ed [Tue, 22 Jul 2008 21:27:22 +0000 (21:27 +0000)]
Move the NFS/RPC code away from lbolt.
The kernel has a special wchan called `lbolt', which is triggered each
second. It doesn't seem to be used a lot and it seems pretty redundant,
because we can specify a timeout value to the *sleep() routines. In an
attempt to eventually remove lbolt, make the NFS/RPC code use a timeout
of `hz' when trying to reconnect.
Only the TTY code (not MPSAFE TTY) and the VFS syncer seem to use lbolt
now.
Reviewed by: attilio, jhb
Approved by: philip (mentor), alfred, dfr
Fixes based on bde's feedback.
1) Unindent and sort variables.
2) Indent struct members.
3) Remove _packed, use guaranteed >128 bytes size and only first 128
bytes from the structure.
4) Reword comment.
In preparation to sync Apple and FreeBSD versions of security audit,
pick up the Apple Computer -> Apple change in their copyright and
license templates.
Increase initially dropped bytes from 512 to 768 (768 is also
suggested in the Ilya Mironov's article). 768 taken from another
research where it treats as default for RC4-drop(768):
http://www.users.zetnet.co.uk/hopwood/crypto/scan/cs.html#RC4-drop
Trying to fix compilation bustage:
- removing 'const' qualifier from an input parameter to conform to the type
required by rw_assert();
- using in_addr->s_addr to retrive 32 bits address value.
- Disallow XFS mounting in write mode. The write support never worked really
and there is no need to maintain it.
- Fix vn_get() in order to let it call vget(9) with a valid locking
request. vget(9) returns the vnode locked in order to prevent recycling,
but in this case internal XFS locks alredy prevent it from happening, so
it is safe to drop the vnode lock before to return by vn_get().
- Add a VNASSERT() in vget(9) in order to catch malformed locking requests.
1) Use __packed attr on rdat structure to make it exact 128 bytes.
2) Use gettimeofday() and getpid() only if reading from /dev/urandom
fails or impossible.
3) Discard N bytes on very first initialization only (i.e. don't
discard on re-stir).
4) Reduce N from 1024 to 512 as really suggested in the
"(Not So) Random Shuffles of RC4" paper:
http://research.microsoft.com/users/mironov/papers/rc4full.pdf
1) Update copyright notice.
2) Eliminate "struct arc4_stream *as" arg since only single arg is
possible.
3) Set rs.j = rs.i after arc4random key schedule to be more like arc4
stream cipher.
Allocate a single CCB at the start of the main loop of the RAID monitoring
kthread of the mpt(4) driver that hangs around for the entire lifetime of
the thread. Previously the driver would allocate a new CCB using M_WAITOK
with a lock held each time it updated its state. While here, use the
CAM API for allocating a CCB rather than raw malloc(9).
After the commit of SVN rev 180236, wilko@ noticed that the approach
doesn't work on the Alpha platform: machine/elf.h doesn't include
sys/elf32.h there.
PR: related to bin/124906
Approved by: bde@
MFC after: 1 week
Add support for actually sending WCCP return packets via GRE.
This MAY be combined by a clever person with the 'key' code recently
added, however a cursary glance suggest that it would be safer to just keep
the patches as it is unlikely that the two modes would be used together
and the separate patch has been extensively tested.
tcsh: fix abort in cleanup_until(), caused by internal xgetpass()
The xgetpass() function pushes a cleanup entry for &osa, but then
attempts to flush the stack until an entry matching &sa. The two
object pointers do not match, so the stack is popped too much,
and then cleanup_until calls abort() when it unexpectedly hits
the bottom of the cleanup stack.
Reported by: Paul B. Mahol, onemda at gmail.com,
Pietro Cerutti, gahr at gahr.ch
PR: bin/124191
Reviewed by: mp, kris
MFC after: 2 days
Rather than simply waiting silently and indefinitely for all
interrupt-driven configuration handlers to complete, print out a
diagnostic message every 60 second indicating which handlers are
still running. Do this at most 5 times per run so as to avoid
scrolling out any useful information from the kernel message
buffer.
The interval of 60 seconds was selected based on a best guess as
to the nature of "long enough" and may want to be tuned higher
or lower depending on real-world tolerances.
With uart(4) default, change /dev/cuad# to /dev/cuau# so that
out-of-the-box FreeBSD is consistent. Add uart[0-7] as a
fingerfriendly shortcut alongside sio[0-7] and com[1-8].
Add DDB "show conifhk" command, which lists hooks currently waiting
for completion in run_interrupt_driven_config_hooks(). This is
helpful when trying to figure out which device drivers have gone
into la-la land during boot-time autoconfiguration.
jeff [Sat, 19 Jul 2008 05:18:29 +0000 (05:18 +0000)]
Fix a race which could result in some timeout buckets being skipped.
- When a tick occurs on a cpu, iterate from cs_softticks until ticks.
The per-cpu tick processing happens asynchronously with the actual
adjustment of the 'ticks' variable. Sometimes the results may
be visible before the local call and sometimes after. Previously this
could cause a one tick window where we didn't evaluate the bucket.
- In softclock fetch curticks before incrementing cc_softticks so we
don't skip insertions which were made for the current time.
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.