]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoSecurity levels are not discussed further in init.8; we moved that
ceri [Thu, 19 Jan 2006 20:01:43 +0000 (20:01 +0000)]
Security levels are not discussed further in init.8; we moved that
discussion to this manual.

18 years agoRevert addtion of assertions in revision 1.99. These assertions cause
jasone [Thu, 19 Jan 2006 19:20:42 +0000 (19:20 +0000)]
Revert addtion of assertions in revision 1.99.  These assertions cause
problems in cases where regions are faked up for the purposes of red-black
tree searches, since those faked region headers reside on the stack, rather
than in a malloc chunk.

18 years agoAdd assertions that detect some forms of region separator corruption.
jasone [Thu, 19 Jan 2006 19:08:11 +0000 (19:08 +0000)]
Add assertions that detect some forms of region separator corruption.

18 years agoRemove loops in arena_coalesce(). They are no longer necessary, now that
jasone [Thu, 19 Jan 2006 18:37:30 +0000 (18:37 +0000)]
Remove loops in arena_coalesce().  They are no longer necessary, now that
internal allocation does not rely on recursive arena use (base_arena was
removed in revision 1.95).

18 years agoAdd myself to the calendar
vd [Thu, 19 Jan 2006 18:35:44 +0000 (18:35 +0000)]
Add myself to the calendar

Approved by: garga (mentor)

18 years agoAdd myself.
anray [Thu, 19 Jan 2006 15:31:37 +0000 (15:31 +0000)]
Add myself.
Feel free to donate me a beer :)

18 years agofix a bug in the fragment cache (used for 'scrub fragment crop/drop-ovl',
dhartmei [Thu, 19 Jan 2006 11:46:45 +0000 (11:46 +0000)]
fix a bug in the fragment cache (used for 'scrub fragment crop/drop-ovl',
but not 'fragment reassemble'), which can cause some fragments to get
inserted into the cache twice, thereby violating an invariant, and panic-
ing the system subsequently.

Reviewed by: mlaier
MFC after: 1 day

18 years agoIn order to maintain interoperability with certain broken FTP servers,
des [Thu, 19 Jan 2006 08:31:47 +0000 (08:31 +0000)]
In order to maintain interoperability with certain broken FTP servers,
ignore a MODE failure if and only if the mode we attempted to set was S
(which is supposed to be the default).

PR: bin/91973

18 years agoMake all internal variables and functions static.
jasone [Thu, 19 Jan 2006 07:23:13 +0000 (07:23 +0000)]
Make all internal variables and functions static.

Reported by: ache

18 years agoAdd the RB_PROTOTYPE_STATIC and RB_GENERATE_STATIC macros.
jasone [Thu, 19 Jan 2006 07:20:20 +0000 (07:20 +0000)]
Add the RB_PROTOTYPE_STATIC and RB_GENERATE_STATIC macros.

Approved by: markm (mentor)

18 years agoReturn NULL if there is an OOM error during initialization, rather than
jasone [Thu, 19 Jan 2006 02:11:05 +0000 (02:11 +0000)]
Return NULL if there is an OOM error during initialization, rather than
allowing the error to be fatal.

Move a label in order to make sure to properly handle errors in malloc(0).

Reported by: Alastair D'Silva, Saneto Takanori

18 years agoRev ioctl to get scan results:
sam [Wed, 18 Jan 2006 22:17:50 +0000 (22:17 +0000)]
Rev ioctl to get scan results:
o lock the list walk
o malloc a results buffer instead of copying out one result at a time
  using an on-stack buffer
o fix definition of ieee80211req_scan_result so size of variable-length
  information elements is large enough to hold all possible ie's
  (still only return wpa+wme, at some point may return all)
o make rssi+noise data signed; they should've been so all along
o add a bit more padding for future additions while we're here
o define a new ioctl for new api and add compat code for old ioctl
  under COMPAT_FREEBSD6 (temporarily enabled local to the file)

Reviewed by: Scott Long
MFC after: 2 weeks

18 years agoFix bio leak in case of malloc(9) failure.
pjd [Wed, 18 Jan 2006 21:44:57 +0000 (21:44 +0000)]
Fix bio leak in case of malloc(9) failure.

Found by: Coverity Prevent(tm)
Coverity ID: CID794
MFC after: 3 days

18 years agoRemove dead code.
pjd [Wed, 18 Jan 2006 21:43:27 +0000 (21:43 +0000)]
Remove dead code.

Found by: Coverity Prevent(tm)
Coverity ID: CID105
MFC after: 3 days

18 years agoRemove dead code.
pjd [Wed, 18 Jan 2006 21:42:19 +0000 (21:42 +0000)]
Remove dead code.

Found by: Coverity Prevent(tm)
Coverity ID: CID104
MFC after: 3 days

18 years ago- Fix typo
pav [Wed, 18 Jan 2006 20:36:58 +0000 (20:36 +0000)]
- Fix typo

Reported by: neologism/#freenode

18 years agoReturn the thread name in the kinfo_proc structure.
julian [Wed, 18 Jan 2006 20:27:43 +0000 (20:27 +0000)]
Return the thread name in the kinfo_proc structure.
Also correct the comment describing what the value is.

18 years agoDuh! put the thread name into the section that is zero'd on allocation
julian [Wed, 18 Jan 2006 20:18:40 +0000 (20:18 +0000)]
Duh! put the thread name into the section that is zero'd on allocation
(by default there is no name)

18 years agoCongratulations, we now have a place for a thread to store its name.
julian [Wed, 18 Jan 2006 20:12:14 +0000 (20:12 +0000)]
Congratulations, we now have a place for a thread to store its name.
Will be needed when we start using real threads in the kernel instead
of whole pseudo-processes.

18 years agobump the scan generation number before iterating so we're guaranteed
sam [Wed, 18 Jan 2006 19:56:17 +0000 (19:56 +0000)]
bump the scan generation number before iterating so we're guaranteed
to have a value that's not been used before; this fixes the problem
where the first traversal of the scan list did nothing because the
entries were initialized with the current generation number (a
separate issue)

MFC after: 1 week

18 years agoNote that the size of /boot has blown oout to 140 MBytes unless the
julian [Wed, 18 Jan 2006 19:54:51 +0000 (19:54 +0000)]
Note that the size of /boot has blown oout to 140 MBytes unless the
right options are used. Tell how to avoid this.

18 years agoRevert previous commit for now, which seems to have (re)introduced some
delphij [Wed, 18 Jan 2006 19:38:43 +0000 (19:38 +0000)]
Revert previous commit for now, which seems to have (re)introduced some
old bugs, as well as some unwanted side effects.  I will do more
investigation and fix these issues first.

Pointed out by: dwmalone

18 years ago- Bring some sysctl descriptions in sync with reality (additional
netchild [Wed, 18 Jan 2006 19:14:33 +0000 (19:14 +0000)]
- Bring some sysctl descriptions in sync with reality (additional
  values/overrides). [1]
- Add some supported cards. [2]

PR: 91916 [1]
Submitted by: mv <mv@hysteria.sk> [1]
Requested by: joel [2] (after reading my commit log)
MFC after: 3 days

18 years agoDo not derefence the ip header pointer in the IPv6 case.
andre [Wed, 18 Jan 2006 18:59:30 +0000 (18:59 +0000)]
Do not derefence the ip header pointer in the IPv6 case.
This fixes a bug in the previous commit.

Found by: Coverity Prevent(tm)
Coverity ID: CID253
Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoRemove ELF dynamic tag definitions that aren't used. Ideally,
marcel [Wed, 18 Jan 2006 18:49:25 +0000 (18:49 +0000)]
Remove ELF dynamic tag definitions that aren't used. Ideally,
those that are used should come from <machine/elf.h>.

18 years agoIn in_delayed_cksum() we can't perform a m_pullup() as it may
andre [Wed, 18 Jan 2006 18:49:16 +0000 (18:49 +0000)]
In in_delayed_cksum() we can't perform a m_pullup() as it may
change the mbuf pointer and we don't have any way of passing
it back to the callers.  Instead just fail silently without
updating the checksum but leaving the mbuf+chain intact.

A search in our GNATS database did not turn up any match for
the existing warning message when this case is encountered.

Found by: Coverity Prevent(tm)
Coverity ID: CID779
Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoThose files are present on -current (but empty). And they are needed
netchild [Wed, 18 Jan 2006 18:48:42 +0000 (18:48 +0000)]
Those files are present on -current (but empty). And they are needed
if you don't want to get some errors (e.g. in the build of
x11/XFree86-4-documents). So they don't belong here.

Noticed by: Dmitry Pryanishnikov <dmitry@atlantis.dp.ua>

18 years agoIn syncache_expand() insert a proper syncache_free() to fix a case
andre [Wed, 18 Jan 2006 18:25:03 +0000 (18:25 +0000)]
In syncache_expand() insert a proper syncache_free() to fix a case
that currently can't be triggered.  But better be safe than sorry
later on.  Additionally it properly silences Coverity Prevent for
future tests.

Found by: Coverity Prevent(tm)
Coverity ID: CID802
Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoAlways include the lock_classes[] array in the kernel. The
jhb [Wed, 18 Jan 2006 18:02:50 +0000 (18:02 +0000)]
Always include the lock_classes[] array in the kernel.  The
"is it a spinlock" test in mtx_destroy() needs it even in non-debug
kernels.

Reported by: danfe

18 years agodon't widen the ssid field by default, this screws up other formatting;
sam [Wed, 18 Jan 2006 17:45:38 +0000 (17:45 +0000)]
don't widen the ssid field by default, this screws up other formatting;
make it available with the -v option

18 years agoFix two accesses to uninitialized variables that a revision 1.27
ru [Wed, 18 Jan 2006 16:09:00 +0000 (16:09 +0000)]
Fix two accesses to uninitialized variables that a revision 1.27
has introduced.

Found with: Coverity Prevent(tm)

18 years agoPrevent dereferencing a NULL route pointer when trying to update the
andre [Wed, 18 Jan 2006 15:05:05 +0000 (15:05 +0000)]
Prevent dereferencing a NULL route pointer when trying to update the
route MTU.

This bug is very difficult to reach and not remotely exploitable.

Found by: Coverity Prevent(tm)
Coverity ID: CID162
Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoBe ready to a case when not a constant number of segments is returned
glebius [Wed, 18 Jan 2006 14:31:21 +0000 (14:31 +0000)]
Be ready to a case when not a constant number of segments is returned
by bus_dmamap_load_mbuf_sg() on jumbo buffer allocation.

Reviewed by: scottl, gallatin

18 years agoReturn mbuf pointer or NULL from ip_fastforward() as the mbuf pointer
andre [Wed, 18 Jan 2006 14:24:39 +0000 (14:24 +0000)]
Return mbuf pointer or NULL from ip_fastforward() as the mbuf pointer
may have changed by m_pullup() during fastforward processing.

While this is a bug it is actually never triggered in real world
situations and it is not remotely exploitable.

Found by: Coverity Prevent(tm)
Coverity ID: CID780
Sponsored by: TCP/IP Optimization Fundraise 2005

18 years agoCorrect two promise entries, these are only 2port SATA.
sos [Wed, 18 Jan 2006 13:34:28 +0000 (13:34 +0000)]
Correct two promise entries, these are only 2port SATA.

18 years agoWhitespace cleanup.
sos [Wed, 18 Jan 2006 13:10:17 +0000 (13:10 +0000)]
Whitespace cleanup.

18 years agoStyle cleanups.
pjd [Wed, 18 Jan 2006 11:03:20 +0000 (11:03 +0000)]
Style cleanups.

X-MFC-after: Already MFCed to RELENG_6 by accident.

18 years agoResolve the mount point's path with realpath(2) before checking if file
pjd [Wed, 18 Jan 2006 11:00:34 +0000 (11:00 +0000)]
Resolve the mount point's path with realpath(2) before checking if file
system is mounted. This prevevents duplicated mounts.

The change I made against the original patch is to fall back to the given
path on realpath(2) failure instead of exiting with an error.

Submitted by: Andreas Kohn <andreas@syndrom23.de>
PR: bin/89782
MFC after: 3 days

18 years agoProperly report SATA connection speed.
sos [Wed, 18 Jan 2006 10:02:23 +0000 (10:02 +0000)]
Properly report SATA connection speed.

18 years agoProperly print the SATA protocal version.
sos [Wed, 18 Jan 2006 10:01:43 +0000 (10:01 +0000)]
Properly print the SATA protocal version.

18 years agoCheck ifp before dereferencing it in xl_detach(). xl_detach() can be called
glebius [Wed, 18 Jan 2006 09:42:21 +0000 (09:42 +0000)]
Check ifp before dereferencing it in xl_detach(). xl_detach() can be called
from xl_attach(), when ifp is not defined yet.

Found with: Coverity Prevent(tm)

18 years agoAdd support for using DMA on dump, greatly speeds up the dump process.
sos [Wed, 18 Jan 2006 09:14:55 +0000 (09:14 +0000)]
Add support for using DMA on dump, greatly speeds up the dump process.
Add dump support in ataraid.

18 years agoCorrect a buffer overflow when scanning for 802.11 wireless networks.
cperciva [Wed, 18 Jan 2006 09:03:15 +0000 (09:03 +0000)]
Correct a buffer overflow when scanning for 802.11 wireless networks.

Security: FreeBSD-SA-06:05.80211

18 years agoFix ioctl to not get swallowed by ioctl code (i.e., make sure the darned
mjacob [Wed, 18 Jan 2006 08:37:27 +0000 (08:37 +0000)]
Fix ioctl to not get swallowed by ioctl code (i.e., make sure the darned
code sticks within 8 bits)

MFC after: 1 week

18 years agoImproves and cleanups over inetd(8):
delphij [Wed, 18 Jan 2006 07:47:46 +0000 (07:47 +0000)]
Improves and cleanups over inetd(8):

 - Teach inetd(8) about kqueue, originally implemented by jmg@[1].
 - Use new C99 style function prototypes instead of K&Rs.
 - Raise WARNS from 2 to 6

Glanced at by: ru
MFC After: 2 weeks

[1] http://people.freebsd.org/~jmg/inetd.kq.patch,
    http://people.freebsd.org/~jmg/inetd.kq.html

18 years agoEliminate a stale instruction introduced in revision 1.136.
davidxu [Wed, 18 Jan 2006 06:42:42 +0000 (06:42 +0000)]
Eliminate a stale instruction introduced in revision 1.136.

18 years agoDisable "tp" support until I figure out why it's breaking the build. <sigh>
kientzle [Wed, 18 Jan 2006 06:26:42 +0000 (06:26 +0000)]
Disable "tp" support until I figure out why it's breaking the build. <sigh>

18 years agoBump date to Sept 20, the date of the last content change
imp [Wed, 18 Jan 2006 05:48:45 +0000 (05:48 +0000)]
Bump date to Sept 20, the date of the last content change

18 years agoTell nextboot to clean up after itself.
wes [Wed, 18 Jan 2006 04:53:48 +0000 (04:53 +0000)]
Tell nextboot to clean up after itself.

18 years agoRevert previous change, "getopts" (with an 's') is a builtin in
wes [Wed, 18 Jan 2006 04:48:45 +0000 (04:48 +0000)]
Revert previous change, "getopts" (with an 's') is a builtin in
/bin/sh.

18 years agoAdd a warning pointing out that incomplete ports trees are not
cperciva [Wed, 18 Jan 2006 03:40:57 +0000 (03:40 +0000)]
Add a warning pointing out that incomplete ports trees are not
supported and that users of REFUSE directives should update their
entire ports tree before asking for help on the mailing lists.

Requested by: kris
MFC after: 3 days

18 years agoo Add missing relocations.
marcel [Wed, 18 Jan 2006 01:45:57 +0000 (01:45 +0000)]
o  Add missing relocations.
o  Minor white-space fixups.

18 years agoRemove Giant / busdma_lock_mutex from busdma_tag_create().
ariff [Wed, 18 Jan 2006 01:01:37 +0000 (01:01 +0000)]
Remove Giant / busdma_lock_mutex from busdma_tag_create().

18 years ago - Count packets discarded by RX/TX MAC (cause of FIFO overflow, etc)
oleg [Tue, 17 Jan 2006 23:01:58 +0000 (23:01 +0000)]
 - Count packets discarded by RX/TX MAC (cause of FIFO overflow, etc)
   as input/output interface errors.
 - Keep values of rx/tx discards & tx collisions inside struct bge_softc.
   So we can keep statistic across ifconfig down/up runs (cause bringing
   bge up will reset chip).

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

18 years agos/R_IA64_/R_IA_64_/g as per the ia64 psABI.
marcel [Tue, 17 Jan 2006 21:03:22 +0000 (21:03 +0000)]
s/R_IA64_/R_IA_64_/g as per the ia64 psABI.

18 years agoSince p_cansee will end up dereferencing p_ucred, don't check for p_ucred
jmallett [Tue, 17 Jan 2006 20:25:01 +0000 (20:25 +0000)]
Since p_cansee will end up dereferencing p_ucred, don't check for p_ucred
equal to NULL several times later.  p_ucred "should probably not" be NULL
if the process isn't PRS_NEW anyway.  This is strongly reinforced by the fact
that we don't see frequent crashes here.  Remove the checks after p_cansee and
add a KASSERT right before it.

Found by: Coverity Prevent (tm)

Also trim one nearby trailing space.

18 years agoUse the .Fx macro.
simon [Tue, 17 Jan 2006 19:37:07 +0000 (19:37 +0000)]
Use the .Fx macro.

MFC after: 3 days

18 years agoRemove the module loading magic again; it's not needed after all.
philip [Tue, 17 Jan 2006 19:29:31 +0000 (19:29 +0000)]
Remove the module loading magic again; it's not needed after all.

Pointy hat to: matteo
Submitted by: matteo
Reviewed by: pjd
MFC after: 3 days

18 years agoI ran into an nfs client panic a couple of times in a row over the
alfred [Tue, 17 Jan 2006 17:29:03 +0000 (17:29 +0000)]
I ran into an nfs client panic a couple of times in a row over the
last few days.  I tracked it down to the fact that nfs_reclaim()
is setting vp->v_data to NULL _before_ calling vnode_destroy_object().
After silence from the mailing list I checked further and discovered
that ufs_reclaim() is unique among FreeBSD filesystems for calling
vnode_destroy_object() early, long before tossing v_data or much
of anything else, for that matter.  The rest, including NFS, appear
to be identical, as if they were just clones of one original routine.

The enclosed patch fixes all file systems in essentially the same
way, by moving the call to vnode_destroy_object() to early in the
routine (before the call to vfs_hash_remove(), if any).  I have
only tested NFS, but I've now run for over eighteen hours with the
patch where I wouldn't get past four or five without it.

Submitted by: Frank Mayhar
Requested by: Mohan Srinivasan
MFC After: 1 week

18 years agoFix a memory leak I introduced with the hostb/vgapci stuff.
jhb [Tue, 17 Jan 2006 17:02:45 +0000 (17:02 +0000)]
Fix a memory leak I introduced with the hostb/vgapci stuff.

Reported by: Coverity (via dfr's clue-bat)

18 years agoBah. Fix 'show lock' to actually be compiled in. I had just fixed this in
jhb [Tue, 17 Jan 2006 16:58:32 +0000 (16:58 +0000)]
Bah.  Fix 'show lock' to actually be compiled in.  I had just fixed this in
p4 but had an older subr_lock.c on the machine I committed to CVS from.

18 years agoAdd a new file (kern/subr_lock.c) for holding code related to struct
jhb [Tue, 17 Jan 2006 16:55:17 +0000 (16:55 +0000)]
Add a new file (kern/subr_lock.c) for holding code related to struct
lock_obj objects:
- Add new lock_init() and lock_destroy() functions to setup and teardown
  lock_object objects including KTR logging and registering with WITNESS.
- Move all the handling of LO_INITIALIZED out of witness and the various
  lock init functions into lock_init() and lock_destroy().
- Remove the constants for static indices into the lock_classes[] array
  and change the code outside of subr_lock.c to use LOCK_CLASS to compare
  against a known lock class.
- Move the 'show lock' ddb function and lock_classes[] array out of
  kern_mutex.c over to subr_lock.c.

18 years agoFix a memory leak.
jkoshy [Tue, 17 Jan 2006 16:53:50 +0000 (16:53 +0000)]
Fix a memory leak.

Found by: Coverity

18 years agoInitialize thread0.td_contested in init_turnstiles() rather than
jhb [Tue, 17 Jan 2006 16:47:42 +0000 (16:47 +0000)]
Initialize thread0.td_contested in init_turnstiles() rather than
mutex_init() as it is used by the turnstile code and is not mutex-specific.

18 years agoUpdate a stale comment.
jhb [Tue, 17 Jan 2006 16:44:34 +0000 (16:44 +0000)]
Update a stale comment.

18 years agoGarbage collect turnstile_empty() since it is unused.
jhb [Tue, 17 Jan 2006 16:40:20 +0000 (16:40 +0000)]
Garbage collect turnstile_empty() since it is unused.

18 years agoFix an 11 year old mistake: Let the hash functions take a void* instead
phk [Tue, 17 Jan 2006 15:35:57 +0000 (15:35 +0000)]
Fix an 11 year old mistake:  Let the hash functions take a void* instead
of unsigned char* argument.

18 years agoMove $FreeBSD$ from comment to __FBSDID().
pjd [Tue, 17 Jan 2006 11:48:16 +0000 (11:48 +0000)]
Move $FreeBSD$ from comment to __FBSDID().

18 years agoAdd two new variables:
phk [Tue, 17 Jan 2006 08:33:53 +0000 (08:33 +0000)]
Add two new variables:

# Size of the /etc ramdisk in 512 bytes sectors
NANO_RAM_ETCSIZE=10240

# Size of the /tmp+/var ramdisk in 512 bytes sectors
NANO_RAM_TMPVARSIZE=10240

And a last_orders() shell function which can be redefined to copy
finished images away, order peanuts or whatever else is appropriate

18 years agoFix a "free(): error: chunk is already free" under certain
dougb [Tue, 17 Jan 2006 08:01:00 +0000 (08:01 +0000)]
Fix a "free(): error: chunk is already free" under certain
circumstances that include circular dependencies.

PR:                     bin/91789
PR submitted by:        Frank Behrens <frank@pinky.sax.de>
Patch submitted by:     Divacky Roman <xdivac02@stud.fit.vutbr.cz>

18 years ago- Use better types.
pjd [Tue, 17 Jan 2006 07:32:43 +0000 (07:32 +0000)]
- Use better types.
- Log problems at level 0 when killing providers.

MFC after: 3 days

18 years agoCheck return value.
pjd [Tue, 17 Jan 2006 07:30:34 +0000 (07:30 +0000)]
Check return value.

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

18 years agoRemove dead code.
pjd [Tue, 17 Jan 2006 07:27:46 +0000 (07:27 +0000)]
Remove dead code.

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

18 years agoRemove unused value.
pjd [Tue, 17 Jan 2006 07:26:48 +0000 (07:26 +0000)]
Remove unused value.

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

18 years agoLog situation when EIO is returned.
pjd [Tue, 17 Jan 2006 07:23:36 +0000 (07:23 +0000)]
Log situation when EIO is returned.

18 years agoRemove bio leak when EIO error is emulated.
pjd [Tue, 17 Jan 2006 07:22:44 +0000 (07:22 +0000)]
Remove bio leak when EIO error is emulated.

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

18 years agoThe number of ticks per usec for YUKON_EC is 125.
yongari [Tue, 17 Jan 2006 06:58:25 +0000 (06:58 +0000)]
The number of ticks per usec for YUKON_EC is 125.

18 years agofix Rx checksum computation on little endian systems.
yongari [Tue, 17 Jan 2006 06:02:22 +0000 (06:02 +0000)]
fix Rx checksum computation on little endian systems.

Reported & Tested by: brad@OpenBSD

18 years agofix interrupt moderation timer frequencies for Yukon
yongari [Tue, 17 Jan 2006 05:57:44 +0000 (05:57 +0000)]
fix interrupt moderation timer frequencies for Yukon

Obtained from: OpenBSD

18 years agoremove trailing spaces
yongari [Tue, 17 Jan 2006 05:41:20 +0000 (05:41 +0000)]
remove trailing spaces

18 years agoIf the attempt to open the archive fails (either the client open
kientzle [Tue, 17 Jan 2006 04:49:04 +0000 (04:49 +0000)]
If the attempt to open the archive fails (either the client open
routine fails or the first read fails), invoke the client close
routine immediately so the client can clean up.  Also, don't store the
client pointers in this case, so that the client close routine can't
accidentally get called more than once.

A minor style fix to archive_read_open_fd.c while I'm here.

PR: 86453
Thanks to: Andrew Turner for reporting this and suggesting a fix.

18 years agoAdd support for "tp" format. tp was the standard system
kientzle [Tue, 17 Jan 2006 03:40:42 +0000 (03:40 +0000)]
Add support for "tp" format.  tp was the standard system
archiver for Fourth Edition through Sixth Edition Unix; it was
replaced by tar in Seventh Edition.  (First Edition through
Third Edition used "tap.")

Unfortunately, tp was not so very standard; there were a
few different variants.  The code here attempts to support
what I believe were the most common variants.

tp support is not yet enabled by archive_read_support_format_all(),
as I'm not yet entirely comfortable with the detection
heuristics.  People interested in experimenting can
add archive_read_support_format_tp() just after any calls
to archive_read_support_format_all() in bsdtar to see how
well this works.

TODO: tp format is roughly similar in structure to dump/restore
   archive formats used by many systems.  It should be possible
   to generalize this code to handle many dump/restore variants.
   Format detection heuristics are going to be rough, though.

Thanks to: Warren Toomey, whose very basic tp extraction programs
   and documentation made this possible.

18 years agoMention that the random time slept by "portsnap cron" is between 1 and
cperciva [Tue, 17 Jan 2006 03:02:44 +0000 (03:02 +0000)]
Mention that the random time slept by "portsnap cron" is between 1 and
3600 seconds.

Suggested by: Niki Denev
MFC after: 3 days

18 years agoSet flag in needsbuffer while still holding bqlock to avoid lost wakeup.
tegge [Mon, 16 Jan 2006 22:09:47 +0000 (22:09 +0000)]
Set flag in needsbuffer while still holding bqlock to avoid lost wakeup.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r154439,
sam [Mon, 16 Jan 2006 20:33:23 +0000 (20:33 +0000)]
This commit was generated by cvs2svn to compensate for changes in r154439,
which included commits to RCS files with non-trunk default branches.

18 years agocorrect dlt buffer alloc; this goes on the vendor branch as it
sam [Mon, 16 Jan 2006 20:33:23 +0000 (20:33 +0000)]
correct dlt buffer alloc; this goes on the vendor branch as it
will be committed there shortly

MFC after: 3 days

18 years agoFix broken capabilites. There are possible calculation errors within
ariff [Mon, 16 Jan 2006 20:01:33 +0000 (20:01 +0000)]
Fix broken capabilites. There are possible calculation errors within
ess_calcspeed8() and ess_calcspeed9() that need to be fixed as well
(TODO).

Reported by: [1] Claude Buisson <cbuisson at nerim.net>
MFC after: 3 days

[1] http://lists.freebsd.org/pipermail/freebsd-multimedia/2006-January/003566.html

18 years agoehci_pci.c:
ariff [Mon, 16 Jan 2006 19:23:59 +0000 (19:23 +0000)]
ehci_pci.c:
ATI EHCI controllers exhibit simmilar stall issues and require
this dropped interrupts workaround. Be verbose about it.
ehci.c:
ehcivar.h:
Slight change in comments to note about issues surrounding both
VIA and ATI EHCI controllers.

Approved by: iedowse

18 years agoAdd PCI vendor id for ATI OHCI USB controllers.
ariff [Mon, 16 Jan 2006 19:07:05 +0000 (19:07 +0000)]
Add PCI vendor id for ATI OHCI USB controllers.

Approved by: iedowse

18 years agoAlthough we check the return value of copyin(9) while determaining how
csjp [Mon, 16 Jan 2006 17:03:21 +0000 (17:03 +0000)]
Although we check the return value of copyin(9) while determaining how
long the string is in userspace, afterwards we call malloc(M_WAITOK),
which could sleep for an unknown amount of time. Check the return
value of copyin(9) just to be sure that nothing has changed during that
time.

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

18 years agoFix standalone module build.
ru [Mon, 16 Jan 2006 17:03:06 +0000 (17:03 +0000)]
Fix standalone module build.

Reported by: Boris Samorodov <bsam@ipt.ru>

18 years agoI couldn't find any traces of what the tags1 file was supposed to do.
ru [Mon, 16 Jan 2006 16:25:17 +0000 (16:25 +0000)]
I couldn't find any traces of what the tags1 file was supposed to do.

18 years agoRemove dependency on getopts because it is on /usr and we want to
wes [Mon, 16 Jan 2006 16:18:54 +0000 (16:18 +0000)]
Remove dependency on getopts because it is on /usr and we want to
use 'nextboot -D' in rc.  Option parsing cribbed from vgrind.sh.

18 years agoLoad the g_md kernel module if needed.
philip [Mon, 16 Jan 2006 13:26:05 +0000 (13:26 +0000)]
Load the g_md kernel module if needed.

Pointed out by: Gianmarco Giovannelli
Submitted by: matteo
MFC after: 3 days

18 years agoRestore use of strncpy(), as there is later unconditional termination
rwatson [Mon, 16 Jan 2006 11:54:07 +0000 (11:54 +0000)]
Restore use of strncpy(), as there is later unconditional termination
of the string, and reliance on the returned pointer.

Found by: bde (tm)

18 years agoRestore old compatibility of feeding directly into /dev/dsp at 8000 hz
ariff [Mon, 16 Jan 2006 11:07:15 +0000 (11:07 +0000)]
Restore old compatibility of feeding directly into /dev/dsp at 8000 hz
while preserving the New World Order.

Discussed with: [1] Michael W. Oliver <michael at gargantuan.com>
MFC after: 1 week

[1] http://lists.freebsd.org/pipermail/freebsd-multimedia/2006-January/003562.html

18 years agoREQUIRE named. On all systems I've examined running HEAD and
dougb [Mon, 16 Jan 2006 06:03:42 +0000 (06:03 +0000)]
REQUIRE named. On all systems I've examined running HEAD and
RELENG_6 this will be a noop, however as we introduce local
startup scripts to the base rcorder, we'll see more cases
where the previous status quo will need to be made explicit
to avoid having it disrupted when random local scripts are
added to the mix.

18 years agoEliminate unused code.
davidxu [Mon, 16 Jan 2006 05:33:48 +0000 (05:33 +0000)]
Eliminate unused code.

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