Sepherosa Ziehau [Wed, 18 Mar 2009 13:11:19 +0000 (21:11 +0800)]
jme(4): Using code logic to create redirect table.
- Avoid using magic number when creating redirect table, apply
code logic instead.
- Add comment about how the redirect table is filled.
Sepherosa Ziehau [Wed, 18 Mar 2009 12:20:27 +0000 (20:20 +0800)]
jme(4): Get key from Toeplitz code
Sepherosa Ziehau [Wed, 18 Mar 2009 12:17:42 +0000 (20:17 +0800)]
toeplitz: Add toeplitz_get_key() and nuke toeplitz_get_keyseed()
toeplitz_get_key() hides more implementation details (i.e. key
seed replication) from callers than toeplitz_get_keyseed().
Sascha Wildner [Wed, 18 Mar 2009 10:39:52 +0000 (11:39 +0100)]
pax(1): Remove argument names from prototypes.
Sascha Wildner [Wed, 18 Mar 2009 10:30:53 +0000 (11:30 +0100)]
pax(1): Fix WARNS regression
Sascha Wildner [Tue, 17 Mar 2009 21:43:45 +0000 (22:43 +0100)]
motd(5) manual page: "Update" the example for DragonFly.
Reported-by: free-coder
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1316>
Sascha Wildner [Tue, 17 Mar 2009 21:40:33 +0000 (22:40 +0100)]
shlock(1) manual page: Fix typo
Reported-by: free-coder
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1316>
Hasso Tepper [Tue, 17 Mar 2009 14:41:03 +0000 (16:41 +0200)]
Create destination on copy if it doesn't exists.
Should conform to SUSv3 and fixes a lot of building failures in pkgsrc.
Digged-out-by: Alex Hornung
Sepherosa Ziehau [Tue, 17 Mar 2009 12:58:18 +0000 (20:58 +0800)]
atphy: Reset PHY after extracting MII capabilities, since reset will
trigger auto negotiation, which depends on MII capabilities.
Sepherosa Ziehau [Tue, 17 Mar 2009 12:53:37 +0000 (20:53 +0800)]
ale(4): Don't read/write 1000baseT specific MII registers on 10/100M only device
It looks like on 10/100M only device they do contain "valid" 1000baseT values.
Sepherosa Ziehau [Tue, 17 Mar 2009 12:41:48 +0000 (20:41 +0800)]
bce/bge(4): Allow txcsum/rxcsum to be set individually
Sepherosa Ziehau [Tue, 17 Mar 2009 11:46:52 +0000 (19:46 +0800)]
tcp_wrapper: Make sure that the buf has more than 1byte, before calling fgets()
FreeBSD-PR: kern/132705
Obtained-from: OpenBSD
Sepherosa Ziehau [Tue, 17 Mar 2009 11:09:27 +0000 (19:09 +0800)]
ether_demux: Add comment
Sepherosa Ziehau [Tue, 17 Mar 2009 11:04:32 +0000 (19:04 +0800)]
bridge: If we could enter bridge_input, then M_ETHER_BRIDGED will not be set.
Sepherosa Ziehau [Tue, 17 Mar 2009 11:00:37 +0000 (19:00 +0800)]
ether_input: Add M_ETHER_VLANCHECKED
- In ether_input_handler(), if ether_vlancheck() has not been called
(M_ETHER_VLANCHECKED) call it there; this could happen with upcoming
RSS changes.
- Clear ethernet protocol mbuf flags before entering upper layer.
Sepherosa Ziehau [Mon, 16 Mar 2009 12:39:53 +0000 (20:39 +0800)]
ether_input: Factor out ether_vlancheck()
Sepherosa Ziehau [Sun, 15 Mar 2009 09:40:24 +0000 (17:40 +0800)]
Add M_LENCHECKED mbuf flag.
This flag indicates that various protocol length fields of the packet
are checked. It is only used by IPv4 currently (ip_lengthcheck()).
Sepherosa Ziehau [Sun, 15 Mar 2009 08:42:07 +0000 (16:42 +0800)]
Define M_ETHER_BRIDGED to M_PROTO1
Sepherosa Ziehau [Sun, 15 Mar 2009 08:31:44 +0000 (16:31 +0800)]
Duplicated M_PROTO1 in M_COPYFLAGS
Sepherosa Ziehau [Sun, 15 Mar 2009 08:19:00 +0000 (16:19 +0800)]
802.11: Clear M_PROTO? flags before calling ifnet.if_input
Sepherosa Ziehau [Sun, 15 Mar 2009 03:20:40 +0000 (11:20 +0800)]
Defer ethernet mcast/bcast address detection to ether_input_handler()
Sepherosa Ziehau [Sat, 14 Mar 2009 12:59:11 +0000 (20:59 +0800)]
ip_demux: If RSS kernel option is set, use Toeplitz hash function
Sascha Wildner [Mon, 16 Mar 2009 21:52:42 +0000 (22:52 +0100)]
Sync zoneinfo database with tzdata2009c from elsie.
northamerica: 8.26 -> 8.27
Cuba started DST one week earlier than previously thought.
Matthias Schmidt [Mon, 16 Mar 2009 14:26:06 +0000 (15:26 +0100)]
Remove CVS Id from DragonFly fortune file
Remove the ID as the information is now misleading. The content is
newer as the ID suggests.
Spotted-by: hasso@
Matthew Dillon [Mon, 16 Mar 2009 04:45:57 +0000 (21:45 -0700)]
HAMMER VFS - rebalancing, zero unused fields in boundary element
The rebalancer was not zeroing certain unused fields in a newly created
boundary element when truncating an existing internal B-Tree node.
While these fields are not used by the filesystem for boundary elements,
they are still expected to be zero'd out.
Matthew Dillon [Sun, 15 Mar 2009 22:28:38 +0000 (15:28 -0700)]
HAMMER VFS - Add a B-Tree rebalancing feature.
This is the initial commit of B-Tree rebalancing support for HAMMER.
The rebalancer may be run using the 'hammer rebalance' utility directive.
The leafs in a HAMMER B-Tree all reside at the same depth. Insertions and
deletions only collapse the B-Tree when a leaf node becomes empty and then
only if any necessary recursion (possibly reaching the root node) succeeds.
No balancing occurs during normal operation and B-Tree nodes can wind up
with wildly different element counts which bloats the tree and makes
searches less efficient.
The rebalancer effectively does a depth-first traversal of the B-Tree,
visiting leaf nodes first and parent nodes as a trailing function on the
way back up the tree. For any given internal node the sum total of
elements contained in its children is divided by the number of children.
The effective number of children is reduced as is practical to obtain a 75%
fill level. The elements are then packed into the children and any
wholely empty children left over are deleted. The rebalancer does not
create new B-Tree nodes.
Element packing is fairly complex, requiring tracked cursors, on-media
parent pointers, mirror TIDs, and boundary elements to be updated. The
rebalancer must hold a large number of B-Tree nodes exclusively locked
while running.
Matthew Dillon [Sun, 15 Mar 2009 19:26:06 +0000 (12:26 -0700)]
Remove conditionals based on #if defined(i386) which wrapped code which was
unused. Gcc does not define 'i386' anyway, it uses __i386__.
This also fixes some clang issues.
Requested-by: Alex Hornung
Matthew Dillon [Sun, 15 Mar 2009 00:35:39 +0000 (17:35 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Sun, 15 Mar 2009 00:22:40 +0000 (17:22 -0700)]
HAMMER Utility: Change B-Tree, inodes, and dir reblocking params, add rebalance
* The normal cleanup operations now reblock all B-Tree, inode, and directory
elements in the normal daily reblock mode instead of only the ones in
fragmented big-blocks. Bulk data is handled by the 30-day recopy mode.
* Add a new directive 'rebalance' (a future VFS ioctl). This directive will
tell the HAMMER VFS to rebalance the B-Tree. HAMMER B-Trees are always
balanced by depth but degenerate cases with minimal elements in a node
can easily build up. The new directive will rebalance the elements in
each B-Tree node.
* The hammer cleanup directive was not reblocking directories. Now it does.
It usually does not take very long to reblock the B-Tree nodes, inodes, or
directory elements. Reblocking these unconditionally, instead of just
reblocking fragmented allocation areas, keeps the B-Tree in a more optimal
layout, though there is still a lack of correlation between inode numbers
and directory scan order.
Sepherosa Ziehau [Sat, 14 Mar 2009 09:02:12 +0000 (17:02 +0800)]
rss: Add RSS_DEBUG kernel option
Sepherosa Ziehau [Sat, 14 Mar 2009 08:14:45 +0000 (16:14 +0800)]
rss: Add Toeplitz hash function
Sepherosa Ziehau [Sat, 14 Mar 2009 06:11:16 +0000 (14:11 +0800)]
ip_demux: Wrap INP_MPORT_HASH_{TCP,UDP}() around INP_MPORT_HASH()
Sepherosa Ziehau [Sat, 14 Mar 2009 06:01:29 +0000 (14:01 +0800)]
ip_demux: Test IN_MULTICAST() if RSS kernel option is not enabled.
Sepherosa Ziehau [Sat, 14 Mar 2009 03:41:35 +0000 (11:41 +0800)]
Add static ARP support.
Obtained-from: FreeBSD
Sepherosa Ziehau [Sat, 14 Mar 2009 02:56:25 +0000 (10:56 +0800)]
emx(4): Fix comment
Sepherosa Ziehau [Sat, 14 Mar 2009 02:54:42 +0000 (10:54 +0800)]
ifconfig(8): Remove string of if_flags, which will never be adopted.
Sepherosa Ziehau [Sat, 14 Mar 2009 02:22:25 +0000 (10:22 +0800)]
jme(4): Correct SIOCSIFCAP ioctl handling
- Upper layer will always check if_capabilities against ifreq,
so we don't need to check if_capabilities again.
- When IFCAP_RSS changes, the jme_init() should be called only if
the interface is running.
- Don't use compile time condition for the code handling IFCAP_RSS.
Sepherosa Ziehau [Sat, 14 Mar 2009 02:20:53 +0000 (10:20 +0800)]
emx(4): Support IFCAP_RSS
Sepherosa Ziehau [Fri, 13 Mar 2009 12:46:37 +0000 (20:46 +0800)]
rss: Add RSS kernel option; Add IFCAP_RSS iface capability
Convert jme(4) to use the new IFCAP_RSS.
Matthew Dillon [Fri, 13 Mar 2009 04:15:46 +0000 (21:15 -0700)]
Update cpdup rev to 1.12.
Matthew Dillon [Thu, 12 Mar 2009 16:15:10 +0000 (09:15 -0700)]
Fix kernel panic on NULL pointer deref w/ certain ktrace commands, allow -c -f.
Fix a kernel panic on a NULL pointer dereference with certain ktrace -c
commands.
Allow the -c -f <file> combination (taken from FreeBSD).
Submitted-by: Stathis Kamperis <beket>
Matthew Dillon [Thu, 12 Mar 2009 16:08:55 +0000 (09:08 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Thu, 12 Mar 2009 16:03:08 +0000 (09:03 -0700)]
HAMMER VFS - return shorter-form softlinks "@@-1:%05d" when appropriate.
The HAMMER VFS supports a short "@@-1:%05d" for master PFSs. Adjust the
HAMMER VFS to return softlinks in that form and adjust the hammer cleanup
code to recognize softlinks in that form.
Note that PFS softlinks are created as "@@PFS%05d", but the HAMMER VFS
presents them in an expanded form which allows the HAMMER VFS to reflect
the latest synchronized transaction id on slave PFSs. This also prevents
slave PFSs from confusing DragonFly's namecache as each snapshot will appear
to be an entirely different path.
Unfortunately this does mean that cpdup/cp/tar will pickup a translated
softlink and not the actual one. It's just something else to remember
about these 'weird' PFS mount points.
Suggested-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
Sepherosa Ziehau [Thu, 12 Mar 2009 12:00:55 +0000 (20:00 +0800)]
emx(4): Move EMX_RSS_DEBUG into opt_emx.h
Sepherosa Ziehau [Thu, 12 Mar 2009 11:24:52 +0000 (19:24 +0800)]
jme(4): Move JME_RSS_DEBUG into opt_jme.h
Sascha Wildner [Wed, 11 Mar 2009 19:02:10 +0000 (20:02 +0100)]
identcpu.c: Add MOVBE feature string.
Matthew Dillon [Wed, 11 Mar 2009 17:45:38 +0000 (10:45 -0700)]
Extract PFS null mounts from anywhere in the filesystem, instead of just
from the /pfs subdirectory.
Submitted-by: pgeorgi (via bugs)
Sepherosa Ziehau [Tue, 10 Mar 2009 13:14:22 +0000 (21:14 +0800)]
em(4): Enable RX after RDH is configured (backport from emx(4)).
According to Intel OpenSDM's RDH description:
"... If software were to write to this register while the receive
function was enabled, the on-chip descriptor buffers can be
invalidated and other indeterminate operations might result ..."
Maybe we were just that lucky, the driver actually worked ...
Sepherosa Ziehau [Tue, 10 Mar 2009 13:10:23 +0000 (21:10 +0800)]
em(4): Backport part of
62e0614e4c59f94c83cdf24ebf5995add53f03bf
Used transmit descriptors do not need to be cleared in txeof and
tx_collect
Sepherosa Ziehau [Tue, 10 Mar 2009 12:00:40 +0000 (20:00 +0800)]
em(4): Backport
cc8fffade4528dd788a4789998ed3d1e85f9644a
Use E1000_ICR_ macros when check against value read from ICR
Sepherosa Ziehau [Tue, 10 Mar 2009 11:58:10 +0000 (19:58 +0800)]
em(4): Backport part of
62e0614e4c59f94c83cdf24ebf5995add53f03bf
em_buffer.m_head is only set/cleared in the last em_buffer associated
with the packet, so we don't need to keep clearing it in transmit
descriptor setup loop.
Sepherosa Ziehau [Tue, 10 Mar 2009 10:56:40 +0000 (18:56 +0800)]
network: Turn on adaptive MPSAFE for network threads, and MPSAFE for IP and ARP
Approved-by dillon@
Hasso Tepper [Mon, 9 Mar 2009 13:52:11 +0000 (15:52 +0200)]
Install ncurses_dll.h to unbreak devel/ncursesw from pkgsrc.
Pkgsrc assumes that our ncurses is good enough to use, but installs
ncursesw/ncurses.h header which includes ncurses_dll.h. So, compiling
every piece of software using ncursesw/ncurses.h just failed.
Simon Schubert [Sat, 7 Mar 2009 18:35:10 +0000 (19:35 +0100)]
signals: rename CURSIGNB to CURSIG_NOBLOCK and unify implementation
CURSIG() is defined as __cursig() and CURSIGNB() as __cursignb().
However both __cursig() and __cursignb() do almost the same thing, so
consolidate the code into one function.
Simon Schubert [Sat, 7 Mar 2009 17:09:13 +0000 (18:09 +0100)]
signals: rework when CURSIG/issignal can stop a thread for tracing
When tracing a process, it can happen that the thread would get stopped
due to the signal and its tracing. In this case the tracing parent
would get notified and it might choose to let the process to serve the
signal.
However if this stop+trace is happening somewhere deep in the kernel due
to a call to CURSIG(), it might happen that the same signal again is the
cause for a stop+trace cycle because of another call to CURSIG() while
the call stack is unwinding.
Introduce CURSIG_TRACE(), which explicitly allows stopping for tracing
signal delivery. This is only called from userret().
All other instances of CURSIG() may still block/sleep because of SA_STOP
signals, but these invocations may not trace + repost signals.
As such, the only place where trace + repost of signals can happen now
is userret(). Nevertheless, CURSIG() still decides not to ignore a
currently ignored signal and rather lets the kernel unwind until this
signal arrives in the CURSIG_TRACE() called from userret().
Simon Schubert [Sat, 7 Mar 2009 16:35:02 +0000 (17:35 +0100)]
signals: stop all threads of a process before taking a checkpoint
If we want to checkpoint a multi-threaded process, we need to wait until
the process is in a quiescent state, otherwise our checkpoint might be
incorrect.
Simon Schubert [Sat, 7 Mar 2009 16:54:37 +0000 (17:54 +0100)]
stay in tstop() as long as the process is stopped
Under some conditions (mainly related to multi-threaded processes and
tracing (gdb)), wakeups, scheduling and stops can lead to a race which
will leave the process stopped and wait()ed, but the P_WAITED flag
cleared. This happens because a thread in tstop() might have been woken
up, but not yet scheduled. If the process in turn would get stopped
again (another bug), the thread in tstop() would be counted as stopped,
but would only be waiting to be scheduled to transition into LSRUN.
Prevent this by checking for SSLEEP in tstop() before returning.
Simon Schubert [Sat, 7 Mar 2009 16:37:16 +0000 (17:37 +0100)]
don't loop printing the diagnostic "waiting for lwps to exit"
This diagnostic never was really useful and only clutters the message
buffer. Also don't poll the process state once a second, but rely on
the wakeups from exiting threads.
Sepherosa Ziehau [Mon, 9 Mar 2009 13:13:59 +0000 (21:13 +0800)]
emx(4): Add tunable to enable/disable multi-RX queue support
Multi-RX queue support is disabled on UP kernel, even if this
tunable is enabled.
Sepherosa Ziehau [Sun, 8 Mar 2009 12:59:27 +0000 (20:59 +0800)]
emx(4): Add multi-RX queue support
Sascha Wildner [Mon, 9 Mar 2009 09:07:10 +0000 (10:07 +0100)]
Add an objcache(9) manual page.
Submitted-by: Stathis Kamperis <ekamperi@gmail.com>
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1251>
Sepherosa Ziehau [Sun, 8 Mar 2009 04:08:30 +0000 (12:08 +0800)]
emx(4): Use E1000_ICR_ macros when check against value read from ICR
Sepherosa Ziehau [Sun, 8 Mar 2009 03:51:38 +0000 (11:51 +0800)]
emx(4): this driver does not support 82543
Sepherosa Ziehau [Sat, 7 Mar 2009 13:09:09 +0000 (21:09 +0800)]
emx(4): Let the hardware strip the ethernet CRC on RX path
Sepherosa Ziehau [Sat, 7 Mar 2009 11:29:59 +0000 (19:29 +0800)]
emx(4): Prepare multi-RX queue support -- use extended RX descriptor
Sepherosa Ziehau [Sat, 7 Mar 2009 05:08:34 +0000 (13:08 +0800)]
emx(4): Prepare multi-RX queue support -- use different struct for RX/TX buffer
Extended RX descriptor will be needed for multi-RX queue support. However,
hardware will write information into RX buffer address field of extended
RX descriptors, so we will need to save "device visible" address in RX buffer.
Sepherosa Ziehau [Sat, 7 Mar 2009 04:01:11 +0000 (12:01 +0800)]
emx(4): Prepare multi-RX queue support -- group RX related fields together
Sepherosa Ziehau [Fri, 6 Mar 2009 11:17:45 +0000 (19:17 +0800)]
emx(4): Factor out emx_dma_{alloc,free}()
Sepherosa Ziehau [Thu, 5 Mar 2009 12:52:02 +0000 (20:52 +0800)]
emx(4): Remove emx_dma_{malloc,free}() and related struct.
Sepherosa Ziehau [Thu, 5 Mar 2009 11:49:28 +0000 (19:49 +0800)]
emx(4): Move TX/RX desc ring alloc/free into ring creation/destroy functions
Sepherosa Ziehau [Thu, 5 Mar 2009 11:25:23 +0000 (19:25 +0800)]
emx(4): Regroup RX/TX ring creation code
Simon Schubert [Fri, 6 Mar 2009 14:41:25 +0000 (15:41 +0100)]
undo: don't run diff between last tid and tid_max
Diff works on pairs of tids and not on only the first tid passed in to
the generator. As a result, there is nothing to do for the last pair of
tids: the last tid and max_tid refer to the same version. Avoid running
the generator in this case.
Simon Schubert [Fri, 6 Mar 2009 14:29:40 +0000 (15:29 +0100)]
undo: use MAX_TID for tid_max, or else HAMMER won't find the latest change
We should anyways use HAMMER_MAX_TID. This also works around the bug in
<http://bugs.dragonflybsd.org/issue1307>.
Simon Schubert [Fri, 6 Mar 2009 14:26:46 +0000 (15:26 +0100)]
undo: for one-shot operations, pass in the last tid and not tid_max
This allows diff to print a sensible time of the last modification and
not print the start of the epoch.
Simon Schubert [Wed, 4 Mar 2009 14:00:05 +0000 (15:00 +0100)]
cleanup: typo in comment
Simon Schubert [Tue, 3 Mar 2009 17:23:48 +0000 (18:23 +0100)]
ktrdump: print ktr data using vfprintf()
The code previously assumed that there is a maximum of 10 arguments which
are all integer (sized). Printing of more or differently sized arguments
seemed to work on occasion, or not.
Sepherosa Ziehau [Wed, 4 Mar 2009 13:31:32 +0000 (21:31 +0800)]
emx(4): Compact transmit path a little bit
- emx_buf.m_head is only set/cleared in the last emx_buf associated with
the packet, so we don't need to keep clearing it in transmit descriptor
setup loop.
- Used transmit descriptors do not need to be cleared in txeof and tx_collect
Sepherosa Ziehau [Wed, 4 Mar 2009 12:56:59 +0000 (20:56 +0800)]
em/emx(4): Remove dead code; add comment.
Sepherosa Ziehau [Wed, 4 Mar 2009 12:04:07 +0000 (20:04 +0800)]
em/emx(4): Don't use __predict_false when test m_len against EM_TXCSUM_MINHL
For 100 bytes UDP datagram written by userspace program, m_len is always
less then EM_TXCSUM_MINHL.
Sepherosa Ziehau [Wed, 4 Mar 2009 11:05:16 +0000 (19:05 +0800)]
em/emx(4): Fix comment
Sepherosa Ziehau [Tue, 3 Mar 2009 11:35:49 +0000 (19:35 +0800)]
Get rid of iwicontrol(8)
Sepherosa Ziehau [Tue, 3 Mar 2009 10:58:03 +0000 (18:58 +0800)]
Update iwi(4) manpage according to recent firmware upgrade.
Sepherosa Ziehau [Tue, 3 Mar 2009 10:42:18 +0000 (18:42 +0800)]
iwi(4): Upgrade to use 3.0 firmware and use firmware(9) API to load firmware.
Submitted-by: Johannes Hofmann <johannes.hofmann@gmx.de>
Hasso Tepper [Mon, 2 Mar 2009 13:07:00 +0000 (15:07 +0200)]
Allow the IP_MINTTL socket option to be set to zero.
This disables the feature and is the default actually.
Reported-by: Nick Hilliard <nick at foobar.org>
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1304>
Matthew Dillon [Mon, 2 Mar 2009 02:56:05 +0000 (18:56 -0800)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Mon, 2 Mar 2009 02:49:06 +0000 (18:49 -0800)]
disklabel64 UFS+HAMMER boot support (was previously just HAMMER boot support)
Create a mini-API for boot2 filesystems, split out the filesystem probe &
initialization code, and adjust boot2 to probe multiple filesystems. While
the coding is fairly generic, only the larger boot2 area in a disklabel64
is big enough to hold a multi-filesystem boot2. 32 bit disklabels can still
only boot from UFS.
As part of this work the BTX loader offset for boot2 had to be adjusted.
boot1 used to load boot2 at 0xA000+0x4000 = 0xE000, but this left only 8KB
available before the segment would overflow in boot1's relocation code.
The BOOT2_VORIGIN was adjusted downward from 0x4000 to 0x2000, reducing
the absolute physical load address for boot2 to 0xC000 and allowing us
to load up to a 16K boot2 without overflowing the segment.
Matthew Dillon [Sun, 1 Mar 2009 18:37:35 +0000 (10:37 -0800)]
Fix checkpt's thaw failure error message.
Sepherosa Ziehau [Sun, 1 Mar 2009 10:04:43 +0000 (18:04 +0800)]
em(4): Give preference to emx(4) based on PCI ids.
Sepherosa Ziehau [Sun, 1 Mar 2009 09:14:58 +0000 (17:14 +0800)]
Copy em(4) to emx(4), which supports only 8257{1,2,3} parts.
The reasons to create another driver for 8257{1,2,3}:
- Various old hardware bug workaround code are removed, so the
code is more straightforward, especially on the transmit path.
- Only 8257{1,2,3} support multi RX queues.
- Only 8257{1,2} support multi TX queues (no plan for it yet).
- It could be a sandbox for me to add multi queues support, while
em(4) always works :).
Sepherosa Ziehau [Sun, 1 Mar 2009 09:35:56 +0000 (17:35 +0800)]
em(4): Put back pkt_txclean KTR
Sepherosa Ziehau [Sun, 1 Mar 2009 09:32:24 +0000 (17:32 +0800)]
em(4): We don't need in_detach trick.
Sepherosa Ziehau [Sun, 1 Mar 2009 09:31:37 +0000 (17:31 +0800)]
em(4): Fix comment
Sepherosa Ziehau [Sun, 1 Mar 2009 09:30:19 +0000 (17:30 +0800)]
em(4): We return 0 or errno, but not Intel e1000's error code.
Sepherosa Ziehau [Sun, 1 Mar 2009 09:29:16 +0000 (17:29 +0800)]
em(4): White space
Sascha Wildner [Sun, 1 Mar 2009 07:34:34 +0000 (08:34 +0100)]
Small adjustment to committer(7): cherry-pick takes one argument.
Matthew Dillon [Sun, 1 Mar 2009 07:26:20 +0000 (23:26 -0800)]
Adjust checkpoint restore to use same note order as coredumps.
Recent changes to the coredump code broke the checkpoint restore
code (checkpt -r) due to reordering of the ELF notes. Adjust the
checkpoint restore code to deal with the new ordering.
Reported-by: Samuel J. Greear <sjg@evilcode.net>
Matthew Dillon [Sun, 1 Mar 2009 03:11:55 +0000 (19:11 -0800)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Sat, 28 Feb 2009 23:30:30 +0000 (15:30 -0800)]
HAMMER VFS - Fix improper cred checks for utimes()
Adjust the kern_utimes() code in the kernel to check for write permissions
prior to diving into the VFS. UFS checks for write perms but HAMMER doesn't.
Generally speaking we want (at least for now) the kernel to do as much of
these checks as possible.
Reported-by: YONETANI Tomokazu <qhwt+dfly@les.ath.cx>
Nicolas Thery [Sat, 28 Feb 2009 23:28:08 +0000 (00:28 +0100)]
Mention commit template in committer.7.
Matthew Dillon [Sat, 28 Feb 2009 18:00:13 +0000 (10:00 -0800)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly