]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
15 years agoUpdate version to 2.6.901a to indicate this is synced up with r745 of
kientzle [Sun, 8 Mar 2009 06:20:35 +0000 (06:20 +0000)]
Update version to 2.6.901a to indicate this is synced up with r745 of
libarchive.googlecode.com (except for the lzma/xz support).

15 years agoMerge r492 from libarchive.googlecode.com: First cut at exposing the
kientzle [Sun, 8 Mar 2009 06:19:28 +0000 (06:19 +0000)]
Merge r492 from libarchive.googlecode.com:  First cut at exposing the
new options mechanism to userland.  Documentation pending...

15 years agoDon't declare bin_search() as an inline function, since there's no
das [Sun, 8 Mar 2009 06:14:33 +0000 (06:14 +0000)]
Don't declare bin_search() as an inline function, since there's no
inline definition of it.

15 years agoMatch a comment to reduce differences with libarchive.googlecode.com.
kientzle [Sun, 8 Mar 2009 06:09:20 +0000 (06:09 +0000)]
Match a comment to reduce differences with libarchive.googlecode.com.

15 years agoMerge r709,r710 from libarchive.googlecode.com: More work on
kientzle [Sun, 8 Mar 2009 06:07:35 +0000 (06:07 +0000)]
Merge r709,r710 from libarchive.googlecode.com:  More work on
Windows support.

15 years agoFix endian conversion from htole16 to htole32.
thompsa [Sun, 8 Mar 2009 06:03:28 +0000 (06:03 +0000)]
Fix endian conversion from htole16 to htole32.

Tested with: ARM xscale

15 years agoMerge r687-689,691,693-701,720 from libarchive.googlecode.com:
kientzle [Sun, 8 Mar 2009 06:03:15 +0000 (06:03 +0000)]
Merge r687-689,691,693-701,720 from libarchive.googlecode.com:
Translate getdate.y into C for portability.  Make the get_date()
function easier to test as well:
 * Have it accept a time_t "now" to use as a reference so that test
   code can verify relative time specifications against known starting
   points.
 * Set up default date after parsing the string so that we
   can use the specified timezone (if any) instead of the local
   default.  Otherwise, local DST makes it almost impossible to
   reliably test time specifications such as "sunday UTC"

15 years agoMerger r629-631,633-646,648,654,678,681,682 from libarchive.googlecode.com:
kientzle [Sun, 8 Mar 2009 05:47:21 +0000 (05:47 +0000)]
Merger r629-631,633-646,648,654,678,681,682 from libarchive.googlecode.com:
Many changes for Windows compatibility.  bsdtar_test now runs successfully
on both POSIX platforms and Windows.

15 years agoMerge r368,496,625,626 from libarchive.googlecode.com: A number of
kientzle [Sun, 8 Mar 2009 05:38:45 +0000 (05:38 +0000)]
Merge r368,496,625,626 from libarchive.googlecode.com: A number of
style and portability tweaks to the test harness.  Most significantly,
don't use getopt().

15 years agoMerge r622 from libarchive.googlecode.com: Avoid warning on platforms
kientzle [Sun, 8 Mar 2009 05:35:59 +0000 (05:35 +0000)]
Merge r622 from libarchive.googlecode.com:  Avoid warning on platforms
that lack regex.h.

15 years agoMerge r552,r559 from libarchive.googlecode.com: High-resolution time
kientzle [Sun, 8 Mar 2009 05:34:20 +0000 (05:34 +0000)]
Merge r552,r559 from libarchive.googlecode.com:  High-resolution time
support on Tru64, AIX, and GNU Hurd, thanks to Björn Jacke.

15 years agoMerge r529 from libarchive.googlecode.com: Fix how we read ext2fs_fs.h
kientzle [Sun, 8 Mar 2009 05:28:52 +0000 (05:28 +0000)]
Merge r529 from libarchive.googlecode.com:  Fix how we read ext2fs_fs.h
headers on Linux.

15 years agoMerge r435,r443 from libarchive.googlecode.com: Let the compiler options
kientzle [Sun, 8 Mar 2009 05:24:37 +0000 (05:24 +0000)]
Merge r435,r443 from libarchive.googlecode.com:  Let the compiler options
determine how to read config.h.

15 years agoMerge r283,r423 from libarchive.googlecode.com: Use libarchive's new
kientzle [Sun, 8 Mar 2009 05:22:50 +0000 (05:22 +0000)]
Merge r283,r423 from libarchive.googlecode.com:  Use libarchive's new
archive_read_disk API to pull metadata off of disk.  This
removes a lot of platform-specific knowledge of things like
ACLs, file flags, and extended attributes from bsdtar.

15 years agoMerge r374 from libarchive.googlecode.com: Stupid typo in open() call. <sigh>
kientzle [Sun, 8 Mar 2009 05:19:36 +0000 (05:19 +0000)]
Merge r374 from libarchive.googlecode.com: Stupid typo in open() call. <sigh>

15 years agoMerge r369 from libarchive.googlecode.com: Test -s option.
kientzle [Sun, 8 Mar 2009 05:17:58 +0000 (05:17 +0000)]
Merge r369 from libarchive.googlecode.com: Test -s option.

15 years agoMerge r278 from libarchive.googlecode.com: Reduce the number of
kientzle [Sun, 8 Mar 2009 05:14:16 +0000 (05:14 +0000)]
Merge r278 from libarchive.googlecode.com:  Reduce the number of
patterns tested here from 200 to 170, which seems to be the
most that Cygwin can handle.

15 years agoMerge r273 from libarchive.googlecode.com: Use open() correctly.
kientzle [Sun, 8 Mar 2009 05:10:51 +0000 (05:10 +0000)]
Merge r273 from libarchive.googlecode.com:  Use open() correctly.

15 years agoSmall comment nit: "run time" -> "run-time".
sobomax [Sun, 8 Mar 2009 05:01:39 +0000 (05:01 +0000)]
Small comment nit: "run time" -> "run-time".

Submitted by: rwatson

15 years agoSet version to 2.6.901a to indicate this now matches
kientzle [Sun, 8 Mar 2009 04:32:38 +0000 (04:32 +0000)]
Set version to 2.6.901a to indicate this now matches
libarchive.googlecode.com r745.  (Except for the lzma/xz support,
which needs a little more attention before it can be merged.)

15 years agoMerge a bunch of changes through r722 from libarchive.googlecode.com:
kientzle [Sun, 8 Mar 2009 04:20:19 +0000 (04:20 +0000)]
Merge a bunch of changes through r722 from libarchive.googlecode.com:
mtree writer now supports a variety of checksum keys; it also provides
option hooks to set what keys get written.

15 years agoAdd static DTrace probes for MAC Framework access control checks and
rwatson [Sun, 8 Mar 2009 00:50:37 +0000 (00:50 +0000)]
Add static DTrace probes for MAC Framework access control checks and
privilege grants so that dtrace can be more easily used to monitor
the security decisions being generated by the MAC Framework following
policy invocation.

Successful access control checks will be reported by:

  mac_framework:kernel:<entrypoint>:mac_check_ok

Failed access control checks will be reported by:

  mac_framework:kernel:<entrypoint>:mac_check_err

Successful privilege grants will be reported by:

  mac_framework:kernel:priv_grant:mac_grant_ok

Failed privilege grants will be reported by:

  mac_framework:kernel:priv_grant:mac_grant_err

In all cases, the return value (always 0 for _ok, otherwise an errno
for _err) will be reported via arg0 on the probe, and subsequent
arguments will hold entrypoint-specific data, in a style similar to
privilege tracing.

Obtained from: TrustedBSD Project
Sponsored by: Google, Inc.

15 years agoWhen resetting a BPF descriptor, properly check that zero-copy buffers
rwatson [Sat, 7 Mar 2009 22:17:44 +0000 (22:17 +0000)]
When resetting a BPF descriptor, properly check that zero-copy buffers
are not currently owned by userspace before clearing or rotating them.

Otherwise we may not play by the rules of the shared memory protocol,
potentially corrupting packet data or causing userspace applications
that are playing by the rules to spin due to being notified that a
buffer is complete but the shared memory header not reflecting that.

This behavior was seen with pflogd by a number of reporters; note that
this fix is not sufficient to get pflogd properly working with
zero-copy BPF, due to pflogd opening the BPF device before forking,
leading to the shared memory buffer not being propery inherited in the
privilege-separated child.  We're still deciding how to fix that
problem.

This change exposes buffer-model specific strategy information in
reset_d(), which will be fixed at a later date once we've decided how
best to improve the BPF buffer abstraction.

Reviewed by: csjp
Reported by: keramida

15 years agoRevert the part of change 107879 that employs the unused bytes after
marcel [Sat, 7 Mar 2009 22:05:58 +0000 (22:05 +0000)]
Revert the part of change 107879 that employs the unused bytes after
the disklabel in the 2nd sector for boot code. Even with both UFS1
and UFS2 supported, there's enough bytes left that we don't have to
nibble from the disklabel.
Thus, the entire 2nd sector is now reserved for the disklabel, which
makes the bootcode compatible again with disklabels that have more
than 8 partitions -- such as those created and supported by gpart.

i386: 135 bytes available
amd64: 151 bytes available

Ok'd by: jhb

15 years agoImport support for ATI Radeon R600 and R700 series chips.
rnoland [Sat, 7 Mar 2009 21:36:57 +0000 (21:36 +0000)]
Import support for ATI Radeon R600 and R700 series chips.

Tested on an HD3850 (RV670) on loan from Warren Block.

Currently, you need one of the following for this to be useful:

x11-drivers/xf86-video-radeonhd-devel (not tested)
xf86-video-ati from git (EXA works, xv is too fast)
xf86-video-radeonhd from git (EXA works, xv works)

There is no 3d support available from dri just yet.

MFC after: 2 weeks

15 years agoReenable ndis in the LINT build now that it has been updated for USB. Thanks to
thompsa [Sat, 7 Mar 2009 19:54:30 +0000 (19:54 +0000)]
Reenable ndis in the LINT build now that it has been updated for USB. Thanks to
HPS and Weongyo.

15 years ago(re)merge r186415,186416 from the old usb stack;
thompsa [Sat, 7 Mar 2009 19:49:47 +0000 (19:49 +0000)]
(re)merge r186415,186416 from the old usb stack;

o add Transaction Translator support (still missing ISOC xfers)
o add EHCI_SCFLG_BIGEMMIO flag to force big-endian byte-select to be
  set in USBMODE
o split reset work into new public routine ehci_reset so bus shim drivers
  can force big-endian byte-select before ehci_init
o enable TT and big-endian MMIO
o force a reset before ehci_init to get byte-select setup

Also go back to using USB_EHCI_BIG_ENDIAN_DESC at compile time to enable the
byteswapping and reduce diffs to the original commits.

This fixes the new USB stack on the Cambria board.

15 years agoOn architectures with strict alignment requirements compensate
marius [Sat, 7 Mar 2009 19:08:58 +0000 (19:08 +0000)]
On architectures with strict alignment requirements compensate
the misalignment of the IP header that prepending the EtherIP
header might have caused.

PR: 131921
MFC after: 1 week

15 years agoFix some missed htole32 conversions to htoehci32.
thompsa [Sat, 7 Mar 2009 18:08:59 +0000 (18:08 +0000)]
Fix some missed htole32 conversions to htoehci32.

Reviewed by: hps

15 years agoMark the bpf stats sysctl as being mpsafe. We do not require
csjp [Sat, 7 Mar 2009 17:07:29 +0000 (17:07 +0000)]
Mark the bpf stats sysctl as being mpsafe. We do not require
Giant here.

15 years agoClarify some comments, fix some types, and rename ZBUF_FLAG_IMMUTABLE to
rwatson [Sat, 7 Mar 2009 10:21:37 +0000 (10:21 +0000)]
Clarify some comments, fix some types, and rename ZBUF_FLAG_IMMUTABLE to
ZBUF_FLAG_ASSIGNED to make it clear why the buffer can't be written to:
it is assigned to userspace.

15 years agoo port NDIS USB support from USB1 to the new usb(USB2).
weongyo [Sat, 7 Mar 2009 07:26:22 +0000 (07:26 +0000)]
o port NDIS USB support from USB1 to the new usb(USB2).
o implement URB_FUNCTION_ABORT_PIPE handling.
o remove unused code related with canceling the timer list for USB
  drivers.
o whitespace cleanup and style(9)

Obtained from: hps's original patch

15 years agoMerge r718 from libarchive.googlecode.com: Some additional
kientzle [Sat, 7 Mar 2009 07:23:04 +0000 (07:23 +0000)]
Merge r718 from libarchive.googlecode.com:  Some additional
tests of restoring files to disk with unusual characters, specifically
to exercise Windows issues.

15 years agoFix spelling.
kientzle [Sat, 7 Mar 2009 07:19:25 +0000 (07:19 +0000)]
Fix spelling.

15 years agoMerge r348 from libarchive.googlecode.com: Suppress testing invalid
kientzle [Sat, 7 Mar 2009 03:41:29 +0000 (03:41 +0000)]
Merge r348 from libarchive.googlecode.com:  Suppress testing invalid
conversions if there aren't any.  In particular, Cygwin's "C" locale
has no invalid inputs for wctomb().

15 years agoMerge r596,r690 from libarchive.googlecode.com: Minor style and compile
kientzle [Sat, 7 Mar 2009 03:34:34 +0000 (03:34 +0000)]
Merge r596,r690 from libarchive.googlecode.com:  Minor style and compile
warning fixes for test_read_pax_truncated.c.

15 years agoMerge r335,653,676 from libarchive.googlecode.com: Instead of
kientzle [Sat, 7 Mar 2009 03:30:35 +0000 (03:30 +0000)]
Merge r335,653,676 from libarchive.googlecode.com: Instead of
conditioning tests on HAVE_ZLIB, etc, just ask libarchive for the
service and handle the failure coming back from libarchive.  This
gives us better test coverage of common client usage where clients
simply try to use libarchive services and handle the errors coming
back instead of trying to second-guess which libarchive services are
compiled in.

15 years agoMerge r280,281,496,595,675,712 from libarchive.googlecode.com: Various
kientzle [Sat, 7 Mar 2009 03:16:16 +0000 (03:16 +0000)]
Merge r280,281,496,595,675,712 from libarchive.googlecode.com: Various
test improvements, including some work on Windows compatibility and an
extra check to verify that no test leaves open file descriptors
around.

15 years agoMerge r540 from libarchive.googlecode.com: Interix doesn't have inttypes.h
kientzle [Sat, 7 Mar 2009 03:04:06 +0000 (03:04 +0000)]
Merge r540 from libarchive.googlecode.com:  Interix doesn't have inttypes.h
either.  <sigh>

15 years agoMerge r723 from libarchive.googlecode.com: Don't try to restore
kientzle [Sat, 7 Mar 2009 03:00:44 +0000 (03:00 +0000)]
Merge r723 from libarchive.googlecode.com:  Don't try to restore
owner or SUID bits on Windows; just ignore them.

15 years agoMerge r511,r513,r607 from libarchive.googlecode.com: Mtree reader
kientzle [Sat, 7 Mar 2009 02:58:15 +0000 (02:58 +0000)]
Merge r511,r513,r607 from libarchive.googlecode.com:  Mtree reader
tweaks:  Support nanosecond timestamps, handle attributes broken
across multiple lines.

15 years agoMerge r585,r669 from libarchive.googlecode.com: If zlib is unavailable,
kientzle [Sat, 7 Mar 2009 02:51:18 +0000 (02:51 +0000)]
Merge r585,r669 from libarchive.googlecode.com:  If zlib is unavailable,
use external "gunzip" instead.  With this in place, we can unconditionally
enable gzip read support.

15 years agoMerge r550,584,587,609,647,674 from libarchive.googlecode.com:
kientzle [Sat, 7 Mar 2009 02:47:04 +0000 (02:47 +0000)]
Merge r550,584,587,609,647,674 from libarchive.googlecode.com:
Refactor the read_compression_program to add two new abilities:
 * Public API:  You can now include a signature string when you
   register a program; the program will run only on input that
   matches the signature string.
 * Internal API: You can use the init() function to instantiate
   an external program as part of a filter pipeline.  This
   can be used for graceful fallback (if zlib is unavailable, use
   external gzip instead) and to use external programs with
   bidders that are more sophisticated than a static signature check.

15 years agoMerge r591 from libarchive.googlecode.com: signed/unsigned fixes.
kientzle [Sat, 7 Mar 2009 02:29:43 +0000 (02:29 +0000)]
Merge r591 from libarchive.googlecode.com: signed/unsigned fixes.

15 years agoMerge r558,567,569,571,581,582,583,598 from libarchive.googlecode.com:
kientzle [Sat, 7 Mar 2009 02:24:32 +0000 (02:24 +0000)]
Merge r558,567,569,571,581,582,583,598 from libarchive.googlecode.com:
Support Joliet extensions.  This currently ignores Rockridge extensions
if both exist on the same disk unless the '!joliet' option is provided.
e.g.: tar -xvf example.iso --options '!joliet'
Thanks to: Andreas Henriksson

15 years agoMerge r658 from libarchive.googlecode.com: Only flush and close the
kientzle [Sat, 7 Mar 2009 02:09:21 +0000 (02:09 +0000)]
Merge r658 from libarchive.googlecode.com: Only flush and close the
file if it was actually opened.  Test for this case.

15 years agoMerge r714,r715 from libarchive.googlecode.com: Fix Debian bug #516577.
kientzle [Sat, 7 Mar 2009 01:21:46 +0000 (01:21 +0000)]
Merge r714,r715 from libarchive.googlecode.com: Fix Debian bug #516577.
Don't crash if client does not provide a skip function.  Extend one
of the test cases to use archive_read_open2() with only a read callback.

15 years agoMerge r608 from libarchive.googlecode.com: Clear a newly-allocated
kientzle [Sat, 7 Mar 2009 01:18:30 +0000 (01:18 +0000)]
Merge r608 from libarchive.googlecode.com: Clear a newly-allocated
bidder object.

15 years agoMerge r564,r566 from libarchive.googlecode.com: Fix segfault when
kientzle [Sat, 7 Mar 2009 01:17:13 +0000 (01:17 +0000)]
Merge r564,r566 from libarchive.googlecode.com:  Fix segfault when
specifying an option and the current format doesn't have an
options handler.

15 years agoMerge the rest of r453 from libarchive.googlecode.com: Test using
kientzle [Sat, 7 Mar 2009 01:12:01 +0000 (01:12 +0000)]
Merge the rest of r453 from libarchive.googlecode.com:  Test using
libarchive APIs to read extended attributes from disk on FreeBSD.

15 years agoMinor whitespace adjustment to reduce diffs with libarchive.googlecode.com.
kientzle [Sat, 7 Mar 2009 01:08:16 +0000 (01:08 +0000)]
Minor whitespace adjustment to reduce diffs with libarchive.googlecode.com.

15 years agoMerge r552,r559 from libarchive.googlecode.com: Support high-resolution
kientzle [Sat, 7 Mar 2009 00:52:02 +0000 (00:52 +0000)]
Merge r552,r559 from libarchive.googlecode.com: Support high-resolution
timestamps on Tru64, AIX, and GNU Hurd.  Thanks to Björn Jacke.

15 years agoMerge r283,r529 from libarchive.googlecode.com: Fix ext2_fs.h includes
kientzle [Sat, 7 Mar 2009 00:36:50 +0000 (00:36 +0000)]
Merge r283,r529 from libarchive.googlecode.com: Fix ext2_fs.h includes
for Linux.

15 years agoMerge r659 from libarchive.googlecode.com: Correctly report "none"
kientzle [Sat, 7 Mar 2009 00:25:33 +0000 (00:25 +0000)]
Merge r659 from libarchive.googlecode.com:  Correctly report "none"
as the compression name when no other read filter bid.  Add some
assertions to various tests to verify that read filters are properly
setting the textual name as well as the compression code.

15 years agoCleanup virtual device mapping some more:
sam [Fri, 6 Mar 2009 23:32:45 +0000 (23:32 +0000)]
Cleanup virtual device mapping some more:
o improves understandability by replacing numerous relative address
  calculations with fixed addresses; everything should now match up
  more easily with the vm layout shown at the top of the file
o move the expansion bus chip select regions to be contiguous with
  the expansion bus configuration area; this is not exploited right
  now but allows map consolidation in the future
o leave a gap between the expansion bus regions and the pci config
  space in case we want to map more exp bus cs regions

Reviewed by: imp, thompsa

15 years agoAdd Bluetooth compatibility shims. Inspired by Linux BlueZ and NetBSD.
emax [Fri, 6 Mar 2009 23:30:07 +0000 (23:30 +0000)]
Add Bluetooth compatibility shims. Inspired by Linux BlueZ and NetBSD.

Discussed with: Iain Hibbert plunky -at- rya-online -dot- net of NetBSD
MFC after: 1 month

15 years agoremove unneeded static mappings for NPE and MAC regions; these are
sam [Fri, 6 Mar 2009 23:29:00 +0000 (23:29 +0000)]
remove unneeded static mappings for NPE and MAC regions; these are
already mapped through the IO region so never used

Reviewed by: imp, thompsa

15 years agoenable tdma support by default; many people using these boards
sam [Fri, 6 Mar 2009 23:27:47 +0000 (23:27 +0000)]
enable tdma support by default; many people using these boards
are using them to setup tdma p2p links

15 years agolegacy USB is required on these platforms at the moment
sam [Fri, 6 Mar 2009 23:26:50 +0000 (23:26 +0000)]
legacy USB is required on these platforms at the moment

15 years agofix legacy usb configuration
sam [Fri, 6 Mar 2009 23:22:09 +0000 (23:22 +0000)]
fix legacy usb configuration

15 years agoo simplify code in ixppcib_conf_setup
sam [Fri, 6 Mar 2009 20:40:09 +0000 (20:40 +0000)]
o simplify code in ixppcib_conf_setup
o fixup debug printfs

15 years ago- Try to handle rcsfile write failures in the same way as cvsup, as they are not
lulf [Fri, 6 Mar 2009 20:17:16 +0000 (20:17 +0000)]
- Try to handle rcsfile write failures in the same way as cvsup, as they are not
  necessarily fatal. If the file was incorrectly written, the checksum will
  detect it and the file will be retransferred.

15 years agoIf the PDE is known, then use the direct mapping instead of the recursive
alc [Fri, 6 Mar 2009 17:40:58 +0000 (17:40 +0000)]
If the PDE is known, then use the direct mapping instead of the recursive
mapping to access the PTE.

15 years agoMFp4 //depot/projects/usb@158692
thompsa [Fri, 6 Mar 2009 17:13:12 +0000 (17:13 +0000)]
MFp4 //depot/projects/usb@158692

Workaround a EHCI performance problem by issuing a doorbell after queueing a
bulk xfer.

Submitted by: Hans Petter Selasky

15 years agoEnsure the cached rq pointer is still valid before waking up the address, the
thompsa [Fri, 6 Mar 2009 17:04:47 +0000 (17:04 +0000)]
Ensure the cached rq pointer is still valid before waking up the address, the
zyd_cmd function may have timed out. It wouldnt cause a panic but could wakeup
someone.

Spotted by: HPS

15 years agoExtract the no_poll() and vop_nopoll() code into the common routine
kib [Fri, 6 Mar 2009 15:35:37 +0000 (15:35 +0000)]
Extract the no_poll() and vop_nopoll() code into the common routine
poll_no_poll().
Return a poll_no_poll() result from devfs_poll_f() when
filedescriptor does not reference the live cdev, instead of ENXIO.

Noted and tested by: hps
MFC after: 1 week

15 years agoAdd a couple of more things to the FTDI driver I came across:
joerg [Fri, 6 Mar 2009 14:53:51 +0000 (14:53 +0000)]
Add a couple of more things to the FTDI driver I came across:

. Dresden Elektronik "Wireless Handheld Terminal"
. Atmel STK541 "Zigbee Controller"

MFC after: 1 week

15 years agoToo good to pass.
phk [Fri, 6 Mar 2009 13:24:54 +0000 (13:24 +0000)]
Too good to pass.

15 years agoInvert the logic error for the MSI/MSIX vs INTx case.
rnoland [Fri, 6 Mar 2009 11:24:42 +0000 (11:24 +0000)]
Invert the logic error for the MSI/MSIX vs INTx case.

Pointyhat to: me

MFC after: 3 days

15 years agoMFi386: 189423
nyan [Fri, 6 Mar 2009 11:15:24 +0000 (11:15 +0000)]
MFi386: 189423

  A better fix for handling different FPU initial control words for different
  ABIs.

15 years agoMFi386: part of 189421
nyan [Fri, 6 Mar 2009 11:10:31 +0000 (11:10 +0000)]
MFi386: part of 189421

  - If there are no syscons hints at all, assume there is a single sc0 device
    anyway.  The console probe will still fail unless a VGA adapter is found.

15 years agoFixes for window probes:
rrs [Fri, 6 Mar 2009 11:03:52 +0000 (11:03 +0000)]
Fixes for window probes:
 1) WP should never be marked unless flight size is 0
 2) When recovering from wp if the peer ack's it we don't mark for retran
 3) When recovering, we must assure a timer is still running.

15 years agoMerge r668 from libarchive.googlecode.com: Style correction to the
kientzle [Fri, 6 Mar 2009 06:14:44 +0000 (06:14 +0000)]
Merge r668 from libarchive.googlecode.com:  Style correction to the
'ar' reader:  Don't redefine 'isdigit' and don't create a macro that's
only used once.

15 years agoMerge r541,r542 from libarchive.googlecode.com: Two sign mismatches
kientzle [Fri, 6 Mar 2009 06:13:25 +0000 (06:13 +0000)]
Merge r541,r542 from libarchive.googlecode.com:  Two sign mismatches
in the Zip reader.

15 years agoMerge r491,493,500,507,510,530,543 from libarchive.googlecode.com:
kientzle [Fri, 6 Mar 2009 05:58:56 +0000 (05:58 +0000)]
Merge r491,493,500,507,510,530,543 from libarchive.googlecode.com:
This implements the new generic options framework that provides a way
to override format- and compression-specific parameters.

15 years agoMerge r448 from libarchive.googlecode.com: Suppress testing write_disk
kientzle [Fri, 6 Mar 2009 05:40:09 +0000 (05:40 +0000)]
Merge r448 from libarchive.googlecode.com:  Suppress testing write_disk
failures on Windows for now.  Someday this will be revisited.

15 years agoMerge r403,702,721 from libarchive.googlecode.com: Handle odd
kientzle [Fri, 6 Mar 2009 05:38:53 +0000 (05:38 +0000)]
Merge r403,702,721 from libarchive.googlecode.com:  Handle odd
pathnames on Windows by mapping '\\' to '/' and converting
illegal characters to '_'.

15 years agoMerge r505 from libarchive.googlecode.com: Fix %ju support. Simplify
kientzle [Fri, 6 Mar 2009 05:14:55 +0000 (05:14 +0000)]
Merge r505 from libarchive.googlecode.com: Fix %ju support.  Simplify
the code here a bit by making the int formatting functions static to
archive_string_sprintf.c, which is the only place this has ever been
used.

15 years agoMerge r457 from libarchive.googlecode.com: Stop appending strerror()
kientzle [Fri, 6 Mar 2009 05:13:12 +0000 (05:13 +0000)]
Merge r457 from libarchive.googlecode.com:  Stop appending strerror()
information to error strings.  This caused a lot of unnecessary
duplication in error messages; in particular, there are a few cases
where error messages get copied from one archive object to another
and this would cause the strerror() info to get appended each time.

15 years agoMerge r420,r494 from libarchive.googlecode.com: Prettify the test
kientzle [Fri, 6 Mar 2009 05:07:03 +0000 (05:07 +0000)]
Merge r420,r494 from libarchive.googlecode.com: Prettify the test
harness a bit:  remove a dead comment, tweak the wording of the
summary report.

15 years agoMerge r435,r443 from libarchive.googlecode.com: Read config files from
kientzle [Fri, 6 Mar 2009 05:04:15 +0000 (05:04 +0000)]
Merge r435,r443 from libarchive.googlecode.com:  Read config files from
include path; this makes it easier to support multiple build frameworks.

15 years agoMerge r416 from libarchive.googlecode.com:
kientzle [Fri, 6 Mar 2009 04:55:51 +0000 (04:55 +0000)]
Merge r416 from libarchive.googlecode.com:

Restoring POSIX.1e Extended Attributes on FreeBSD, part 1

This implements the basic ability to restore extended attributes
on FreeBSD, including a test suite.

15 years agoMerge r407,r508 from libarchive.googlecode.com: Correctly handle
kientzle [Fri, 6 Mar 2009 04:50:39 +0000 (04:50 +0000)]
Merge r407,r508 from libarchive.googlecode.com:  Correctly handle
Zip entries that are zero length but stored with deflate.  This
is arguably a silly thing to do (deflating a zero-length file actually
makes it bigger) but apparently quite a few Zip writers do this.
This was broken in two places:  archive_write_disk disliked being asked
to write data to zero-length files (even if the write was zero-length)
and zip_read_file_header tripped over itself when non-regular files
had compressed bodies.

15 years agoMerge r399,401,402,405,415,430,440,452,453,458,506,533,536,538,544,590
kientzle [Fri, 6 Mar 2009 04:35:31 +0000 (04:35 +0000)]
Merge r399,401,402,405,415,430,440,452,453,458,506,533,536,538,544,590
from libarchive.googlecode.com:  Add a new "archive_read_disk" API
that provides the important service of reading metadata from the
disk.  In particular, this will make it possible to remove all
knowledge of extended attributes, ACLs, etc, from clients such
as bsdtar and bsdcpio.

Closely related, this API also provides pluggable uid->uname
and gid->gname lookup and caching services similar to
the uname->uid and gname->gid services provided by archive_write_disk.
Remember this is also required for correct ACL management.

Documentation is still pending...

15 years agoMerge r398 from libarchive.googlecode.com: Check that bidder
kientzle [Fri, 6 Mar 2009 04:22:34 +0000 (04:22 +0000)]
Merge r398 from libarchive.googlecode.com: Check that bidder
object was allocated before we try to use it.

15 years agoMerge r394,r396 from libarchive.googlecode.com: Plug some memory
kientzle [Fri, 6 Mar 2009 04:21:23 +0000 (04:21 +0000)]
Merge r394,r396 from libarchive.googlecode.com: Plug some memory
leaks in the ACL test, correctly mark that FreeBSD has acl_get_perm_np().

15 years agoMerge r390,r391,r392,r397 from libarchive.googlecode.com: Virtualize
kientzle [Thu, 5 Mar 2009 21:18:10 +0000 (21:18 +0000)]
Merge r390,r391,r392,r397 from libarchive.googlecode.com:  Virtualize
"close" and "finish" across both read and write interfaces.
(Someday, "finish" should be renamed to "free" to better reflect
what it actually does...)

15 years agoA better fix for handling different FPU initial control words for different
jhb [Thu, 5 Mar 2009 19:42:11 +0000 (19:42 +0000)]
A better fix for handling different FPU initial control words for different
ABIs:
- Store the FPU initial control word in the pcb for each thread.
- When first using the FPU, load the initial control word after restoring
  the clean state if it is not the standard control word.
- Provide a correct control word for Linux/i386 binaries under
  FreeBSD/amd64.
- Adjust the control word returned for fpugetregs()/npxgetregs() when a
  thread hasn't used the FPU yet to reflect the real initial control
  word for the current ABI.
- The Linux/i386 ABI for FreeBSD/i386 now properly sets the right control
  word instead of trashing whatever the current state of the FPU is.

Reviewed by: bde

15 years agoFix usb2_poll not to return an error number as the function return value is a
thompsa [Thu, 5 Mar 2009 19:20:17 +0000 (19:20 +0000)]
Fix usb2_poll not to return an error number as the function return value is a
bitmask of events.

Pointed out by: HPS

15 years agoAllow syscons to work on amd64 and i386 without any hints:
jhb [Thu, 5 Mar 2009 19:10:17 +0000 (19:10 +0000)]
Allow syscons to work on amd64 and i386 without any hints:
- Enable keyboard autodetection by default for ISA syscons attachments.
- If there are no syscons hints at all, assume there is a single sc0 device
  anyway.  The console probe will still fail unless a VGA adapter is found.

MFC after: 2 weeks

15 years agoRemove unused arg from npxinit(). Forgot to commit this file in the
jhb [Thu, 5 Mar 2009 18:43:54 +0000 (18:43 +0000)]
Remove unused arg from npxinit().  Forgot to commit this file in the
last i386 FPU change.

15 years agoMerge r389 from libarchive.googlecode.com: Fix a memory
kientzle [Thu, 5 Mar 2009 18:38:36 +0000 (18:38 +0000)]
Merge r389 from libarchive.googlecode.com: Fix a memory
leak in ISO9660 handler structure whenever a file entry
has a nonsensical CE offset.

15 years agoSome cleanups to the i386 FPU support:
jhb [Thu, 5 Mar 2009 18:32:43 +0000 (18:32 +0000)]
Some cleanups to the i386 FPU support:
- Remove the control word parameter to npxinit().  It was always set
  to __INITIAL_NPXCW__.
- Remove npx_cleanstate_ready as the cleanstate is always initalized
  when it is used.
- Improve the handling of the case when the FPU isn't present.  Now
  the npx0 device no longer succeeds in its probe so all of npx_attach()
  is skipped.  Also, we allow this case with SMP (though that shouldn't
  actually occur as all i386 systems that support SMP have FPUs) now.
  SMP was only an issue back when we had an FPU emulator which was not
  per-CPU.
- MFamd64: Clear some of the state in npx_cleanstate rather than leaving
  it as garbage.
- MFamd64: When a user thread first uses the FPU, use npx_cleanstate for
  the initial FPU state.

Reviewed by: bde

15 years agoMerge r386,r395,r451 from libarchive.googlecode.com: On Windows, break
kientzle [Thu, 5 Mar 2009 18:30:50 +0000 (18:30 +0000)]
Merge r386,r395,r451 from libarchive.googlecode.com: On Windows, break
into the debugger on test setup failures (otherwise, the console window
just goes away and you can't see what went wrong).  On all platforms,
clean up a stray buffer before exiting.

15 years agolibc_r_* library is no more required, so just axe it.
attilio [Thu, 5 Mar 2009 18:27:16 +0000 (18:27 +0000)]
libc_r_* library is no more required, so just axe it.

Approved by: marcel, emaste
Sponsored by: Sandvine Incorporated

15 years agoMake pmap_copy() more TLB friendly. Specifically, make it use the kernel's
alc [Thu, 5 Mar 2009 18:11:26 +0000 (18:11 +0000)]
Make pmap_copy() more TLB friendly.  Specifically, make it use the kernel's
direct map instead of the pmap's recursive mapping to access the lowest
level in the page table.

MFC after: 6 weeks

15 years agoA few cleanups to the FPU code on amd64:
jhb [Thu, 5 Mar 2009 16:56:16 +0000 (16:56 +0000)]
A few cleanups to the FPU code on amd64:
- fpudna() always returned 1 since amd64 CPUs always have FPUs.  Change
  the function to return void and adjust the calling code in trap() to
  assume the return 1 case is the only case.
- Remove fpu_cleanstate_ready as it is always true when it is tested.
  Also, only initialize fpu_cleanstate when fpuinit() is called on the BSP.

Reviewed by: bde

15 years agoMove the PCB flag macros up next to the 'pcb_flags' member in the struct.
jhb [Thu, 5 Mar 2009 16:52:50 +0000 (16:52 +0000)]
Move the PCB flag macros up next to the 'pcb_flags' member in the struct.

15 years agoAdd support for the single-port NetMos NM9835 serial adapter. The puc(4)
jhb [Thu, 5 Mar 2009 16:43:33 +0000 (16:43 +0000)]
Add support for the single-port NetMos NM9835 serial adapter.  The puc(4)
entry is a specific entry to override the generic NetMos entry so that
puc(4) will leave this device alone and let uart(4) claim it.

Submitted by: Navdeep Parhar  nparhar @ gmail
Reviewed by: marcel
MFC after: 1 week

15 years agoSAs are valid (but dying) when they reached soft lifetime,
vanhu [Thu, 5 Mar 2009 16:22:32 +0000 (16:22 +0000)]
SAs are valid (but dying) when they reached soft lifetime,
even if they have never been used.

Approved by: gnn(mentor)
MFC after: 2 weeks