jeff [Sat, 19 Jul 2008 05:13:47 +0000 (05:13 +0000)]
- Check whether we've recorded this tick in ts_ticks on another cpu in
sched_tick() to prevent multiple increments for one tick. This pushes
the value out of range and breaks priority calculation.
Reviewed by: kib
Found by: pho/nokia
Sponsored by: Nokia
MFC after: 3 days
Disable SSP on arm for the time being.
The segfaults when using SSP seem to be a gcc bug, a patch is available
in the gcc bugzilla, and will be imported once it's committed
into the official gcc tree.
Correct an error in pmap_change_attr()'s initial loop that verifies that the
given range of addresses are mapped. Previously, the loop was testing the
same address every time.
Enhance arena_chunk_map_t to directly support run coalescing, and use
the chunk map instead of red-black trees where possible. Remove the
red-black trees and node objects that are obsoleted by this change. The
net result is a ~1-2% memory savings, and a substantial allocation speed
improvement.
Fix a longstanding bug, from Otto Moerbeck:
if we're reducing a rule that has an empty
right hand side and the yacc stackpointer is pointing at the very
end of the allocated stack, we end up accessing the stack out of
bounds by the implicit $$ = $1 action
Eliminate use of the global ripsrc which was being used to pass address
information from rip_input() to rip_append(). Instead, pass the source
address for an IP datagram to rip_append() using a stack-allocated
sockaddr_in, similar to udp_input() and udp_append().
Prior to the move to rwlocks for inpcbinfo, this was not a problem, as
use of the global was synchronized using the ripcbinfo mutex, but with
read-locking there is the potential for a race during concurrent
receive.
This problem is not present in the IPv6 raw IP socket code, which
already used a stack variable for the address.
das [Fri, 18 Jul 2008 02:18:34 +0000 (02:18 +0000)]
Sort the .PATH entries to give a more reasonable order of precedence:
1. architecture-specific files
2. long double format-specific files
3. bsdsrc
4. src
5. man
The original order was virtually the opposite of this.
This should not cause any functional changes at this time. The
difference is only significant when one wants to override, say, a
generic foo.c with a more specialized foo.c (as opposed to foo.S).
Fix a benign typo that would give out an incorrect warning message.
Change a get-or-set sequence on OHCI_COMMAND_STATUS register which
is "write to set" to a simple set.
Pair the VOP_OPEN call from do_execve() with the reciprocal VOP_CLOSE.
This was unnoticed because local filesystems usually do nothing
non-trivial in the close vop.
Reported and tested by: Rick Macklem
MFC after: 2 weeks
Clean up mxge's use of callouts as pointed out by jhb,
and handle NIC hardware watchdog resets.
- remove buggy code at the top of mxge_tick() which tried
to detect a race which is already detected in the kernel's
callout code.
- move callout_stop() and callout_reset() into mxge_close()
mxge_open() rather than doing the callout manipulation
all over the place.
- use callout_drain(), rather than callout_stop() to prevent
a potential race between mxge_tick() and mxge_detach()
which could lead to softclock using a destroyed mutex
- restructure the mxge_tick() and mxge_watchdog_reset()
routines to avoid resetting a callout, and then
immediately stopping it if the watchdog reset routine
is called, and fails.
- enable the driver to handle NIC hardware watchdog
resets by restoring the NIC's PCI config space, which is
lost when the NIC hardware watchdog triggers.
Before updating the password database, the pw(8) utility first performs a
sanity check by invoking "pwd_mkdb -C". However, if this failed it
silently returned success. Fix this so it fails the update operation
instead.
Add the shutdown KEYWORD to those scripts that start persistent services
to allow them to do a "clean" shutdown.
I purposely avoided making changes to network-related stuff since the
system shutting down is pretty conclusive, and there may be complicated
dependencies on the network that I would rather not try to unravel.
I also skipped kerberos-related stuff for the reasons above, and
because I have no way to test it.
ed [Wed, 16 Jul 2008 12:36:39 +0000 (12:36 +0000)]
Move the TCSA* definitions out of _KERNEL. They are processed in libc.
The tcsetattr() routine already converts the TCSA* arguments to their
respective TIOCSETA* ioctl's in the C library. There is no need to have
these definitions inside the kernel.
ed [Wed, 16 Jul 2008 11:23:15 +0000 (11:23 +0000)]
Sort the ioctl's in <sys/ttycom.h> by number.
I think one of the reasons why we have so many conflicts in the TTY
ioctl category, is because the ioctl's aren't ordered logically. This
commit only sorts them by number. The comments may still be inaccurate.
ed [Wed, 16 Jul 2008 11:20:04 +0000 (11:20 +0000)]
Remove OTTYDISC, NETLDISC and NTTYDISC definitions.
When I ported most applications away from <sgtty.h>, I noticed none of
them were actually using these definitions. I kept them in place,
because I didn't want to touch tools like pstat(8) and stty(1).
In preparation for the MPSAFE TTY layer, remove these definitions. This
doesn't have any impact with respect to binary compatibility (see
tty_conf.c).
We couldn now add an #error to <sys/ioctl_compat.h> when included
outside the kernel. Unfortunately, kdump's mkioctls includes this file
unconditionally.
Fix a multicast handling regression on VT6105M introduced in
vr(4) overhauling(r177050).
It seems that filtering multicast addresses with multicast CAM
entries require accessing 'CAM enable bit' for each CAM entry.
Subsequent accessing multicast CAM control register without
toggling the 'CAM enable bit' seem to no effects.
In order to fix that separate CAM setup from CAM mask configuration
and CAM entry modification. While I'm here add VLAN CAM filtering
feature which will be enabled in future(FreeBSD now can receive
VLAN id insertion/removal event from vlan(4) on the fly).
For VT6105M hardware, explicitly disable VLAN hardware tag
insertion/stripping and enable VLAN CAM filtering for VLAN id 0.
This shall make non-VLAN frames set VR_RXSTAT_VIDHIT bit in Rx
status word.
Added multicast/VLAN CAM address definition to header file.
Fix VR_RXSTAT_RX_OK bit definition which lasted for more than 9
years. All datasheet I have indicates the bit 15 is the
VR_RXSTAT_RX_OK. The bit 14 is reserved for all Rhine family
except VT6105M. VT6105M uses that bit to indicate a VLAN frame
with matching CAM VLAN id.
Use the VR_RXSTAT_RX_OK instead of VR_RXSTAT_RXERR when vr(4)
checks the validity of received frame.
This should fix occasional dropping frames on VT6105M.
Tested by: Goran Lowkrantz ( goran.lowkrantz at ismobile dot com )
MFC after: 1 week
Merge last of a series of rwlock conversion changes to UDP, which
completes the move to a fully parallel UDP transmit path by using
global read, rather than write, locking of inpcbinfo in further
semi-connected cases:
- Add macros to allow try-locking of inpcb and inpcbinfo.
- Always acquire an incpcb read lock in udp_output(), which stablizes the
local inpcb address and port bindings in order to determine what further
locking is required:
- If the inpcb is currently not bound (at all) and are implicitly
connecting, we require inpcbinfo and inpcb write locks, so drop the
read lock and re-acquire.
- If the inpcb is bound for at least one of the port or address, but an
explicit source or destination is requested, trylock the inpcbinfo
lock, and if that fails, drop the inpcb lock, lock the global lock,
and relock the inpcb lock.
- Otherwise, no further locking is required (common case).
- Update comments.
In practice, this means that the vast majority of consumers of UDP sockets
will not acquire any exclusive locks at the socket or UDP levels of the
network stack. This leads to a marked performance improvement in several
important workloads, including BIND, nsd, and memcached over UDP, as well
as significant improvements in pps microbenchmarks.
The plan is to MFC all of the rwlock changes to RELENG_7 once they have
settled for a weeks in the tree.
Add an event handler to the vlan driver so the NIC driver
becomes aware of it, and gets the VLAN ID. This will allow
the easy use of VLAN hardware filtering by adapters that
support it.
Handle a race between pmap_kextract() and pmap_promote_pde(). This race
caused ZFS to crash when restoring a snapshot with superpage promotion
enabled.
The libc acl_valid(3) function validates the contents of a POSIX.1e ACL.
This change removes the requirement that an ACL contain no ACL_USER
entries with a uid the same as those of a file, or ACL_GROUP entries
with a gid the same as those of a file. This requirement is not in the
specification, and not enforced by the kernel's ACL implementation.
Reported by: Iustin Pop <iusty at k1024 dot org>
MFC after: 1 week
ed [Sun, 13 Jul 2008 07:20:14 +0000 (07:20 +0000)]
Make uart(4) the default serial port driver on i386 and amd64.
The uart(4) driver has the advantage of supporting a wider variety of
hardware on a greater amount of platforms. This driver has already been
the standard on platforms such as ia64, powerpc and sparc64.
I've decided not to change anything on pc98. I'd rather let people from
the pc98 team look at this.
Refine the changes made in SVN rev 180430. Specifically, instantiate a new
page table page only if the 2MB page mapping has been used. Also, refactor
some assertions.
In nmount(), if we see "update" in the mount options,
set MNT_UPDATE in fsflags, and delete the
"update" option from the global mount options.
MNT_UPDATE is a command, and not a property of a mount
that should persist after the command is executed.
We need to do similar things for MNT_FORCE and MNT_RELOAD.
All mount flags are prefixed by MNT_..... it would
be nice if flags which were commands were named differently
from flags which are persistent properties of a mount.
This was not such a big deal in the pre-nmount() days,
but with nmount() it is more important.
Strongly discourage the use of the query-source option, and explain why.
Give a better example if a user absolutely must use this option, and
suggest they pick something from the ephemeral port range rather than
port 53. This means that the example will not work if it is merely
uncommented, but this will hopefully encourage users to read the comment.
Merge from vendor/bind9/dist as of the 9.4.2-P1 import, including
the patch from ISC for lib/bind9/check.c and deletion of unused
files in lib/bind.
This version will by default randomize the UDP query source port
(and sequence number of course) for every query.
In order to take advantage of this randomization users MUST have an
appropriate firewall configuration to allow UDP queries to be sent and
answers to be received on random ports; and users MUST NOT specify a
port number using the query-source[-v6] options.
The avoid-v[46]-udp-ports options exist for users who wish to eliminate
certain port numbers from being chosen by named for this purpose. See
the ARM Chatper 6 for more information.
Also please note, this issue applies only to UDP query ports. A random
ephemeral port is always chosen for TCP queries.
This issue applies primarily to name servers whose main purpose is to
resolve random queries (sometimes referred to as "caching" servers, or
more properly as "resolving" servers), although even an "authoritative"
name server will make some queries, primarily at startup time.
All users of BIND are strongly encouraged to upgrade to the latest
version, and to utilize the source port randomization feature.
This update addresses issues raised in:
http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2008-1447
http://www.kb.cert.org/vuls/id/800113
http://tools.ietf.org/html/draft-ietf-dnsext-forgery-resilience
A number of significant enhancements to the ciss driver:
1. The FreeBSD driver was setting an interrupt coalesce delay of 1000us
for reasons that I can only speculate on. This was hurting everything
from lame sequential I/O "benchmarks" to legitimate filesystem metadata
operations that relied on serialized barrier writes. One of my
filesystem tests went from 35s to complete down to 6s.
2. Implemented the Performant transport method. Without the fix in
(1), I saw almost no difference. With it, my filesystem tests showed
another 5-10% improvement in speed. It was hard to measure CPU
utilization in any meaningful way, so it's not clear if there was a
benefit there, though there should have been since the interrupt handler
was reduced from 2 or more PCI reads down to 1.
3. Implemented MSI-X. Without any docs on this, I was just taking a
guess, and it appears to only work with the Performant method. This
could be a programming or understanding mistake on my part. While this
by itself made almost no difference to performance since the Performant
method already eliminated most of the synchronous reads over the PCI
bus, it did allow the CISS hardware to stop sharing its interrupt with
the USB hardware, which in turn allowed the driver to become decoupled
from the Giant-locked USB driver stack. This increased performance by
almost 20%. The MSI-X setup was done with 4 vectors allocated, but only
1 vector used since the performant method was told to only use 1 of 4
queues. Fiddling with this might make it work with the simpleq method,
not sure. I did not implement MSI since I have no MSI-specific hardware
in my test lab.
4. Improved the locking in the driver, trimmed some data structures.
This didn't improve test times in any measurable way, but it does look
like it gave a minor improvement to CPU usage when many
processes/threads were doing I/O in parallel. Again, this was hard to
accurately test.
Dust off old code for support of USB isochronous transfers.
USB isochronous transfer support is required for Bluetooth SCO.
While i'm here change u_int to uint and update TODO.
This should produce no visible changes unless the device is
broken (or really old).
Use the VM_ALLOC_INTERRUPT for the page requests when allocating memory
for the bio for swapout write. It allows the page allocator to drain
free page list deeper. As result, a deadlock where pageout deamon sleeps
waiting for bio to be allocated for swapout is no more reproducable in
practice.
Alan said that M_USE_RESERVE shall be ressurrected and used there, but
until this is implemented, M_NOWAIT does exactly what is needed.
Make it atomic for the devfs_populate_loop() to see the setting of
SI_ALIAS flag and initialization of the si_parent when alias is created.
Assert that supplied parent device is not NULL.
Both situations could cause NULL dereference in the
devfs_populate_loop() when creating a symlink for SI_ALIAS'ed device.
Namely, cdp->cdp_c.si_parent may be NULL.
id_t is a 64-bit integer and thus is passed as two arguments like off_t is.
As a result, those arguments must be recombined before calling the real
syscal implementation. This change fixes 32-bit compatibility for
cpuset_getid(), cpuset_setid(), cpuset_getaffinity(), and
cpuset_setaffinity().
Use 'CSCOPE_ARCHDIR' to change the default architecture directories to
cscope. After the addition of sys/modules/dtrace/dtrace, setting
'ARCHDIR' in /etc/src.conf breaks the build.
Apply the MAC label to an outgoing UDP packet when other inpcb properties are
processed, meaning that we avoid the cost of MAC label assignment if we're
going to drop the packet due to mbuf exhaustion, etc.
peter [Thu, 10 Jul 2008 02:08:00 +0000 (02:08 +0000)]
Merge gnu cpio 2.6 -> 2.8 changes. Unfortunately, we have massive
conflicts due to radically different approaches to security and bug fixes.
In some cases I re-started from the vendor version and reimplemented our
patches. Fortunately, this is not enabled by default in -current.
Add a new program to the multicast test suite. The mcgrab program
is used to grab and hold some number of multicast addresses in order
to test what happens when an interface goes over the number of multicast
addresses it can filter in hardware.