (and thus IPv6),
TCP_SIGNATURE applies only to IPv4. For the vast majority of prospective
users of this feature, this will not pose any problem.

This implementation is output-only; that is, the option is honoured when
responding to a host initiating a TCP session, but no effort is made
[yet] to authenticate inbound traffic. This is, however, sufficient to
interwork with Cisco equipment.

Tested with a Cisco 2501 running IOS 12.0(27), and Quagga 0.96.4 with
local patches. Patches for tcpdump to validate TCP-MD5 sessions are also
available from me upon request.

Sponsored by: sentex.net

20 years agoPrefer buttons defined in the AML over the ones in the FADT. Some
Nate Lawson [Wed, 11 Feb 2004 02:57:33 +0000 (02:57 +0000)]
Prefer buttons defined in the AML over the ones in the FADT.  Some
systems define power/sleep buttons in both places but only deliver
notifies to the ones defined in the AML.

Also, reduce length of various function handler names.

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:
MFC after:

20 years agoDocument the functions m_apply() and m_getptr().
Bruce M Simpson [Wed, 11 Feb 2004 02:39:24 +0000 (02:39 +0000)]
Document the functions m_apply() and m_getptr().

Requested by: Maxim Konovalov

20 years agoCorrectly create interrupt key for PCI, which is the OpenFirmware
Peter Grehan [Tue, 10 Feb 2004 23:57:35 +0000 (23:57 +0000)]
Correctly create interrupt key for PCI, which is the OpenFirmware
pci-hi/med/lo + node 'interrupts' property. This worked by
accident until recent notebooks required correct operation.

Tested by:  Suleiman Souhlal <refugee@segfaulted.com>

20 years agoAdd some bulletproofing: don't allow the ndis_get_info() or ndis_set_info()
Bill Paul [Tue, 10 Feb 2004 23:01:53 +0000 (23:01 +0000)]
Add some bulletproofing: don't allow the ndis_get_info() or ndis_set_info()
routines to do anything except return error if the miniport adapter context
is not set (meaning we either having init'ed the driver yet, or the
initialization failed).

Also, be sure to NULL out the adapter context along with the
miniport characteristics pointers if calling the MiniportInitialize()
method fails.

20 years agoFlush stdout when -A option is used.
Munechika SUMIKAWA [Tue, 10 Feb 2004 22:50:57 +0000 (22:50 +0000)]
Flush stdout when -A option is used.

Obtained from: KAME
MFC after: 1 week

20 years agoSince loader(8) on SPARC64 is a pure ELF executable (as opposed
Ruslan Ermilov [Tue, 10 Feb 2004 22:11:36 +0000 (22:11 +0000)]
Since loader(8) on SPARC64 is a pure ELF executable (as opposed
to other architectures), there is no reason not to strip(1) it.

Tested by: kensmith

20 years agoOnly reset the phy when it is absolutely required.
Prafulla Deuskar [Tue, 10 Feb 2004 21:31:09 +0000 (21:31 +0000)]
Only reset the phy when it is absolutely required.
This should fix the issues with long *init* times when
you do ifconfig em0 alias.

MFC after: 3 days

20 years agoFix a panic in pseudofs(9) that could occur when doing an I/O
Jacques Vidrine [Tue, 10 Feb 2004 21:06:47 +0000 (21:06 +0000)]
Fix a panic in pseudofs(9) that could occur when doing an I/O
operation with a large request or large offset.

Reported by: Joel Ray Holveck <joelh@piquan.org>
Submitted by: des

20 years agostyle cleanup: Remove duplicate $FreeBSD$ tags.
Colin Percival [Tue, 10 Feb 2004 20:48:08 +0000 (20:48 +0000)]
style cleanup: Remove duplicate $FreeBSD$ tags.

These files had tags after the copyright notice,
inside the comment block (incorrect, removed),
and outside the comment block (correct).

Approved by: rwatson (mentor)

20 years agostyle cleanup: Remove duplicate $FreeBSD$ tags.
Colin Percival [Tue, 10 Feb 2004 20:45:28 +0000 (20:45 +0000)]
style cleanup: Remove duplicate $FreeBSD$ tags.

These files had tags after teh copyright notice,
inside the comment block (incorrect, removed),
and outside the comment block (correct).

Approved by: rwatson (mentor)

20 years agostyle cleanup: Remove duplicate $FreeBSD$ tags.
Colin Percival [Tue, 10 Feb 2004 20:42:33 +0000 (20:42 +0000)]
style cleanup: Remove duplicate $FreeBSD$ tags.

These files had tags after the copyright notice,
inside the comment block (incorrect, removed),
and outside the comment block (correct).

Approved by: rwatson (mentor)

20 years agostyle cleanup: Remove duplicate $FreeBSD$ tags.
Colin Percival [Tue, 10 Feb 2004 20:40:17 +0000 (20:40 +0000)]
style cleanup: Remove duplicate $FreeBSD$ tags.

These files had tags at the start of the file
(incorrect, removed), and after the copyright
notices (correct).

Approved by: rwatson (mentor)

20 years agoadd verbage about how once BUS_TEARDOWN_INTR returns, the interrupt will
John-Mark Gurney [Tue, 10 Feb 2004 20:34:44 +0000 (20:34 +0000)]
add verbage about how once BUS_TEARDOWN_INTR returns, the interrupt will
no longer be active or called..

Also document requirement that no mutexes be held across calls to these
functions..

Reviewed by: jhb, rwatson

20 years agoOfficial patches 001-005
Andrey A. Chernov [Tue, 10 Feb 2004 20:17:58 +0000 (20:17 +0000)]
Official patches 001-005

PR:             62555

20 years agoRun /etc/rc.d/routing at the appropriate time.
Dag-Erling Smørgrav [Tue, 10 Feb 2004 19:10:34 +0000 (19:10 +0000)]
Run /etc/rc.d/routing at the appropriate time.

Spotted by: mat

20 years agoAdd some presidential birthdays.
Wes Peters [Tue, 10 Feb 2004 18:41:52 +0000 (18:41 +0000)]
Add some presidential birthdays.

20 years agoCorrect a long-standing race condition in the inactive queue scan. (See
Alan Cox [Tue, 10 Feb 2004 18:34:27 +0000 (18:34 +0000)]
Correct a long-standing race condition in the inactive queue scan.  (See
the added comment for low-level details.)  The effect of this race
condition is a panic "vm_page_cache: caching a dirty page, ..."

Reviewed by: tegge
MFC after: 7 days

20 years agoAdd my birthday (for those interested)
Max Laier [Tue, 10 Feb 2004 16:23:17 +0000 (16:23 +0000)]
Add my birthday (for those interested)

Approved by: bms (mentor)

20 years agoSpelling.
Bruce M Simpson [Tue, 10 Feb 2004 16:11:08 +0000 (16:11 +0000)]
Spelling.

20 years agoNow we have g_topology_assert_not(), so use it to detect deadlocks.
Pawel Jakub Dawidek [Tue, 10 Feb 2004 15:55:17 +0000 (15:55 +0000)]
Now we have g_topology_assert_not(), so use it to detect deadlocks.

Approved by: phk, scottl (mentor)

20 years agoAdded macro which will be used to assert, that the topology lock is not held.
Pawel Jakub Dawidek [Tue, 10 Feb 2004 15:53:28 +0000 (15:53 +0000)]
Added macro which will be used to assert, that the topology lock is not held.

Approved by: phk, scottl (mentor)

20 years agoChange the kernels definition of NODEV from ((dev_t)-1) to NULL.
Poul-Henning Kamp [Tue, 10 Feb 2004 15:23:47 +0000 (15:23 +0000)]
Change the kernels definition of NODEV from ((dev_t)-1) to NULL.

20 years agoDocument the '-h hostname' option, which seems to have been present
Peter Pentchev [Tue, 10 Feb 2004 15:12:01 +0000 (15:12 +0000)]
Document the '-h hostname' option, which seems to have been present
ever since rev. 1.1 of bootpd.c.
While I'm here, rearrange the synopsis a bit: sort the options and
clarify that -i and -s are mutually exclusive.

Reported by: Atanas Buchvarov <nasko@nove.bg>
MFC after: 3 days

20 years agodon't call sbuf_clear() right after sbuf_new(), it is not necessary.
Poul-Henning Kamp [Tue, 10 Feb 2004 10:54:19 +0000 (10:54 +0000)]
don't call sbuf_clear() right after sbuf_new(), it is not necessary.

20 years agoFix numerous constness and aliasing issues.
Dag-Erling Smørgrav [Tue, 10 Feb 2004 10:13:21 +0000 (10:13 +0000)]
Fix numerous constness and aliasing issues.

20 years agoThis commit was generated by cvs2svn to compensate for changes in r125647,
Dag-Erling Smørgrav [Tue, 10 Feb 2004 10:11:23 +0000 (10:11 +0000)]
This commit was generated by cvs2svn to compensate for changes in r125647,
which included commits to RCS files with non-trunk default branches.

20 years agoVendor import of OpenPAM Eelgrass.
Dag-Erling Smørgrav [Tue, 10 Feb 2004 10:11:23 +0000 (10:11 +0000)]
Vendor import of OpenPAM Eelgrass.

20 years agoRemove a redundant command. No MFC, this was removed in STABLE.
Tom Rhodes [Tue, 10 Feb 2004 07:29:04 +0000 (07:29 +0000)]
Remove a redundant command.  No MFC, this was removed in STABLE.
Update the date.

20 years agoRemove 'broken in some cases' Xref to Kerberos.
Tom Rhodes [Tue, 10 Feb 2004 07:08:54 +0000 (07:08 +0000)]
Remove 'broken in some cases' Xref to Kerberos.

PR: 30443
Cool with: markm (who approved this ages ago)

20 years agoReorder Xrefs in SEE ALSO.
Tom Rhodes [Tue, 10 Feb 2004 06:44:41 +0000 (06:44 +0000)]
Reorder Xrefs in SEE ALSO.

20 years agoUpdate to inform users of acls and multilabel options. Add Xrefs to the
Tom Rhodes [Tue, 10 Feb 2004 06:33:29 +0000 (06:33 +0000)]
Update to inform users of acls and multilabel options.  Add Xrefs to the
more relevant manual pages.

PR: 62394
Submitted by: Marc Silver <marcs@draenor.org>

20 years agoFixes problems that occurred when a file was removed and a directory
Tim J. Robbins [Tue, 10 Feb 2004 05:53:02 +0000 (05:53 +0000)]
Fixes problems that occurred when a file was removed and a directory
created with the same name, and vice versa:
- Immediately recycle vnodes of files & directories that have been deleted
or renamed.
- When looking an entry in the VFS name cache or smbfs's private
cache, make sure the vnode type is consistent with the type of file
the server thinks it is, and re-create the vnode if it isn't.

The alternative to this is to recycle vnodes unconditionally when their
use count drops to 0, but this would make all the caching we do
mostly useless.

PR: 62342
MFC after: 2 weeks

20 years agoAdd sem_timedwait to the symbol map, otherwise the libc version will
Daniel Eischen [Tue, 10 Feb 2004 05:36:09 +0000 (05:36 +0000)]
Add sem_timedwait to the symbol map, otherwise the libc version will
get used.

20 years agoInitial import of libarchive.
Tim Kientzle [Mon, 9 Feb 2004 23:22:54 +0000 (23:22 +0000)]
Initial import of libarchive.

What it is:
   A library for reading and writing various streaming archive
   formats, especially tar and cpio.  Being a library, it should
   be easy to incorporate into pkg_* tools, sysinstall, and any
   other place that needs to read or write such archives.

Features:
  * Full automatic detection of both compression and archive format.
  * Extensible internal architecture to make it easy to add new formats.
  * Support for "pax interchange format," a new POSIX-standard tar format
    that eliminates essentially all of the restrictions of historic formats.
  * BSD license

Thanks to: jkh for pushing me to start this work, gordon for
  encouraging me to commit it, bde for answering endless style
  questions, and many others for feedback and encouragement.

Status: Pretty good overall, though there are still a few rough edges and
  the library could always use more testing.  Feedback eagerly solicited.

20 years agoLoader makefile cleanup.
Peter Grehan [Mon, 9 Feb 2004 21:54:54 +0000 (21:54 +0000)]
Loader makefile cleanup.

Submitted by:  ru

20 years agoAdded an option -ldi<N> to control indentation of local variable names.
Bruce Evans [Mon, 9 Feb 2004 21:48:51 +0000 (21:48 +0000)]
Added an option -ldi<N> to control indentation of local variable names.
The default is to be backwards compatible and non-KNF (use the same
indentation for locals as for globals; -ldi0 gives KNF indentation
for locals (none)).  The indentation for globals also applies to struct
member names in local declatations.  The indentation of variable names
in multi-line declarations is broken in various ways and this commit
gives some new variations.

indent.1:
Also clarified the description of -di<N>.

20 years agoOnly use tabs to indent variable names if the declaration indent is
Bruce Evans [Mon, 9 Feb 2004 20:42:08 +0000 (20:42 +0000)]
Only use tabs to indent variable names if the declaration indent is
nonzero (so that the 1-char indentation given by -di0 is never rendered
by a tab).

Removed garbage commented out code for setting the indentation of variable
names.

20 years agoRemove VFS_STATFS() call which violated the lock order and wasn't
Dag-Erling Smørgrav [Mon, 9 Feb 2004 20:33:42 +0000 (20:33 +0000)]
Remove VFS_STATFS() call which violated the lock order and wasn't
really required anyway.

PR: kern/61994
Submitted by: Bjoern Groenvall <bg@sics.se>

20 years agoBacked out a part of previous commit that wasn;t mentioned in the log
Bruce Evans [Mon, 9 Feb 2004 20:26:27 +0000 (20:26 +0000)]
Backed out a part of previous commit that wasn;t mentioned in the log
message and wasn't quite ready (it avoided indenting the names of local
variables).

20 years agoAdd stub implementations of KfLowerIrql() and KfRaiseIrql() (both of
Bill Paul [Mon, 9 Feb 2004 19:13:58 +0000 (19:13 +0000)]
Add stub implementations of KfLowerIrql() and KfRaiseIrql() (both of
which are _fastcall).

20 years agoToo good to miss.
Wes Peters [Mon, 9 Feb 2004 18:45:25 +0000 (18:45 +0000)]
Too good to miss.

Wisdumb by: Danno Coppock <danno@ipinc.com>

20 years agofix build with FAST_IPSEC.
Hajimu UMEMOTO [Mon, 9 Feb 2004 16:02:16 +0000 (16:02 +0000)]
fix build with FAST_IPSEC.

Reported by: cjc

20 years agoMFi386: Factor out -nostdlib and use PROG.
Yoshihiro Takahashi [Mon, 9 Feb 2004 16:01:45 +0000 (16:01 +0000)]
MFi386: Factor out -nostdlib and use PROG.

20 years agoFixed tab lossage in indented variable names. The -diN option was
Bruce Evans [Mon, 9 Feb 2004 15:57:35 +0000 (15:57 +0000)]
Fixed tab lossage in indented variable names.  The -diN option was
worse than useless because it caused things like mangling of
"^int\tfoo" to "int     foo" (this for N = 8).  This quick fix breaks
the invariant that characters between s_code and e_code have width 1,
but nothing seems to depend on this.

20 years agoFixed misformatting of "struct foo *bar" in function parameter lists. It
Bruce Evans [Mon, 9 Feb 2004 15:27:02 +0000 (15:27 +0000)]
Fixed misformatting of "struct foo *bar" in function parameter lists.  It
was mangled to "struct foo * bar".  There should be an option to control
this, but no space is normal.  This finishes fixing the bugs in rev.1.4.

indent(1) still doesn't really understand types in parameter lists.  It
thinks keywords inside parentheses are for casts or sizeofs.  This works
accidentally for scalar types and this quick fix makes it work similarly
but not so accidentally for struct/union/enum types.

20 years agoMFi386.
Ruslan Ermilov [Mon, 9 Feb 2004 14:17:02 +0000 (14:17 +0000)]
MFi386.

- Factor out common settings and put them in an upper level Makefile.inc.
- Properly use PROG for real programs, not their products.
- Further reduce diffs to i386 versions.

Tested on: sparc64 (panther)

20 years ago- Factor out -nostdlib to an upper level Makefile.inc.
Ruslan Ermilov [Mon, 9 Feb 2004 14:11:58 +0000 (14:11 +0000)]
- Factor out -nostdlib to an upper level Makefile.inc.

- Now that bsd.prog.mk deals with programs linked with -nostdlib
  better, and has a notion of an "internal" program, use PROG
  where possible.  This has a good impact on the contents of
  .depend files and causes programs to be linked with cc(1).

XXX: boot2 couldn't be converted as it's actually two programs.

Tested on: i386, amd64

20 years agoTwo changes to aid in cleaning up sys/boot/ makefiles:
Ruslan Ermilov [Mon, 9 Feb 2004 14:05:17 +0000 (14:05 +0000)]
Two changes to aid in cleaning up sys/boot/ makefiles:

- Don't put libc.a as a dependency if program is linked with -nostdlib.

- Added INTERNALPROG (by analogy to INTERNALLIB) for programs which are
  built only for its side effect and shold not be installed.

20 years agoExpanded the size of the keyword table from 100 to 1000 entries so that
Bruce Evans [Mon, 9 Feb 2004 13:13:03 +0000 (13:13 +0000)]
Expanded the size of the keyword table from 100 to 1000 entries so that
the number of typedef-names is not so limited.  Same as in rev.1.4.

Added the "const" and "volatile" to the keyword table.  Rev.1.4 added
these but they were misclassified so they were not formatted as types.
indent still doesn't really understand them.  E.g., it mangles
"char * const *foo" and "char *const *foo".  This change mainly stops
it mangling "char const foo" to "char<declaration-indent>const foo".

20 years agoBacked out rev.1.4 and 1.7 so they they can be implemented and committed
Bruce Evans [Mon, 9 Feb 2004 12:52:15 +0000 (12:52 +0000)]
Backed out rev.1.4 and 1.7 so they they can be implemented and committed
properly.  Of the 3 changes mentioned in the log message for rev.1.4,
the first (implementing -[n]fcb) was correct but didn't touch this
file, the second (no-space-after-sizeof) was not actually done (it is
the default and is controlled by the undcoumented -[n]bs options), and
the third (no-space-after 'struct foo *') was very buggy and was reduced
to wrong comments and other style bugs by backing out the main part
of it in rev.1.6.  Rev.1.4 had 2 changes which were not mentioned in
its commit log: expand specials[] so that more than -83 typedef-names
can be specified (this was the one working change in rev.1.4), and add
"const" and "volatile" to specials[] (this was buggy).

20 years agoDisable branch-target instruction cache on MPC7457 as outlined
Peter Grehan [Mon, 9 Feb 2004 07:04:01 +0000 (07:04 +0000)]
Disable branch-target instruction cache on MPC7457 as outlined
in Motorola processor errata.

Submitted by:  Suleiman Souhlal <refugee@segfaulted.com>

20 years agoFall back to INTR_MPSAFE if INTR_FAST registration fails.
Scott Long [Mon, 9 Feb 2004 05:29:05 +0000 (05:29 +0000)]
Fall back to INTR_MPSAFE if INTR_FAST registration fails.

PR: kern/62276

20 years agoRecognize MPC7547 (aka G4+)
Peter Grehan [Mon, 9 Feb 2004 02:25:24 +0000 (02:25 +0000)]
Recognize MPC7547 (aka G4+)

20 years agoDefinitions for MPC7457 CPU type and HID0 bits
Peter Grehan [Mon, 9 Feb 2004 00:12:50 +0000 (00:12 +0000)]
Definitions for MPC7457 CPU type and HID0 bits

20 years agoFix a typo that stopped ypset's -h option from working with a
Ian Dowse [Mon, 9 Feb 2004 00:10:18 +0000 (00:10 +0000)]
Fix a typo that stopped ypset's -h option from working with a
hostname rather than an IP. The code was copying the pointer to the
IP address instead of the IP address itself. The bug has existed
ever since ypset was first imported in 1994.

PR: bin/62550
Submitted by: aardvark@saintaardvarkthecarpeted.com
MFC after: 1 week

20 years agoMFp4 @46705:
Juli Mallett [Sun, 8 Feb 2004 23:57:19 +0000 (23:57 +0000)]
MFp4 @46705:

Support "uprocp" exactly like "paddr" with the former having been
documented in the manual but not implemented.

PR: 42484

20 years agoUse int rather than size_t storage for printf field widths to avoid
Ian Dowse [Sun, 8 Feb 2004 23:42:09 +0000 (23:42 +0000)]
Use int rather than size_t storage for printf field widths to avoid
many casts.

Reviewed by: bde

20 years agoFurther simplify the code for printing the message buffer:
Ian Dowse [Sun, 8 Feb 2004 23:08:53 +0000 (23:08 +0000)]
Further simplify the code for printing the message buffer:
 - Ensure that the buffer ends with "\n\0" to avoid special cases
   and allow the use of strtol().
 - Use strvisx() on each complete line instead of character by
   character.

Submitted by: bde
MFC after: 1 week

20 years agoNew errata: SA-04:02.
Bruce A. Mah [Sun, 8 Feb 2004 22:16:29 +0000 (22:16 +0000)]
New errata:  SA-04:02.

20 years agoCertain ICMP error replies cause ping to perform a reverse DNS
Ian Dowse [Sun, 8 Feb 2004 21:59:17 +0000 (21:59 +0000)]
Certain ICMP error replies cause ping to perform a reverse DNS
lookup on an IP address from the packet (such as the IP that sent
a TTL exceeded error). If the DNS lookup takes a long time, ^C will
appear to be ineffective since the SIGINT handler just sets a flag
and returns. Work around this by exiting immediately on receipt of
a second SIGINT when DNS lookups are enabled.

PR: bin/4696
MFC after: 1 week

20 years agoUpdate for the 2004/02/07 import.
Ruslan Ermilov [Sun, 8 Feb 2004 21:39:18 +0000 (21:39 +0000)]
Update for the 2004/02/07 import.

20 years agoThis commit was generated by cvs2svn to compensate for changes in r125601,
Ruslan Ermilov [Sun, 8 Feb 2004 21:32:21 +0000 (21:32 +0000)]
This commit was generated by cvs2svn to compensate for changes in r125601,
which included commits to RCS files with non-trunk default branches.

20 years agoVendor import of bwk's 7-Feb-2004 release.
Ruslan Ermilov [Sun, 8 Feb 2004 21:32:21 +0000 (21:32 +0000)]
Vendor import of bwk's 7-Feb-2004 release.

20 years agoRewrite fabs.S to use pure SSE2 operations. I got the clues how to do
Peter Wemm [Sun, 8 Feb 2004 21:21:45 +0000 (21:21 +0000)]
Rewrite fabs.S to use pure SSE2 operations.  I got the clues how to do
this from looking at code generated by gcc.

20 years agoMake NdisMMapIoSpace() guard against NULL/uninitialized resource pointers too.
Bill Paul [Sun, 8 Feb 2004 20:39:35 +0000 (20:39 +0000)]
Make NdisMMapIoSpace() guard against NULL/uninitialized resource pointers too.

20 years agoMake NdisMMapIoSpace() handle the case where a device has both mem
Bill Paul [Sun, 8 Feb 2004 20:32:41 +0000 (20:32 +0000)]
Make NdisMMapIoSpace() handle the case where a device has both mem
and altmem ranges mapped.

20 years agoRemove condition variables and status associated with target mode
Matt Jacob [Sun, 8 Feb 2004 19:17:56 +0000 (19:17 +0000)]
Remove condition variables and status associated with target mode
enabling. Instead, go to an interrupt/polled model.

Fix get_lun_statep so we don't panic if there are no wildcard luns enabled.

MFC after: 6 days

20 years agoRemove condition variables and status associated with target mode
Matt Jacob [Sun, 8 Feb 2004 19:16:01 +0000 (19:16 +0000)]
Remove condition variables and status associated with target mode
enabling. Instead, go to an interrupt/polled model.

MFC after: 6 days

20 years ago- obey ip6po_minmtu.
Hajimu UMEMOTO [Sun, 8 Feb 2004 18:22:27 +0000 (18:22 +0000)]
- obey ip6po_minmtu.
- notify a proper path MTU to applications.

Obtained from: KAME

20 years agoAllow amr(4) to get a dynamic major number instead of a static one.
Scott Long [Sun, 8 Feb 2004 16:07:22 +0000 (16:07 +0000)]
Allow amr(4) to get a dynamic major number instead of a static one.

Submitted by: Andre Guibert de Bruet