]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoAdd support for the HTTP_TIMEOUT environment variable (integer number
cperciva [Tue, 7 Nov 2006 09:18:09 +0000 (09:18 +0000)]
Add support for the HTTP_TIMEOUT environment variable (integer number
of seconds, just like in fetch(1)).

Submitted by: rdivacky

17 years agoAdd priv.h include required to build FAST_IPSEC, which is not present in
rwatson [Tue, 7 Nov 2006 08:58:06 +0000 (08:58 +0000)]
Add priv.h include required to build FAST_IPSEC, which is not present in
LINT due to a conflict with KAME IPSEC.

Submitted by: Pawel Worach <pawel dot worach at gmail dot com>

17 years agoFix the hanging chad if the NEW_TRAN_CODE change and change the
mjacob [Tue, 7 Nov 2006 05:51:40 +0000 (05:51 +0000)]
Fix the hanging chad if the NEW_TRAN_CODE change and change the
qualifier to force async from cur_spi to spi.

17 years agoDon't drop reference to tty in tty_close() if TS_ISOPEN is already cleared.
tegge [Mon, 6 Nov 2006 22:12:43 +0000 (22:12 +0000)]
Don't drop reference to tty in tty_close() if TS_ISOPEN is already cleared.

Reviewed by: bde

17 years agoInstead of re-implementing hton[ls] and friends for each arch, add a new MI
cognet [Mon, 6 Nov 2006 22:07:47 +0000 (22:07 +0000)]
Instead of re-implementing hton[ls] and friends for each arch, add a new MI
file, net/ntoh.c, which just implement them using the inline functions from
<sys/endian.h>.

Suggested by: bde

17 years agoHandle early errors in kern_sendfile() by introducing a new goto 'out'
andre [Mon, 6 Nov 2006 21:53:19 +0000 (21:53 +0000)]
Handle early errors in kern_sendfile() by introducing a new goto 'out'
label after the sbunlock() part.

This correctly handles calls to sendfile(2) without valid parameters
that was broken in rev. 1.240.

Coverity error: 272162

17 years agoProvide definitions suitable for arm big-endian.
cognet [Mon, 6 Nov 2006 20:49:23 +0000 (20:49 +0000)]
Provide definitions suitable for arm big-endian.

17 years agoSometimes the vty switching has to be delayed; the vty
ru [Mon, 6 Nov 2006 19:06:07 +0000 (19:06 +0000)]
Sometimes the vty switching has to be delayed; the vty
to be switched to is saved in sc->delayed_next_scr and
the actual switch is performed later.  It was possible
to get into the endless loop when attempting to switch
to a closed vty (which is not allowed and beep-alerted
when attempted) and when the visual beep was in effect.
This caused sc->delayed_next_scr to never be reset and
endless attempts to switch to a closed vty and endless
visual beeping.  How to repeat:

- boot into single-user
- run "kbdcontrol -b visual"
- quickly press Alt+F2 two times

PR: kern/68016
X-MFC after: 6.2-RELEASE

17 years agoAdd missing includes of priv.h.
rwatson [Mon, 6 Nov 2006 17:43:10 +0000 (17:43 +0000)]
Add missing includes of priv.h.

17 years agoStyle facelift.
dds [Mon, 6 Nov 2006 15:58:35 +0000 (15:58 +0000)]
Style facelift.
- Reduce the number of global variables
- Make global objects static
- Use bool consistently
- Sort getopt arguments and their processing
- Add function comments
- Change notlast != 0 into !last

17 years agoAdd needed hyphens, note the KTRACE kernel option, bump doc date.
trhodes [Mon, 6 Nov 2006 15:17:50 +0000 (15:17 +0000)]
Add needed hyphens, note the KTRACE kernel option, bump doc date.

PR: 85186
Submitted by: garys

17 years agoRemove sys/uio.h include -- this is no longer required by the extattr
rwatson [Mon, 6 Nov 2006 15:12:43 +0000 (15:12 +0000)]
Remove sys/uio.h include -- this is no longer required by the extattr
system call API.

MFC after: 3 weeks

17 years agoUse a more sensible default of 1 or -1 when only the start and
dds [Mon, 6 Nov 2006 15:11:50 +0000 (15:11 +0000)]
Use a more sensible default of 1 or -1 when only the start and
end values are specified.

PR: bin/68981
Submitted by: Stefan `Sec` Zehl
MFC after: 2 weeks

17 years agoAdd auditd_program variable to defaults, in order to make it more clear
rwatson [Mon, 6 Nov 2006 15:11:24 +0000 (15:11 +0000)]
Add auditd_program variable to defaults, in order to make it more clear
how to change the auditd instance.  When using a port/package-based
OpenBSM, changing the auditd pointer may be desirable.

Obtained from: TrustedBSD Project
MFC after: 3 weeks

17 years agoAdd the examples from the manual page.
dds [Mon, 6 Nov 2006 15:05:03 +0000 (15:05 +0000)]
Add the examples from the manual page.

17 years agoUpdated results for bin/68981
dds [Mon, 6 Nov 2006 15:00:37 +0000 (15:00 +0000)]
Updated results for bin/68981

PR: bin/68981

17 years agoConvert three new suser(9) calls introduced between when the priv(9)
rwatson [Mon, 6 Nov 2006 14:54:06 +0000 (14:54 +0000)]
Convert three new suser(9) calls introduced between when the priv(9)
patch was prepared and committed to priv(9) calls.  Add XXX comments
as, in each case, the semantics appear to differ from the TCP/UDP
versions of the calls with respect to jail, and because cr_canseecred()
is not used to validate the query.

Obtained from: TrustedBSD Project

17 years agoThis changes tracks down the EEOR->NonEEOR mode failure
rrs [Mon, 6 Nov 2006 14:34:21 +0000 (14:34 +0000)]
This changes tracks down the EEOR->NonEEOR mode failure
to wakeup on close of the sender. It basically moves
the return (when the asoc has a reader/writer) further
down and gets the wakeup and assoc appending (of the
PD-API event) moved up before the return.  It also
moves the flag set right before the return so we can
assure only once adding the PD-API events.

Approved by: gnn

17 years agoFix markup.
ru [Mon, 6 Nov 2006 14:28:09 +0000 (14:28 +0000)]
Fix markup.

17 years agoBump document date.
ru [Mon, 6 Nov 2006 14:26:43 +0000 (14:26 +0000)]
Bump document date.

17 years agoDo What I Mean when the user asks for random integers or characters.
dds [Mon, 6 Nov 2006 13:55:11 +0000 (13:55 +0000)]
Do What I Mean when the user asks for random integers or characters.
Up to now jot would fail to generate the last character in the range
or skew the integer distribution in a way that would generate the numbers
in the range's limits with half the probability of the rest.

This modification fixes the program, rather than documenting the
strange behavior, as suggested in docs/54879.

Also, correctly specify the range of random(3).

PR: docs/54879
MFC after: 2 weeks

17 years agoAdd stub entry point implementations of mpo_priv_check and mpo_priv_grant to
rwatson [Mon, 6 Nov 2006 13:45:45 +0000 (13:45 +0000)]
Add stub entry point implementations of mpo_priv_check and mpo_priv_grant to
the mac_stub policy.

Obtained from: TrustedBSD Project

17 years agoSweep kernel replacing suser(9) calls with priv(9) calls, assigning
rwatson [Mon, 6 Nov 2006 13:42:10 +0000 (13:42 +0000)]
Sweep kernel replacing suser(9) calls with priv(9) calls, assigning
specific privilege names to a broad range of privileges.  These may
require some future tweaking.

Sponsored by:           nCircle Network Security, Inc.
Obtained from:          TrustedBSD Project
Discussed on:           arch@
Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri,
                        Alex Lyashkov <umka at sevcity dot net>,
                        Skip Ford <skip dot ford at verizon dot net>,
                        Antoine Brodin <antoine dot brodin at laposte dot net>

17 years agoAdd a new priv(9) kernel interface for checking the availability of
rwatson [Mon, 6 Nov 2006 13:37:19 +0000 (13:37 +0000)]
Add a new priv(9) kernel interface for checking the availability of
privilege for threads and credentials.  Unlike the existing suser(9)
interface, priv(9) exposes a named privilege identifier to the privilege
checking code, allowing more complex policies regarding the granting of
privilege to be expressed.  Two interfaces are provided, replacing the
existing suser(9) interface:

suser(td)                 ->   priv_check(td, priv)
suser_cred(cred, flags)   ->   priv_check_cred(cred, priv, flags)

A comprehensive list of currently available kernel privileges may be
found in priv.h.  New privileges are easily added as required, but the
comments on adding privileges found in priv.h and priv(9) should be read
before doing so.

The new privilege interface exposed sufficient information to the
privilege checking routine that it will now be possible for jail to
determine whether a particular privilege is granted in the check routine,
rather than relying on hints from the calling context via the
SUSER_ALLOWJAIL flag.  For now, the flag is maintained, but a new jail
check function, prison_priv_check(), is exposed from kern_jail.c and used
by the privilege check routine to determine if the privilege is permitted
in jail.  As a result, a centralized list of privileges permitted in jail
is now present in kern_jail.c.

The MAC Framework is now also able to instrument privilege checks, both
to deny privileges otherwise granted (mac_priv_check()), and to grant
privileges otherwise denied (mac_priv_grant()), permitting MAC Policy
modules to implement privilege models, as well as control a much broader
range of system behavior in order to constrain processes running with
root privilege.

The suser() and suser_cred() functions remain implemented, now in terms
of priv_check() and the PRIV_ROOT privilege, for use during the transition
and possibly continuing use by third party kernel modules that have not
been updated.  The PRIV_DRIVER privilege exists to allow device drivers to
check privilege without adopting a more specific privilege identifier.

This change does not modify the actual security policy, rather, it
modifies the interface for privilege checks so changes to the security
policy become more feasible.

Sponsored by: nCircle Network Security, Inc.
Obtained from: TrustedBSD Project
Discussed on: arch@
Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri,
Alex Lyashkov <umka at sevcity dot net>,
Skip Ford <skip dot ford at verizon dot net>,
Antoine Brodin <antoine dot brodin at laposte dot net>

17 years agoAdded PCI ID's for:
rink [Mon, 6 Nov 2006 12:19:43 +0000 (12:19 +0000)]
Added PCI ID's for:

- 0x1065: Intel 82562ET/EZ/GT/GZ PRO/100 VE Ethernet [1], as found on
  Tyan GS14 barebones.
- 0x1094: Intel Pro/100 946GZ (ICH7) Network Connection [2], as found on
  Intel 946GZis motherboards.

[1] Submitted by: myself
[2] Submitted by: Mike Tancsa <mike@sentex.net>
Reviewed by: imp (mentor), jfv
Approved by: imp (mentor)
MFC after: 3 days

17 years agoPrevent freeing wild pointer when bailing out.
takawata [Mon, 6 Nov 2006 12:14:27 +0000 (12:14 +0000)]
Prevent freeing wild pointer when bailing out.

17 years agoSet up the context for the dbbe_trace callback in the ddb. Otherwise,
kib [Mon, 6 Nov 2006 11:10:57 +0000 (11:10 +0000)]
Set up the context for the dbbe_trace callback in the ddb. Otherwise,
trap caused by backtracing would lead to panic.

Noted and reviewed by: bde

17 years agoAvoid negative array indices: an empty string can also be used
dds [Mon, 6 Nov 2006 11:03:43 +0000 (11:03 +0000)]
Avoid negative array indices: an empty string can also be used
to specify a default value.

17 years agoSee also arc4random
dds [Mon, 6 Nov 2006 10:39:49 +0000 (10:39 +0000)]
See also arc4random

PR: docs/54879
MFC after: 2 weeks

17 years agoRestore jot's ability to use a seed for producing a deterministic
dds [Mon, 6 Nov 2006 10:30:29 +0000 (10:30 +0000)]
Restore jot's ability to use a seed for producing a deterministic
sequence of random numbers.
This functionality was lost in revision 1.9 when the random number
generator was switched to arc4random.

PR: docs/54879
MFC after: 2 weeks

17 years agoReplace obscure aliases through pointers with plain variables.
dds [Mon, 6 Nov 2006 09:15:21 +0000 (09:15 +0000)]
Replace obscure aliases through pointers with plain variables.

MFC after: 2 weeks

17 years agoRegress.out is now regress.x.out.
dds [Mon, 6 Nov 2006 08:49:43 +0000 (08:49 +0000)]
Regress.out is now regress.x.out.

17 years agoMerge code in common cases.
dds [Mon, 6 Nov 2006 08:47:41 +0000 (08:47 +0000)]
Merge code in common cases.

Verified by: New regression tests in tools/regression/usr.bin/jot
MFC after: 2 weeks

17 years agoTest the handling of supplied and default parameters.
dds [Mon, 6 Nov 2006 08:39:52 +0000 (08:39 +0000)]
Test the handling of supplied and default parameters.

17 years agoReplace opaque numeric bit flag values with #defined identifiers.
dds [Mon, 6 Nov 2006 07:26:16 +0000 (07:26 +0000)]
Replace opaque numeric bit flag values with #defined identifiers.
While there, add some missing FALLTHROUGH comments.

Verified with: cmp(1) on the executable
MFC after: 2 weeks

17 years agoRemove gratuitous white space change.
obrien [Mon, 6 Nov 2006 02:49:19 +0000 (02:49 +0000)]
Remove gratuitous white space change.

17 years agoAllow one to force with issue with 'TARGET_BIG_ENDIAN'.
obrien [Mon, 6 Nov 2006 02:32:29 +0000 (02:32 +0000)]
Allow one to force with issue with 'TARGET_BIG_ENDIAN'.

17 years agoSwitch default proto to TCP.
obrien [Mon, 6 Nov 2006 01:42:11 +0000 (01:42 +0000)]
Switch default proto to TCP.

17 years agoEliminate documentation references to a non-existent function.
kientzle [Mon, 6 Nov 2006 00:28:46 +0000 (00:28 +0000)]
Eliminate documentation references to a non-existent function.

17 years agoComputing SHLIB_MAJOR is not a good idea. It's really a FreeBSD
kientzle [Mon, 6 Nov 2006 00:24:57 +0000 (00:24 +0000)]
Computing SHLIB_MAJOR is not a good idea.  It's really a FreeBSD
system value that has no real relation to the libarchive version.
(Except, of course, that any ABI breakage will force both to be
incremented.)

17 years agoRemove an unused declaration.
kientzle [Mon, 6 Nov 2006 00:16:40 +0000 (00:16 +0000)]
Remove an unused declaration.

17 years agoChange the type of ar_arg_sockaddr from struct sockaddr to struct
csjp [Mon, 6 Nov 2006 00:15:44 +0000 (00:15 +0000)]
Change the type of ar_arg_sockaddr from struct sockaddr to struct
sockaddr_storage.  This structure is defined in RFC 2553 and is a more
semantically correct structure for holding IP and IP6 sockaddr information.
struct sockaddr is not big enough to hold all the required information for
IP6, resulting in truncated addresses et al when auditing IP6 sockaddr
information.

We also need to assume that the sa->sa_len has been validated before the call to
audit_arg_sockaddr() is made, otherwise it could result in a buffer overflow.
This is being done to accommodate auditing of network related arguments (like
connect, bind et al) that will be added soon.

Discussed with: rwatson
Obtained from: TrustedBSD Project
MFC after: 2 weeks

17 years agoMajor rework of the ia64 loaders. The two primary objectives are:
marcel [Sun, 5 Nov 2006 22:03:04 +0000 (22:03 +0000)]
Major rework of the ia64 loaders. The two primary objectives are:
1. Make libefi portable by removing ia64 specific code and build
   it on i386 and amd64 by default to prevent regressions. These
   changes include fixes and improvements over previous code to
   establish or improve APIs where none existed or when the amount
   of kluging was unacceptably high.
2. Increase the amount of sharing between the efi and ski loaders
   to improve maintainability of the loaders and simplify making
   changes to the loader-kernel handshaking in the future.

The version of the efi and ski loaders are now both changed to 1.2
as user visible improvements and changes have been made.

17 years agoForced commit to document that the following repo copy has been
marcel [Sun, 5 Nov 2006 21:27:43 +0000 (21:27 +0000)]
Forced commit to document that the following repo copy has been
performed:

from: src/sys/boot/efi/libefi/bootinfo.c
to: src/sys/boot/ia64/common/bootinfo.c

from: src/sys/boot/efi/libefi/copy.c
to: src/sys/boot/ia64/common/copy.c

from: src/sys/boot/efi/libefi/devicename.c
to: src/sys/boot/ia64/common/devicename.c

from: src/sys/boot/efi/libefi/elf_freebsd.c
to: src/sys/boot/ia64/common/exec.c

ncvs meister: simon (thanks!)

17 years ago- I forgot to bump a date.
danger [Sun, 5 Nov 2006 20:12:32 +0000 (20:12 +0000)]
- I forgot to bump a date.

Approved by: keramida (mentor), trhodes (mentor) (implicit)

17 years ago- capitalize the first letter of the first word in the sentence
danger [Sun, 5 Nov 2006 20:08:58 +0000 (20:08 +0000)]
- capitalize the first letter of the first word in the sentence

Approved by: trhodes (mentor), keramida (mentor)

17 years ago- add some files to FILES section and provide their description
danger [Sun, 5 Nov 2006 19:37:27 +0000 (19:37 +0000)]
- add some files to FILES section and provide their description
- reference devfs.conf and devfs.rules in SEE ALSO section

Approved by: keramida (mentor), trhodes (mentor)
PR: docs/103347
MFC-after: 3 days

17 years agoFix a typo: "the give login class" to "the given login class."
trhodes [Sun, 5 Nov 2006 19:00:08 +0000 (19:00 +0000)]
Fix a typo: "the give login class" to "the given login class."

PR: 75577
Submitted by: Nobuyuki Koganemaru

17 years agoAdd regression tests for PR 105078.
stefanf [Sun, 5 Nov 2006 18:41:23 +0000 (18:41 +0000)]
Add regression tests for PR 105078.

17 years agoWhen parsing an invalid parameter expansion (eg. ${} or ${foo@bar}) do not
stefanf [Sun, 5 Nov 2006 18:36:05 +0000 (18:36 +0000)]
When parsing an invalid parameter expansion (eg. ${} or ${foo@bar}) do not
issue a syntax error immediately but save the information that it is erroneous
for later when the parameter expansion is actually done.  This means eg. "false
&& ${}" will not generate an error which seems to be required by POSIX.
Include the invalid parameter expansion in the error message (sometimes
abbreviated with ... because recovering it would require a lot of code).

PR: 105078
Submitted by: emaste

17 years agoFix possible leak when bridge is in monitor mode. Use m_freem() which will
csjp [Sun, 5 Nov 2006 17:56:25 +0000 (17:56 +0000)]
Fix possible leak when bridge is in monitor mode. Use m_freem() which will
free the entire chain, instead of using m_free() which will free just the
mbuf that was passed.

Discussed with: thompsa
MFC after: 3 days

17 years agoo Add NetBSD 3.0.2 and NetBSD 3.1.
maxim [Sun, 5 Nov 2006 15:56:45 +0000 (15:56 +0000)]
o Add NetBSD 3.0.2 and NetBSD 3.1.

17 years agoChange one more ARM_BIG_ENDIAN to TARGET_BIG_ENDIAN.
cognet [Sun, 5 Nov 2006 15:33:26 +0000 (15:33 +0000)]
Change one more ARM_BIG_ENDIAN to TARGET_BIG_ENDIAN.

17 years agoSync up with rhyolite routed 2.31 which fixes the handling of varargs.
trhodes [Sun, 5 Nov 2006 14:49:47 +0000 (14:49 +0000)]
Sync up with rhyolite routed 2.31 which fixes the handling of varargs.
Remove -p from usage, it's gone completely now.

PR: 83387
Submited by: arved

17 years agoRevert previous commit, and instead make the expression in rev. 1.2
ru [Sun, 5 Nov 2006 14:36:59 +0000 (14:36 +0000)]
Revert previous commit, and instead make the expression in rev. 1.2
match the style of this file.

OK'ed by: rrs

17 years agoTons of fixes to get all the 64bit issues removed.
rrs [Sun, 5 Nov 2006 13:25:18 +0000 (13:25 +0000)]
Tons of fixes to get all the 64bit issues removed.
This also moves two 16 bit int's to become 32 bit
values so we do not have to use atomic_add_16.
Most of the changes are %p, casts and other various
nasty's that were in the orignal code base. With this
commit my machine will now do a build universe.. however
I as yet have not tested on a 64bit machine .. it may not work :-(

17 years agoo Document ServeRAID 7x support.
maxim [Sun, 5 Nov 2006 08:55:21 +0000 (08:55 +0000)]
o Document ServeRAID 7x support.

Submitted by: Andrey V. Elsukov
MFC after: 3 days

17 years agoCreate a bidirectional mapping of the DOS 'read only' attribute
bp [Sun, 5 Nov 2006 06:38:42 +0000 (06:38 +0000)]
Create a bidirectional mapping of the DOS 'read only' attribute
to the 'w' flag.

PR: kern/77958
Submitted by: ghozzy gmail com
MFC after: 1 month

17 years agoIt seems to be safe to ignore 'file not locked' error
bp [Sun, 5 Nov 2006 06:31:08 +0000 (06:31 +0000)]
It seems to be safe to ignore 'file not locked' error
from server.  This effectively suppresses 'Unmapped error 1:158'.

MFC after: 1 month

17 years agoThe rune interface was removed by tjr over two years ago.
trhodes [Sun, 5 Nov 2006 00:39:56 +0000 (00:39 +0000)]
The rune interface was removed by tjr over two years ago.

17 years agoWhen the source and destination objects for VOP_RENAME() are the same,
pjd [Sat, 4 Nov 2006 23:59:51 +0000 (23:59 +0000)]
When the source and destination objects for VOP_RENAME() are the same,
POSIX advice to just ignore the operation and return 0.
We already handle this case in kern_rename(), but don't give bad example -
the source is not removed.

17 years agoVOP_REMOVE() doesn't unlock vnodes nor decrease reference counts.
pjd [Sat, 4 Nov 2006 23:58:15 +0000 (23:58 +0000)]
VOP_REMOVE() doesn't unlock vnodes nor decrease reference counts.

17 years agoTypo, 'from' vnode is locked here, not 'to' vnode.
pjd [Sat, 4 Nov 2006 23:57:02 +0000 (23:57 +0000)]
Typo, 'from' vnode is locked here, not 'to' vnode.

17 years agoRemove the KDTRACE option again because of the complaints about having
jb [Sat, 4 Nov 2006 23:50:12 +0000 (23:50 +0000)]
Remove the KDTRACE option again because of the complaints about having
it as a default.

For the record, the KDTRACE option caused _no_ additional source files
to be compiled in; certainly no CDDL source files. All it did was to
allow existing BSD licensed kernel files to include one or more CDDL
header files.

By removing this from DEFAULTS, the onus is on a kernel builder to add
the option to the kernel config, possibly by including GENERIC and
customising from there. It means that DTrace won't be a feature
available in FreeBSD by default, which is the way I intended it to be.

Without this option, you can't load the dtrace module (which contains
the dtrace device and the DTrace framework). This is equivalent to
requiring an option in a kernel config before you can load the linux
emulation module, for example.

I think it is a mistake to have DTrace ported to FreeBSD, but not
to have it available to everyone, all the time. The only exception
to this is the companies which distribute systems with FreeBSD embedded.
Those companies will customise their systems anyway. The KDTRACE
option was intended for them, and only them.

17 years agoCurrently, we initialize "error" to zero when it's declared, then
csjp [Sat, 4 Nov 2006 20:54:37 +0000 (20:54 +0000)]
Currently, we initialize "error" to zero when it's declared, then
we never initialize it to anything else. However, in the case that
m_uiotombuf fails, we return error (effectively reporting success).

This appears to be a relic of an older revision of this file, where
"error" used to be doing something useful. (See revision 1.1, where
error is used in a loop with uiomove() instead of using m_uiotomubf).

So instead on unconditionally reporting success in the case there is
a failure in m_uiotombuf, explicitly return ENOBUFS. While we are
here, garbage collect the error variable since it's no longer required.

MFC after: 2 weeks

17 years agoWhen the packet is for the bridge then note which interface to send the reply
thompsa [Sat, 4 Nov 2006 10:40:59 +0000 (10:40 +0000)]
When the packet is for the bridge then note which interface to send the reply
to, previously it was always broadcast to all interfaces (a bug). This is
useful when the bridge is the default gateway and vlans are used to isolate
each client, the reply is now kept private to the vlan which the client
resides.

Reported by: Jon Otterholm
Tested by: Jon Otterholm
MFC after: 3 days

17 years agoXref pthread_condattr.3, pthread_attr.3, and pthread_mutexattr.3.
trhodes [Sat, 4 Nov 2006 10:08:16 +0000 (10:08 +0000)]
Xref pthread_condattr.3, pthread_attr.3, and pthread_mutexattr.3.

PR: 60679
Submitted by: roam (with minor modifications)

17 years agoDocument return values for res_query and res_search.
lawrance [Sat, 4 Nov 2006 09:42:22 +0000 (09:42 +0000)]
Document return values for res_query and res_search.
h_errno is not an extern int, just a macro providing an integer lvalue.

PR: doc/50573
Submitted by: Ronald F.Guilmette <rfg@monkeys.com>
Reviewed by: trhodes
MFC after: 3 days

17 years ago- clock_gettime is in time.h instead of sys/time.h
danger [Sat, 4 Nov 2006 09:35:36 +0000 (09:35 +0000)]
- clock_gettime is in time.h instead of sys/time.h

Approved by: trhodes (mentor), keramida (mentor)
PR: docs/104707
MFC-after: 3 days

17 years agoFix pointer arithmetic to be 64-bit friendly.
ru [Sat, 4 Nov 2006 08:45:50 +0000 (08:45 +0000)]
Fix pointer arithmetic to be 64-bit friendly.

17 years agoRemove bogus casts that Randall for some reason didn't borrow
ru [Sat, 4 Nov 2006 08:19:01 +0000 (08:19 +0000)]
Remove bogus casts that Randall for some reason didn't borrow
from my supplied patch.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r163976,
dougb [Sat, 4 Nov 2006 07:53:25 +0000 (07:53 +0000)]
This commit was generated by cvs2svn to compensate for changes in r163976,
which included commits to RCS files with non-trunk default branches.

17 years agoUpdate to version 9.3.2-P2, which addresses the vulnerability
dougb [Sat, 4 Nov 2006 07:53:25 +0000 (07:53 +0000)]
Update to version 9.3.2-P2, which addresses the vulnerability
announced by ISC dated 31 October (delivered via e-mail to the
bind-announce@isc.org list on 2 November):

Description:
        Because of OpenSSL's recently announced vulnerabilities
        (CAN-2006-4339, CVE-2006-2937 and CVE-2006-2940) which affect named,
        we are announcing this workaround and releasing patches.  A proof of
        concept attack on OpenSSL has been demonstrated for CAN-2006-4339.

        OpenSSL is required to use DNSSEC with BIND.

Fix for version 9.3.2-P1 and lower:
        Upgrade to BIND 9.3.2-P2, then generate new RSASHA1 and
        RSAMD5 keys for all old keys using the old default exponent
        and perform a key rollover to these new keys.

        These versions also change the default RSA exponent to be
        65537 which is not vulnerable to the attacks described in
        CAN-2006-4339.

17 years agoDoh! Actually commit checking against NULL for res.
imp [Sat, 4 Nov 2006 06:56:51 +0000 (06:56 +0000)]
Doh!  Actually commit checking against NULL for res.

Noticed by: dougb@

17 years agoRemove a bogus cast in an attempt to fix the tinderbox builds on
jb [Sat, 4 Nov 2006 05:39:39 +0000 (05:39 +0000)]
Remove a bogus cast in an attempt to fix the tinderbox builds on
lots of arches.

17 years agoBackout the previous change. It was not intended to be part of the
jb [Sat, 4 Nov 2006 05:27:21 +0000 (05:27 +0000)]
Backout the previous change. It was not intended to be part of the
commit and, while something like that is probably required for sparc64,
it hadn't been tested.

17 years agoBuild in kernel support for loading DTrace modules by default. This
jb [Sat, 4 Nov 2006 04:58:10 +0000 (04:58 +0000)]
Build in kernel support for loading DTrace modules by default. This
adds the hooks that DTrace modules register with, and adds a few functions
which have the dtrace_ prefix to allow the DTrace FBT (function boundary
trace) provider to avoid tracing because they are called from the DTtrace
probe context.

Unlike other forms of tracing and debug, DTrace support in the kernel
incurs negligible run-time cost.

I think the only reason why anyone wouldn't want to have kernel support
enabled for DTrace would be due to the license (CDDL) under which DTrace
is released.

17 years agoAllow a makefile to set IGNORE_PRAGMA so that OpenSolaris code can
jb [Sat, 4 Nov 2006 04:39:05 +0000 (04:39 +0000)]
Allow a makefile to set IGNORE_PRAGMA so that OpenSolaris code can
be built with other gcc warnings enabled.

Every Solaris source file has a #pragma ident in it. We can just
ignore those definitions.

17 years agoUpdated release note: em(4) 6.2.9.
bmah [Sat, 4 Nov 2006 02:01:52 +0000 (02:01 +0000)]
Updated release note:  em(4) 6.2.9.

17 years ago- change ABI for user trap handler for sun4v (eventually will sync sparc64 no
kmacy [Fri, 3 Nov 2006 23:43:18 +0000 (23:43 +0000)]
- change ABI for user trap handler for sun4v (eventually will sync sparc64 no
  hardware set up to test at the moment)

Approved by: scottl (standing in for mentor rwatson)

17 years ago- map hardware trap numbers to those used by by sparc64 for inter-compatibility
kmacy [Fri, 3 Nov 2006 23:41:53 +0000 (23:41 +0000)]
- map hardware trap numbers to those used by by sparc64 for inter-compatibility
  and to make user-level trap handlers work
- add new trap entry to trap table to enable fast fetching of floating point trap
  context
- remove unused debug code
- map unimplemented floating point trap to SIGFPE

Approved by: scottl (standing in for mentor rwatson)

17 years agomake pcb pad area accessible from asm
kmacy [Fri, 3 Nov 2006 23:33:40 +0000 (23:33 +0000)]
make pcb pad area accessible from asm
Approved by: scottl (standing in for rwatson as mentor)

17 years agoMore 64 bit pointer fun.
rrs [Fri, 3 Nov 2006 23:04:34 +0000 (23:04 +0000)]
More 64 bit pointer fun.
%p changed in multiple prints
the mtod() was also fixed.

17 years agoThe relocation definitions are now defined in the machine independent
jb [Fri, 3 Nov 2006 23:03:46 +0000 (23:03 +0000)]
The relocation definitions are now defined in the machine independent
elf_common.h so that one arch can identify relocations on another
arch.

17 years agoRemove no longer existing opt_bdg.h and opt_bge.h.
ru [Fri, 3 Nov 2006 21:41:19 +0000 (21:41 +0000)]
Remove no longer existing opt_bdg.h and opt_bge.h.

17 years agoRegen.
ru [Fri, 3 Nov 2006 21:23:33 +0000 (21:23 +0000)]
Regen.

17 years agoFix build breakage introduced in previous commit (redeclatation
ru [Fri, 3 Nov 2006 21:21:28 +0000 (21:21 +0000)]
Fix build breakage introduced in previous commit (redeclatation
of sctp functions).

17 years agoFix two of the 64bit errors on the printfs.
rrs [Fri, 3 Nov 2006 21:19:54 +0000 (21:19 +0000)]
Fix two of the 64bit errors on the printfs.

17 years agoNitpicking.
ru [Fri, 3 Nov 2006 21:09:03 +0000 (21:09 +0000)]
Nitpicking.

17 years agoSomehow I missed this one. The sys/cdef.h was out
rrs [Fri, 3 Nov 2006 19:48:56 +0000 (19:48 +0000)]
Somehow I missed this one. The sys/cdef.h was out
of order with respect to the FSBID..

17 years agoThis commits the remake in kern/ make sysent to get
rrs [Fri, 3 Nov 2006 18:57:49 +0000 (18:57 +0000)]
This commits the remake in kern/ make sysent to get
the correct syscalls.master's $FreeBSD$ tag record and
a make sysent in sys/compat/freebsd32. Thanks Ruslan
for pointing out the steps I missed :-0
Approved by: gnn

17 years agoOpps... in my fix up of all the $FreeBSD:$-> $FreeBSD$ I
rrs [Fri, 3 Nov 2006 17:21:53 +0000 (17:21 +0000)]
Opps... in my fix up of all the $FreeBSD:$-> $FreeBSD$ I
inserted a few to the new files.. but I falied to
add the #include <sys/cdef.h>

Which causes a compile error.. sorry about that... got it
now :-)

Approved by:gnn

17 years agoOk, here it is, we finally add SCTP to current. Note that this
rrs [Fri, 3 Nov 2006 15:23:16 +0000 (15:23 +0000)]
Ok, here it is, we finally add SCTP to current. Note that this
work is not just mine, but it is also the works of Peter Lei
and Michael Tuexen. They both are my two key other developers
working on the project.. and they need ata-boy's too:
****
peterlei@cisco.com
tuexen@fh-muenster.de
****
I did do a make sysent which updated the
syscall's and sysproto.. I hope that is correct... without
it you don't build since we have new syscalls for SCTP :-0

So go out and look at the NOTES, add
option SCTP (make sure inet and inet6 are present too)
and play with SCTP.

I will see about comitting some test tools I have after I
figure out where I should place them. I also have a
lib (libsctp.a) that adds some of the missing socketapi
functions that I need to put into lib's.. I will talk
to George about this :-)

There may still be some 64 bit issues in here, none of
us have a 64 bit processor to test with yet.. Michael
may have a MAC but thats another beast too..

If you have a mac and want to use SCTP contact Michael
he maintains a web site with a loadable module with
this code :-)

Reviewed by: gnn
Approved by: gnn

17 years agoRemove the -C option as it does more harm than good. To be fully
ru [Fri, 3 Nov 2006 12:02:24 +0000 (12:02 +0000)]
Remove the -C option as it does more harm than good.  To be fully
compatible, it would have to (at least):

- support the "compat-compat" -T option,
- *not* support the -l, -O, and -v options,
- default to soft updates being disabled.

Worse, the compatibility mode makes it impossible to mount_mfs(8)
a file system from fstab(5) with soft updates disabled (-S).  [1]

Now, the only difference when called as "mount_mfs" or "mfs" (as
opposed to "mdmfs") is that the file mode of the mount point is
set by default to 01777.  All options available to mdmfs(8) are
also available to mount_mfs(8); the -C option is still recognized
but ignored for backward compatibility.

PR: bin/98860 [1]
MFC after: 2 weeks

17 years agoMFp4:
imp [Fri, 3 Nov 2006 07:39:37 +0000 (07:39 +0000)]
MFp4:
o Fix the packet statistics
o Make sure we set the FD bit when in full duplex
o Improve TX side efficency by eliminating a data copy for
  unfragmented mbufs (the hardware can't do s/g).
o Minor busdma pedantry
o better comments in some places, more XXX in others
o Minor style nits.

This solves a problem I was seeing where I'd get no ethernet when not
booting with a NFS root.  Well, unless I unplugged the cable and
plugged it back in first so I'd get the same up down up messages I get
for NFS root...

Thanks to sam and scottl for suggestions on making this driver more
efficient through better use of approrpiate APIs.

17 years agoFix initialization sequence for console
kmacy [Fri, 3 Nov 2006 07:29:09 +0000 (07:29 +0000)]
Fix initialization sequence for console
Fix commenting convention slightly
Approved by: rwatson (mentor)
Reviewed by: jb

17 years agomake sure physmem is initialized
kmacy [Fri, 3 Nov 2006 07:27:55 +0000 (07:27 +0000)]
make sure physmem is initialized
add clarifying comments
Reviewed by: jb
Approved by: rwatson (mentor)

17 years agoDocument IBM ServeRAID 8k support.
brueffer [Fri, 3 Nov 2006 06:50:03 +0000 (06:50 +0000)]
Document IBM ServeRAID 8k support.

17 years agoAdd a low level function to write a string to the hypervisor
jb [Fri, 3 Nov 2006 06:31:56 +0000 (06:31 +0000)]
Add a low level function to write a string to the hypervisor
console directly.

Discussed with: kmacy

17 years agoAlways init the console before trying to cnadd it to
jb [Fri, 3 Nov 2006 06:23:53 +0000 (06:23 +0000)]
Always init the console before trying to cnadd it to
avoid the case where the console name isn't set and
cnadd wants to use printf to complain about it.