]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoMerge from NetBSD:
joe [Mon, 21 Jan 2002 04:24:33 +0000 (04:24 +0000)]
Merge from NetBSD:

    ohci.c: -r1.68
    ohcireg.h: -r1.13

date: 2000/01/31 20:17:25;  author: augustss;  state: Exp;
Fiddle with over-current protect when turning on port power to make
things work for some OHCI controllers.

22 years agoMerge from NetBSD:
joe [Mon, 21 Jan 2002 04:15:39 +0000 (04:15 +0000)]
Merge from NetBSD:

revision 1.65
date: 2000/01/25 12:06:21;  author: augustss;  state: Exp;  lines: +10 -2
Add done method for root control transfers.

22 years agoMerge from NetBSD:
joe [Mon, 21 Jan 2002 03:44:00 +0000 (03:44 +0000)]
Merge from NetBSD:

revision 1.84
date: 2000/01/28 00:44:27;  author: augustss;  state: Exp;  lines: +9 -2
Add uhci_root_ctrl_done() method.

22 years agoMerge from NetBSD:
joe [Mon, 21 Jan 2002 03:35:55 +0000 (03:35 +0000)]
Merge from NetBSD:

    uhci.c: -r1.82
    uhcivar.h: -r1.22

date: 2000/01/26 10:04:39;  author: augustss;  state: Exp;
Try to avoid accessing the HC if it is dead.  Suggested by mycroft.

22 years agoAllow dump device be configured as early as possible using loader(8) tunable.
sobomax [Mon, 21 Jan 2002 01:16:11 +0000 (01:16 +0000)]
Allow dump device be configured as early as possible using loader(8) tunable.
This allows obtaining crash dumps from the panics occured during late stages
of kernel initialisation before system enters into single-user mode.

MFC after: 2 weeks

22 years agoAdd support for Linksys WDT11 PCI adaptors.
brooks [Mon, 21 Jan 2002 00:59:59 +0000 (00:59 +0000)]
Add support for Linksys WDT11 PCI adaptors.

Submitted by: Eric Liedtke <eliedtke@apogeetelecom.com>

22 years agoRemove some unused code, in line with NetBSD's version.
joe [Sun, 20 Jan 2002 23:56:46 +0000 (23:56 +0000)]
Remove some unused code, in line with NetBSD's version.

22 years agoMerge from NetBSD:
joe [Sun, 20 Jan 2002 23:48:43 +0000 (23:48 +0000)]
Merge from NetBSD:

revision 1.125
date: 2000/09/23 21:00:10;  author: augustss;  state: Exp;  lines: +19 -3
Avoid "bandwidth reclamation" for control transfers.  The kue device chokes
on it.

22 years agoMerge from NetBSD:
joe [Sun, 20 Jan 2002 23:38:33 +0000 (23:38 +0000)]
Merge from NetBSD:

    uhci.c: -r1.124
    uhcireg.h: -r1.13

date: 2000/08/13 18:20:14;  author: augustss;  state: Exp;
Fix race condition when unlinking xfers.  Thanks to IWAMOTO Toshihiro
<iwamoto@sat.t.u-tokyo.ac.jp> for analyzing the problem and suggesting a fix.
Fixes PR 10662.

22 years agoMFS 1.536.2.52 (partial) and 1.536.2.70: kgzip loader if i386 (only).
matusita [Sun, 20 Jan 2002 23:17:52 +0000 (23:17 +0000)]
MFS 1.536.2.52 (partial) and 1.536.2.70: kgzip loader if i386 (only).

5.0-CURRENT-20020121-JPSNAP was failed to build since kern.flp has no
space.  There is already a hack to kgzip loader, and it makes:

-r-xr-xr-x  1 root  wheel  163840 Jan 21 07:40 loader*
-r-xr-xr-x  1 root  wheel   90121 Jan 21 08:06 loader.kgz*

73719 bytes of free spaces to us.

22 years agouse mutex pools for "struct file" locking.
alfred [Sun, 20 Jan 2002 22:58:08 +0000 (22:58 +0000)]
use mutex pools for "struct file" locking.
fix indentation of FILE_LOCK/UNLOCK macros while I'm here.

22 years agouse mutex pool mutexes for uidinfo locking.
alfred [Sun, 20 Jan 2002 22:48:49 +0000 (22:48 +0000)]
use mutex pool mutexes for uidinfo locking.
replace mutex_lock calls on uidinfo with macro calls:
  mtx_lock(&uidp->ui_mtx) -> UIDINFO_LOCK(uidp)

Terry Lambert <tlambert2@mindspring.com> helped with this.

22 years agosnprintf bloat -> strlcpy
ache [Sun, 20 Jan 2002 20:56:47 +0000 (20:56 +0000)]
snprintf bloat -> strlcpy
Add getpwnam return check

Approved by: des, markm

22 years agoMerge from NetBSD:
joe [Sun, 20 Jan 2002 20:12:25 +0000 (20:12 +0000)]
Merge from NetBSD:

    uhci.c: -r1.123 (and a tiny bit of -r1.92)
    uhcivar.h: -r1.32

date: 2000/08/13 16:18:09;  author: augustss;  state: Exp;
Implement what in Intel-speech is known as "bandwidth
reclamation".  It means that we continously poll USB devices
that have a pending transfer instead of polling just once
every ms.  This speeds up some transfers at the expense of
using more PCI bandwidth.

22 years agoCheck the return value from read() when reading the CR/LF at the end of a
des [Sun, 20 Jan 2002 19:53:12 +0000 (19:53 +0000)]
Check the return value from read() when reading the CR/LF at the end of a
chunk.

PR: bin/33608
MFC after: 2 weeks

22 years agoMark uploads as O_WRONLY, not O_RDONLY.
des [Sun, 20 Jan 2002 19:52:25 +0000 (19:52 +0000)]
Mark uploads as O_WRONLY, not O_RDONLY.

PR: misc/34043
MFC after: 2 weeks

22 years ago o Remove the unused vestiges of JOBST_JOBQPROC and
alc [Sun, 20 Jan 2002 18:59:58 +0000 (18:59 +0000)]
 o Remove the unused vestiges of JOBST_JOBQPROC and
   the per-thread jobtorun queue.
 o Use TAILQ_EMPTY() instead of TAILQ_FIRST(...) == NULL.

22 years agoWARNS=4 fixes.
markm [Sun, 20 Jan 2002 17:54:33 +0000 (17:54 +0000)]
WARNS=4 fixes.

OK'ed by: des

22 years agoAdd `-h' to the chown command so that symbolic links copied from
iedowse [Sun, 20 Jan 2002 17:05:07 +0000 (17:05 +0000)]
Add `-h' to the chown command so that symbolic links copied from
the skeleton directory are chown'd to the new user.

PR: bin/10601
Submitted by: Adrian Filipi-Martin <adrian2ubergeeks.com@gosub.cstone.net>
MFC after: 1 month

22 years agoMinor typo fix: uquad_t -> u_quad_t.
yar [Sun, 20 Jan 2002 16:50:29 +0000 (16:50 +0000)]
Minor typo fix: uquad_t -> u_quad_t.

22 years agoRemove very outdated comment from the Bugs section that states that
scottl [Sun, 20 Jan 2002 16:09:18 +0000 (16:09 +0000)]
Remove very outdated comment from the Bugs section that states that
the driver cannot do U160.

Submitted by: ladisalv.kostal@fem.uniag.sk
MFC after: 1 day

22 years agofrom select(2):
billf [Sun, 20 Jan 2002 12:13:28 +0000 (12:13 +0000)]
from select(2):
Any of readfds, writefds, and exceptfds may be given as nil
pointers if no descriptors are of interest.

neither wfds nor efds were of interest so now they are nil.

also, do a little better then making an educated guess for nfds.

22 years agoMerge from NetBSD:
joe [Sun, 20 Jan 2002 12:08:09 +0000 (12:08 +0000)]
Merge from NetBSD:

    uchireg.h: -r1.12
    uchi.c: -r1.121

    date: 2000/07/23 19:43:38;  author: augustss;  state: Exp;
    Be a little more explicit and careful about setting links in TDs and QHs.

22 years agoCorrect the majors entry for 'iir'. 163 is vacated now.
msmith [Sun, 20 Jan 2002 09:01:07 +0000 (09:01 +0000)]
Correct the majors entry for 'iir'.  163 is vacated now.

22 years agoAdd the 'iir' driver, for the Intel Integrated RAID controllers and
msmith [Sun, 20 Jan 2002 08:51:08 +0000 (08:51 +0000)]
Add the 'iir' driver, for the Intel Integrated RAID controllers and
prior ICP Vortex models.  This driver was developed by Achim Leubner
of Intel (previously with ICP Vortex) and Boji Kannanthanam of Intel.

Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com>
MFC after: 2 weeks

22 years agoICP have been acquired by Intel, and their driver is now the Intel
msmith [Sun, 20 Jan 2002 06:21:33 +0000 (06:21 +0000)]
ICP have been acquired by Intel, and their driver is now the Intel
Integrated RAID driver, supported by <boji.t.kannanthanam@intel.com> and
<achim.leubner@intel.com>.

Submitted by: "Kannanthanam, Boji T" <boji.t.kannanthanam@intel.com>

22 years agoThe Libretto L series has no $PIR table, but does have a _PIR table.
imp [Sun, 20 Jan 2002 03:28:29 +0000 (03:28 +0000)]
The Libretto L series has no $PIR table, but does have a _PIR table.
This typo keeps us from properly routing an interrupt for CardBus
bridges on this machine.  So, now we look for $PIR and then _PIR to
cope.  With these changes, the Libretto L1 now works properly.
Evidentally, the idea comes from patch that the Japanese version of
RedHat (or against a Japanese version of Red Hat), but my Japanese
isn't good enough to to know for sure.

Reported by: Hiroyuki Aizu-san <eyes@navi.org>

# This may be an MFC candidate, but I'm not yet sure.

22 years agoPrevent overflowing the buffer that stores the command arguments.
mikeh [Sun, 20 Jan 2002 01:30:40 +0000 (01:30 +0000)]
Prevent overflowing the buffer that stores the command arguments.

PR: bin/19422
Not objected to by: -audit
MFC after: 3 weeks

22 years ago o Revision 1.99 ("KSE Milestone 2") left the aio daemons
alc [Sun, 20 Jan 2002 00:52:44 +0000 (00:52 +0000)]
 o Revision 1.99 ("KSE Milestone 2") left the aio daemons
   sleeping on a process object but changed the corresponding
   wakeup()s to the thread object.  The result was that non-raw
   aio ops waited for an aio daemon to timeout before action
   was taken.  Now, we sleep on the thread object.

PR: kern/34016

22 years agoLink mail(1) to mailx per POSIX. Mail(1) isn't 100% POSIX compliant,
mikeh [Sat, 19 Jan 2002 23:47:21 +0000 (23:47 +0000)]
Link mail(1) to mailx per POSIX. Mail(1) isn't 100% POSIX compliant,
but it is pretty close.

Not objected to by: -standards

22 years agoI've been meaning to do this for a while. Add an underscore to the
dillon [Sat, 19 Jan 2002 23:20:02 +0000 (23:20 +0000)]
I've been meaning to do this for a while.  Add an underscore to the
time_to_xxx() and xxx_to_time() functions.  e.g. _time_to_xxx()
instead of time_to_xxx(), to make it more obvious that these are
stopgap functions & placemarkers and not meant to create a defacto
standard.  They will eventually be replaced when a real standard
comes out of committee.

22 years agoReally back out ache's commits. These files are now precisely as they were
des [Sat, 19 Jan 2002 18:29:50 +0000 (18:29 +0000)]
Really back out ache's commits.  These files are now precisely as they were
twentyfour hours ago, except for RCS ids.

22 years agoBack out PAM_CRED_ERR addition
ache [Sat, 19 Jan 2002 18:06:05 +0000 (18:06 +0000)]
Back out PAM_CRED_ERR addition

22 years agoBack out recent changes
ache [Sat, 19 Jan 2002 18:03:11 +0000 (18:03 +0000)]
Back out recent changes

22 years agoMFen (1.261 --> 1.263).
kuriyama [Sat, 19 Jan 2002 15:53:23 +0000 (15:53 +0000)]
MFen (1.261 --> 1.263).

22 years agoRemove bogus entry.
kuriyama [Sat, 19 Jan 2002 15:12:40 +0000 (15:12 +0000)]
Remove bogus entry.

22 years agoMFen (1.260 --> 1.261; order changes only).
kuriyama [Sat, 19 Jan 2002 15:12:05 +0000 (15:12 +0000)]
MFen (1.260 --> 1.261; order changes only).

22 years agoNew release notes: ARCNET support, cm driver.
bmah [Sat, 19 Jan 2002 13:38:48 +0000 (13:38 +0000)]
New release notes:  ARCNET support, cm driver.

22 years agoNew release notes: dc(4) supports VLANs, man(1) no longer SUID man,
bmah [Sat, 19 Jan 2002 13:02:32 +0000 (13:02 +0000)]
New release notes:  dc(4) supports VLANs, man(1) no longer SUID man,
/etc/pam.d/.

22 years agoAlphabetize the kernel changes section by manpage reference or
bmah [Sat, 19 Jan 2002 12:50:37 +0000 (12:50 +0000)]
Alphabetize the kernel changes section by manpage reference or
keyword, where possible.

22 years agoTurn on pam_opie by default. It should not affect non-OPIE users.
ache [Sat, 19 Jan 2002 10:31:32 +0000 (10:31 +0000)]
Turn on pam_opie by default. It should not affect non-OPIE users.

22 years agoIf user not exist in OPIE system, return failure immediately instead
ache [Sat, 19 Jan 2002 10:09:05 +0000 (10:09 +0000)]
If user not exist in OPIE system, return failure immediately instead
of producing fake prompts with random numbers which can be detected by
potential intruder in two tries and totally confuse non-OPIE users.

22 years agoBack out second right-now-expired password check in pam_sm_chauthtok,
ache [Sat, 19 Jan 2002 09:23:36 +0000 (09:23 +0000)]
Back out second right-now-expired password check in pam_sm_chauthtok,
old expired password assumed there

22 years agoTurn on pam_opie by default. It not affect non-OPIE users
ache [Sat, 19 Jan 2002 09:06:45 +0000 (09:06 +0000)]
Turn on pam_opie by default. It not affect non-OPIE users

22 years agoAdd PAM_CRED_ERR as valid failure case
ache [Sat, 19 Jan 2002 09:01:17 +0000 (09:01 +0000)]
Add PAM_CRED_ERR as valid failure case

22 years agoPrevious commit was incomplete, use
ache [Sat, 19 Jan 2002 08:39:35 +0000 (08:39 +0000)]
Previous commit was incomplete, use
"[default=ignore success=done cred_err=die]"
options instead of "required"

22 years agoPrevious commit was incomplete, use new error code PAM_CRED_ERR to
ache [Sat, 19 Jan 2002 08:36:47 +0000 (08:36 +0000)]
Previous commit was incomplete, use new error code PAM_CRED_ERR to
indicate die case, different from PAM_SUCCESS and PAM_AUTH_ERR

22 years agoAdd flags 0x10000 to IO Data WN-B11/PCM's entry. Evidentally, they
imp [Sat, 19 Jan 2002 08:11:39 +0000 (08:11 +0000)]
Add flags 0x10000 to IO Data WN-B11/PCM's entry.  Evidentally, they
changed firmware and the new cards don't work without this.

Submitted by: ume
MFC after: 3 days

22 years agoRemove explaining comment and pam_unix commented out, now pam_unix can be
ache [Sat, 19 Jan 2002 07:32:47 +0000 (07:32 +0000)]
Remove explaining comment and pam_unix commented out, now pam_unix can be
chained with pam_opie

22 years agoRewrite 'pwok' fallback in the way it can be properly chained with pam_unix
ache [Sat, 19 Jan 2002 07:23:48 +0000 (07:23 +0000)]
Rewrite 'pwok' fallback in the way it can be properly chained with pam_unix

Replace snprintf %s with strlcpy

Check for NULL returned from getpwnam()

22 years agoLock the caller process if the pid passed to getsid() or getpgid()
tanimura [Sat, 19 Jan 2002 06:34:58 +0000 (06:34 +0000)]
Lock the caller process if the pid passed to getsid() or getpgid()
equals to zero.

22 years agoCall opieunlock() only if we skip opieverify() part
ache [Sat, 19 Jan 2002 05:59:24 +0000 (05:59 +0000)]
Call opieunlock() only if we skip opieverify() part

22 years agoMFen (1.259 --> 1.260).
kuriyama [Sat, 19 Jan 2002 05:41:17 +0000 (05:41 +0000)]
MFen (1.259 --> 1.260).

22 years agoFor getsid(), return the sid stored in struct session. This prevents
tanimura [Sat, 19 Jan 2002 05:31:51 +0000 (05:31 +0000)]
For getsid(), return the sid stored in struct session. This prevents
panic in case where a session has no session leader.

Inspired by: Solaris 8

22 years agoNew release notes: bpf(4) works better with select(2) and poll(2),
bmah [Sat, 19 Jan 2002 05:23:17 +0000 (05:23 +0000)]
New release notes:  bpf(4) works better with select(2) and poll(2),
natd(8) -log_ipfw_denied.

22 years agoAdd yet one expired-right-now password check, in pam_sm_chauthtok
ache [Sat, 19 Jan 2002 04:58:51 +0000 (04:58 +0000)]
Add yet one expired-right-now password check, in pam_sm_chauthtok

srandomdev() can't be used in libraries, replace srandomdev()+random()
by arc4random()

22 years agoMFen (1.257 --> 1.259).
kuriyama [Sat, 19 Jan 2002 04:29:55 +0000 (04:29 +0000)]
MFen (1.257 --> 1.259).

22 years agoMake compile, remove extra fdrop() calls.
alfred [Sat, 19 Jan 2002 03:52:17 +0000 (03:52 +0000)]
Make compile, remove extra fdrop() calls.
Change name of function to what it's supposed to be (s/sys/do)

22 years agomake compile, add missing { and variable declaration.
alfred [Sat, 19 Jan 2002 03:48:38 +0000 (03:48 +0000)]
make compile, add missing { and variable declaration.

22 years agoSemi-backout previous fgetvp change, we need the struct file pointer
alfred [Sat, 19 Jan 2002 03:45:14 +0000 (03:45 +0000)]
Semi-backout previous fgetvp change, we need the struct file pointer
to perform relative offset calculations, so use fget instead.

22 years agoo Clarify that various sysctl timing values are in milliseconds.
rwatson [Sat, 19 Jan 2002 03:44:42 +0000 (03:44 +0000)]
o Clarify that various sysctl timing values are in milliseconds.

22 years agoChange comment since fallback provided now not by ftpd but by pam_opie
ache [Sat, 19 Jan 2002 03:35:39 +0000 (03:35 +0000)]
Change comment since fallback provided now not by ftpd but by pam_opie

22 years agoSet pwok to 1 for non-OPIE users
ache [Sat, 19 Jan 2002 03:31:39 +0000 (03:31 +0000)]
Set pwok to 1 for non-OPIE users

22 years agoRemove conditional 'pwok' fallback for PAM which now
ache [Sat, 19 Jan 2002 03:18:33 +0000 (03:18 +0000)]
Remove conditional 'pwok' fallback for PAM which now
is implemented in pam_opie module

For non-PAM variant rewrite empty password checking code to do the right thing
and not disallow empty passwords in all cases.

22 years agoAdd missing check for right-now-expired password
ache [Sat, 19 Jan 2002 02:45:24 +0000 (02:45 +0000)]
Add missing check for right-now-expired password

22 years agoImplement 'pwok', i.e. conditional fallback to unix password
ache [Sat, 19 Jan 2002 02:38:43 +0000 (02:38 +0000)]
Implement 'pwok', i.e. conditional fallback to unix password
as supposed by opieaccessfile() and opiealways()

22 years agoRemove 'VXLOCK: interlock avoided' warnings. This can now occur in normal
dillon [Sat, 19 Jan 2002 02:14:45 +0000 (02:14 +0000)]
Remove 'VXLOCK: interlock avoided' warnings.  This can now occur in normal
operation.  The vgonel() code has always called vclean() but until we
started proactively freeing vnodes it would never actually be called with
a dirty vnode, so this situation did not occur prior to the vnlru() code.
Now that we proactively free vnodes when kern.maxvnodes is hit, however,
vclean() winds up with work to do and improperly generates the warnings.

Reviewed by: peter
Approved by: re (for MFC)
MFC after: 1 day

22 years agoundo a bit of the Giant pushdown.
alfred [Sat, 19 Jan 2002 01:03:54 +0000 (01:03 +0000)]
undo a bit of the Giant pushdown.

fdrop isn't SMP safe as it may call into the file's close routine which
definetly is not SMP safe right now, so we hold Giant over calls to
fdrop now.

22 years agoExplain that the admin can safely power down the system as well as
nik [Fri, 18 Jan 2002 22:45:29 +0000 (22:45 +0000)]
Explain that the admin can safely power down the system as well as
rebooting.

22 years agoBy popular demand, also include the "devlist" subcommand into the set
joerg [Fri, 18 Jan 2002 22:42:51 +0000 (22:42 +0000)]
By popular demand, also include the "devlist" subcommand into the set
of commands available in the boot floppy environment.

MFC after: 1 week

22 years agoWell, RELEASE_BUILD_FIXIT has now been renamed into RELEASE_CRUNCH.
joerg [Fri, 18 Jan 2002 22:17:35 +0000 (22:17 +0000)]
Well, RELEASE_BUILD_FIXIT has now been renamed into RELEASE_CRUNCH.
At least, the old version is still good for the MFC though (where
everything is still going the old way). ;-)

22 years agoCut more than 500 bytes off the size of the alpha boot1 by adding
iedowse [Fri, 18 Jan 2002 21:54:02 +0000 (21:54 +0000)]
Cut more than 500 bytes off the size of the alpha boot1 by adding
a simple version of bcopy() so we avoid picking up the overly-complex
implementation in libc (via libstand). This is not necessary on
-current, but RELENG_4 has apparently just exceeded the 15-sector
limit for boot1.

Reviewed by: wilko

22 years agoCoalesce several release notes dealing with the aac(4) driver,
bmah [Fri, 18 Jan 2002 19:10:06 +0000 (19:10 +0000)]
Coalesce several release notes dealing with the aac(4) driver,
clean-up, and mention that it is "fully qualified and sanctioned by
Adaptec".

Requested by: scottl (the last part, that is)
Approved by: re (implicitly)

22 years agoChange ich_calibrate to busy wait on buffer fill level and use a more
orion [Fri, 18 Jan 2002 18:44:41 +0000 (18:44 +0000)]
Change ich_calibrate to busy wait on buffer fill level and use a more
likely looking rate calculation.

Install interrupt handler before calling ich_init as the initialization
occasionally generates spurious interrupts.

These changes are derived from cg's work in progress version of this
driver.

22 years agoAdd 'camcontrol rescan all' and 'camcontrol reset all' functionality to
ken [Fri, 18 Jan 2002 18:00:00 +0000 (18:00 +0000)]
Add 'camcontrol rescan all' and 'camcontrol reset all' functionality to
camcontrol.

This enables rescanning all busses or resetting all busses in a system.
The current implementation is not the ideal way to do it -- the ideal way
to do it would be for the transport layer to handle wildcarded busses on
bus rescan and reset operations.  The current implementation enumerates all
the busses and sends a rescan or reset CCB individually.  Handling this
behavior in the transport layer will happen later.

Reviewed by: imp
Tested by: joerg
MFC after: 1 week

22 years agoWe don't use two-letters terminal names, so strip them out from vt520* entries
ache [Fri, 18 Jan 2002 17:41:54 +0000 (17:41 +0000)]
We don't use two-letters terminal names, so strip them out from vt520* entries

22 years agoUpdate PCMCIA cards (fe(4), ex(4), and ep(4) drivers).
bmah [Fri, 18 Jan 2002 16:55:22 +0000 (16:55 +0000)]
Update PCMCIA cards (fe(4), ex(4), and ep(4) drivers).

Submitted by: Tom Rhodes <darklogik@pittgoth.com>

22 years agomdoc(7) police: tidy up.
ru [Fri, 18 Jan 2002 16:46:43 +0000 (16:46 +0000)]
mdoc(7) police: tidy up.

Reviewed by: chris

22 years agoNew release note: SA-02:07.
bmah [Fri, 18 Jan 2002 16:42:43 +0000 (16:42 +0000)]
New release note:  SA-02:07.

22 years agoGC spurious -m in the usage().
ru [Fri, 18 Jan 2002 16:08:29 +0000 (16:08 +0000)]
GC spurious -m in the usage().

22 years agoMFV: Initialize %I register.
ru [Fri, 18 Jan 2002 15:28:32 +0000 (15:28 +0000)]
MFV: Initialize %I register.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r89500,
ru [Fri, 18 Jan 2002 15:25:01 +0000 (15:25 +0000)]
This commit was generated by cvs2svn to compensate for changes in r89500,
which included commits to RCS files with non-trunk default branches.

22 years agoFrom ChangeLog:
ru [Fri, 18 Jan 2002 15:25:01 +0000 (15:25 +0000)]
From ChangeLog:

: 2002-01-17  Ruslan Ermilov  <ru@FreeBSD.org>
:
:  * tmac/doc.common: Initialize %I register for the %I macro to
:  avoid (harmless) warning.
:  * tmac/doc.tmac (Bd): There is no reason to enforce -compact
:  when in the SYNOPSIS section.

22 years agoIntroduce an interface announcement message for the routing
ru [Fri, 18 Jan 2002 14:33:04 +0000 (14:33 +0000)]
Introduce an interface announcement message for the routing
socket so that routing daemons and other interested parties
know when an interface is attached/detached.

PR: kern/33747
Obtained from: NetBSD
MFC after: 2 weeks

22 years agoInvert the test of sx_xholder for SX_LOCKED. We need to warn if a
tanimura [Fri, 18 Jan 2002 09:21:15 +0000 (09:21 +0000)]
Invert the test of sx_xholder for SX_LOCKED. We need to warn if a
thread other than the curthread holds an sx.

While I am here, break a line at the end of warning.

22 years agoAdded kgzldr for pc98.
nyan [Fri, 18 Jan 2002 09:00:38 +0000 (09:00 +0000)]
Added kgzldr for pc98.
crt.s is based on btx.s.

22 years agoAdd a definition of ddb_regs. ddb_regs is declared as extern in
marcel [Fri, 18 Jan 2002 07:55:23 +0000 (07:55 +0000)]
Add a definition of ddb_regs. ddb_regs is declared as extern in
db_machdep.h to fix the link failure (multiple definitions)
caused by disabling the emission of common symbols. As a result,
there were no definitions at all. While here, remove useless
declarations.

22 years agoRemove the definition of bootverbose. This fixes the link failure
marcel [Fri, 18 Jan 2002 07:49:37 +0000 (07:49 +0000)]
Remove the definition of bootverbose. This fixes the link failure
caused by disabling the emission of common symbols.

22 years agoDeclare ddb_regs as extern to avoid creating a tentative definition.
marcel [Fri, 18 Jan 2002 07:47:38 +0000 (07:47 +0000)]
Declare ddb_regs as extern to avoid creating a tentative definition.
This fixes the link failure caused by disabling the emission of
common symbols.

22 years agoAvoid __func__ string concatenation
peter [Fri, 18 Jan 2002 04:41:23 +0000 (04:41 +0000)]
Avoid __func__ string concatenation

22 years agoAdd:
imp [Fri, 18 Jan 2002 03:49:03 +0000 (03:49 +0000)]
Add:
Accton airDirect WN3301
Melco WLI-CF-S11G
GeoWave GW-NS11S

Submitted by: Shigeru Ishida-san on bsd-nomads (16142,16143,16144)

22 years agoMFi386: revision 1.52: add camcontrol
nyan [Fri, 18 Jan 2002 03:35:51 +0000 (03:35 +0000)]
MFi386: revision 1.52: add camcontrol

22 years agoMFi386: revision 1.64
nyan [Fri, 18 Jan 2002 03:33:52 +0000 (03:33 +0000)]
MFi386: revision 1.64

22 years agoMerged from sys/dev/sio/sio.c revisions from 1.360 to 1.362.
nyan [Fri, 18 Jan 2002 03:30:22 +0000 (03:30 +0000)]
Merged from sys/dev/sio/sio.c revisions from 1.360 to 1.362.

22 years agoMFi386: revision 1.487
nyan [Fri, 18 Jan 2002 03:28:02 +0000 (03:28 +0000)]
MFi386: revision 1.487

22 years agoMFi386: revision 1.388
nyan [Fri, 18 Jan 2002 03:26:38 +0000 (03:26 +0000)]
MFi386: revision 1.388

22 years agoChange "From Id:" to the correct "$Id$" string, as there's no danger
chris [Thu, 17 Jan 2002 23:08:43 +0000 (23:08 +0000)]
Change "From Id:" to the correct "$Id$" string, as there's no danger
of it being expanded.

Sponsored by: DARPA, NAI Labs

22 years agoOops, the previous revision (1.35) broke booting from floppies
iedowse [Thu, 17 Jan 2002 22:39:19 +0000 (22:39 +0000)]
Oops, the previous revision (1.35) broke booting from floppies
because the buffers we use could end up spanning a 64k boundary.
Unfortunately it causes too much bloat (228 -> 72 bytes free) to
just reinstate the old malloc() function.

Instead, define a structure that contains all 4 buffers which must
not cross 64k boundaries. We allocate a 64k-aligned instance in
main() using the magic that was in the old boot2 malloc() function.
This brings the free space down to 168 bytes, but that is still
better than it was before revision 1.35 (136 bytes).

Reported by: Mike Brancato <funnyguy@digitalsmackdown.net>
Pointy-hat to: iedowse

22 years agoRe-add a call to "camcontrol rescan" after insertion of an aic pccard.
joerg [Thu, 17 Jan 2002 20:38:04 +0000 (20:38 +0000)]
Re-add a call to "camcontrol rescan" after insertion of an aic pccard.
We now do it as a "camcontrol rescan all" which is something ken
promised to implement; for the time being it's not worse than the old
"camcontrol rescan $device" which ended up in something like
"camcontrol rescan aic1".  Currently, camcontrol misinterprets the
third non-numeric arg as number 0, and rescans bus 0, which is about
the best we could get at this time.

Approved by: imp
MFC after: 1 week

22 years agoBack out the hack from rev 1.13 that was done to initiate a bus rescan
joerg [Thu, 17 Jan 2002 20:34:58 +0000 (20:34 +0000)]
Back out the hack from rev 1.13 that was done to initiate a bus rescan
at insert time.  When asking gibbs for approval for an MFC, this was
his reply:

1) It leaks memory if it can't allocate a path.

2) It defers allocation of aic->path until the call to scan the
   bus.  This means the path may be NULL when an interrupt occurs
   prior to the call to scan the bus (stray bus reset for instance),
   which will lead to a panic.

3) The driver in current doesn't recover from the failure to allocate
   aic->path.  The driver doesn't check during normal operation if
   the path is NULL, so again a panic will result.

4) aic_cam_rescan calls malloc with M_WAITOK.  aic_cam_rescan is called
   from attach where it isn't necessarily safe to sleep.

5) And most importantly, it co-opts the xpt_periph from the driver level.
   This was never part of the design (xpt_periph used to be static).  Making
   a call of this type may completely confuse the XPT if other XPT operations
   are ongoing.

In the long term, Justin and Warner agreed to implement solution where
CAM itself will initiate the bus rescan if a new bus is added.  For
the time being (and in particular in light of the upcoming 4.5
release), we now have camcontrol available on the boot floppy, and can
have pccardd initiate the rescan through it.