]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoAdd uart(4).
marcel [Tue, 7 Feb 2006 03:25:18 +0000 (03:25 +0000)]
Add uart(4).

MFC after: 3 days

18 years agoUse ps_linfo to retrieve LWP info, current it is used to retrieve
davidxu [Tue, 7 Feb 2006 02:55:34 +0000 (02:55 +0000)]
Use ps_linfo to retrieve LWP info, current it is used to retrieve
signal mask and pending signals.

18 years agoUse ps_linfo to retrieve LWP info.
davidxu [Tue, 7 Feb 2006 02:51:25 +0000 (02:51 +0000)]
Use ps_linfo to retrieve LWP info.

18 years agoIntroduce a new proc service routine ps_linfo() to get LWP info.
davidxu [Tue, 7 Feb 2006 02:29:55 +0000 (02:29 +0000)]
Introduce a new proc service routine ps_linfo() to get LWP info.

18 years agoReplace ptrace syscall with ps_lgetgregs to check a LWP's existence.
davidxu [Tue, 7 Feb 2006 02:12:33 +0000 (02:12 +0000)]
Replace ptrace syscall with ps_lgetgregs to check a LWP's existence.

18 years agoAllocate memory for the MCA state information with M_NOWAIT. We can
marcel [Tue, 7 Feb 2006 02:02:14 +0000 (02:02 +0000)]
Allocate memory for the MCA state information with M_NOWAIT. We can
get a MCA event at any moment and it may not be safe to sleep.

MFC after: 3 days

18 years agoAdd support for audit pipe special devices, which allow user space
rwatson [Mon, 6 Feb 2006 22:50:39 +0000 (22:50 +0000)]
Add support for audit pipe special devices, which allow user space
applications to insert a "tee" in the live audit event stream.  Records
are inserted into a per-clone queue so that user processes can pull
discreet records out of the queue.  Unlike delivery to disk, audit pipes
are "lossy", dropping records in low memory conditions or when the
process falls behind real-time events.  This mechanism is appropriate
for use by live monitoring systems, host-based intrusion detection, etc,
and avoids applications having to dig through active on-disk trails that
are owned by the audit daemon.

Obtained from: TrustedBSD Project

18 years agoAlphabetize.
rwatson [Mon, 6 Feb 2006 22:34:29 +0000 (22:34 +0000)]
Alphabetize.

18 years agoManage audit record memory with the slab allocator, turning
rwatson [Mon, 6 Feb 2006 22:30:54 +0000 (22:30 +0000)]
Manage audit record memory with the slab allocator, turning
initialization routines into a ctor, tear-down to a dtor, cleaning
up, etc.  This will allow audit records to be allocated from
per-cpu caches.

On recent FreeBSD, dropping the audit_mtx around freeing to UMA is
no longer required (at one point it was possible to acquire Giant
on that path), so a mutex-free thread-local drain is no longer
required.

Obtained from: TrustedBSD Project

18 years ago- Call mii_phy_probe() after we allocated an ifp. mii has this evil
cognet [Mon, 6 Feb 2006 22:17:42 +0000 (22:17 +0000)]
- Call mii_phy_probe() after we allocated an ifp. mii has this evil
hack where it assumes the first field of the driver softc is the struct
ifnet, and it copies its value in mii_phy_probe().
- In the interrupt handler, set the mbuf m_len field on packet receive.

18 years agoRegenerate.
rwatson [Mon, 6 Feb 2006 22:15:00 +0000 (22:15 +0000)]
Regenerate.

18 years agoAssign audit event identifiers to ibcs2 system calls.
rwatson [Mon, 6 Feb 2006 22:14:50 +0000 (22:14 +0000)]
Assign audit event identifiers to ibcs2 system calls.

Obtained from: TrustedBSD Project

18 years ago- Always call exec_free_args() in kern_execve() instead of doing it in all
jhb [Mon, 6 Feb 2006 22:06:54 +0000 (22:06 +0000)]
- Always call exec_free_args() in kern_execve() instead of doing it in all
  the callers if the exec either succeeds or fails early.
- Move the code to call exit1() if the exec fails after the vmspace is
  gone to the bottom of kern_execve() to cut down on some code duplication.

18 years agoAdd a kern_eaccess() function and use it to implement xenix_eaccess()
jhb [Mon, 6 Feb 2006 22:00:53 +0000 (22:00 +0000)]
Add a kern_eaccess() function and use it to implement xenix_eaccess()
rather than kern_access().

Suggested by: rwatson

18 years ago- Move the wakeup() for exiting kthreads out of exit1() and into
jhb [Mon, 6 Feb 2006 21:56:13 +0000 (21:56 +0000)]
- Move the wakeup() for exiting kthreads out of exit1() and into
  kthread_exit() as that is cleaner and less obscured.  It also does the
  wakeup sooner.
- Add some comments to kthread_exit().

18 years agoWe don't need the proc lock to check P_KTHREAD on curthread since it is
jhb [Mon, 6 Feb 2006 21:54:47 +0000 (21:54 +0000)]
We don't need the proc lock to check P_KTHREAD on curthread since it is
only set before the kthread starts executing and is never cleared.

18 years agoClarify and expand on some of the points about audit pipe devices.
rwatson [Mon, 6 Feb 2006 20:27:00 +0000 (20:27 +0000)]
Clarify and expand on some of the points about audit pipe devices.

Discussed with: remko

18 years agoUpdate man page for some booting and settings stuff.
mjacob [Mon, 6 Feb 2006 19:28:46 +0000 (19:28 +0000)]
Update man page for some booting and settings stuff.

Remove a bunch of cards from vendors who are long since defunct.
Add a note about 2322 support.

Obtained from: Marcus

18 years agoAdd a missing word and use the .Qq macro for quotes.
brueffer [Mon, 6 Feb 2006 19:28:02 +0000 (19:28 +0000)]
Add a missing word and use the .Qq macro for quotes.

18 years agoCorrect RFC for NTP.
jcamou [Mon, 6 Feb 2006 19:22:34 +0000 (19:22 +0000)]
Correct RFC for NTP.

PR: docs/92629
Submitted by: Daniel Gerzo <danger@rulez.sk>
Noticed by: Michal F. Hanula <f@7f000001.org>
Approved by: trhodes (mentor)

18 years agoUnbreak DMA dumo on Intel 31224.
sos [Mon, 6 Feb 2006 19:17:48 +0000 (19:17 +0000)]
Unbreak DMA dumo on Intel 31224.

18 years agoAdd information on audit pipe special devices, which allow user processes
rwatson [Mon, 6 Feb 2006 18:41:00 +0000 (18:41 +0000)]
Add information on audit pipe special devices, which allow user processes
to "tee" the BSM record stream for the purposes of live monitoring,
intrusion detection, etc.  Support for audit pipes will be committed in
the near future.

Obtained from: TrustedBSD Project

18 years agoUse memory clobbers, to be on the safe side.
cognet [Mon, 6 Feb 2006 18:29:05 +0000 (18:29 +0000)]
Use memory clobbers, to be on the safe side.
Suggested by: jhb

18 years agorwlock expects the struct thread to be aligned on 8 bytes, so make sure
cognet [Mon, 6 Feb 2006 16:03:10 +0000 (16:03 +0000)]
rwlock expects the struct thread to be aligned on 8 bytes, so make sure
thread0 is.

18 years agoTwo fixes:
ru [Mon, 6 Feb 2006 14:30:21 +0000 (14:30 +0000)]
Two fixes:

- Run send queue down to completion, not just one packet.
  It has been observed to cause a stall queue otherwise.

- Prevent queueing multiple function calls to a node.

MFC after: 3 days

18 years agoAlways clear thread info buffer to zero.
davidxu [Mon, 6 Feb 2006 11:54:19 +0000 (11:54 +0000)]
Always clear thread info buffer to zero.

18 years ago - Add a ref count to the mount structure. Sleep for up to 3 seconds in
jeff [Mon, 6 Feb 2006 10:19:50 +0000 (10:19 +0000)]
 - Add a ref count to the mount structure.  Sleep for up to 3 seconds in
   vfs_mount_destroy waiting for this ref to hit 0.  We don't print an
   error if we are rebooting as the root mount always retains some refernces
   by init proc.
 - Acquire a mnt ref for every vnode allocated to a mount point.  Drop this
   ref only once vdestroy() has been called and the mount has been freed.
 - No longer NULL the v_mount pointer in delmntque() so that we may release
   the ref after vgone() has been called.  This allows us to guarantee
   that the mount point structure will be valid until the last vnode has
   lost its last ref.
 - Fix a few places that rely on checking v_mount to detect recycling.

Sponsored by: Isilon Systems, Inc.
MFC After: 1 week

18 years ago - Don't check v_mount for NULL to determine if a vnode has been recycled.
jeff [Mon, 6 Feb 2006 10:15:27 +0000 (10:15 +0000)]
 - Don't check v_mount for NULL to determine if a vnode has been recycled.
   Use the more appropriate VI_DOOMED flag instead.

Sponsored by: Isilon Systems, Inc.
MFC After: 1 week

18 years ago - Fix silly VI locking that is used to check a single flag. The vnode
jeff [Mon, 6 Feb 2006 10:14:12 +0000 (10:14 +0000)]
 - Fix silly VI locking that is used to check a single flag.  The vnode
   lock also protects this flag so it is not necessary.
 - Don't rely on v_mount to detect whether or not we've been recycled, use
   the more appropriate VI_DOOMED instead.

Sponsored by: Isilon Systems, Inc.
MFC After: 1 week

18 years ago - Add the global 'rebooting' variable that is used to detect when
jeff [Mon, 6 Feb 2006 10:12:00 +0000 (10:12 +0000)]
 - Add the global 'rebooting' variable that is used to detect when
   boot() has been called.

Sponsored by: Isilon Systems, Inc.
MFC After: 1 week

18 years ago - Remove ifdef disabled code that doesn't have a chance of working anymore.
jeff [Mon, 6 Feb 2006 10:10:42 +0000 (10:10 +0000)]
 - Remove ifdef disabled code that doesn't have a chance of working anymore.

18 years agoAdd members pl_sigmask and pl_siglist into ptrace_lwpinfo to get lwp's
davidxu [Mon, 6 Feb 2006 09:41:56 +0000 (09:41 +0000)]
Add members pl_sigmask and pl_siglist into ptrace_lwpinfo to get lwp's
signal mask and pending signals.

18 years agoAttempt to fix DHCP address acquisition which was broken by rev. 1.133.
delphij [Mon, 6 Feb 2006 06:49:57 +0000 (06:49 +0000)]
Attempt to fix DHCP address acquisition which was broken by rev. 1.133.

MFC After: 3 days

18 years ago1. Eliminate compile warnings.
davidxu [Mon, 6 Feb 2006 06:07:56 +0000 (06:07 +0000)]
1. Eliminate compile warnings.
2. Add command 'thread signal' to print out current thread's signal mask
   and pending signals.

18 years agoRegenerate.
rwatson [Mon, 6 Feb 2006 02:00:32 +0000 (02:00 +0000)]
Regenerate.

18 years agoPrefer AUE_FOO audit identifiers to AUE_O_FOO, which are largely left
rwatson [Mon, 6 Feb 2006 02:00:06 +0000 (02:00 +0000)]
Prefer AUE_FOO audit identifiers to AUE_O_FOO, which are largely left
over from the Darwin implementation.

When we implement a system call as a wrapper to sysctl(), audit it as
AUE_SYSCTL.  This leads to greater compatibility with Solaris audit
trails as sysctl() argument tokens are not the same as the ones for
the originaly system calls (i.e., setdomainname()).

Replace references to AUE_ events that are equivilent to AUE_NULL with
AUE_NULL.  In the case of process signal configuration, this is
because these events do not require auditing.

Move from the Darwin spelling of getsockopt() to the FreeBSD/Solaris
one.

Audit nmount().

Obtained from: TrustedBSD Project

18 years agoWhen exiting a thread, submit any pending record. Today, we don't
rwatson [Mon, 6 Feb 2006 01:51:08 +0000 (01:51 +0000)]
When exiting a thread, submit any pending record.  Today, we don't
audit thread exit, but should that happen, this will prevent
unhappiness, as the thread exit system call will never return, and
hence not commit the record.

Pointed out by/with: cognet
Obtained from: TrustedBSD Project

18 years agoRegenerate.
rwatson [Mon, 6 Feb 2006 01:40:48 +0000 (01:40 +0000)]
Regenerate.

18 years agoAssign audit event identifiers to Linux i386 system calls.
rwatson [Mon, 6 Feb 2006 01:40:30 +0000 (01:40 +0000)]
Assign audit event identifiers to Linux i386 system calls.

Obtained from: TrustedBSD Project

18 years agoRegenerate.
rwatson [Mon, 6 Feb 2006 01:16:00 +0000 (01:16 +0000)]
Regenerate.

18 years agoReflect fix in Linux setfsgid() event name from OpenBSM in the alpha
rwatson [Mon, 6 Feb 2006 01:13:47 +0000 (01:13 +0000)]
Reflect fix in Linux setfsgid() event name from OpenBSM in the alpha
linux system call table.

Obtained from: TrustedBSD Project

18 years agoMerge OpenBSM 1.0 alpha 3 include file changes from contrib/openbsm/bsm
rwatson [Mon, 6 Feb 2006 01:12:46 +0000 (01:12 +0000)]
Merge OpenBSM 1.0 alpha 3 include file changes from contrib/openbsm/bsm
to sys/bsm:

- Correct error in definition of audit event for Linux setfsgid().
- Add audit event identifier for sysarch().

Obtained from: TrustedBSD Project

18 years agoAudit the arguments (user/group IDs) for the system calls that set these IDs.
wsalamon [Mon, 6 Feb 2006 00:32:33 +0000 (00:32 +0000)]
Audit the arguments (user/group IDs) for the system calls that set these IDs.

Obtained from: TrustedBSD Project
Approved by: rwatson (mentor)

18 years agoAudit the args to rfork(), and the child PID for all fork system calls.
wsalamon [Mon, 6 Feb 2006 00:28:50 +0000 (00:28 +0000)]
Audit the args to rfork(), and the child PID for all fork system calls.

Obtained from: TrustedBSD Project
Approved by: rwatson (mentor)

18 years agoAudit the pid being requested in wait4().
wsalamon [Mon, 6 Feb 2006 00:19:09 +0000 (00:19 +0000)]
Audit the pid being requested in wait4().

Obtained from: TrustedBSD Project
Approved by: rwatson (mentor)

18 years agoAnother update for reality: "dangerously dedicated" mode is now
ceri [Mon, 6 Feb 2006 00:06:39 +0000 (00:06 +0000)]
Another update for reality:  "dangerously dedicated" mode is now
achieved by hitting 'F', which is no longer undocumented.

PR: bin/92533
Submitted by: Rudolf Cejka <cejkar at fit.vutbr dot cz>

18 years agoThis commit was generated by cvs2svn to compensate for changes in r155364,
rwatson [Mon, 6 Feb 2006 00:06:04 +0000 (00:06 +0000)]
This commit was generated by cvs2svn to compensate for changes in r155364,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor branch import of OpenBSM 1.0 alpha 3:
rwatson [Mon, 6 Feb 2006 00:06:04 +0000 (00:06 +0000)]
Vendor branch import of OpenBSM 1.0 alpha 3:

- Man page formatting, cross reference, mlinks, and accuracy improvements.
- auditd and tools now compile and run on FreeBSD/arm.
- auditd will now fchown() the trail file to the audit review group, if
  defined at compile-time.
- Added AUE_SYSARCH for FreeBSD.
- Definition of AUE_SETFSGID fixed for Linux.

Many thanks to: brueffer, cognet
Obtained from: TrustedBSD Project

18 years agoCorrect typo in sample CVS import line: must specify full path to the
rwatson [Mon, 6 Feb 2006 00:03:39 +0000 (00:03 +0000)]
Correct typo in sample CVS import line: must specify full path to the
CVS repository on repoman.

18 years agoAdd auditing of arguments to the close() and fstat() system calls. Much more
wsalamon [Sun, 5 Feb 2006 23:57:32 +0000 (23:57 +0000)]
Add auditing of arguments to the close() and fstat() system calls. Much more
argument auditing yet to come, for remaining system calls in this file.

Obtained from: TrustedBSD Project
Approved by: rwatson (mentor)

18 years agoRegenerate.
rwatson [Sun, 5 Feb 2006 23:28:46 +0000 (23:28 +0000)]
Regenerate.

18 years agoAssign audit event identfiers to Xenix system calls. Note: AUE_EACCESS
rwatson [Sun, 5 Feb 2006 23:28:01 +0000 (23:28 +0000)]
Assign audit event identfiers to Xenix system calls.  Note: AUE_EACCESS
is assigned to xenix_eaccess() instead of AUE_ACCESS, as that is the
intended meaning of the system call.  xenix_eaccess() should be
reimplemented using our native eaccess() implementation so that it
works as intended.

Obtained from: TrustedBSD Project

18 years agoCorrect help line: list targets, not names of files generated by targets
rwatson [Sun, 5 Feb 2006 23:25:19 +0000 (23:25 +0000)]
Correct help line: list targets, not names of files generated by targets
when no argument is provided to make.

MFC after: 1 week

18 years agoRegenerate (accidentally also committed in commit that updated
rwatson [Sun, 5 Feb 2006 23:16:20 +0000 (23:16 +0000)]
Regenerate (accidentally also committed in commit that updated
syscalls.isc).

18 years agoAssign audit event identifiers to ibcs2 ISC system calls.
rwatson [Sun, 5 Feb 2006 23:15:22 +0000 (23:15 +0000)]
Assign audit event identifiers to ibcs2 ISC system calls.

Obtained from: TrustedBSD Project

18 years agopflog is a separate module now.
mlaier [Sun, 5 Feb 2006 22:38:08 +0000 (22:38 +0000)]
pflog is a separate module now.

Submitted by: Antoine Brodin
PR: kern/88271
MFC after: 1 week

18 years agoBackout rev 1.12. It would have been a good thing, if gcc was smart enough
cognet [Sun, 5 Feb 2006 22:06:12 +0000 (22:06 +0000)]
Backout rev 1.12. It would have been a good thing, if gcc was smart enough
not to generate bad code.

18 years agoOn process exit, audit the return value of the process, and commit the
rwatson [Sun, 5 Feb 2006 21:08:25 +0000 (21:08 +0000)]
On process exit, audit the return value of the process, and commit the
record immediately, as this system call never returns.

Obtained from: TrustedBSD Project

18 years agoWhen GC'ing a thread, assert that it has no active audit record.
rwatson [Sun, 5 Feb 2006 21:06:09 +0000 (21:06 +0000)]
When GC'ing a thread, assert that it has no active audit record.
This should not happen, but with this assert, brueffer and I would
not have spent 45 minutes trying to figure out why he wasn't
seeing audit records with the audit version in CVS.

Obtained from: TrustedBSD Project

18 years agoAdd some new release notes for 2006 security advisories:
bmah [Sun, 5 Feb 2006 20:54:53 +0000 (20:54 +0000)]
Add some new release notes for 2006 security advisories:
SA-06:{01,02,03,04,05,06,07}.

(SA-06:08 does not appear to apply to CURRENT.)

Bump copyright date while here.

18 years agoAssign gid 77 to audit instead of gid 73. The ports group list did not
rwatson [Sun, 5 Feb 2006 19:34:09 +0000 (19:34 +0000)]
Assign gid 77 to audit instead of gid 73.  The ports group list did not
include '73', which was assigned in a ports passwd entry to ircservices.

Pointed out by: ceri

18 years agoThe rpc.pcnfsd server was in the base for a little over seven minutes
ceri [Sun, 5 Feb 2006 19:23:05 +0000 (19:23 +0000)]
The rpc.pcnfsd server was in the base for a little over seven minutes
back in 1994.  Change the example entry to point at the port, as per
the entries for uucpd et al.

18 years agoCorrect a user-visible typo.
ceri [Sun, 5 Feb 2006 18:57:01 +0000 (18:57 +0000)]
Correct a user-visible typo.

18 years agoGratuitous "typo in comment" fix.
ceri [Sun, 5 Feb 2006 18:49:52 +0000 (18:49 +0000)]
Gratuitous "typo in comment" fix.

18 years agoBring these files somewhat into the present.
ceri [Sun, 5 Feb 2006 18:46:46 +0000 (18:46 +0000)]
Bring these files somewhat into the present.
Perform some rewording while here.

Remove register.hlp, since the code that deals with it was removed
nearly 7 years ago.

18 years agoChange group for /var/audit to audit, so that audit review can be
rwatson [Sun, 5 Feb 2006 18:22:56 +0000 (18:22 +0000)]
Change group for /var/audit to audit, so that audit review can be
delegated to non-administrators.

Obtained from: TrustedBSD Project

18 years agoAllocate an 'audit' group, membership in which will grant the audit
rwatson [Sun, 5 Feb 2006 18:04:39 +0000 (18:04 +0000)]
Allocate an 'audit' group, membership in which will grant the audit
review right by virtue of read file permission on /var/audit and its
contents.

Obtained from: TrustedBSD Project

18 years agoConvert NULL checks into KASSERT (and move them before the first
netchild [Sun, 5 Feb 2006 17:47:26 +0000 (17:47 +0000)]
Convert NULL checks into KASSERT (and move them before the first
dereferencing) since a NULL value would be a bug here.

Note: Both affected functions look very similar. A refactoring may
be beneficial.

CID: 483, 485
Found with: Coverity Prevent(tm)
Discussed with: ariff
MFC after: 5 days

18 years agoan driver not endian clean, so don't enable on sparc64
imp [Sun, 5 Feb 2006 17:38:28 +0000 (17:38 +0000)]
an driver not endian clean, so don't enable on sparc64

18 years agoforced commit: previous commit was CID 429
netchild [Sun, 5 Feb 2006 17:35:13 +0000 (17:35 +0000)]
forced commit: previous commit was CID 429

18 years agoforced commit: previous commit was CID 483
netchild [Sun, 5 Feb 2006 17:34:13 +0000 (17:34 +0000)]
forced commit: previous commit was CID 483

18 years agoA pointer was checked for NULL after dereferencing it. The check is not
netchild [Sun, 5 Feb 2006 17:33:18 +0000 (17:33 +0000)]
A pointer was checked for NULL after dereferencing it. The check is not
needed here, except there's a bug which results in detaching the device
twice.

Move the NULL pointer check to the beginning of the function and convert
it into a KASSERT.

CID: 420
Found with: Coverity Prevent(tm)
Discussed with: ariff
MFC after: 5 days

18 years agoMake pflog a seperate module. As a result pflog_packet() becomes a function
mlaier [Sun, 5 Feb 2006 17:17:32 +0000 (17:17 +0000)]
Make pflog a seperate module.  As a result pflog_packet() becomes a function
pointer that is declared in pf_ioctl.c

Requested by: yar (as part of the module build reorg)
MFC after: 1 week
X-MFC with: yar's module reorg

18 years agoFix memory leak in some failure cases.
netchild [Sun, 5 Feb 2006 17:10:52 +0000 (17:10 +0000)]
Fix memory leak in some failure cases.

CID: 420
Found with: Coverity Prevent(tm)

18 years agoAdd AUDITVNODE[12] flags to namei(), which cause namei() to audit path
rwatson [Sun, 5 Feb 2006 15:42:01 +0000 (15:42 +0000)]
Add AUDITVNODE[12] flags to namei(), which cause namei() to audit path
and vnode attribute information for looked up vnodes during the lookup
operation.  This will allow consumers of namei() to specify that this
information be added to the in-process audit record.

Submitted by: wsalamon
Obtained from: TrustedBSD Project

18 years agoshut up strict-aliasing rules warning.
ume [Sun, 5 Feb 2006 09:52:40 +0000 (09:52 +0000)]
shut up strict-aliasing rules warning.

18 years agoMove asr driver from global NOTES to i386-specific NOTES. Requestor
kensmith [Sun, 5 Feb 2006 05:06:04 +0000 (05:06 +0000)]
Move asr driver from global NOTES to i386-specific NOTES.  Requestor
reports it is neither endian-clean or 64-bit clean.  :-)

Requested by: scottl

18 years agoSquash a couple more invalid long casts.
scottl [Sun, 5 Feb 2006 05:05:27 +0000 (05:05 +0000)]
Squash a couple more invalid long casts.

18 years agoNow, thread name is stored in kernel, userland no longer has to keep it.
davidxu [Sun, 5 Feb 2006 03:04:54 +0000 (03:04 +0000)]
Now, thread name is stored in kernel, userland no longer has to keep it.

18 years agouse syscall thr_set_name to implement pthread_set_name_np.
davidxu [Sun, 5 Feb 2006 02:26:17 +0000 (02:26 +0000)]
use syscall thr_set_name to implement pthread_set_name_np.

18 years agoRegenerate.
davidxu [Sun, 5 Feb 2006 02:23:41 +0000 (02:23 +0000)]
Regenerate.

18 years agoImplement thr_set_name to set a name for thread.
davidxu [Sun, 5 Feb 2006 02:18:46 +0000 (02:18 +0000)]
Implement thr_set_name to set a name for thread.

Reviewed by: julian

18 years agoImport support for the Atmel AT91RM9200 CPU/Microcontroller. This SoC
imp [Sat, 4 Feb 2006 23:32:13 +0000 (23:32 +0000)]
Import support for the Atmel AT91RM9200 CPU/Microcontroller.  This SoC
is a ARM920T based CPU with a bunch of built-in peripherals.  The
inital import supports the SPI bus, the TWI bus (although iicbus
integration is not complete), the uarts, the system timer and the
onboard ethernet.  Support for the Kwikbyte KB9202
(http://www.kwikbyte.com) board is also included, although there's no
reason why the 9200 and the 9201 wouldn't also work.  Primitive
support for running under the skyeye emulator is also provided
(although skyeye's support for the AT91RM9200 is a little weak).

The code has been structured so that other members of Atmel's arm family can
be supported in the future.  The AT91SAM9260 is not presently supported
due to lack of hardware.  The arm7tdmi families are also not supported
becasue they lack an MMU.

Many thanks to cognet@ for his help and assistance in bringing up this
board.  He did much of the vm work and wrote parts of the uart and
system timer code as well as the bus space implementation.

The system boots to single user w/o problem, although the serial
console is a little slow and the ethernet driver is still in flux.

This work was sponsored by Timing Solutions, Corporation.  I am
grateful to their support of the FreeBSD project in this manner.

18 years agoEnable getty(8) on ttyu2 by default in order to get machines that use a
marius [Sat, 4 Feb 2006 23:30:09 +0000 (23:30 +0000)]
Enable getty(8) on ttyu2 by default in order to get machines that use a
RSC (Remote System Control) connected via uart2 as console working out
of the box. On machines that use uart2 to connect a keyboard and thus
the ttyu2 node doesn't exist this will trigger a warning from getty(8)
but cause no real harm.

MFC after: 1 week

18 years ago- Add support for using LOM (Lights Out Management) and RSC (Remote System
marius [Sat, 4 Feb 2006 23:27:16 +0000 (23:27 +0000)]
- Add support for using LOM (Lights Out Management) and RSC (Remote System
  Control) devices as console. These are microcontrollers which are either
  on-board or part of an add-on card and provide terminal server, remote
  power switch and monitoring functionality. For console usage these are
  connected to the rest of the system via a SCC or an UART. This commit adds
  support for the following variants (corresponds to what 'input-device' and
  'output-device' have to be set to):
  rsc found on-board in E250 and supposedly some Netra, connected
via a SAB82532, com. parameters can be determined via OFW
  rsc-console RSC card found in E280R, Fire V4x0, Fire V8x0, connected
via a NS16550, hardwired to 115200 8N1
  lom-console LOMlite2 card found in Netra 20/T4, connected via a NS16550,
hardwired to 9600 8N1

- Add my copyright to uart_cpu_sparc64.c as I've rewritten about one third
  of that file over time.

Tested on: E250, E280R
Thanks to: dwhite@ for providing access to an E280R
OK'ed by: marcel
MFC after: 1 week

18 years agoSilence the strict-alias warnings. Make a trip through (void *) when
imp [Sat, 4 Feb 2006 22:51:03 +0000 (22:51 +0000)]
Silence the strict-alias warnings.  Make a trip through (void *) when
casting a structure to a uint32_t *.  Many drivers in the tree do this, but
I'll not update them until these changes can be reviewed by the pedantic
standards folks.

18 years agoRemove an unnecessary call to pmap_remove_all(). The given page is not
alc [Sat, 4 Feb 2006 22:37:10 +0000 (22:37 +0000)]
Remove an unnecessary call to pmap_remove_all().  The given page is not
mapped because its contents are invalid.

18 years agoType of overrun_buf doesn't matter to this code, but does to gcc.
imp [Sat, 4 Feb 2006 22:33:08 +0000 (22:33 +0000)]
Type of overrun_buf doesn't matter to this code, but does to gcc.
Make it a void *.

18 years agoPerform minor rewording and grammatical improvement. Add a missing Xr.
ceri [Sat, 4 Feb 2006 22:17:38 +0000 (22:17 +0000)]
Perform minor rewording and grammatical improvement.  Add a missing Xr.

18 years agoUse void * for pointer rather than u_int8_t *, since it doesn't
imp [Sat, 4 Feb 2006 22:14:20 +0000 (22:14 +0000)]
Use void * for pointer rather than u_int8_t *, since it doesn't
matter.  Well it does for type punning warnings.

18 years agoNow that TUNABLE_ULONG is in the tree, and has been for a while, use it.
imp [Sat, 4 Feb 2006 21:56:14 +0000 (21:56 +0000)]
Now that TUNABLE_ULONG is in the tree, and has been for a while, use it.
This also avoids typepunning.

18 years agoDon't type pun accidentally. Instead, be explicit that we're type
imp [Sat, 4 Feb 2006 21:37:39 +0000 (21:37 +0000)]
Don't type pun accidentally.  Instead, be explicit that we're type
punning with an union so that the compiler knows.

18 years agoRemove reference to non-existant manpage.
brueffer [Sat, 4 Feb 2006 21:10:48 +0000 (21:10 +0000)]
Remove reference to non-existant manpage.

Confirmed by: rwatson

18 years agoCall the audit syscall enter/exit functions for the amd64 architecture,
wsalamon [Sat, 4 Feb 2006 20:37:20 +0000 (20:37 +0000)]
Call the audit syscall enter/exit functions for the amd64 architecture,
both 32-bit and 64-bit paths. System calls will now be audited.

Obtained from: TrustedBSD Project
Approved by: rwatson (mentor)

18 years agoMake login audit-enabled, submitting audit records for the login and logout
wsalamon [Sat, 4 Feb 2006 20:20:02 +0000 (20:20 +0000)]
Make login audit-enabled, submitting audit records for the login and logout
events. The specifics of submitting the records is contained within
login_audit.c.
Document the auditing behavior in the man page.

Obtained from: TrustedBSD Project, Apple Computer, Inc.
Approved by: rwatson (mentor)

18 years agoWhen ndis_attach() runs, it has to very briefly initialize the card
wpaul [Sat, 4 Feb 2006 19:42:49 +0000 (19:42 +0000)]
When ndis_attach() runs, it has to very briefly initialize the card
in order to query the underlying Windows driver for the station address
and some other properties. There is a slim chance that the card may
receive a packet and indicate it up to us before ndis_attach() can call
ndis_halt_nic(). This is bad, because both the softc structure and
the ifnet structure aren't fully initialized yet: many pointers are
still NULL, so if we make it into ndis_rxeof(), we will panic.

To fix this, we need to do the following:

- Move the calls to IoAllocateWorkItem() to before the call to ndis_init_nic().
- Move the initialization of the RX DPC and status callback function pointers
  to before ndis_init_nic() as well.
- Modify ndis_rxeof() to check if the IFF_DRV_RUNNING flag is set. If it
  isn't, we return any supplied NDIS_PACKETs to the NIC without processing
  them.

This fixes a crash than can occur when activating a wireless NIC in
close proximity to a very busy wireless network, reported by Ryan
Beasley (ryan%^$!ATgoddamnbastard-****!!!DOTorg.

MFC after: 3 days

18 years agoAdd a brief FREEBSD-upgrade file to provide direction on how to perform
rwatson [Sat, 4 Feb 2006 18:29:51 +0000 (18:29 +0000)]
Add a brief FREEBSD-upgrade file to provide direction on how to perform
OpenBSM upgrades.  Right now, this is very easy, but in the future it
will probably become more complicated.

18 years agoAdd a -A argument to mergemaster to allow explicitly specifying an
rwatson [Sat, 4 Feb 2006 18:24:06 +0000 (18:24 +0000)]
Add a -A argument to mergemaster to allow explicitly specifying an
architecture to pass through to the underlying makefiles.  This is
quite useful when building on an i386 box to populate an amd64 NFS
root.

Head nod: dougb
MFC after: 1 week

18 years agoDon't forget to set the address of the next descriptor to 0 when we're
cognet [Sat, 4 Feb 2006 18:01:15 +0000 (18:01 +0000)]
Don't forget to set the address of the next descriptor to 0 when we're
zeroing a physical page, or we could end up re-zeroing portions of
memory we have zeroed before, which is clearly not wanted.

18 years agoSquash another use of vtophys. Instead of creating separate busdma objects
scottl [Sat, 4 Feb 2006 17:56:17 +0000 (17:56 +0000)]
Squash another use of vtophys.  Instead of creating separate busdma objects
for doing static memory transfers, start collecting them into a single
object.