]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
15 years agoUnbreak the build without INET6.
bz [Tue, 25 Nov 2008 09:49:05 +0000 (09:49 +0000)]
Unbreak the build without INET6.

15 years agoUnbreak the build until alfred@ is back to fix it correctly.
bz [Tue, 25 Nov 2008 09:17:01 +0000 (09:17 +0000)]
Unbreak the build until alfred@ is back to fix it correctly.
For now disconnect the nonexistent serial_3g.

15 years agosrc/lib/libusb20/libusb20_compat01.c
alfred [Tue, 25 Nov 2008 08:04:40 +0000 (08:04 +0000)]
src/lib/libusb20/libusb20_compat01.c

Fix some issues about re-scanning of the devices.

src/lib/libusb20/libusb20_ugen20.c

Fix issue about libusb20 having to release the
USB transfers before doing a SET_CONFIG, else
the kernel will kill the file handle.

src/sys/dev/usb2/core/usb2_device.
src/sys/dev/usb2/core/usb2_generic.c
src/sys/dev/usb2/core/usb2_generic.h

Add support for U3G devices.
Improve and cleanup FIFO free handling.
Improve device re-enumeration.

src/sys/dev/usb2/core/usb2_msctest.c
src/sys/dev/usb2/core/usb2_msctest.h

Fix some problems in the USB Mass Storage Test.
Add Huawei vendor specific quirks.

src/sys/dev/usb2/core/usb2_request.c

Improve device re-enumeration.

src/sys/dev/usb2/ethernet/if_aue2.c
src/sys/dev/usb2/include/usb2_devid.h
src/sys/dev/usb2/include/usb2_devtable.h
src/sys/dev/usb2/quirk/usb2_quirk.c

Integrate changes from the old USB driver.

src/sys/dev/usb2/include/usb2_standard.h

Add definition of USB3.0 structures from USB.org.

src/sys/dev/usb2/serial/u3g2.c
src/sys/dev/usb2/serial/ugensa2.c
src/sys/modules/usb2/Makefile
src/sys/modules/usb2/serial_3g/Makefile

Import U3G driver.

Submitted by: Hans Petter Selasky (usb4bsd)

15 years agoBig update to the iSCSI initiator code. Highlights include IPv6 support,
scottl [Tue, 25 Nov 2008 07:17:11 +0000 (07:17 +0000)]
Big update to the iSCSI initiator code.  Highlights include IPv6 support,
many bugs fixes, many more performance improvements.

Submitted by: Danny Braniss

M    sbin/iscontrol/iscsi.conf.5
M    sbin/iscontrol/iscontrol.8
M    sbin/iscontrol/iscontrol.h
M    sbin/iscontrol/config.c
M    sbin/iscontrol/fsm.c
M    sbin/iscontrol/login.c
M    sbin/iscontrol/pdu.c
M    sbin/iscontrol/misc.c
M    sbin/iscontrol/auth_subr.c
M    sbin/iscontrol/iscontrol.c
M    sys/dev/iscsi/initiator/isc_cam.c
M    sys/dev/iscsi/initiator/iscsi.h
M    sys/dev/iscsi/initiator/isc_soc.c
M    sys/dev/iscsi/initiator/iscsi_subr.c
M    sys/dev/iscsi/initiator/iscsivar.h
M    sys/dev/iscsi/initiator/isc_subr.c
M    sys/dev/iscsi/initiator/iscsi.c
M    sys/dev/iscsi/initiator/isc_sm.c

15 years agoSave boot args.
imp [Tue, 25 Nov 2008 05:17:39 +0000 (05:17 +0000)]
Save boot args.

15 years agomore silly whitespace changes.
imp [Tue, 25 Nov 2008 05:12:19 +0000 (05:12 +0000)]
more silly whitespace changes.

15 years agoSimplify Tx checksum offload configuration. Also clear CSUM_IP if
yongari [Tue, 25 Nov 2008 04:33:02 +0000 (04:33 +0000)]
Simplify Tx checksum offload configuration. Also clear CSUM_IP if
we've computed IP checksum with software method.

15 years ago- Allow fxp_encap() enqueue failed transmissions and set
yongari [Tue, 25 Nov 2008 04:16:16 +0000 (04:16 +0000)]
- Allow fxp_encap() enqueue failed transmissions and set
  IFF_DRV_OACTIVE to note resource shortage to upper stack.
- Don't count number of mbuf chains. Default 32 DMA segments for a
  frame is enough for most cases. If bus_dmamap_mbuf_sg fails use
  m_collapse(9) to collapse the mbuf chain instead of relying on
  expensive m_defrag(9).
- Move bpf handling to fxp_start_body() which is supposed to be
  more appropriate place.
- Always arm watchdog timer whenever a new Tx request is made.
  Previously fxp(4) used to arm watchdog timer only when
  FXP_CXINT_THRESH-th Tx request is made. Because fxp(4) does not
  rely on Tx interrupt to reclaim transmitted mbufs it's better to
  arm watchdog timer to detect potential lockups.
- Add more aggresive Tx buffer reclaiming in fxp_start_body to make
  room for new Tx requests. Since fxp(4) does not request Tx
  completion interrupt for every frames it's necessary to clean
  TXCBs in advance to saturate link.
- Make fxp(4) try to start more packets transmitting regardless of
  interrupt type in fxp_intr_body.

15 years agoSimplify mode_t check treatment (suggested by trasz).
daichi [Tue, 25 Nov 2008 03:49:41 +0000 (03:49 +0000)]
Simplify mode_t check treatment (suggested by trasz).
By semantical view, trasz's code is better than prior one.

Submitted by: trasz
Reviewed by: Masanori OZAWA <ozawa@ongs.co.jp>

15 years agoFixes Unionfs socket issue reported as kern/118346.
daichi [Tue, 25 Nov 2008 03:18:35 +0000 (03:18 +0000)]
Fixes Unionfs socket issue reported as kern/118346.

PR: 118346
Submitted by: Masanori OZAWA <ozawa@ongs.co.jp>
Discussed at: devsummit Strassburg, EuroBSDCon2008
Discussed with: rwatson, gnn, hrs
MFC after: 2 week

15 years ago- improve AL2230 RF handling when the device type is zd1211. After this
weongyo [Tue, 25 Nov 2008 02:19:02 +0000 (02:19 +0000)]
- improve AL2230 RF handling when the device type is zd1211.  After this
  patch the RX/TX performance becomes about 17~18 Mbps comparing with
  the previous whose values were RX 7~8Mbps and TX 13~14Mbps.
- improve AL2230 RF handling in zd1211b
- support AL2230S RF that PV2000 is renamed to AL2230S
- use register ZYD_CR244, ZYD_CR243, ZYD_CR242 when the driver writes
  values on RF.  This routine is more faster than the original one
- use private TX lock to avoid LOR at zyd_raw_xmit()
- increase TX slots from 1 to 5
- needs to set the channel at IEEE80211_S_AUTH not IEEE80211_S_RUN
- detailed error handling.  In previous the next command was sent to the
  device even if there was errors
- setting ZYD_MAC_RX_THRESHOLD value should be different between 1211
  and 1211b
- only try to stop the device at zyd_init_locked() if the device is
  UPed
- do not use MTX_RECURSE
- do not try to grap Giant lock when the channel is changing
- move the device initialization routines from zyd_attach to zyd_init to
  give a device full-reset chance to the driver.
- code cleanup at zyd_raw_xmit()
- simplify zyd_attach() routines
- resort functions and clean up variables
- DPRINTF style change.
- style(9)

Reviewed by: sam

15 years agoFixing !INET6 builds.
avatar [Tue, 25 Nov 2008 02:15:09 +0000 (02:15 +0000)]
Fixing !INET6 builds.

15 years agoMove unarming watchdog timer and pending multicast configuration
yongari [Tue, 25 Nov 2008 02:05:01 +0000 (02:05 +0000)]
Move unarming watchdog timer and pending multicast configuration
check to fxp_txeof(). While I'm here unarm watchdog timer only if
there are no pending queued Tx requests.
Previously the watchdog timer was unarmed whenever Tx interrupt is
raised. This could be resulted in hiding root cause of watchdog
timeouts.

15 years agoFix Tx/Rx checksum offload ioctl handling and make Rx handler honor
yongari [Tue, 25 Nov 2008 01:33:16 +0000 (01:33 +0000)]
Fix Tx/Rx checksum offload ioctl handling and make Rx handler honor
checksum offload configuration. Now checksum offload can be
controlled by ifconfig(8).
While I'm here add an additional check for interface capabilities
before applying user's request.

15 years agoMake fxp(4) build with FXP_IP_CSUM_WAR.
yongari [Tue, 25 Nov 2008 01:01:48 +0000 (01:01 +0000)]
Make fxp(4) build with FXP_IP_CSUM_WAR.

15 years agoSort head files and removed ununsed header file.
yongari [Tue, 25 Nov 2008 00:59:11 +0000 (00:59 +0000)]
Sort head files and removed ununsed header file.

15 years agoFix various whitespace botches, mostly having them at the end of a line.
imp [Tue, 25 Nov 2008 00:48:15 +0000 (00:48 +0000)]
Fix various whitespace botches, mostly having them at the end of a line.

15 years agoWhitespace fix.
yongari [Tue, 25 Nov 2008 00:48:05 +0000 (00:48 +0000)]
Whitespace fix.

15 years agoAdd support for Marvell PATA M88SX6121.
rpaulo [Tue, 25 Nov 2008 00:39:03 +0000 (00:39 +0000)]
Add support for Marvell PATA M88SX6121.

PR: 129154
Submitted by: João Barros <joao.barros at gmail.com>
MFC after: 1 month

15 years agoReplace disclaimer with the one from COPYRIGHT. Joint authors aggreed
imp [Tue, 25 Nov 2008 00:14:49 +0000 (00:14 +0000)]
Replace disclaimer with the one from COPYRIGHT.  Joint authors aggreed
to the change.

15 years agoReplace three magic constants with L1_S_SIZE, since that's what is
imp [Tue, 25 Nov 2008 00:14:14 +0000 (00:14 +0000)]
Replace three magic constants with L1_S_SIZE, since that's what is
really meant in those places.

15 years agoTweak the disclaimer section of the license to match COPYRIGHT, for
imp [Tue, 25 Nov 2008 00:13:26 +0000 (00:13 +0000)]
Tweak the disclaimer section of the license to match COPYRIGHT, for
better or worse.  Ok'd by folks that have additional copyrights to the
files in cases where there's joint authorship.

15 years agoCorrect typo in comment: thier -> their
emaste [Mon, 24 Nov 2008 19:28:52 +0000 (19:28 +0000)]
Correct typo in comment: thier -> their

15 years agoRestore sfence semantics in mxge after the introduction
gallatin [Mon, 24 Nov 2008 19:00:57 +0000 (19:00 +0000)]
Restore sfence semantics in mxge after the introduction
of a global mfence based mb() in r185162

15 years agoMake sure arc4random(9) is properly seeded when /etc/rc.d/initrandom returns.
cperciva [Mon, 24 Nov 2008 17:39:39 +0000 (17:39 +0000)]
Make sure arc4random(9) is properly seeded when /etc/rc.d/initrandom returns.

Approved by: so (cperciva)
Approved by: re (kensmith)
Security: FreeBSD-SA-08:11.arc4random

15 years agouse consistent style
sam [Mon, 24 Nov 2008 17:34:00 +0000 (17:34 +0000)]
use consistent style

15 years agoThis actually works on Linux, I just had wrong directory permission.
pjd [Mon, 24 Nov 2008 16:33:23 +0000 (16:33 +0000)]
This actually works on Linux, I just had wrong directory permission.

Found by: trasz

15 years agoInclude TODO messages even if tests succeeds, so we can detect when something
pjd [Mon, 24 Nov 2008 16:32:39 +0000 (16:32 +0000)]
Include TODO messages even if tests succeeds, so we can detect when something
suddenly started to work.

15 years agoChange the universe target to warn the user for every world or kernel that
des [Mon, 24 Nov 2008 11:23:14 +0000 (11:23 +0000)]
Change the universe target to warn the user for every world or kernel that
fails.  The error message includes a reference to the relevant log file.

Approved by: phk
MFC after: 1 week

15 years agoCorrection: these PMCs do not support a "umask" modifier.
jkoshy [Mon, 24 Nov 2008 08:14:01 +0000 (08:14 +0000)]
Correction: these PMCs do not support a "umask" modifier.

15 years ago- Document the rules used to determine when spellings of events
jkoshy [Mon, 24 Nov 2008 08:11:33 +0000 (08:11 +0000)]
- Document the rules used to determine when spellings of events
  are equivalent.
- Reorder text to make the manual page more coherent.

15 years agoSave the value read from TX queue CSR, so we don't test against a
yongari [Mon, 24 Nov 2008 02:21:50 +0000 (02:21 +0000)]
Save the value read from TX queue CSR, so we don't test against a
stale one.

Submitted by: sephe

15 years agoprint the extended tx/rx descriptor for 5416 and later parts
sam [Mon, 24 Nov 2008 01:34:56 +0000 (01:34 +0000)]
print the extended tx/rx descriptor for 5416 and later parts

15 years agonuke special handling of RXORN interrupt; the hal marks the FATAL
sam [Mon, 24 Nov 2008 01:31:10 +0000 (01:31 +0000)]
nuke special handling of RXORN interrupt; the hal marks the FATAL
bit in the interrupt status when RXORN is hit and the chip requires
a reset so our special handling was causing useless resets

15 years agoDocument ale(4).
delphij [Mon, 24 Nov 2008 01:26:22 +0000 (01:26 +0000)]
Document ale(4).

15 years agoFixed style issues with variable ordering and naming, spacing and
scf [Sun, 23 Nov 2008 23:26:12 +0000 (23:26 +0000)]
Fixed style issues with variable ordering and naming, spacing and
parentheses.

Fixed alignment issue in gr_dup() in its assignment of gr_mem using a
struct to force alignment without performing alignment mathematics.  This
was noticed recently with libutil was built with WARNS=6 on platform such
as sparc64.

Added checks to gr_dup(), gr_equal() and gr_make() to prevent segfaults
when examining struct group's with the struct members pointing to NULL's.

With fix of alignment issue, restore WARNS?=6.

Reviewed by: des
MFC after: 1 week

15 years agoRemove DELAY() on hdac_audio_commit().
mav [Sun, 23 Nov 2008 23:19:31 +0000 (23:19 +0000)]
Remove DELAY() on hdac_audio_commit().
I don't see the reason why it is needed.

15 years agoEliminate an unused variable reported by coverity
imp [Sun, 23 Nov 2008 23:04:28 +0000 (23:04 +0000)]
Eliminate an unused variable reported by coverity

Submitted by: Ganbold

15 years agoIt's possible that the dump device has gone away after it was
dwmalone [Sun, 23 Nov 2008 21:05:22 +0000 (21:05 +0000)]
It's possible that the dump device has gone away after it was
configured, change the message to let people know this is a
possibility. I've slightly changed the message from the one
submitted by Pekka to keep the printf on one line.

Submitted by: Pekka Savola <pekkas@netcore.fi>

15 years agoMake the new CCEQ macro a little more like the old one - first do
dwmalone [Sun, 23 Nov 2008 21:01:28 +0000 (21:01 +0000)]
Make the new CCEQ macro a little more like the old one - first do
the comparison between c and val and then compare val to _POSIX_VDISABLE.
This avoids comparing c (which is usually of type char) to
_POSIX_VDISABLE (which has value 0xff and may not be representable
as a char).

Reviewed by: ed

15 years agoAdd a test for r185231.
stefanf [Sun, 23 Nov 2008 20:27:03 +0000 (20:27 +0000)]
Add a test for r185231.

15 years agoFix $? at the first command of a function. The previous exit status was saved
stefanf [Sun, 23 Nov 2008 20:23:57 +0000 (20:23 +0000)]
Fix $? at the first command of a function.  The previous exit status was saved
twice and thus lost.

15 years agoDo not use soft PCM volume for digital outputs to allow SPDIF AC3 bypass
mav [Sun, 23 Nov 2008 20:19:35 +0000 (20:19 +0000)]
Do not use soft PCM volume for digital outputs to allow SPDIF AC3 bypass
to work. Soft PCM volume (and vchans) unable to bypass AC3 stream now.

15 years agoMark all the places where Linux is not POSIX-compilant. Tested on ext3.
pjd [Sun, 23 Nov 2008 20:17:17 +0000 (20:17 +0000)]
Mark all the places where Linux is not POSIX-compilant. Tested on ext3.

15 years agoDon't use PRIV_ROOT. Here we check if user can share ZFS file system, so
pjd [Sun, 23 Nov 2008 20:14:19 +0000 (20:14 +0000)]
Don't use PRIV_ROOT. Here we check if user can share ZFS file system, so
PRIV_NFS_DAEMON seems best choice.

Discussed with: rwatson

15 years agofstest for Linux:
pjd [Sun, 23 Nov 2008 20:07:00 +0000 (20:07 +0000)]
fstest for Linux:
- Use -- when needed so Linux getopt(3) won't get confused.
- Follow POSIX more closely.

Submitted by: Szabolcs Szakacsits <szaka@ntfs-3g.org>

15 years agoRemove vfs_mountedon(9) manual page. The vfs_mountedon(9) has been
hrs [Sun, 23 Nov 2008 20:05:12 +0000 (20:05 +0000)]
Remove vfs_mountedon(9) manual page.  The vfs_mountedon(9) has been
removed since 2004.

15 years agoStrictly differentiate digital and analog PCM devices according to codec
mav [Sun, 23 Nov 2008 20:03:58 +0000 (20:03 +0000)]
Strictly differentiate digital and analog PCM devices according to codec
nodes capabilities. Add "Analog"/"Digital" marks to the pcm device names.

I hope it will help new users easier accept concept of several PCM devices
and understand exact purposes of that devices.

15 years agovn_fullpath(9) does not assume vp is locked by the caller now.
hrs [Sun, 23 Nov 2008 20:02:58 +0000 (20:02 +0000)]
vn_fullpath(9) does not assume vp is locked by the caller now.

15 years agofstest for Linux:
pjd [Sun, 23 Nov 2008 20:02:14 +0000 (20:02 +0000)]
fstest for Linux:
Automatically detect file system type.

15 years agoFreeBSD's way of handling rmdir("..") is not POSIX-compilant.
pjd [Sun, 23 Nov 2008 19:58:26 +0000 (19:58 +0000)]
FreeBSD's way of handling rmdir("..") is not POSIX-compilant.

15 years agoShorter version.
pjd [Sun, 23 Nov 2008 19:57:16 +0000 (19:57 +0000)]
Shorter version.

15 years agoAdd support for pathconf(2).
pjd [Sun, 23 Nov 2008 19:56:40 +0000 (19:56 +0000)]
Add support for pathconf(2).

15 years agoDetect operating system automatically.
pjd [Sun, 23 Nov 2008 19:56:09 +0000 (19:56 +0000)]
Detect operating system automatically.

15 years agoEnhance the explanation of using filesystem-specific mount options
keramida [Sun, 23 Nov 2008 19:52:56 +0000 (19:52 +0000)]
Enhance the explanation of using filesystem-specific mount options
in /etc/fstab.  We do support passing special options on a per
filesystem type basis, like `-u UID -g GID' for mount_msdosfs, but
the syntax of these options in fstab is non-obvious and a lot of
users have asked about it.

PR: docs/128816
Submitted by: Roland Smith, rsmith at xs4all dot nl
MFC after: 2 days

15 years agofstest for Linux:
pjd [Sun, 23 Nov 2008 19:50:20 +0000 (19:50 +0000)]
fstest for Linux:
- Use /dev/urandom, it is more portable.
- Implement todo() function which allows to mark known failures.

15 years agoReport card erase sector size as disk stripe size.
mav [Sun, 23 Nov 2008 14:32:40 +0000 (14:32 +0000)]
Report card erase sector size as disk stripe size.

15 years agoImprove detach handling: close races, flush queue.
mav [Sun, 23 Nov 2008 14:02:06 +0000 (14:02 +0000)]
Improve detach handling: close races, flush queue.

15 years agoImplement -T option as found in Linux df(1). When given, file system type
pjd [Sun, 23 Nov 2008 13:03:44 +0000 (13:03 +0000)]
Implement -T option as found in Linux df(1). When given, file system type
will be included in output.

Reviewed and tested by: keramida
MFC after: 2 weeks

15 years agoAdd backward compatibility ifdefs for non-multiq kernels
kmacy [Sun, 23 Nov 2008 07:30:07 +0000 (07:30 +0000)]
Add backward compatibility ifdefs for non-multiq kernels

15 years agocorrect sorting
sam [Sun, 23 Nov 2008 01:51:14 +0000 (01:51 +0000)]
correct sorting

15 years agowork around periodic leak on queue overrun by enabling coalescing of packets in to
kmacy [Sun, 23 Nov 2008 00:22:52 +0000 (00:22 +0000)]
work around periodic leak on queue overrun by enabling coalescing of packets in to
work requests by default

15 years agobuf_ring_peek should return NULL if the ring is empty rather than
kmacy [Sun, 23 Nov 2008 00:20:51 +0000 (00:20 +0000)]
buf_ring_peek should return NULL if the ring is empty rather than
whatever happened to be at cons_tail last time it was in use

15 years agointr_machdep.h breaks build on some arches and is not needed
kmacy [Sun, 23 Nov 2008 00:13:25 +0000 (00:13 +0000)]
intr_machdep.h breaks build on some arches and is not needed

15 years agoUnbreak previous commit.
marcel [Sat, 22 Nov 2008 22:15:34 +0000 (22:15 +0000)]
Unbreak previous commit.

15 years agoInclude the QUICC backend in the kernel module.
marcel [Sat, 22 Nov 2008 21:24:07 +0000 (21:24 +0000)]
Include the QUICC backend in the kernel module.

PR: 127120

15 years agoCast to uintptr_t before casting to void*. This allows the
marcel [Sat, 22 Nov 2008 21:22:53 +0000 (21:22 +0000)]
Cast to uintptr_t before casting to void*. This allows the
QUICC backend to be built on LP64 platforms. This makes it
possible to include the QUICC backend in the kernel module.

PR: 127120

15 years agoAllow multiple makeoption lines to be used with the += operator, this permits
thompsa [Sat, 22 Nov 2008 21:12:47 +0000 (21:12 +0000)]
Allow multiple makeoption lines to be used with the += operator, this permits
the following syntax in the kernel config.

 makeoptions MODULES_OVERRIDE=foo
 makeoptions MODULES_OVERRIDE+=bar
 makeoptions MODULES_OVERRIDE+=baz

Bump config minor version to 600007.

15 years agoRemove unused variable.
mav [Sat, 22 Nov 2008 16:55:55 +0000 (16:55 +0000)]
Remove unused variable.

Found with:     Coverity Prevent(tm)
CID:            3682

15 years agoFix typo. Clear session stats instead of config and part of stats.
mav [Sat, 22 Nov 2008 16:40:12 +0000 (16:40 +0000)]
Fix typo. Clear session stats instead of config and part of stats.

Found with: Coverity Prevent(tm)
CID: 2472

15 years agoRemove unneeded NULL check. At first msg can't be null here and and at second
mav [Sat, 22 Nov 2008 16:03:18 +0000 (16:03 +0000)]
Remove unneeded NULL check. At first msg can't be null here and and at second
NG_FREE_MSG() also checks it.

Found with: Coverity Prevent(tm)

15 years agoWhitespace cleanup for previous commit.
mav [Sat, 22 Nov 2008 15:38:24 +0000 (15:38 +0000)]
Whitespace cleanup for previous commit.

15 years agoHandle device_get_children() errors in more correct way.
mav [Sat, 22 Nov 2008 15:35:21 +0000 (15:35 +0000)]
Handle device_get_children() errors in more correct way.
We shouldn't detach until all children are surely destroyed.

Found with:     Coverity Prevent(tm)
CID:            2137

15 years agoChange check order to avoid array overrun.
mav [Sat, 22 Nov 2008 14:48:40 +0000 (14:48 +0000)]
Change check order to avoid array overrun.

Found with: Coverity Prevent(tm)
CID: 2497

15 years agoFix amd64 build and re-enable gptzfsboot.
dfr [Sat, 22 Nov 2008 14:24:55 +0000 (14:24 +0000)]
Fix amd64 build and re-enable gptzfsboot.

15 years agoIFp4: Don't rely on disk IDs and always use vdev guids, which means always look
pjd [Sat, 22 Nov 2008 13:33:06 +0000 (13:33 +0000)]
IFp4: Don't rely on disk IDs and always use vdev guids, which means always look
up for components by reading metadata. This might be slower when there are big
number of disks in the system, but is definiately more reliable.

15 years agoIFp4: Regression tests for FreeBSD/ZFS chflags(2)/lchflags(2).
pjd [Sat, 22 Nov 2008 13:27:15 +0000 (13:27 +0000)]
IFp4: Regression tests for FreeBSD/ZFS chflags(2)/lchflags(2).

15 years agoIFp4: Finish implemnetation of chflags(2) for ZFS. While doing this I found
pjd [Sat, 22 Nov 2008 13:24:44 +0000 (13:24 +0000)]
IFp4: Finish implemnetation of chflags(2) for ZFS. While doing this I found
that zfs_access() can only handle VREAD, VWRITE and VEXEC, for the rest we need
to use vaccess(9).

15 years agoIFp4: Don't free pathname too soon, debugging code is still using it.
pjd [Sat, 22 Nov 2008 13:22:24 +0000 (13:22 +0000)]
IFp4: Don't free pathname too soon, debugging code is still using it.

15 years agoBusy ufs filesystem around block of code that does ".." lookup. Since
kib [Sat, 22 Nov 2008 13:11:11 +0000 (13:11 +0000)]
Busy ufs filesystem around block of code that does ".." lookup. Since
mnt_lock is before lock of any vnode on the mp, it uses LK_NOWAIT. Since
MNTK_UNMOUNT may be transient, pdp lock is dropped when vfs_busy()
failed, and operation is retried after some time. This way, ffs_vget()
is not called on the mp that may be in the process of being destroyed by
unmount.

Check for the VI_DOOMED flag on pdp after its lock is reacquired, to
better detect some situations where directory containing ".."
entry is removed during the lookup.

Reviewed by: tegge, attilio (previous version)
Tested by: pho
MFC after: 1 month

15 years agoAdd sv_flags field to struct sysentvec with intention to provide description
kib [Sat, 22 Nov 2008 12:36:15 +0000 (12:36 +0000)]
Add sv_flags field to struct sysentvec with intention to provide description
of the ABI of the currently executing image. Change some places to test
the flags instead of explicit comparing with address of known sysentvec
structures to determine ABI features.

Discussed with: dchagin, imp, jhb, peter

15 years agoUnbreak LINT.
jkoshy [Sat, 22 Nov 2008 12:34:49 +0000 (12:34 +0000)]
Unbreak LINT.

15 years agoAdd flowid to mbuf to allow drivers to uniquely
kmacy [Sat, 22 Nov 2008 08:46:16 +0000 (08:46 +0000)]
Add flowid to mbuf to allow drivers to uniquely
identify connection flows to guarantee ordering
across queues

15 years ago- enable multiple transmit queues
kmacy [Sat, 22 Nov 2008 08:05:05 +0000 (08:05 +0000)]
- enable multiple transmit queues
- invert sense of hw.cxgb.singleq tunable to hw.cxgb.multiq
- don't wake up transmitting thread by default
- add per tx queue ifaltq to handle ALTQ
- remove several unused functions in cxgb_multiq.c
- add several sysctls: multiq_tx_enable, coalesce_tx_enable,
  and wakeup_tx_thread
- this obsoletes the hw.cxgb.snd_queue_len as ifq is replaced
  by a buf_ring

15 years agoconvert calls to IFQ_HANDOFF to if_transmit
kmacy [Sat, 22 Nov 2008 07:35:45 +0000 (07:35 +0000)]
convert calls to IFQ_HANDOFF to if_transmit

15 years agoDefine mb(), rmb() and wmb() for real.
marcel [Sat, 22 Nov 2008 06:56:49 +0000 (06:56 +0000)]
Define mb(), rmb() and wmb() for real.

15 years ago- bump __FreeBSD version to reflect added buf_ring, memory barriers,
kmacy [Sat, 22 Nov 2008 05:55:56 +0000 (05:55 +0000)]
- bump __FreeBSD version to reflect added buf_ring, memory barriers,
  and ifnet functions

- add memory barriers to <machine/atomic.h>
- update drivers to only conditionally define their own

- add lockless producer / consumer ring buffer
- remove ring buffer implementation from cxgb and update its callers

- add if_transmit(struct ifnet *ifp, struct mbuf *m) to ifnet to
  allow drivers to efficiently manage multiple hardware queues
  (i.e. not serialize all packets through one ifq)
- expose if_qflush to allow drivers to flush any driver managed queues

This work was supported by Bitgravity Inc. and Chelsio Inc.

15 years agoSeveral small additions to the Chelsio 10G driver.
gnn [Fri, 21 Nov 2008 19:22:25 +0000 (19:22 +0000)]
Several small additions to the Chelsio 10G driver.

1) Fix a bug in dealing with the Alerus 1006 PHY which prevented the
device from ever coming back up once it had been set to down.

2) Add a kernel tunable (hw.cxgb.snd_queue_len) which makes it possible
to give the device more than IFQ_MAXLEN entries in its send queue.  The
default remains 50.

3) Add code to place the card'd identification and serial number into
its description (%desc) so that users can tell which card they have
installed.

15 years ago'Strict EAP conformance' makes more sense here than 'String EAP
imp [Fri, 21 Nov 2008 18:15:39 +0000 (18:15 +0000)]
'Strict EAP conformance' makes more sense here than 'String EAP
conformance.'

15 years agouse the correct variable in a debug message.
luigi [Fri, 21 Nov 2008 09:14:29 +0000 (09:14 +0000)]
use the correct variable in a debug message.

15 years agoAdd definitions for ZFS pool version 13.
dfr [Fri, 21 Nov 2008 09:10:35 +0000 (09:10 +0000)]
Add definitions for ZFS pool version 13.

15 years agoDisconnect gptzfsboot from the build until dfr@ gets his act together.
des [Fri, 21 Nov 2008 08:10:07 +0000 (08:10 +0000)]
Disconnect gptzfsboot from the build until dfr@ gets his act together.

15 years agoCreate a /dev/cardbus%d.cis, to be compatible with older versions of
imp [Fri, 21 Nov 2008 03:03:57 +0000 (03:03 +0000)]
Create a /dev/cardbus%d.cis, to be compatible with older versions of
the software.  This is a trivial amount of code to keep wireless
monitoring software working...  I plan on removing it in 9.0.

15 years ago- According to OpenSolaris, CDMA flushing/syncing for Tomatillos
marius [Thu, 20 Nov 2008 18:44:09 +0000 (18:44 +0000)]
- According to OpenSolaris, CDMA flushing/syncing for Tomatillos
  and XMITS has to be basically done in the same manner as for
  the Sabres, i.e. only for devices behind PCI-PCI-bridges and
  after a PIO read on the far side of the farest PCI-PCI-bridge.
  Given that the Tomatillo documentation mentions no difference
  to the Schizo bridges in this regard and this is also still
  part of the procedure described Schizo documentation this
  seems about right so adjust accordingly (the unconditional
  CDMA flushing/syncing previously done was based on how Linux
  behaves).
- Implement CDMA flushing/syncing for Schizo version >= 5,
  which requires the workaround described in Schizo Errata I-23.
  According to Schizo Errata I-13 it's just unusable with
  version < 5 though. [1]
- Don't register the Schizo streaming buffer for now until it's
  usage is sorted out according to the erratas.
- Register our interrupt filters with the revived INTR_FAST so
  they these interrupts can even interrupt filters of device
  drivers as necessary.
- Remove the comment regarding lack of newbus'ified bus_dma(9)
  as being able to associate a DMA tag with a device would
  allow to implement CDMA flushing/syncing in bus_dmamap_sync(9)
  but that would totally kill performance. Given that for devices
  not behind a PCI-PCI bridge the host-to-PCI bridges also only
  do CDMA flushing/syncing based on interrupts there's no
  additional disadvantage for polling(4) callbacks in the case
  schizo(4) has to do the CDMA flushing/syncing but rather a
  general problem.

Reported by: Michael Moll [1]

15 years agoAs reported in kern/118222, pxeboot in RELENG7 (and presumably
luigi [Thu, 20 Nov 2008 14:57:09 +0000 (14:57 +0000)]
As reported in kern/118222, pxeboot in RELENG7 (and presumably
above) exhibits some misbehaviours on machines with AMD64 CPUs,
which at least in some cases I have tracked down to a heap overflow.

It is unclear whether it depends on the CPU or on the pxe bios
itself which may use more memory on AMD machines.

Noticeably a pxeboot compiled from 6.x sources works fine on all
machines I have tried so far, while a pxeboot compiled from 7.x
sources does not.

This patch is a first step in reducing the amount of memory used
while processing the configuration files read by the loader at boot
(some of them are quite large, 1700+ lines), and it does so by:
+ moving a buffer to static memory instead of allocating in the heap;
+ skipping empty lines;
+ reducing the amount of memory used for line descriptors;

Unfortunately there are several changes between 6.x and above,
affecting the compiler, the loader code itself, and libstand,
and it is not so straightforward to
These changes fix the behaviour on one motherboard with a
single-core AMD cpu, but are still not enough e.g on an Asus
M2N-VM (with a dual-core CPU).

I need to investigate the problem a bit more before figuring
out what should be committed to RELENG_7

PR: kern/118222

15 years agodamn. Always do make depend. Forgot to recompile main because of it,
imp [Thu, 20 Nov 2008 08:32:19 +0000 (08:32 +0000)]
damn.  Always do make depend.  Forgot to recompile main because of it,
so the changes for the struct cis -> struct tuple_list didn't get
made.  They have been now.

15 years agoFix check for link target so we don't print cardbus CIS information twice.
imp [Thu, 20 Nov 2008 08:30:15 +0000 (08:30 +0000)]
Fix check for link target so we don't print cardbus CIS information twice.
Also, eliminate some magic constants and replace them with values from cis.h.

15 years agoRestore now-useless ioctl as a roadmap. The original dumpcis code
imp [Thu, 20 Nov 2008 08:20:53 +0000 (08:20 +0000)]
Restore now-useless ioctl as a roadmap.  The original dumpcis code
assumed it had to toggle between attribute and common memory in the
cards.  The kernel is supposed to cope with that automatically and
give us a tuple list.  However, there's a number of details of how
that happens that's currently, ummm, magical and/or not implemented
for 16-bit PC Cards that have CIS_LONGLINK_C tuples in them (eg, mix
both attribute memory and common memory).  Also, CIS_LOGNLINK_A
entries might not be handled completely correctly either, since there
can be gaps in the attribute vs common stuff.

All this will need to be corrected in the kernel.  Once it is
corrected, dumpcis can be made even simpler in some ways, a little
more complicated in others once an API for presentation of CIS to
userland in these weird cases is settled upon.

15 years agoMore dead code removal.
imp [Thu, 20 Nov 2008 08:14:14 +0000 (08:14 +0000)]
More dead code removal.

15 years agoThe original programs that this code was lifted from (pccardd and
imp [Thu, 20 Nov 2008 08:12:26 +0000 (08:12 +0000)]
The original programs that this code was lifted from (pccardd and
pccardc) parsed data to make decisions about stuff related to card
configuration.

The purely CIS dumping aspect of this program obviates the need for
such parsing.  Save some space and don't parse the data anymore for
configuration purposes.  Just parse it to print an interpreatation of
it.