]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoDeny the SIZE command on large files when in ASCII mode.
Yaroslav Tykhiy [Wed, 31 Jul 2002 10:55:31 +0000 (10:55 +0000)]
Deny the SIZE command on large files when in ASCII mode.
This eliminates an opportunity for DoS attack.

Pointed out by: maxim
Inspired by: lukemftpd, OpenBSD
MFC after: 2 weeks

21 years agoIt should be "ucom", not "usio".
Josef Karthauser [Wed, 31 Jul 2002 10:52:46 +0000 (10:52 +0000)]
It should be "ucom", not "usio".

21 years agoRegen
Josef Karthauser [Wed, 31 Jul 2002 10:05:58 +0000 (10:05 +0000)]
Regen

21 years agoCorrect URL to the Handbook
Marc Fonvieille [Wed, 31 Jul 2002 10:05:37 +0000 (10:05 +0000)]
Correct URL to the Handbook

MFC after: 1 week

21 years agoMFNetBSD: uvisor.c (1.14, 1.15)
Josef Karthauser [Wed, 31 Jul 2002 10:05:26 +0000 (10:05 +0000)]
MFNetBSD: uvisor.c (1.14, 1.15)

Add support for Palm (M500, M505, M125) and Sony devices (Clie 4.0 and 4.1).

21 years agoCorrect URL to the FAQ
Marc Fonvieille [Wed, 31 Jul 2002 10:05:00 +0000 (10:05 +0000)]
Correct URL to the FAQ

MFC after: 1 week

21 years agoCorrect a few grammar bogons.
Jens Schweikhardt [Wed, 31 Jul 2002 09:32:22 +0000 (09:32 +0000)]
Correct a few grammar bogons.
MFC after: 3 days

21 years agoRemove trailing white spaces
Josef Karthauser [Wed, 31 Jul 2002 09:24:41 +0000 (09:24 +0000)]
Remove trailing white spaces

Obtained from: NetBSD (1.16)

21 years agoRegen
Josef Karthauser [Wed, 31 Jul 2002 09:09:50 +0000 (09:09 +0000)]
Regen

21 years agoAdd the HP ScanJet 2200C.
Josef Karthauser [Wed, 31 Jul 2002 09:07:56 +0000 (09:07 +0000)]
Add the HP ScanJet 2200C.

21 years agoEnable the HP 2200C and Visioneer Scanport 3000 scanners, and remove
Josef Karthauser [Wed, 31 Jul 2002 09:05:52 +0000 (09:05 +0000)]
Enable the HP 2200C and Visioneer Scanport 3000 scanners, and remove
some trailing spaces whilst I'm here.

Obtained from: NetBSD.

21 years agoAdd the Primax (Visioneer) 6200 scanner.
Josef Karthauser [Wed, 31 Jul 2002 08:51:32 +0000 (08:51 +0000)]
Add the Primax (Visioneer) 6200 scanner.

Obtained from: OpenBSD (via NetBSD)

21 years ago o Lock page accesses by vm_page_io_start() with the page queues lock.
Alan Cox [Wed, 31 Jul 2002 07:27:08 +0000 (07:27 +0000)]
 o Lock page accesses by vm_page_io_start() with the page queues lock.
 o Assert that the page queues lock is held in vm_page_io_start().

21 years agoI forgot this bit of uglyness in the fsck_ffs cleanup.
Poul-Henning Kamp [Wed, 31 Jul 2002 07:01:18 +0000 (07:01 +0000)]
I forgot this bit of uglyness in the fsck_ffs cleanup.

21 years agoGrammar (`was' -> `were')
Chris Costello [Wed, 31 Jul 2002 06:40:34 +0000 (06:40 +0000)]
Grammar (`was' -> `were')

21 years agoNew release notes: uaudio, uvisor, gpt, mca.
Bruce A. Mah [Wed, 31 Jul 2002 04:01:12 +0000 (04:01 +0000)]
New release notes:  uaudio, uvisor, gpt, mca.

21 years ago- Define NO_CPU_CFLAGS during BMAKE and TMAKE (and thus XMAKE) so that
John Baldwin [Wed, 31 Jul 2002 03:56:03 +0000 (03:56 +0000)]
- Define NO_CPU_CFLAGS during BMAKE and TMAKE (and thus XMAKE) so that
  bsd.cpu.mk doesn't have to worry about compilers other than the current
  version.
- Allow TARGET_CPUTYPE to override CPUTYPE in bsd.cpu.mk.
- Treat an empty CPUTYPE the same as an undefined CPUTYPE.
- For buildworld, buildkernel, etc., define TARGET_CPUTYPE to CPUTYPE for
  native builds and define it to be empty for cross-builds.
  TARGET_CPUTYPE is only defined if it is not already defined via the
  commandline or environment.

21 years agoFix markup nits and a (not related) typo.
Bruce A. Mah [Wed, 31 Jul 2002 03:25:25 +0000 (03:25 +0000)]
Fix markup nits and a (not related) typo.

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 03:03:22 +0000 (03:03 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Invoke the necessary MAC entry points to maintain labels on sockets.
In particular, invoke entry points during socket allocation and
destruction, as well as creation by a process or during an
accept-scenario (sonewconn).  For UNIX domain sockets, also assign
a peer label.  As the socket code isn't locked down yet, locking
interactions are not yet clear.  Various protocol stack socket
operations (such as peer label assignment for IPv4) will follow.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoUpdated release notes: ACPI CA 20020725, OpenSSL 0.9.6e (+MFC).
Bruce A. Mah [Wed, 31 Jul 2002 02:48:20 +0000 (02:48 +0000)]
Updated release notes:  ACPI CA 20020725, OpenSSL 0.9.6e (+MFC).

21 years agoNote that the privilege indicating flag to vaccess() originally used
Robert Watson [Wed, 31 Jul 2002 02:05:12 +0000 (02:05 +0000)]
Note that the privilege indicating flag  to vaccess() originally used
by the process accounting system is now deprecated.

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 02:03:46 +0000 (02:03 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Invoke the necessary MAC entry points to maintain labels on vnodes.
In particular, initialize the label when the vnode is allocated or
reused, and destroy the label when the vnode is going to be released,
or reused.  Wow, an object where there really is exactly one place
where it's allocated, and one other where it's freed.  Amazing.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 01:51:34 +0000 (01:51 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Invoke additional MAC entry points when an mbuf packet header is
copied to another mbuf: release the old label if any, reinitialize
the new header, and ask the MAC framework to copy the header label
data.  Note that this requires a potential allocation operation,
but m_copy_pkthdr() is not permitted to fail, so we must block.
Since we now use interrupt threads, this is possible, but not
desirable.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 01:42:19 +0000 (01:42 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Invoke the necessary MAC entry points to maintain labels on header
mbufs.  In particular, invoke entry points during the two mbuf
header allocation cases, and the mbuf freeing case.  Pass the "how"
argument at allocation time to the MAC framework so that it can
determine if it is permitted to block (as with policy modules),
and permit the initialization entry point to fail if it needs to
allocate memory but is not permitted to, failing the mbuf
allocation.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 01:27:33 +0000 (01:27 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Implement MAC framework access control entry points relating to
operations on mountpoints.  Currently, this consists only of
access control on mountpoint listing using the various statfs()
variations.  In the future, it might also be desirable to
implement checks on mount() and unmount().

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 01:11:29 +0000 (01:11 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Invoke the necessary MAC entry points to maintain labels on
mount structures.  In particular, invoke entry points for
intialization and destruction in various scenarios (root,
non-root).  Also introduce an entry point in the boot procedure
following the mount of the root file system, but prior to the
start of the userland init process to permit policies to
perform further initialization.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 00:48:24 +0000 (00:48 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Implement inter-process access control entry points for the MAC
framework.  This permits policy modules to augment the decision
making process for process and socket visibility, process debugging,
re-scheduling, and signaling.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoFix a problem with sendfile() syscall by always doing I/O via bread() in
Semen Ustimenko [Wed, 31 Jul 2002 00:42:57 +0000 (00:42 +0000)]
Fix a problem with sendfile() syscall by always doing I/O via bread() in
ntfs_read(). This guarantee that requested cache pages will be valid if
UIO_NOCOPY specifed.

PR: bin/34072, bin/36189
MFC after: 1 week

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Wed, 31 Jul 2002 00:39:19 +0000 (00:39 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Invoke the necessary MAC entry points to maintain labels on
process credentials.  In particular, invoke entry points for
the initialization and destruction of struct ucred, the copying
of struct ucred, and permit the initial labels to be set for
both process 0 (parent of all kernel processes) and process 1
(parent of all user processes).

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoRegen.
Robert Watson [Wed, 31 Jul 2002 00:16:58 +0000 (00:16 +0000)]
Regen.

21 years agoReduce the memory footprint of MAC in the base system by halving
Robert Watson [Wed, 31 Jul 2002 00:03:26 +0000 (00:03 +0000)]
Reduce the memory footprint of MAC in the base system by halving
the number of policy slots to 4.

(Having run a quick errand, time to start on phase 2 of the MAC
integration)

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoRegen
Josef Karthauser [Tue, 30 Jul 2002 23:30:56 +0000 (23:30 +0000)]
Regen

21 years agoAdd support for a couple of network adapters; the 3Com 3C460B and
Josef Karthauser [Tue, 30 Jul 2002 23:30:15 +0000 (23:30 +0000)]
Add support for a couple of network adapters; the 3Com 3C460B and
the Belkin USB2LAN.

Obtained from: NetBSD

21 years agoRemove some strange code that allocates memory and then immediately
Ian Dowse [Tue, 30 Jul 2002 23:26:22 +0000 (23:26 +0000)]
Remove some strange code that allocates memory and then immediately
frees it again. The idea was to perform M_WAITOK allocations in a
process context to reduce the risk of later interrupt-context
M_NOWAIT allocations failing, but in fact this code can be called
from contexts where it is not desirable to sleep (e.g. if_start
routines), so it causes lots of witness "could sleep" warnings.

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 23:12:37 +0000 (23:12 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Label devfs directory entries, permitting labels to be maintained
on device nodes in devfs instances persistently despite vnode
recycling.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 23:09:20 +0000 (23:09 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Label IP fragment reassembly queues, permitting security features to
be maintained on those objects.  ipq_label will be used to manage
the reassembly of fragments into IP datagrams using security
properties.  This permits policies to deny the reassembly of fragments,
as well as influence the resulting label of a datagram following
reassembly.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 23:06:07 +0000 (23:06 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Label network interface structures, permitting security features to
be maintained on those objects.  if_label will be used to authorize
data flow using the network interface.  if_label will be protected
using the same synchronization primitives as other mutable entries
in struct ifnet.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 23:03:29 +0000 (23:03 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Label BPF descriptor objects, permitting security features to be
maintained on those objects.  bd_label will be used to authorize
data flow from network interfaces to user processes.  BPF
labels are protected using the same synchronization model as other
mutable data in the BPF descriptor.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 22:43:20 +0000 (22:43 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Replace 'void *' with 'struct mac *' now that mac.h is in the base
tree.  The current POSIX.1e-derived userland MAC interface is
schedule for replacement, but will act as a functional placeholder
until the replacement is done.  These system calls allow userland
processes to get and set labels on both the current process, as well
as file system objects and file descriptor backed objects.

21 years agoIntroduce support for Mandatory Access Control and extensible kernel
Robert Watson [Tue, 30 Jul 2002 22:39:28 +0000 (22:39 +0000)]
Introduce support for Mandatory Access Control and extensible kernel
access control.

Label socket IPC objects, permitting security features to be maintained
at the granularity of the socket.  Two labels are stored for each
socket: the label of the socket itself, and a cached peer label
permitting interogation of the remote endpoint.  Since socket locking
is not yet present in the base tree, these objects are not locked,
but are assumed to follow the same semantics as other modifiable
entries in the socket structure.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoIntroduce support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 22:33:28 +0000 (22:33 +0000)]
Introduce support for Mandatory Access Control and extensible
kernel access control.

Label pipe IPC objects, permitting security information to be
maintained at the granularity of the pipe object.  The label is
shared between the two pipe endpoints in the style of the
pipe mutex, and is maintained using similar conventions.  The
label is protected by the pipe mutex.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoFix path to other files.<arch> files.
Warner Losh [Tue, 30 Jul 2002 22:28:43 +0000 (22:28 +0000)]
Fix path to other files.<arch> files.
Fix disordering of libkern/crc32.c entry.

MFC after: 1 day

21 years agoIntroduce support for Mandatory Access Control and extensible kernel
Robert Watson [Tue, 30 Jul 2002 22:28:28 +0000 (22:28 +0000)]
Introduce support for Mandatory Access Control and extensible kernel
access control.

Label process credentials, permitting security information to be
maintained at the granularity of processes and cached credential
objects.  cr_label follows the semantics of other entries in struct
ucred: when a credential is exclusively referenced, it may be
modified.  Otherwise, it must be treated as immutable.  As with
other interesting entries in struct ucred, failing to use the
documented credential management APIs (such as crcopy, crdup, ...)
can result in data corruption or incorrect behavior.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoBegin committing support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 22:22:35 +0000 (22:22 +0000)]
Begin committing support for Mandatory Access Control and extensible
kernel access control.  The MAC framework permits loadable kernel
modules to link to the kernel at compile-time, boot-time, or run-time,
and augment the system security policy.  This commit includes the
initial kernel implementation, although the interface with the userland
components of the oeprating system is still under work, and not all
kernel subsystems are supported.  Later in this commit sequence,
documentation of which kernel subsystems will not work correctly with
a kernel compiled with MAC support will be added.

Label file system mount points, permitting security information to be
maintained at the granularity of the file system.  Two labels are
currently maintained: a security label for the mount itself, and
a default label for objects in the file system (in particular, for
file systems not supporting per-vnode labeling directly).

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoBegin committing support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 22:15:09 +0000 (22:15 +0000)]
Begin committing support for Mandatory Access Control and extensible
kernel access control.  The MAC framework permits loadable kernel
modules to link to the kernel at compile-time, boot-time, or run-time,
and augment the system security policy.  This commit includes the
initial kernel implementation, although the interface with the userland
components of the operating system is still under work, and not all
kernel subsystems are supported.  Later in this commit sequence,
documentation of which kernel subsystems will not work correctly with
a kernel compiled with MAC support will be added.

Introduce two node vnode operations required to support MAC.  First,
VOP_REFRESHLABEL(), which will be invoked by callers requiring that
vp->v_label be sufficiently "fresh" for access control purposes.
Second, VOP_SETLABEL(), which be invoked by callers requiring that
the passed label contents be updated.  The file system is responsible
for updating v_label if appropriate in coordination with the MAC
framework, as well as committing to disk.  File systems that are
not MAC-aware need not implement these VOPs, as the MAC framework
will default to maintaining a single label for all vnodes based
on the label on the file system mount point.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoBegin committing support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 22:08:12 +0000 (22:08 +0000)]
Begin committing support for Mandatory Access Control and extensible
kernel access control.  The MAC framework permits loadable kernel
modules to link to the kernel at compile-time, boot-time, or run-time,
and augment the system security policy.  This commit includes the
initial kernel implementation, although the interface with the userland
components of the oeprating system is still under work, and not all
kernel subsystems are supported.  Later in this commit sequence,
documentation of which kernel subsystems will not work correctly with
a kernel compiled with MAC support will be added.

Label vnodes, permitting security information to maintained at the
granularity of the individual file, directory (et al).  This data is
protected by the vnode lock and may be read only when holding a shared
lock, or modified only when holding an exclusive lock.  Label
information may be considered either the primary copy, or a cached
copy.  Individual file systems or kernel services may use the
VCACHEDLABEL flag for accounting purposes to determine which it is.
New VOPs will be introduced to refresh this label on demand, or to
set the label value.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoBegin committing support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 22:03:57 +0000 (22:03 +0000)]
Begin committing support for Mandatory Access Control and extensible
kernel access control.  The MAC framework permits loadable kernel
modules to link to the kernel at compile-time, boot-time, or run-time,
and augment the system security policy.  This commit includes the
initial kernel implementation, although the interface with the userland
components of the oeprating system is still under work, and not all
kernel subsystems are supported.  Later in this commit sequence,
documentation of which kernel subsystems will not work correctly with
a kernel compiled with MAC support will be added.

Label mbuf's with packet header data, permitting in-flight datagrams
to be labeled in the TrustedBSD MAC implementation.  Add a questionable
recursive #include of sys/mac.h to maintain the current API for
applications and kernel code including mbuf.h to get 'struct mbuf'
definition.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoCall bpf_mtap() on output, to catch outgoing packets for e.g. tcpdump .
Bill Fenner [Tue, 30 Jul 2002 21:47:14 +0000 (21:47 +0000)]
Call bpf_mtap() on output, to catch outgoing packets for e.g. tcpdump .

21 years agoBegin committing support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 21:36:05 +0000 (21:36 +0000)]
Begin committing support for Mandatory Access Control and extensible
kernel access control.  The MAC framework permits loadable kernel
modules to link to the kernel at compile-time, boot-time, or run-time,
and augment the system security policy.  This commit includes the
initial kernel implementation, although the interface with the userland
components of the oeprating system is still under work, and not all
kernel subsystems are supported.  Later in this commit sequence,
documentation of which kernel subsystems will not work correctly with
a kernel compiled with MAC support will be added.

kern_mac.c contains the body of the MAC framework.  Kernel and
user APIs defined in mac.h are implemented here, providing a front end
to loaded security modules.  This code implements a module registration
service, state (label) management, security configuration and policy
composition.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoBegin committing support for Mandatory Access Control and extensible
Robert Watson [Tue, 30 Jul 2002 21:32:34 +0000 (21:32 +0000)]
Begin committing support for Mandatory Access Control and extensible
kernel access control.  The MAC framework permits loadable kernel
modules to link to the kernel at compile-time, boot-time, or run-time,
and augment the system security policy.  This commit includes the
initial kernel implementation, although the interface with the userland
components of the oeprating system is still under work, and not all
kernel subsystems are supported.  Later in this commit sequence,
documentation of which kernel subsystems will not work correctly with
a kernel compiled with MAC support will be added.

Include files to declare MAC userland interface (mac.h), MAC subsystem
entry points (mac.h), and MAC policy entry points (mac_policy.h).  These
files define the interface between the kernel and the MAC framework,
and between the MAC framework and each registered policy module.  These
APIs and ABIs may not be assumed to be stable until following FreeBSD
5.1-RELEASE.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoCorrect URLs to Handbook & FAQ's pages
Marc Fonvieille [Tue, 30 Jul 2002 21:14:15 +0000 (21:14 +0000)]
Correct URLs to Handbook & FAQ's pages

MFC after: 1 week

21 years agoDon't need to hold schedlock specifically for stop() ans it calls wakeup()
Julian Elischer [Tue, 30 Jul 2002 21:13:48 +0000 (21:13 +0000)]
Don't need to hold schedlock specifically for stop() ans it calls wakeup()
that locks it anyhow.

Reviewed by: jhb@freebsd.org

21 years agoMake reference counting for mbuf clusters [only] work like in RELENG_4.
Bosko Milekic [Tue, 30 Jul 2002 21:06:27 +0000 (21:06 +0000)]
Make reference counting for mbuf clusters [only] work like in RELENG_4.
While I don't think this is the best solution, it certainly is the
fastest and in trying to find bottlenecks in network related code
I want this out of the way, so that I don't have to think about it.
What this means, for mbuf clusters anyway is:
- one less malloc() to do for every cluster allocation (replaced with
  a relatively quick calculation + assignment)
- no more free() in the cluster free case (replaced with empty space) :-)

This can offer a substantial throughput improvement, but it may not for
all cases.  Particularly noticable for larger buffer sends/recvs.
See http://people.freebsd.org/~bmilekic/code/measure2.txt for a rough
idea.

21 years agoCorrect links to Handbook's pages, old URLs does not work anymore.
Marc Fonvieille [Tue, 30 Jul 2002 21:04:26 +0000 (21:04 +0000)]
Correct links to Handbook's pages, old URLs does not work anymore.

MFC after: 1 week

21 years agoRemove some additional paranoia which Kirk forgot to remove from his
Poul-Henning Kamp [Tue, 30 Jul 2002 20:49:29 +0000 (20:49 +0000)]
Remove some additional paranoia which Kirk forgot to remove from his
UFS2 commit.

These bits in essence made any instance of "softupdates expected
corrution", (ie blocks marked allocated but not referenced by an
inode etc) result in a exit value for fsck_ffs of 2.

2 is part of the magic and appearantly undocumented protocol between
fsck_FOO and fsck and means "dump into single user mode ASAP.

Sponsored by: DARPA & NAI Labs.

21 years ago o Replace vm_page_sleep_busy() with vm_page_sleep_if_busy()
Alan Cox [Tue, 30 Jul 2002 20:41:10 +0000 (20:41 +0000)]
 o Replace vm_page_sleep_busy() with vm_page_sleep_if_busy()
   in vfs_busy_pages().

21 years agoRemove code that removes thread from sleep queue before
Julian Elischer [Tue, 30 Jul 2002 20:34:30 +0000 (20:34 +0000)]
Remove code that removes thread from sleep queue before
adding it to a condvar wait.
We do not have asleep() any more so this can not happen.

21 years agoFix some bugs in in-place editing:
Tony Finch [Tue, 30 Jul 2002 19:42:18 +0000 (19:42 +0000)]
Fix some bugs in in-place editing:
(1) errors from freopen were not reported correctly
(2) large files were not handled correctly
(3) read-only files broke things

MFC after: 1 week

21 years agoResolve conflicts arising from the ACPI CA 20020725 import.
Mitsuru IWASAKI [Tue, 30 Jul 2002 19:35:32 +0000 (19:35 +0000)]
Resolve conflicts arising from the ACPI CA 20020725 import.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r100966,
Mitsuru IWASAKI [Tue, 30 Jul 2002 19:33:39 +0000 (19:33 +0000)]
This commit was generated by cvs2svn to compensate for changes in r100966,
which included commits to RCS files with non-trunk default branches.

21 years agoVendor import of the Intel ACPI CA 20020725 drop.
Mitsuru IWASAKI [Tue, 30 Jul 2002 19:33:39 +0000 (19:33 +0000)]
Vendor import of the Intel ACPI CA 20020725 drop.

21 years ago o In do_sendfile(), replace vm_page_sleep_busy() by vm_page_sleep_if_busy()
Alan Cox [Tue, 30 Jul 2002 18:51:07 +0000 (18:51 +0000)]
 o In do_sendfile(), replace vm_page_sleep_busy() by vm_page_sleep_if_busy()
   and extend the scope of the page queues lock to cover all accesses
   to the page's flags and busy fields.

21 years agoWhen referencing nd_cnp after namei(), always pass SAVENAME into
Robert Watson [Tue, 30 Jul 2002 18:48:25 +0000 (18:48 +0000)]
When referencing nd_cnp after namei(), always pass SAVENAME into
NDINIT() operation flags.

Submitted by: green
Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoMake M_COPY_PKTHDR() macro into a wrapper for a m_copy_pkthdr()
Robert Watson [Tue, 30 Jul 2002 18:28:58 +0000 (18:28 +0000)]
Make M_COPY_PKTHDR() macro into a wrapper for a m_copy_pkthdr()
function.  This permits conditionally compiled extensions to the
packet header copying semantic, such as extensions to copy MAC
labels.

Reviewed by: bmilekic
Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoCommit a version of the uvisor driver for connecting Handspring
Josef Karthauser [Tue, 30 Jul 2002 17:44:28 +0000 (17:44 +0000)]
Commit a version of the uvisor driver for connecting Handspring
Visors via USB.

Submitted by: Chia-liang Kao <clkao@clkao.org>

21 years agoIf we get 0xffff back when reading the status register, assume the card
John Baldwin [Tue, 30 Jul 2002 17:31:42 +0000 (17:31 +0000)]
If we get 0xffff back when reading the status register, assume the card
has gone away instead of spinning in the interrupt handler.  This stops
my machine from hanging when I eject a rl(4)-based cardbus card.

Reviewed by: imp

21 years agoRegen.
Robert Watson [Tue, 30 Jul 2002 16:52:22 +0000 (16:52 +0000)]
Regen.

21 years agoIntroduce a mac_policy() system call that will provide MAC policies
Robert Watson [Tue, 30 Jul 2002 16:50:25 +0000 (16:50 +0000)]
Introduce a mac_policy() system call that will provide MAC policies
with a general purpose front end entry point for user applications
to invoke.  The MAC framework will route the system call to the
appropriate policy by name.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoAdd support for controlling line1 mixer device, which on some cards represents
Maxim Sobolev [Tue, 30 Jul 2002 16:24:00 +0000 (16:24 +0000)]
Add support for controlling line1 mixer device, which on some cards represents
onboard FM tuner.

MFC after: 2 weeks

21 years agoFor processes which are set-user-ID or set-group-ID, the kernel performs a few
Jacques Vidrine [Tue, 30 Jul 2002 15:38:29 +0000 (15:38 +0000)]
For processes which are set-user-ID or set-group-ID, the kernel performs a few
special actions for safety.  One of these is to make sure that file descriptors
0..2 are in use, by opening /dev/null for those that are not already open.
Another is to close any file descriptors 0..2 that reference procfs.  However,
these checks were made out of order, so that it was still possible for a
set-user-ID or set-group-ID process to be started with some of the file
descriptors 0..2 unused.

Submitted by: Georgi Guninski <guninski@guninski.com>

21 years agoUpdate list of installed manual pages after regenerating them.
Jacques Vidrine [Tue, 30 Jul 2002 14:47:24 +0000 (14:47 +0000)]
Update list of installed manual pages after regenerating them.

21 years agoImport the regenerated OpenSSL man pages after import of OpenSSL 0.9.6e.
Jacques Vidrine [Tue, 30 Jul 2002 14:34:51 +0000 (14:34 +0000)]
Import the regenerated OpenSSL man pages after import of OpenSSL 0.9.6e.

21 years agoCorrect links to Handbook's pages:
Marc Fonvieille [Tue, 30 Jul 2002 14:08:16 +0000 (14:08 +0000)]
Correct links to Handbook's pages:
http://www.freebsd.org/handbook/mirrors.html is not working anymore,
http://www.freebsd.org/doc/handbook/mirrors.html is the new link

MFC after: 1 week

21 years agoIgnore leading semicolons on commands; required by SUSv3.
Tim J. Robbins [Tue, 30 Jul 2002 14:07:30 +0000 (14:07 +0000)]
Ignore leading semicolons on commands; required by SUSv3.

Obtained from: NetBSD (kleink, Aymeric Vincent)

21 years agoResolve conflicts after import of OpenSSL 0.9.6e.
Jacques Vidrine [Tue, 30 Jul 2002 13:58:53 +0000 (13:58 +0000)]
Resolve conflicts after import of OpenSSL 0.9.6e.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r100936,
Jacques Vidrine [Tue, 30 Jul 2002 13:38:06 +0000 (13:38 +0000)]
This commit was generated by cvs2svn to compensate for changes in r100936,
which included commits to RCS files with non-trunk default branches.

21 years agoImport of OpenSSL 0.9.6e.
Jacques Vidrine [Tue, 30 Jul 2002 13:38:06 +0000 (13:38 +0000)]
Import of OpenSSL 0.9.6e.

21 years agoWarning cleanup.
Poul-Henning Kamp [Tue, 30 Jul 2002 13:01:25 +0000 (13:01 +0000)]
Warning cleanup.

Format changes by peter

21 years agoThis man page has not been referenced by anything for a while,
Jacques Vidrine [Tue, 30 Jul 2002 12:54:03 +0000 (12:54 +0000)]
This man page has not been referenced by anything for a while,
and is not part of the OpenSSL distribution.  Remove it.

21 years agoUpdate to match reality (i.e. reference libcrypto headers and
Jacques Vidrine [Tue, 30 Jul 2002 12:53:15 +0000 (12:53 +0000)]
Update to match reality (i.e. reference libcrypto headers and
libraries, not the no-longer-existent libdes).

21 years agoRemove many obsolete files. The majority of these are simply no
Jacques Vidrine [Tue, 30 Jul 2002 12:51:09 +0000 (12:51 +0000)]
Remove many obsolete files.  The majority of these are simply no
longer included as part of the OpenSSL distribution.  However, a few
we just don't need and are explicitly excluded in FREEBSD-Xlist.

21 years agoResolve conflicts after import of OpenSSL 0.9.6d.
Jacques Vidrine [Tue, 30 Jul 2002 12:46:49 +0000 (12:46 +0000)]
Resolve conflicts after import of OpenSSL 0.9.6d.

21 years agoImport of OpenSSL 0.9.6d.
Jacques Vidrine [Tue, 30 Jul 2002 12:44:15 +0000 (12:44 +0000)]
Import of OpenSSL 0.9.6d.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r100928,
Jacques Vidrine [Tue, 30 Jul 2002 12:44:15 +0000 (12:44 +0000)]
This commit was generated by cvs2svn to compensate for changes in r100928,
which included commits to RCS files with non-trunk default branches.

21 years agoUpdate list of files to remove prior to import of OpenSSL 0.9.6d
Jacques Vidrine [Tue, 30 Jul 2002 12:38:41 +0000 (12:38 +0000)]
Update list of files to remove prior to import of OpenSSL 0.9.6d

21 years agoFix braino in last commit.
Poul-Henning Kamp [Tue, 30 Jul 2002 12:02:41 +0000 (12:02 +0000)]
Fix braino in last commit.

21 years agoMove ffs_isfreeblock() to ffs_alloc.c and make it static.
Poul-Henning Kamp [Tue, 30 Jul 2002 11:54:48 +0000 (11:54 +0000)]
Move ffs_isfreeblock() to ffs_alloc.c and make it static.

Sponsored by: DARPA & NAI Labs.

21 years agoRegen after usbdevs rev.1.95.
Maxim Sobolev [Tue, 30 Jul 2002 11:03:50 +0000 (11:03 +0000)]
Regen after usbdevs rev.1.95.

21 years agoAdd ID for Luwen EasyDisk USB flash memory drive.
Maxim Sobolev [Tue, 30 Jul 2002 11:02:35 +0000 (11:02 +0000)]
Add ID for Luwen EasyDisk USB flash memory drive.

PR: 41081
Submitted by: demon
MFC after: 5 days

21 years agoAdd a quick check that device actually has sound capabilities. The problem
Maxim Sobolev [Tue, 30 Jul 2002 10:54:19 +0000 (10:54 +0000)]
Add a quick check that device actually has sound capabilities. The problem
is that some cards built around fm801 chip have the same device ID, only
have radio tuner onboard, but no sound capabilities. Therefore, with such
card inserted and `device pcm' in kernel the user has a big problem, as
the fm801 driver effectively hangs the machine when trying to initialise
nonexistent ac97 codecs (it does 500 retries with 1 second interval!).
It would be better if MediaForte's engeneers were smart enough to put
different device ID into such cards, but it isn't an option.

MFC after: 2 weeks

21 years agoIn endtsleep() and cv_timedwait_end(), a thread marked TDF_TIMEOUT may
Seigo Tanimura [Tue, 30 Jul 2002 10:12:11 +0000 (10:12 +0000)]
In endtsleep() and cv_timedwait_end(), a thread marked TDF_TIMEOUT may
be swapped out.  Do not put such the thread directly back to the run
queue.

Spotted by: David Xu <davidx@viasoft.com.cn>

While I am here, s/PS_TIMEOUT/TDF_TIMEOUT/.

21 years ago - Add vfs_badlock_{print,panic} support to the remaining VOP_ASSERT_*
Jeff Roberson [Tue, 30 Jul 2002 09:57:37 +0000 (09:57 +0000)]
 - Add vfs_badlock_{print,panic} support to the remaining VOP_ASSERT_*
   macros.

21 years ago - Add automatic post vop debug checks. These work in both the success and
Jeff Roberson [Tue, 30 Jul 2002 08:52:00 +0000 (08:52 +0000)]
 - Add automatic post vop debug checks.  These work in both the success and
   failure cases.

21 years ago - Acknowledge recursive vnode locks in the vop_unlock specification. The
Jeff Roberson [Tue, 30 Jul 2002 08:50:52 +0000 (08:50 +0000)]
 - Acknowledge recursive vnode locks in the vop_unlock specification.  The
   vnode may not be unlocked even if the operation succeeded.

21 years agoSince pam_get_authtok(3) doesn't know about our options structure, setting
Dag-Erling Smørgrav [Tue, 30 Jul 2002 08:32:03 +0000 (08:32 +0000)]
Since pam_get_authtok(3) doesn't know about our options structure, setting
the PAM_ECHO_PASS option on-the-fly is a NOP (though it wasn't with the
old pam_get_pass(3) code).  Instead, call pam_prompt(3) directly.  This
actually simplifies the code a bit.

MFC after: 3 days

21 years agoDo a case insensitive comparison when comparing the ms-chap response
Brian Somers [Tue, 30 Jul 2002 08:09:26 +0000 (08:09 +0000)]
Do a case insensitive comparison when comparing the ms-chap response
string.

21 years ago o In vm_object_madvise() and vm_object_page_remove() replace
Alan Cox [Tue, 30 Jul 2002 07:23:04 +0000 (07:23 +0000)]
 o In vm_object_madvise() and vm_object_page_remove() replace
   vm_page_sleep_busy() with vm_page_sleep_if_busy().  At the same time,
   increase the scope of the page queues lock.  (This should significantly
   reduce the locking overhead in vm_object_page_remove().)
 o Apply some style fixes.

21 years agoAdd a bunch more cards that are known to work.
Warner Losh [Tue, 30 Jul 2002 06:58:27 +0000 (06:58 +0000)]
Add a bunch more cards that are known to work.
Add a warning about the Symbol LA-4100 series of cards.  You gotta
load special firmware, and the wi driver in the tree doesn't support
that yet.

21 years ago- Optimize wakeup() and its friends; if a thread waken up is being
Seigo Tanimura [Tue, 30 Jul 2002 06:54:05 +0000 (06:54 +0000)]
- Optimize wakeup() and its friends; if a thread waken up is being
  swapped in, we do not have to ask for the scheduler thread to do
  that.

- Assert that a process is not swapped out in runq functions and
  swapout().

- Introduce thread_safetoswapout() for readability.

- In swapout_procs(), perform a test that may block (check of a
  thread working on its vm map) first.  This lets us call swapout()
  with the sched_lock held, providing a better atomicity.

21 years ago o Lock page queue accesses by pmap_release_free_page().
Alan Cox [Tue, 30 Jul 2002 06:45:39 +0000 (06:45 +0000)]
 o Lock page queue accesses by pmap_release_free_page().