]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoFix two PRs by adding vendor IDs:
imp [Thu, 21 Jun 2007 17:55:28 +0000 (17:55 +0000)]
Fix two PRs by adding vendor IDs:

PR 108097: ADMtek 851X USB-to-LAN adapter
PR 74849: Samsung SPH-i500 does not attach properly as usb modem/Palm device

PR: 108097, 74849
Approved by: re@ (kensmith trivial usb blanket)

17 years agoRemove duplicate xref and add missing commas.
brueffer [Thu, 21 Jun 2007 16:39:25 +0000 (16:39 +0000)]
Remove duplicate xref and add missing commas.

PR: 113891
Submitted by: pluknet <pluknet@gmail.com>
Approved by: re (blanket)

17 years agoLatest round of usb cleanups:
imp [Thu, 21 Jun 2007 14:42:34 +0000 (14:42 +0000)]
Latest round of usb cleanups:

o Consistently use device_foo_t and bus_foo_t for functions implementing
  device_foo and bus_foo respectively.  Adjust those routines that were wrong
  (we should do this throughout the tree).
o make all the modules depend on usb.  Otherwise these modules won't
  load.
o ucycom doesn't need usb_port.h
o Minor unifdefing
o uhub, umass, ums, urio, uscanner conversion complete.
o ukbd: Remove the NO_SET_PROTO quirk (fixes a PR 77940).  NetBSD removed
  their check and setting the proto a long time ago.
o umodem panic fixed.  UQ_ASSUME_CM_OVER_DATA quirk removed because I've never
  seen a umodem that needed this rejection for proection (this gets rid of
  ~20% of the quirks).

Approved by: re@ (kensmith)
PR: 77940

17 years agoRestore approximately FreeBDS 6.x's level of compatibility macros.
imp [Thu, 21 Jun 2007 14:36:22 +0000 (14:36 +0000)]
Restore approximately FreeBDS 6.x's level of compatibility macros.
Older drivers that do not wish to convert to the native API (which
will work with both 6.x and 7.x) can simply include
<dev/usb/usb_port.h>.  Drivers in the tree shouldn't these macros,
unless they actually work on other OSes and are actively maintained.

Approved by: re@

17 years agoResolve conflicts after less(1) v406 import.
delphij [Thu, 21 Jun 2007 10:44:50 +0000 (10:44 +0000)]
Resolve conflicts after less(1) v406 import.

Approved by: re (hrs)

17 years agoThis commit was generated by cvs2svn to compensate for changes in r170964,
delphij [Thu, 21 Jun 2007 10:42:55 +0000 (10:42 +0000)]
This commit was generated by cvs2svn to compensate for changes in r170964,
which included commits to RCS files with non-trunk default branches.

17 years agoVirgin import of less v406.
delphij [Thu, 21 Jun 2007 10:42:55 +0000 (10:42 +0000)]
Virgin import of less v406.

Approved by: re (hrs)

17 years agoRestore a historical behavior that +foo is considered as a
delphij [Thu, 21 Jun 2007 10:39:24 +0000 (10:39 +0000)]
Restore a historical behavior that +foo is considered as a
filename by more(1).  The less(1) behavior is keep intact.

PR: bin/51488
Prodded by: demon
Approved by: re (hrs)

17 years agos/logprintf/printf/g
imp [Wed, 20 Jun 2007 05:11:37 +0000 (05:11 +0000)]
s/logprintf/printf/g

Approved by: re@

17 years agoUse .CURDIR when referring source code. This fixes build
delphij [Wed, 20 Jun 2007 00:29:51 +0000 (00:29 +0000)]
Use .CURDIR when referring source code.  This fixes build
when 'make obj' was done first.  I found this when fixing
a problem reported by tinderbox, but forgot to send the
patchset to re@ altogether.

Approved by: re (kensmith)

17 years agoClean up after previous commit: new sentence -> new line, no empty lines,
brueffer [Tue, 19 Jun 2007 16:29:46 +0000 (16:29 +0000)]
Clean up after previous commit: new sentence -> new line, no empty lines,
.Fa macro usage.

Approved by: re (blanket)

17 years ago- Fix the signature of sctp_connectx to match the function and adds some
rrs [Tue, 19 Jun 2007 16:18:43 +0000 (16:18 +0000)]
- Fix the signature of sctp_connectx to match the function and adds some
  text about the last argument.
Approved by: re (bmah@freebsd.org)

17 years ago- Fix RFC reference to SCTP to be correct.
rrs [Tue, 19 Jun 2007 15:56:32 +0000 (15:56 +0000)]
- Fix RFC reference to SCTP to be correct.
Approved by: re (bmah@freebsd.org)

17 years agoUpdated release notes: Note that diffutils was formerly version 2.7 [1].
bmah [Tue, 19 Jun 2007 15:08:43 +0000 (15:08 +0000)]
Updated release notes:  Note that diffutils was formerly version 2.7 [1].
Add some more verbage with respect to jemalloc and resource limits [2].

Submitted by: delphij [1]
Collaboration with: jasone [2]
Approved by: re (implicit)

17 years agoFix build problem caused by a set of typos.
delphij [Tue, 19 Jun 2007 14:56:35 +0000 (14:56 +0000)]
Fix build problem caused by a set of typos.

Reported by: tinderbox
Approved by: re (mux)

17 years agoNew release notes:
brueffer [Tue, 19 Jun 2007 14:13:46 +0000 (14:13 +0000)]
New release notes:
- tmpfs on i386,amd64
- gem(4) altq support
- AUDIT and SCTP in GENERIC
- FireWire MPSAFE
- netstat SCTP support
- dhclient RFC 3442 support (1)

Modified release notes:
- MFC markers: mxge(4) and netcat
- touch -A entry grammar fix (2)
- More manpage references used

Submitted by: Andrey V. Elsukov (1)
ceri (2)
Approved by: re (blanket)

17 years agodevfs_free() calls free_unr(), that may sleep.
kib [Tue, 19 Jun 2007 13:19:23 +0000 (13:19 +0000)]
devfs_free() calls free_unr(), that may sleep.
Postpone call to devfs_free() after cdev mutex is dropped. Reuse
cdp_list link for queuing devices awaiting deletion in the
cdevp_free_list.

Reported by: Hans Petter Selasky <hselasky c2i net>
Tested by: Peter Holm
Approved by: re (kensmith)
MFC after: 2 weeks

17 years agoAdd the witness warning for free_unr. Function could sleep, thus callers
kib [Tue, 19 Jun 2007 13:13:17 +0000 (13:13 +0000)]
Add the witness warning for free_unr. Function could sleep, thus callers
shall not have any non-sleepable locks held.

Submitted by: Hans Petter Selasky <hselasky c2i net>
Approved by: re (kensmith)

17 years agoInclude usb_port.h to fix build.
delphij [Tue, 19 Jun 2007 10:28:08 +0000 (10:28 +0000)]
Include usb_port.h to fix build.

Reported by: tinderbox
Approved by: re (mux)

17 years ago- Replace rather inefficient bubble sort with a recursive depth-first search.
pav [Mon, 18 Jun 2007 22:49:13 +0000 (22:49 +0000)]
- Replace rather inefficient bubble sort with a recursive depth-first search.
  This speeds up registration of packages considerably.
- style(9) police welcome!

PR: bin/112630
Submitted by: Stephen Montgomery-Smith <stephen@cauchy.math.missouri.edu>
Tested by: bento i386 experimental run
MFC after: 14 days

17 years agoFix headphone jack sensing support for Olivetti Olibook 610-430 XPSE.
ariff [Mon, 18 Jun 2007 22:39:27 +0000 (22:39 +0000)]
Fix headphone jack sensing support for Olivetti Olibook 610-430 XPSE.

Tested by: Gonzalo Lionel Rodriguez

17 years ago- Two more static analisys bugs found by cisco's tool on a subsequent
rrs [Mon, 18 Jun 2007 22:36:52 +0000 (22:36 +0000)]
- Two more static analisys bugs found by cisco's tool on a subsequent
  run.

17 years agoMark the few remaining drivers that haven't been converted as needing
imp [Mon, 18 Jun 2007 22:32:09 +0000 (22:32 +0000)]
Mark the few remaining drivers that haven't been converted as needing
usb_port.h by explicitly including it.

17 years agoExpand usb_port macros
imp [Mon, 18 Jun 2007 22:31:35 +0000 (22:31 +0000)]
Expand usb_port macros

17 years agoRemove usb_port.h from usb.h, except for one or two useful things that
imp [Mon, 18 Jun 2007 22:30:53 +0000 (22:30 +0000)]
Remove usb_port.h from usb.h, except for one or two useful things that
don't belong there (these moved into usb.h).

17 years agoExpand usb_port.h macros.
imp [Mon, 18 Jun 2007 22:30:08 +0000 (22:30 +0000)]
Expand usb_port.h macros.

17 years agoExpand usb_callout_init
imp [Mon, 18 Jun 2007 22:28:32 +0000 (22:28 +0000)]
Expand usb_callout_init

17 years agoFinish removal of usb_port.h macros.
imp [Mon, 18 Jun 2007 22:27:57 +0000 (22:27 +0000)]
Finish removal of usb_port.h macros.

17 years agoExplicitly use usb_port.h to get compat macros.
imp [Mon, 18 Jun 2007 22:26:41 +0000 (22:26 +0000)]
Explicitly use usb_port.h to get compat macros.

17 years agoUse device_foo_t to declare kobj methods
imp [Mon, 18 Jun 2007 22:25:46 +0000 (22:25 +0000)]
Use device_foo_t to declare kobj methods
Fix shutdown type to return int rather than void.

17 years agofinish removing usb_port.h macros.
imp [Mon, 18 Jun 2007 22:24:32 +0000 (22:24 +0000)]
finish removing usb_port.h macros.
Use device_foo_t to declare functions
a little de-k&r

17 years agoFinish removing usb_port.h compat macros.
imp [Mon, 18 Jun 2007 22:23:20 +0000 (22:23 +0000)]
Finish removing usb_port.h compat macros.

17 years agoAccording to the default font size on sparc64 provide a 12 x 22
marius [Mon, 18 Jun 2007 22:15:17 +0000 (22:15 +0000)]
According to the default font size on sparc64 provide a 12 x 22
mouse pointer instead of a 8 x 16 one so device drivers don't
need to bring there own one there and in gfb_mouse() (ab)use
the pixel_mask argument of putm() to pass along on/off info as
erasing the mouse cursor image by redrawing the text underneath
doesn't work as we use hardware cursors on sparc64.

17 years ago- Fixes cstatic issues found by cisco sa tool (missing frees and such
rrs [Mon, 18 Jun 2007 21:59:15 +0000 (21:59 +0000)]
- Fixes cstatic issues found by cisco sa tool (missing frees and such
  on error legs)
- align sctp_sockstore to 64 bit boundary ..

17 years ago- Move ofw_pci_alloc_busno() to the ofw_pci KOBJ interface,
marius [Mon, 18 Jun 2007 21:49:42 +0000 (21:49 +0000)]
- Move ofw_pci_alloc_busno() to the ofw_pci KOBJ interface,
  allowing the driver for the host-PCI-bridge to indicate that
  reenumeration of the PCI busses isn't supported by returning
  -1 instead of a valid PCI bus number. This is needed in order
  support both Tomatillo, which don't support reenumeration and
  thus are apparently intended to be used for independently
  numbered PCI domains only, and Psycho bridges, whose busses
  need to be reenumerated on at least some E450, without the
  #ifndef currently used for sun4v in order to support multiple
  independently PCI domains. The actual allocation/incrementation
  of the PCI bus numbers is now done in psycho(4), though it
  no longer establish a mapping between bus numbers and device
  nodes like ofw_pci_alloc_busno() did as that functionality
  wasn't used (but can easily brought back if really needed).
  The now no longer used sys/sparc64/pci/ofw_pci.c is also
  removed from sys/conf/files.sun4v as ofw_pci_alloc_busno()
  wasn't used there in the first place.
- In ofw_pci_default_{adjust_busrange,intr_pending}() sanity
  check that the device has a parent before passing it on.
- Make psycho_softcs static to sys/sparc64/pci/psycho.c as
  it's not used outside of that module.
- In sys/sparc64/pci/ofw_pcib_subr.c remove the superfluous
  inclusion of opt_global.h and correct the debug output for
  adjusting the subordinate bus number.

17 years agoFor sun4u also add PCI busses with a device unit number of -1
marius [Mon, 18 Jun 2007 21:46:07 +0000 (21:46 +0000)]
For sun4u also add PCI busses with a device unit number of -1
instead of using the PCI bus number, like it's already done for
sun4v in order to deal properly with independently numbered PCI
domains which can't be reenumerated (in the case of sun4u f.e.
Tomatillo bridges). For machines where we need to reenumerate
all PCI busses this change obviously introduces the theoretical
cosmetic problem that the device number of the PCI bus no longer
equals to its PCI bus number. In practice this doesn't happen
as both are assigned linearly and in parallel.

17 years agoRemove some stuff I accidently left in the previous commit.
brueffer [Mon, 18 Jun 2007 19:19:02 +0000 (19:19 +0000)]
Remove some stuff I accidently left in the previous commit.

17 years agoComplete SYNOPSIS and some cleanup.
brueffer [Mon, 18 Jun 2007 18:58:15 +0000 (18:58 +0000)]
Complete SYNOPSIS and some cleanup.

17 years ago- Add liblwres.so.10 which is now .30 after latest bind import
rafan [Mon, 18 Jun 2007 18:50:59 +0000 (18:50 +0000)]
- Add liblwres.so.10 which is now .30 after latest bind import
- Add obsolete libraries which got a version bump

17 years ago- Bump share library version which were missed in last bump
rafan [Mon, 18 Jun 2007 18:47:54 +0000 (18:47 +0000)]
- Bump share library version which were missed in last bump

Reported by:       jhb
Discussed with:      deischen, des, doubg, harti
Approved by:      re (kensmith)

17 years agoFix some debugging code that crept in accidentally.
scottl [Mon, 18 Jun 2007 18:28:43 +0000 (18:28 +0000)]
Fix some debugging code that crept in accidentally.

17 years agoo Make ipfw set more robust -- now it is possible:
maxim [Mon, 18 Jun 2007 17:52:37 +0000 (17:52 +0000)]
o Make ipfw set more robust -- now it is possible:
    - to show a specific set: ipfw set 3 show
    - to delete rules from the set: ipfw set 9 delete 100 200 300
    - to flush the set: ipfw set 4 flush
    - to reset rules counters in the set: ipfw set 1 zero

PR: kern/113388
Submitted by: Andrey V. Elsukov
Approved by: re (kensmith)
MFC after: 6 weeks

17 years agoUse vfs_timestamp() instead of nanotime() - make it up to
delphij [Mon, 18 Jun 2007 14:40:19 +0000 (14:40 +0000)]
Use vfs_timestamp() instead of nanotime() - make it up to
the user to make decisions about how detail they wanted
timestamps to have.

17 years agoAdd additional logging level mask for packet_logging too.
rrs [Mon, 18 Jun 2007 13:57:37 +0000 (13:57 +0000)]
Add additional logging level mask for packet_logging too.

17 years agoNote PAM support in atrun(8) and cron(8).
yar [Mon, 18 Jun 2007 12:03:02 +0000 (12:03 +0000)]
Note PAM support in atrun(8) and cron(8).

17 years agoMajor cleanup: mdoc macros, style, typos etc.
brueffer [Mon, 18 Jun 2007 10:20:32 +0000 (10:20 +0000)]
Major cleanup: mdoc macros, style, typos etc.

17 years agoWe only flush entries related to the given file system. Currently there are
pjd [Mon, 18 Jun 2007 09:28:24 +0000 (09:28 +0000)]
We only flush entries related to the given file system. Currently there are
no 'invalid' cache entires - file system is responsible for keeping it that
way. The comment should have been updated in rev.1.25.

17 years ago/usr/lib/libvgl.so.4 was bumped last month (20070521).
rafan [Mon, 18 Jun 2007 09:00:01 +0000 (09:00 +0000)]
/usr/lib/libvgl.so.4 was bumped last month (20070521).

Approved by: delphij (mentor)

17 years agoUpdate comment: kernel privileges are, in fact sorted by subsytem.
rwatson [Mon, 18 Jun 2007 07:54:27 +0000 (07:54 +0000)]
Update comment: kernel privileges are, in fact sorted by subsytem.

17 years agoDrop the default zones that are now covered by the new zones that
dougb [Mon, 18 Jun 2007 06:29:45 +0000 (06:29 +0000)]
Drop the default zones that are now covered by the new zones that
were added in the last revision.

17 years agoBring our default named configuration more in line with current
dougb [Mon, 18 Jun 2007 05:58:23 +0000 (05:58 +0000)]
Bring our default named configuration more in line with current
best practices:

1. The old way of generating the localhost zones was not optimal both
because they did not exist by default, and because they were not really
aligned with BCP. There is no need to have the dynamic data that the
make-localhost script generated, and good reasons to do this more
"by the book."

2. In named.conf
a. Clean up white space
b. Add/clarify a few comments
c. Slave zones from the root servers instead of using a hints
file. This has several advantages, as described in the comments.
d. Significantly revamp the default zones, including the
forward localhost zone, and the reverse zones for IPv4 and IPv6
loopback addresses. There are extensive comments describing what
is included and why. Interested readers should take the time to
review the RFCs mentioned in the comments. There is also relevant
information about the motivations for hosting these zones in the
"work in progress" Internet-Draft,
http://www.ietf.org/internet-drafts/draft-ietf-dnsop-default-local-zones-02.txt
or its successor.
It's also worth noting that a significant number of these
empty zones are already included by default in the named binary
without any user configuration.
e. Because we're including a lot of examples of both local
forward zones and slave zones in the default configuration,
eliminate some of those examples.

3. Add new localhost-{forward|reverse} zone files, and an "empty" zone
to support the changes in 2.d. above. The empty zone file isn't really
empty in order to avoid a warning from BIND about a zone file that
doesn't contain any A or AAAA records.

17 years agoAdd a namedb/master directory for the zone files I'm about to add,
dougb [Mon, 18 Jun 2007 05:44:38 +0000 (05:44 +0000)]
Add a namedb/master directory for the zone files I'm about to add,
and switch to the more "normal" way of installing files for the
namedb directory so that we can pick up the new subdir.

17 years agominor style(9) polishing
imp [Mon, 18 Jun 2007 02:15:58 +0000 (02:15 +0000)]
minor style(9) polishing

# but we need a usb_match function, if we don't already have one...

17 years agoAdd rewind() to the list of functions which may fail and set errno
ache [Mon, 18 Jun 2007 02:13:04 +0000 (02:13 +0000)]
Add rewind() to the list of functions which may fail and set errno

17 years agoAdd mbstate clear missed in one of the cases.
ache [Mon, 18 Jun 2007 02:09:08 +0000 (02:09 +0000)]
Add mbstate clear missed in one of the cases.
Move overflow check for fseek as early as needed.

17 years agoEliminate unnecessary checks from vm_pageout_clean(): The page that is
alc [Mon, 18 Jun 2007 02:04:38 +0000 (02:04 +0000)]
Eliminate unnecessary checks from vm_pageout_clean(): The page that is
passed to vm_pageout_clean() cannot possibly be PG_UNMANAGED because
it came from the inactive queue and PG_UNMANAGED pages are not in any
page queue.  Moreover, PG_UNMANAGED pages only exist in OBJT_PHYS
objects, and all pages within a OBJT_PHYS object are PG_UNMANAGED.
So, if the page that is passed to vm_pageout_clean() is not
PG_UNMANAGED, then it cannot be from an OBJT_PHYS object and its
neighbors from the same object cannot themselves be PG_UNMANAGED.

Reviewed by: tegge

17 years agoFix library names.
davidxu [Mon, 18 Jun 2007 01:50:03 +0000 (01:50 +0000)]
Fix library names.

17 years agoMFp4: fix two locking problems:
delphij [Mon, 18 Jun 2007 01:43:13 +0000 (01:43 +0000)]
MFp4: fix two locking problems:

 - Hold TMPFS_LOCK while updating tm_pages_used.
 - Hold vm page while doing uiomove.

This will hopefully fix all known panics.

Submitted by: Howard Su

17 years agoTrack whether the current read stream supports seek(). For now, we
kientzle [Mon, 18 Jun 2007 00:36:54 +0000 (00:36 +0000)]
Track whether the current read stream supports seek().  For now, we
assume yes unless seek has previously failed, but I fear I'll have to
avoid seeks under other circumstances.  (For instance, tape drives on
FreeBSD seem to return garbage from lseek().)  Also, optimize away
zero-byte skips.

17 years ago- The packet log needs to copy all of the buffer not to the end.
rrs [Sun, 17 Jun 2007 23:43:37 +0000 (23:43 +0000)]
- The packet log needs to copy all of the buffer not to the end.

17 years agoApply a vendor patch that fixes very annoying bug
delphij [Sun, 17 Jun 2007 23:20:43 +0000 (23:20 +0000)]
Apply a vendor patch that fixes very annoying bug
which caused the cursor not to return to the place
it should when -w is used with -e.

Note that this is a commit against -HEAD because
we already have the file off the vendor branch.

Reported by: Tom Uffner (on -current@)
Obtained from: http://www.greenwoodsoftware.com/less/less-405.tar.gz

17 years agoHave gpart synthesize a disk geometry if the underlying provider
marcel [Sun, 17 Jun 2007 22:19:19 +0000 (22:19 +0000)]
Have gpart synthesize a disk geometry if the underlying provider
don't have it. Some partitioning schemes, as well as file systems,
operate on the geometry and without it such schemes (e.g. MBR)
and file systems (e.g. FAT) can't be created. This is useful for
memory disks.

17 years agoSilence some gcc 4 warnings. It is expected that the bpf_movein() routine
csjp [Sun, 17 Jun 2007 21:51:43 +0000 (21:51 +0000)]
Silence some gcc 4 warnings. It is expected that the bpf_movein() routine
will intialize the the header length and re-initialize the mbuf pointer
to reference the mbuf that is allocated after moving user supplied packet
data in.

17 years agoFix a (very) longstanding bug in moused(8) affecting high-resolution rodents
philip [Sun, 17 Jun 2007 20:27:54 +0000 (20:27 +0000)]
Fix a (very) longstanding bug in moused(8) affecting high-resolution rodents
when linear acceleration (-a) was enabled with a <1 value to slow them down.

Previously, rounding errors would eat small movements so the mouse had to be
moved a certain distance to get any movement at all.  We now calculate the
rounding errors and take them into account when reporting movement.

PR: bin/113749
Submitted by: Oliver Fromme <olli -at- secnetix.de>
MFC after: 3 days

17 years agoBack out last change to inpcb_free. Turns out we need
rrs [Sun, 17 Jun 2007 19:27:46 +0000 (19:27 +0000)]
Back out last change to inpcb_free. Turns out we need
to hold off freeing if there is data pending ... someone
might do send/close. Which means we want the data to
go and then close it after startup. Added comments to
the code as well to note that this is done for a reason.

17 years agoSpelling nit due to my lamenglishness.
ariff [Sun, 17 Jun 2007 19:02:05 +0000 (19:02 +0000)]
Spelling nit due to my lamenglishness.

Noticed by: brueffer

17 years agoRemove USBGETSOFTC, USB_ATTACH_START, USB_DETACH_START and
imp [Sun, 17 Jun 2007 18:46:26 +0000 (18:46 +0000)]
Remove USBGETSOFTC, USB_ATTACH_START, USB_DETACH_START and
USB_DECLARE_DRIVER_INIT from the usb network drivers.

17 years agoAdd PAM support to cron(8). Now cron(8) will skip commands scheduled
yar [Sun, 17 Jun 2007 17:25:53 +0000 (17:25 +0000)]
Add PAM support to cron(8).  Now cron(8) will skip commands scheduled
by unavailable accounts, e.g., those locked, expired, not allowed in at
the moment by nologin(5), or whatever, depending on cron's pam.conf(5).
This applies to personal crontabs only, /etc/crontab is unaffected.

In other words, now the account management policy will apply to
commands scheduled by users via crontab(1) so that a user can no
longer use cron(8) to set up a delayed backdoor and run commands
during periods when the admin doesn't want him to.

The PAM check is done just before running a command, not when loading
a crontab, because accounts can get locked, expired, and re-enabled
any time with no changes to their crontabs.  E.g., imagine that you
provide a system with payed access, or better a cluster of such
systems with centralized account management via PAM.  When a user
pays for some days of access, you set his expire field respectively.
If the account expires before its owner pays more, its crontab
commands won't run until the next payment is made.  Then it'll be
enough to set the expire field in future for the commands to run
again.  And so on.

Document this change in the cron(8) manpage, which includes adding
a FILES section and touching the document date.

X-Security: should benefit as users have access to cron(8) by default

17 years agoThis is a forced (and belated) commit to record that these files
phk [Sun, 17 Jun 2007 16:53:45 +0000 (16:53 +0000)]
This is a forced (and belated) commit to record that these files
were repocopied from src/tools/tools/recoverdisk

17 years agoRemove unused softc.
marius [Sun, 17 Jun 2007 16:44:08 +0000 (16:44 +0000)]
Remove unused softc.

17 years agoExpand USB_MATCH_START
imp [Sun, 17 Jun 2007 16:24:49 +0000 (16:24 +0000)]
Expand USB_MATCH_START

17 years agoAdd sysctl/tunable "hw.snd.default_auto", which is useful (especially
ariff [Sun, 17 Jun 2007 16:15:56 +0000 (16:15 +0000)]
Add sysctl/tunable "hw.snd.default_auto", which is useful (especially
for non-root users) to automatically assign default unit to a newly
attach device like USB audio.

17 years agoReassign default unit to a valid unit, be it during attach or detach.
ariff [Sun, 17 Jun 2007 15:53:11 +0000 (15:53 +0000)]
Reassign default unit to a valid unit, be it during attach or detach.
If nothing is available, set to something that is purely ridiculous
so the next valid attach will notice it.

Tested by: chibis

17 years agoFix a compile error from the last change.
scottl [Sun, 17 Jun 2007 15:21:09 +0000 (15:21 +0000)]
Fix a compile error from the last change.

17 years ago o style(9) nit.
rrs [Sun, 17 Jun 2007 14:45:28 +0000 (14:45 +0000)]
o style(9) nit.
o shorten explainations which are over 80 columns in console.
o group rows
o clean up and change explanations a little bit.
Obtained from: weongyo.jeong@gmail.com

17 years agoNote PAM support in atrun(8).
yar [Sun, 17 Jun 2007 14:02:31 +0000 (14:02 +0000)]
Note PAM support in atrun(8).
Sort FILES section while there.
Touch document date.

17 years agoDocument hw.firewire.phydma_enable knob.
simokawa [Sun, 17 Jun 2007 10:40:25 +0000 (10:40 +0000)]
Document hw.firewire.phydma_enable knob.

17 years agoIncrease buffer size of DV stream to prevent buffer
simokawa [Sun, 17 Jun 2007 10:20:55 +0000 (10:20 +0000)]
Increase buffer size of DV stream to prevent buffer
overrun caused by long blocking of file I/O (i.e. zfs).

MFC after: 3 days

17 years agoRemove USB_DO_ATTACH
imp [Sun, 17 Jun 2007 07:34:34 +0000 (07:34 +0000)]
Remove USB_DO_ATTACH

17 years agoExpand USB_DO_ATTACH inline.
imp [Sun, 17 Jun 2007 07:33:59 +0000 (07:33 +0000)]
Expand USB_DO_ATTACH inline.
Remove device_t dv, since it is no longer needed.
Add sizeof(device_t) to replace sizeof dv.
Change device_detach(dev) to device_detach(dev->subdevs[i]) since the type
of dev isn't right!  Not sure when this was introduced, but it likely would
lead to a crash on disconnect.

MFC After: 1 week

17 years agoUse bus_dma to get a page in the first 4 GB. Since the physical address
njl [Sun, 17 Jun 2007 07:18:23 +0000 (07:18 +0000)]
Use bus_dma to get a page in the first 4 GB.  Since the physical address
of the magic string is passed in a 32-bit register, we can't use high
memory in the PAE case.  This also eliminates a use of vtophys().

Tested by: Jeff Shimbo <jts767 / gmail.com>
MFC after: 1 week

17 years agoFlush remaining malloc() cleanups (M_NOWAIT -> M_WAITOK).
ariff [Sun, 17 Jun 2007 06:10:43 +0000 (06:10 +0000)]
Flush remaining malloc() cleanups (M_NOWAIT -> M_WAITOK).

17 years agoPrepare for future integration between CAM and newbus. xpt_bus_register
scottl [Sun, 17 Jun 2007 05:55:54 +0000 (05:55 +0000)]
Prepare for future integration between CAM and newbus.  xpt_bus_register
now takes a device_t to be the parent of the bus that is being created.
Most SIMs have been updated with a reasonable argument, but a few exceptions
just pass NULL for now.  This argument isn't used yet and the newbus
integration likely won't be ready until after 7.0-RELEASE.

17 years agoInitialize key to zero.
mjacob [Sun, 17 Jun 2007 04:40:33 +0000 (04:40 +0000)]
Initialize key to zero.

17 years agoTry a cheap way to get around gcc4.2 believing that user arguments
mjacob [Sun, 17 Jun 2007 04:37:57 +0000 (04:37 +0000)]
Try a cheap way to get around gcc4.2 believing that user arguments
to system calls can change across intervening functions.

17 years ago- switch adapter and port lock over to using sx so that resources
kmacy [Sun, 17 Jun 2007 04:33:38 +0000 (04:33 +0000)]
- switch adapter and port lock over to using sx so that resources
  can be allocated atomically
- add debug macros for printing lock initialization / teardown
- add buffers to port_info and adapter to allow each lock to have a
  unique name
- destroy mutexes initialized by cxgb_offload_init
- remove recursive calls to ADAPTER_LOCK
- move callout_drain calls so that they don't occur with the lock held
- ensure that only as many qsets as are needed are initialized and
  destroyed

MFC after: 3 days
Sponsored by: Chelsio Inc.

17 years agoInitialize mouse resolution to zero if converting from
mjacob [Sun, 17 Jun 2007 04:32:18 +0000 (04:32 +0000)]
Initialize mouse resolution to zero if converting from
OLD to NEW.

17 years agoCheck for pte being NULL in return from pmap_pte_pde- unlikely or
mjacob [Sun, 17 Jun 2007 04:27:45 +0000 (04:27 +0000)]
Check for pte being NULL in return from pmap_pte_pde- unlikely or
even impossible, but it's better ot have a panic and a quiesced
gcc4.2.

17 years agoInitialize lastaddr to zero to make gcc4.2 happy.
mjacob [Sun, 17 Jun 2007 04:21:58 +0000 (04:21 +0000)]
Initialize lastaddr to zero to make gcc4.2 happy.

17 years agoDon't declare inline a function which isn't.
mjacob [Sun, 17 Jun 2007 04:19:05 +0000 (04:19 +0000)]
Don't declare inline a function which isn't.

17 years agoMake sure object is NULL- there is a possible case where you could
mjacob [Sun, 17 Jun 2007 04:17:48 +0000 (04:17 +0000)]
Make sure object is NULL- there is a possible case where you could
fall through to it being used w/o being set. Put a break in the default
case.

17 years agoInitialize reqpage to zero.
mjacob [Sun, 17 Jun 2007 04:14:27 +0000 (04:14 +0000)]
Initialize reqpage to zero.

17 years agogcc4.2 somehow doesn't believe that finaldst can stay stable between
mjacob [Sun, 17 Jun 2007 04:12:21 +0000 (04:12 +0000)]
gcc4.2 somehow doesn't believe that finaldst can stay stable between
where it's initialized and where it's checked twice such that the
origingal destination address is saved. Make it happier and trim
things down a bit.

17 years agoMake gcc4.2 happy and zero save_ip for the unlikely (blackhole != 0)
mjacob [Sun, 17 Jun 2007 04:07:11 +0000 (04:07 +0000)]
Make gcc4.2 happy and zero save_ip for the unlikely (blackhole != 0)
codepath.

17 years ago- Forced commit to update who actually did this code (I forgot
rrs [Sun, 17 Jun 2007 01:57:08 +0000 (01:57 +0000)]
- Forced commit to update who actually did this code (I forgot
  the obtained from in the original line)
Obtained from: Weongyo Jeong (weongyo.jeong@gmail.com)

17 years ago - For sctp_input/sctp6_input add announcment when a packet arrives (debug)
rrs [Sun, 17 Jun 2007 01:36:02 +0000 (01:36 +0000)]
 - For sctp_input/sctp6_input add announcment when a packet arrives (debug)
 - re-factor the packet drop in sctp_output a bit more, we don't need the
   trim after all, but the size calc is now corrected.
 - When a assoc is in the COOKIE-ECHO/COOKIE-WAIT state and the user
   closes, it should not matter if data is queued, the assoc should be
   purged.
 - In error leg a missing free_chunk when iph comes in NULL (should not
   happen but just in case).

17 years agoFix build on arm: time_t on arm is 64 bits while long is
delphij [Sun, 17 Jun 2007 00:43:50 +0000 (00:43 +0000)]
Fix build on arm: time_t on arm is 64 bits while long is
32 bits, so subsequent compile time assertion:

sizeof inf->stat.st_mtime <= sizeof sec

Would fail because of that.  This change is suitable for
general consumption as well, but fix it in our local
patchset as we are near a code freeze.

Submitted by: cognet

17 years agoReplace incorrect local OFFSET_OF macro with the correct and generic
mjacob [Sun, 17 Jun 2007 00:33:34 +0000 (00:33 +0000)]
Replace incorrect local OFFSET_OF macro with the correct and generic
offsetof macro.

17 years agoSimplification to quiet a gcc4.2 warning. Just by setting match.s_addr
mjacob [Sun, 17 Jun 2007 00:31:24 +0000 (00:31 +0000)]
Simplification to quiet a gcc4.2 warning. Just by setting match.s_addr
to nonzero you fulfill the same function as the variable 'cmp'. so you
might as well zero match and test against it later.

Reviewed by: timeout on review request

17 years ago- Make better use of the global chosen, memory and mmu handles instead
marius [Sun, 17 Jun 2007 00:17:15 +0000 (00:17 +0000)]
- Make better use of the global chosen, memory and mmu handles instead
  of obtaining them over and over again and pretending we could do
  anything useful without them (for chosen this includes adding a
  declaration and initializing it in OF_init()).
- In OF_init() if obtaining the memory or mmu handle fails just call
  OF_exit() instead of panic() as the loader hasn't initialized the
  console at these early stages yet and trying to print out something
  causes a hang. With OF_exit() one at least has a change to get back
  to the OFW boot monitor and debug the problem.
- Fix OF_call_method() on 64-bit machines (this is a merge of
  sys/dev/ofw/openfirm.c rev 1.6).
- Replace OF_alloc_phys(), OF_claim_virt(), OF_map_phys() and
  OF_release_phys() in the MI part of the loader with wrappers around
  OF_call_method() in the sparc64. Beside the fact that they duplicate
  OF_call_method() the formers should never have been in the MI part
  of the loader as contrary to the OFW spec they use two-cell physical
  addresses.
- Remove unused functions which are also MD dupes of OF_call_method().
- In sys/boot/sparc64/loader/main.c add __func__ to panic strings as
  different functions use otherwise identical panic strings and make
  some of the panic strings a tad more user-friendly instead of just
  mentioning the name of the function that returned an unexpected
  result.