]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoAdd a separate simple internal base allocator and remove base_arena, so that
jasone [Mon, 16 Jan 2006 05:13:49 +0000 (05:13 +0000)]
Add a separate simple internal base allocator and remove base_arena, so that
there is never any need to recursively call the main allocation functions.

Remove recursive spinlock support, since it is no longer needed.

Allow chunks to be as small as the page size.

Correctly propagate OOM errors from arena_new().

18 years agoCorrect typos (s/OFERFLOW/OVERFLOW/).
kris [Mon, 16 Jan 2006 01:35:25 +0000 (01:35 +0000)]
Correct typos (s/OFERFLOW/OVERFLOW/).

Reviewed by: jhb

18 years agoRemove scary warning, since nullfs works fine thesedays.
kris [Mon, 16 Jan 2006 01:30:52 +0000 (01:30 +0000)]
Remove scary warning, since nullfs works fine thesedays.

MFC after: 1 week

18 years agoReplace strncpy() with strlcpy() when parsing login time limit strings
rwatson [Mon, 16 Jan 2006 00:52:20 +0000 (00:52 +0000)]
Replace strncpy() with strlcpy() when parsing login time limit strings
from /etc/login.conf, or an unterminated string buffer could result.
Probably, login_times.c should reject excessively long time strings as
unparseable, rather than truncating, which might render an invalid
string valid.

Found with: Coverity Prevent (tm)
Reviewed by: csjp
MFC after: 3 days

18 years agoRemove unnecessary and undesirable 'static' from function-local keg
rwatson [Mon, 16 Jan 2006 00:37:20 +0000 (00:37 +0000)]
Remove unnecessary and undesirable 'static' from function-local keg
list, which could cause problems for multi-threaded applications
using libmemstat to monitor UMA in more than one thread
simultaneously.

MFC after: 3 days

18 years agoDefine NO_TLS on ia64. The dynamic TLS implementation on ia64 is
marcel [Mon, 16 Jan 2006 00:32:46 +0000 (00:32 +0000)]
Define NO_TLS on ia64. The dynamic TLS implementation on ia64 is
broken for non-threaded shared processes in that __tls_get_addr()
assumes the thread pointer is always initialized. This is not the
case. When arenas_map is referenced in choose_arena() and it is
defined as a thread-local variable, it will result in a SIGSEGV.

PR: ia64/91846 (describes the TLS/ia64 bug).

18 years agoFix typo in comment.
rwatson [Mon, 16 Jan 2006 00:28:11 +0000 (00:28 +0000)]
Fix typo in comment.

MFC after: 3 days

18 years agoAdd ntp.cgi.cz.
ceri [Sun, 15 Jan 2006 23:15:41 +0000 (23:15 +0000)]
Add ntp.cgi.cz.

PR: bin/90095
Submitted by: Jakub Chromy <hicks at cgi dott cz>
Approved by: jhb (proxy mentor)
MFC after: 3 days

18 years agoMarius Nuennerich pointed out that nextboot(8) configured boot options
ceri [Sun, 15 Jan 2006 23:12:26 +0000 (23:12 +0000)]
Marius Nuennerich pointed out that nextboot(8) configured boot options
were now sticky.  This script was deleting /boot/nextkernel on boot, but
there is no code in the tree that creates that file since revision 1.15
of src/sbin/reboot/reboot.c.
nextboot(8) creates /boot/nextboot.conf, so remove that instead.

Approved by: jhb (proxy mentor)
MFC after: 1 week

18 years agoPreserve any additional arguments to ifconfig that were specified in the
ceri [Sun, 15 Jan 2006 23:00:07 +0000 (23:00 +0000)]
Preserve any additional arguments to ifconfig that were specified in the
DHCP case too.

PR: bin/74406
Approved by: jhb (proxy mentor)
MFC after: 3 days

18 years agoIf sysinstall was used to configure an anonymous FTP area, it was
ceri [Sun, 15 Jan 2006 22:57:17 +0000 (22:57 +0000)]
If sysinstall was used to configure an anonymous FTP area, it was
possible for information to be copied from the group file to the group
file in the FTP area.  This patch based on a patch from Zak Johnson
<zakj at nox dot cx>.

PR: bin/25851
Submitted by: Ted Mittelstaedt <tedm at toybox punkt placo period com>
Approved by: jhb (proxy mentor)
MFC after: 3 days
MFC to: RELENG_5, RELENG_6
Security: Prevents possible group information leakage

18 years agoCorrect obsolete chown syntax.
ceri [Sun, 15 Jan 2006 22:42:38 +0000 (22:42 +0000)]
Correct obsolete chown syntax.

PR: bin/90057
Submitted by: Charlie M. McDonald <BoredOutkast at yahoo punkt com>
Approved by: jhb (mentor by proxy)
MFC to: RELENG_5, RELENG_6
MFC After: 3 days

18 years agoHappy new year, a little late
imp [Sun, 15 Jan 2006 22:06:10 +0000 (22:06 +0000)]
Happy new year, a little late

18 years agoWork around a problem seen on VIA EHCI controllers where occasionally
iedowse [Sun, 15 Jan 2006 21:03:19 +0000 (21:03 +0000)]
Work around a problem seen on VIA EHCI controllers where occasionally
an interrupt appears to occur before the transfer has been marked
as completed. This caused umass transfers to get stuck, especially
when writing large files. The workaround sets up a timer that
rechecks for missed completed transfers if some operations are still
pending. Other suggested workarounds, such as performing a PCI read
immediately after acknowledging the interrupts, do not appear to
help.

Obtained from: OpenBSD

18 years agoSet sc_dying to 1 when detaching. In NetBSD and OpenBSD this was
iedowse [Sun, 15 Jan 2006 20:41:04 +0000 (20:41 +0000)]
Set sc_dying to 1 when detaching. In NetBSD and OpenBSD this was
done by the DVACT_DEACTIVATE case in *hci_activate(), but we don't
use that code in FreeBSD so it was never set.

18 years agoThe ehci driver doesn't use the transfer `hcpriv' field, so don't
iedowse [Sun, 15 Jan 2006 20:32:52 +0000 (20:32 +0000)]
The ehci driver doesn't use the transfer `hcpriv' field, so don't
bother setting it to NULL in ehci_root_ctrl_done().

18 years agoValidate that the supplied file is not empty before trying mmap(2) it
csjp [Sun, 15 Jan 2006 20:30:13 +0000 (20:30 +0000)]
Validate that the supplied file is not empty before trying mmap(2) it
and access the pages associated with it.

Submitted by: Wojciech A. Koszek
PR: bin/91411
MFC after: 1 week

18 years agovfs_busy can only return something useful if MNTK_UNMOUNT has been set.
csjp [Sun, 15 Jan 2006 20:14:11 +0000 (20:14 +0000)]
vfs_busy can only return something useful if MNTK_UNMOUNT has been set.
Since we are using vfs_busy() on a freshly allocated mount structure, use
(void) to show that we do not care about the return value.

Found with: Coverity Prevent (tm)
MFC after: 2 weeks

18 years agoCast VFS_STATFS() in vfs_domount() to (void) to indicate that ignoring the
rwatson [Sun, 15 Jan 2006 20:01:05 +0000 (20:01 +0000)]
Cast VFS_STATFS() in vfs_domount() to (void) to indicate that ignoring the
return value is intentional: this is simply an attempt to pre-cache the
statfs state.

Found with: Coverity Prevent (tm)
MFC after: 3 days

18 years agoFix an off-by-one bug.
ru [Sun, 15 Jan 2006 19:17:13 +0000 (19:17 +0000)]
Fix an off-by-one bug.

Submitted by: Ulrich Spoerlein

18 years agoModify the IP fragment reassembly code so that it uses a new UMA zone,
rwatson [Sun, 15 Jan 2006 18:58:21 +0000 (18:58 +0000)]
Modify the IP fragment reassembly code so that it uses a new UMA zone,
ipq_zone, to allocate fragment headers from, rather than using cast mbuf
storage.  This was one of the few remaining uses of mbuf storage for
local data structures that relied on dtom().  Implement the resource
limit on ipq's using UMA zone limits, but preserve current sysctl
semantics using a sysctl proc.

MFC after: 3 weeks

18 years agoBump date for previous commit before ru can hit me with his mdoc-stick.
brueffer [Sun, 15 Jan 2006 18:40:52 +0000 (18:40 +0000)]
Bump date for previous commit before ru can hit me with his mdoc-stick.

18 years agoMention the fire screensaver.
brueffer [Sun, 15 Jan 2006 18:30:40 +0000 (18:30 +0000)]
Mention the fire screensaver.

Submitted by: rionda

18 years agoStaticize ipqlock, since it is local to ip_input.c.
rwatson [Sun, 15 Jan 2006 17:05:48 +0000 (17:05 +0000)]
Staticize ipqlock, since it is local to ip_input.c.

MFC after: 3 days

18 years agoDon't leak mbufs and mbuf clusters in several error-handling situations
rwatson [Sun, 15 Jan 2006 12:09:03 +0000 (12:09 +0000)]
Don't leak mbufs and mbuf clusters in several error-handling situations
in the if_an receive routine.

Found with: Coverity Prevent (tm)
MFC after: 1 week

18 years agoIf frame length is excessive, don't leak an mbuf and cluster when
rwatson [Sun, 15 Jan 2006 12:06:09 +0000 (12:06 +0000)]
If frame length is excessive, don't leak an mbuf and cluster when
abandoning processing.

Found with: Coverity Prevent (tm)
MFC after: 1 week

18 years agoI don't believe these are used at all, and can be safely removed
imp [Sun, 15 Jan 2006 06:49:28 +0000 (06:49 +0000)]
I don't believe these are used at all, and can be safely removed

18 years agoAdd support for the Compaq LTE docking station. It includes a plug
imp [Sun, 15 Jan 2006 04:10:47 +0000 (04:10 +0000)]
Add support for the Compaq LTE docking station.  It includes a plug
and play device with the ID of PNP8160.  Sotr them while I'm here.

Submitted by: Sean Shapira    sds at jazzie dotty com
MFC After: 1 week

18 years agoRename uid and gid arguments to vaccess() prototype to match vaccess()
rwatson [Sun, 15 Jan 2006 02:01:51 +0000 (02:01 +0000)]
Rename uid and gid arguments to vaccess() prototype to match vaccess()
implementation in vfs_subr.c.  No functional change.

MFC after: 3 days

18 years agoInitialize ki to p->p_aioinfo after we know it's going to be referencing
csjp [Sun, 15 Jan 2006 01:55:45 +0000 (01:55 +0000)]
Initialize ki to p->p_aioinfo after we know it's going to be referencing
a valid kaioinfo structure. This avoids a potential NULL pointer dereference.

Found with: Coverity Prevent(tm)
MFC after: 2 weeks

18 years agoRemove unused code.
bz [Sun, 15 Jan 2006 01:39:01 +0000 (01:39 +0000)]
Remove unused code.

Found with: Coverity Prevent(tm)

18 years agoFix potential overrun of static stack allocated array which stores
csjp [Sun, 15 Jan 2006 01:02:20 +0000 (01:02 +0000)]
Fix potential overrun of static stack allocated array which stores
the rules. If an array is N elements large, we can only access
elements 0..(N-1).

MFC after: 1 week
Found with: Coverity Prevent(tm)

18 years agoI missed a whitespace in the generated commit message.
flz [Sun, 15 Jan 2006 01:01:46 +0000 (01:01 +0000)]
I missed a whitespace in the generated commit message.

Approved by: ssouhlal (implicit)

18 years agoAdd another inverted EAPD quirk for Gateway 7326GZ.
ariff [Sat, 14 Jan 2006 23:37:08 +0000 (23:37 +0000)]
Add another inverted EAPD quirk for Gateway 7326GZ.

Tested by: Jeff Cross <jeff.cross at averageadmins.com>
MFC after: 3 days

18 years agoMove m_adj after checking that m_dup succeeded.
mlaier [Sat, 14 Jan 2006 22:19:17 +0000 (22:19 +0000)]
Move m_adj after checking that m_dup succeeded.

Found with: Coverity Prevent(tm)
MFC after: 3 days

18 years agoInitialize variable.
glebius [Sat, 14 Jan 2006 21:49:31 +0000 (21:49 +0000)]
Initialize variable.

Found with: Coverity Prevent(tm)

18 years agoMake code simplier fixing memory leak.
glebius [Sat, 14 Jan 2006 21:28:30 +0000 (21:28 +0000)]
Make code simplier fixing memory leak.

Found with: Coverity Prevent(tm)

18 years agoDon't pollute output when "make" is run with -s.
ru [Sat, 14 Jan 2006 20:48:50 +0000 (20:48 +0000)]
Don't pollute output when "make" is run with -s.

Approved by: kan

18 years agoAMD64 also supports disk slices.
ru [Sat, 14 Jan 2006 20:47:11 +0000 (20:47 +0000)]
AMD64 also supports disk slices.

18 years agoIgnore spurious '\0' first character read on a serial console.
ru [Sat, 14 Jan 2006 20:45:02 +0000 (20:45 +0000)]
Ignore spurious '\0' first character read on a serial console.
This allows me to "boot -a" over a serial console.  Tested on
several machines.

18 years agoCheck the return value of copyin.
scottl [Sat, 14 Jan 2006 17:59:28 +0000 (17:59 +0000)]
Check the return value of copyin.

Found by: Coverity Prevent (tm)

18 years agoWhen SC_DISABLE_KDBKEY or SC_DISABLE_REBOOT are not defined allow the
brooks [Sat, 14 Jan 2006 17:57:17 +0000 (17:57 +0000)]
When SC_DISABLE_KDBKEY or SC_DISABLE_REBOOT are not defined allow the
same behavior to be controlled by the sysctls, hw.syscons.kbd_kbdkey
and hw.syscons.kbd_reboot respectively.

Apologies to the submitter for taking so long to commit this simple
change.

PR: kern/72728
Submitted by: Luca Morettoni <morettoni at libero dot it>
MFC After: 3 days

18 years agoDon't base the number of jumbo segments on page size, instead base it on the
scottl [Sat, 14 Jan 2006 17:42:22 +0000 (17:42 +0000)]
Don't base the number of jumbo segments on page size, instead base it on the
fact that jumbo desriptors are defined to have 3 segments.

Found by: Coverity Prevent(tm)

18 years agoFree the newtag if we exit with a failure from alloc_bounce_zone().
scottl [Sat, 14 Jan 2006 17:22:47 +0000 (17:22 +0000)]
Free the newtag if we exit with a failure from alloc_bounce_zone().

Found by: Coverity Prevent(tm)

18 years agoCheck the correct TTL in both the IPv6 and IPv4 cases.
gnn [Sat, 14 Jan 2006 16:39:31 +0000 (16:39 +0000)]
Check the correct TTL in both the IPv6 and IPv4 cases.

Submitted by: glebius
Reviewed by: gnn, bz
Found with:     Coverity Prevent(tm)

18 years agoDon't allocate an asr_ccb in asr_attach, it hasn't been needed for years.
scottl [Sat, 14 Jan 2006 16:01:01 +0000 (16:01 +0000)]
Don't allocate an asr_ccb in asr_attach, it hasn't been needed for years.

18 years agoReplace bogus code with real code.
scottl [Sat, 14 Jan 2006 15:59:54 +0000 (15:59 +0000)]
Replace bogus code with real code.

18 years agoRename driver_t variables. Use device_printf() instead of homerolled printf.
scottl [Sat, 14 Jan 2006 15:51:42 +0000 (15:51 +0000)]
Rename driver_t variables.  Use device_printf() instead of homerolled printf.
Use the provided softc instead of rolling our own.

18 years agoThe LinkSys EG1032 is supported by re(4) not nge(4) [1].
bz [Sat, 14 Jan 2006 15:35:21 +0000 (15:35 +0000)]
The LinkSys EG1032 is supported by re(4) not nge(4) [1].
I couldn't find the ID for the EG1064 anywhere in our sources
so I removed the reference for now.

Pointed out by: Robert Huff <roberthuffi at rcn dot com> [1]
Reviewed by: simon

18 years agoIncorporate the O_NONBLOCK open semantics of Linux and Solaris. This allows
mjacob [Sat, 14 Jan 2006 14:32:41 +0000 (14:32 +0000)]
Incorporate the O_NONBLOCK open semantics of Linux and Solaris. This allows
an application to upon a tape (yea, even the non-control device) even if
it cannot establish a mount session. If the open cannot establish a mount
session and O_NONBLOCK was specified, the tape becomes 'open pending mount'.
All I/O operations that would require access to a tape thereafter until
a close attempt to initiate the mount session. If the mount session succeeds,
the tape driver transitions to full open state, else returns an appropriate
I/O error (ENXIO).

At the same time, add a change that remembers whether tape is being opened
read-only. If so, disallow 'write' operations like writing filemarks that
bypass the normal read-only filtering operations that happen in the write(2)
syscall.

Reviewed by: ken, justin, grog
MFC after: 2 weeks
Suggested by: The Bacula Team

18 years agoRemove dead code.
glebius [Sat, 14 Jan 2006 14:17:27 +0000 (14:17 +0000)]
Remove dead code.

Found with: Coverity Prevent(tm)

18 years ago- Add comments about snd_au88x0.
joel [Sat, 14 Jan 2006 13:22:12 +0000 (13:22 +0000)]
-  Add comments about snd_au88x0.
-  Clarify that snd_audiocs is for sparc64 only.
-  Expand snd_ich and snd_t4dwave comments.

Reviewed by: ariff

18 years agoUMA can return NULL not only in case when our zone is full, but
glebius [Sat, 14 Jan 2006 13:04:08 +0000 (13:04 +0000)]
UMA can return NULL not only in case when our zone is full, but
also in case of generic memory shortage. In the latter case we may
not find an old entry.

Found with: Coverity Prevent(tm)

18 years agoCorrect off-by-one errors.
glebius [Sat, 14 Jan 2006 12:26:32 +0000 (12:26 +0000)]
Correct off-by-one errors.

Found with: Coverity Prevent(tm)

18 years agoFix two memory leakages.
glebius [Sat, 14 Jan 2006 12:16:39 +0000 (12:16 +0000)]
Fix two memory leakages.

Found with: Coverity Prevent(tm)

18 years agoIn ncp_sysctl_connstat(), the SLIST_FOREACH() logic to check 'error'
rwatson [Sat, 14 Jan 2006 11:40:32 +0000 (11:40 +0000)]
In ncp_sysctl_connstat(), the SLIST_FOREACH() logic to check 'error'
resulted in deadcode, as 'error' could never be 0.  What this logic
was originally meant to handle is not clear -- it's been this way
(broken) since at least RELENG_4.

Found with: Coverity Prevent(tm)
MFC after: 3 days

18 years agoEnable mutex inheritance code in mutex_fork, I forgot to turn on it.
davidxu [Sat, 14 Jan 2006 11:33:43 +0000 (11:33 +0000)]
Enable mutex inheritance code in mutex_fork, I forgot to turn on it.
while here, add some comments about process shared mutex.

18 years agoRemove dead code: 'opts' is not used in udp_append(), only in udp_input(),
rwatson [Sat, 14 Jan 2006 11:18:32 +0000 (11:18 +0000)]
Remove dead code: 'opts' is not used in udp_append(), only in udp_input(),
so no need to assign it to NULL or conditionally free it.

Found with: Coverity Prevent(tm)
MFC after: 3 days

18 years agoMake resources do the right thing by design instead of accident.
phk [Sat, 14 Jan 2006 09:46:27 +0000 (09:46 +0000)]
Make resources do the right thing by design instead of accident.

Found with:   Coverity Prevent(tm)

18 years agoCorrect STAILQ usage in purge of resourcelist.
phk [Sat, 14 Jan 2006 09:41:35 +0000 (09:41 +0000)]
Correct STAILQ usage in purge of resourcelist.

Found with:   Coverity Prevent(tm)

18 years ago- Change the exit message from "make cleandepend; make depend" to
flz [Sat, 14 Jan 2006 08:12:25 +0000 (08:12 +0000)]
- Change the exit message from "make cleandepend; make depend" to
"make cleandepend && make depend".

PR: bin/91765
Submitted by: Soeren Straarup <xride@x12.dk>
Approved by: cpercival
MFC after: 3 days

18 years agoAdd code that clears certain capabilities from the member interface, these are
thompsa [Sat, 14 Jan 2006 03:51:31 +0000 (03:51 +0000)]
Add code that clears certain capabilities from the member interface, these are
restored when its removed from the bridge.

At the moment we only clear IFCAP_TXCSUM. Since a locally generated packet on
the bridge may be sent out any one or more interfaces it cant be assumed that
every card does hardware csums. Most bridges don't generate a lot of traffic
themselves so turning off offloading won't hurt, bridged packets are
unaffected.

Tested by: Bruce Walker (bmw borderware.com)
MFC after: 5 days

18 years agoAdd a reminder to remove obsolete files from the vendor branch.
dougb [Sat, 14 Jan 2006 02:45:49 +0000 (02:45 +0000)]
Add a reminder to remove obsolete files from the vendor branch.

18 years agoAdd the following to the taskqueue api:
scottl [Sat, 14 Jan 2006 01:55:24 +0000 (01:55 +0000)]
Add the following to the taskqueue api:

taskqueue_start_threads(struct taskqueue **, int count, int pri,
const char *name, ...);

This allows the creation of 1 or more threads that will service a single
taskqueue.  Also rework the taskqueue_create() API to remove the API change
that was introduced a while back.  Creating a taskqueue doesn't rely on
the presence of a process structure, and the proc mechanics are much better
encapsulated in taskqueue_start_threads().  Also clean up the
taskqueue_terminate() and taskqueue_free() functions to safely drain
pending tasks and remove all associated threads.

The TASKQUEUE_DEFINE and TASKQUEUE_DEFINE_THREAD macros have been changed
to use the new API, but drivers compiled against the old definitions will
still work.  Thus, recompiling drivers is not a strict requirement.

18 years agoAdd the HP NC7771 adapter to the list of known products.
thompsa [Sat, 14 Jan 2006 00:52:32 +0000 (00:52 +0000)]
Add the HP NC7771 adapter to the list of known products.

18 years ago- Document the latest changes for controlling the fan
markus [Sat, 14 Jan 2006 00:39:26 +0000 (00:39 +0000)]
- Document the latest changes for controlling the fan
- s/Thinklight/ThinkLight/ in two cases

Approved by: brueffer
MFC after: 1 week

18 years ago- Add support for setting the fan control mode to manual or automatic
markus [Sat, 14 Jan 2006 00:39:10 +0000 (00:39 +0000)]
- Add support for setting the fan control mode to manual or automatic
- Add support for adjusting the fan speed if the fan control mode is manual

Documentation for the relevant embedded controller register was obtained from
http://www.thinkwiki.org/wiki/Patch_for_controlling_fan_speed

Tested on: R51  by Fabian Keil
T41p by markus
Requested by: many
Approved by: philip
MFC after: 1 week

18 years agoWhen storing the results of malloc() in a pointer to a pointer, check
rwatson [Sat, 14 Jan 2006 00:09:41 +0000 (00:09 +0000)]
When storing the results of malloc() in a pointer to a pointer, check
the pointer to a pointer for NULL, not the pointer for NULL.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoIn spx_attach() and spx_detach(), there is no need to check whether the
rwatson [Sat, 14 Jan 2006 00:05:44 +0000 (00:05 +0000)]
In spx_attach() and spx_detach(), there is no need to check whether the
ipxpcb is NULL or not: in attach it will be, and on detach it won't be.
If for any reason these invariants don't hold true, panicking is a good
idea.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoIn ipcomp6_input(), check 'md' not 'm' after a call to m_pulldown(): 'm'
rwatson [Fri, 13 Jan 2006 23:53:23 +0000 (23:53 +0000)]
In ipcomp6_input(), check 'md' not 'm' after a call to m_pulldown(): 'm'
may be a stale pointer at this point, and we're interested in whether or
not m_pulldown() failed.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years ago- Correct amr_enquiry3 structure[1].
jkim [Fri, 13 Jan 2006 23:50:21 +0000 (23:50 +0000)]
- Correct amr_enquiry3 structure[1].
- Remove redundant AMR_CONFIG_ENQ3_SOLICITED_NOTIFY from the previous
commit while I am here.

PR: kern/29727 [1]

18 years agoRemove dead code associated with 'mcopy' in ipx_forward(): at no point
rwatson [Fri, 13 Jan 2006 23:47:55 +0000 (23:47 +0000)]
Remove dead code associated with 'mcopy' in ipx_forward(): at no point
are the contents of the forwarded mbuf ever copied into mcopy, so there's
no need to have mcopy, conditionally look at mcopy, or conditionally free
it.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoWhen calling bioq_first() to see if a queue is empty in bioq_disksort(),
rwatson [Fri, 13 Jan 2006 23:27:12 +0000 (23:27 +0000)]
When calling bioq_first() to see if a queue is empty in bioq_disksort(),
don't save the return value as we won't use it.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoCheck the right ifnet pointer to see if if_alloc() failed or not in
rwatson [Fri, 13 Jan 2006 23:24:09 +0000 (23:24 +0000)]
Check the right ifnet pointer to see if if_alloc() failed or not in
ef_clone(); we were testing the original ifnet, not the one allocated.

When aborting ef_clone() due to if_alloc() failing, free the allocated
efnet structure rather than leaking it.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoWhen freeing the chain of if_ef devices on an aborted load, use
rwatson [Fri, 13 Jan 2006 23:20:46 +0000 (23:20 +0000)]
When freeing the chain of if_ef devices on an aborted load, use
SLIST_FOREACH_SAFE() rather than SLIST_FOREACH(), as elements are
freed on each iteration of the loop.  This prevents use-after-free.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoIn nfs_dolock(), GC now under-used ioflg, rendered obsolete when we moved
rwatson [Fri, 13 Jan 2006 23:16:29 +0000 (23:16 +0000)]
In nfs_dolock(), GC now under-used ioflg, rendered obsolete when we moved
from using a fifo to talk to rpc.lockd to using a special device node.

Noticed by: Coverity Prevent analysis tool
MFC after: 3 days

18 years agoNote that libc's malloc implementation has been replaced.
jasone [Fri, 13 Jan 2006 22:37:48 +0000 (22:37 +0000)]
Note that libc's malloc implementation has been replaced.

Approved by: markm (mentor)

18 years agoBump version.
flz [Fri, 13 Jan 2006 20:43:00 +0000 (20:43 +0000)]
Bump version.

Approved by: ssouhlal (implicit)

18 years agoNow with multi-commit support.
flz [Fri, 13 Jan 2006 20:41:04 +0000 (20:41 +0000)]
Now with multi-commit support.

Use: `./mfc.pl -m "msg-id-1 msg-id-2" -i hurray'
or   `./mfc.pl -f - -i yippee' and copy/paste two commit logs

Approved by: ssouhlal (implicit)

18 years agoAdd Time DPA20B MP3 Player (1Gb) [1]
mnag [Fri, 13 Jan 2006 19:44:44 +0000 (19:44 +0000)]
Add Time DPA20B MP3 Player (1Gb) [1]
Add Qware BeatZkey! Pro [2]
Add Merlin SM300 MP3/WMA Player (256Mb)

PR: usb/81846 [1], usb/79164 [2]
Approved by:    njl
MFC:            1 week

18 years agoReplace malloc(), calloc(), posix_memalign(), realloc(), and free() with
jasone [Fri, 13 Jan 2006 18:38:56 +0000 (18:38 +0000)]
Replace malloc(), calloc(), posix_memalign(), realloc(), and free() with
a scalable concurrent allocator implementation.

Reviewed by: current@
Approved by: phk, markm (mentor)

18 years agoWorkaround for sb16 behave poorly when running at 45000 hz while
ariff [Fri, 13 Jan 2006 18:10:43 +0000 (18:10 +0000)]
Workaround for sb16 behave poorly when running at 45000 hz while
vchan is enabled.

Reported by: many sb16, AWE64 users.
MFC after: 1 week

18 years agoAdd a daily script to show the status of gmirror(8) devices.
wollman [Fri, 13 Jan 2006 18:07:52 +0000 (18:07 +0000)]
Add a daily script to show the status of gmirror(8) devices.

18 years ago"sd(4)" -> "da(4)", "st(4)" -> "sa(4)".
jkoshy [Fri, 13 Jan 2006 17:41:41 +0000 (17:41 +0000)]
"sd(4)" -> "da(4)", "st(4)" -> "sa(4)".

MFC after: 3 days

18 years agoYesterday netgraph ABI has been changed.
glebius [Fri, 13 Jan 2006 17:32:22 +0000 (17:32 +0000)]
Yesterday netgraph ABI has been changed.

18 years agoForget about ipfw1 and ipfw2. We aren't in RELENG_4 anymore.
glebius [Fri, 13 Jan 2006 16:44:56 +0000 (16:44 +0000)]
Forget about ipfw1 and ipfw2. We aren't in RELENG_4 anymore.

18 years agoDocument 'tablearg' keyword.
glebius [Fri, 13 Jan 2006 15:48:38 +0000 (15:48 +0000)]
Document 'tablearg' keyword.

Wording by: emaste

18 years agoAllow to use Time DPA20B 1GB MP3 Player [1]
mnag [Fri, 13 Jan 2006 11:33:40 +0000 (11:33 +0000)]
Allow to use Time DPA20B 1GB MP3 Player [1]
Allow to use Qware BeatZkey! Pro [2]

PR:             usb/81846 [1], usb/79164 [2]
Submitted by:   Ruben de Groot <rdg@bzerk.org> [1], Radek Kozlowski <radek@raadradd.com> [2]
Approved by:    njl
MFC:            1 week

18 years agoFix a misspelled reference.
ru [Fri, 13 Jan 2006 11:24:08 +0000 (11:24 +0000)]
Fix a misspelled reference.

18 years agoAdd sosend_dgram(), a greatly reduced and simplified version of sosend()
rwatson [Fri, 13 Jan 2006 10:22:01 +0000 (10:22 +0000)]
Add sosend_dgram(), a greatly reduced and simplified version of sosend()
intended for use solely with atomic datagram socket types, and relies
on the previous break-out of sosend_copyin().  Changes to allow UDP to
optionally use this instead of sosend() will be committed as a
follow-up.

18 years ago1) move all link state detection code from bge_tick_locked() to bge_link_upd()
oleg [Fri, 13 Jan 2006 08:59:40 +0000 (08:59 +0000)]
1) move all link state detection code from bge_tick_locked() to bge_link_upd()
2) use more robust way of link state handling for BCM5700 rev.B2 chip
3) workaround bug of some BCM570x chips which cause spurious "link up" messages
4) fix bug: some BCM570x chips was unable to detect link state changes after
   ifconfig down/up sequence until any 'non-link related' interrupt generated.
   (this happened due to pending internal link state attention which blocked
   interrupt generation)

Approved by: glebius (mentor)
MFC after: 1 week

18 years agoFix the interrupt race for real. Don't register the interrupt until after
scottl [Fri, 13 Jan 2006 08:18:04 +0000 (08:18 +0000)]
Fix the interrupt race for real.  Don't register the interrupt until after
the the interface has been configured.  I'm not sure how this could ever
have worked before, but it should be fixed now.  Also break out the interrupt
degresitration function into it's own step.

18 years agoDocument commit constraints for RELENG_6_*.
ru [Fri, 13 Jan 2006 06:51:43 +0000 (06:51 +0000)]
Document commit constraints for RELENG_6_*.

Approved by: core (jhb)

18 years agoIncrease the number of spinlocks, since libc's malloc implementation is
jasone [Fri, 13 Jan 2006 06:14:04 +0000 (06:14 +0000)]
Increase the number of spinlocks, since libc's malloc implementation is
about to significantly increase the number of spinlocks used.

Approved by: markm (mentor)

18 years agoDisable interrupts while we are setting up the handler. The interrupt really
scottl [Fri, 13 Jan 2006 05:04:27 +0000 (05:04 +0000)]
Disable interrupts while we are setting up the handler.  The interrupt really
shouldn't be set up or enabled until much later, but that will be investigated
at a later time.

18 years agoJoyport blacklist. Either we're facing with broken hardware
ariff [Fri, 13 Jan 2006 04:54:34 +0000 (04:54 +0000)]
Joyport blacklist. Either we're facing with broken hardware
or because this hardware need special (unknown) initialization
procedures.

Reported by: [1] Rob Clark <vx2 at tds.net>
MFC after: 3 days

[1] http://lists.freebsd.org/pipermail/freebsd-multimedia/2006-January/003535.html

18 years agoUnbreak the AMD64 build by forgoing the 'snmp_hostres' functionality.
obrien [Fri, 13 Jan 2006 02:09:15 +0000 (02:09 +0000)]
Unbreak the AMD64 build by forgoing the 'snmp_hostres' functionality.

Approved by: harti

18 years agoMention that NETWORKING is probably the right value for
dougb [Fri, 13 Jan 2006 01:09:55 +0000 (01:09 +0000)]
Mention that NETWORKING is probably the right value for
early_late_divider in a jail.

Add an explcit Xr for jail(8)

18 years ago- Fix issue when X-FreeBSD-CVS-Branch is not HEAD.
flz [Fri, 13 Jan 2006 00:47:42 +0000 (00:47 +0000)]
- Fix issue when X-FreeBSD-CVS-Branch is not HEAD.

Submitted by: jkim
Approved by: jkim

18 years agoXXX a comment in uipc_usrreq.c that requires updating.
rwatson [Fri, 13 Jan 2006 00:00:32 +0000 (00:00 +0000)]
XXX a comment in uipc_usrreq.c that requires updating.