glebius [Mon, 3 Oct 2005 14:15:26 +0000 (14:15 +0000)]
o Move a lot of parameter checking from netisr_poll() to
dedicated sysctl handlers. Protect manipulations with
poll_mtx. The affected sysctls are:
- kern.polling.burst_max
- kern.polling.each_burst
- kern.polling.user_frac
- kern.polling.reg_frac
o Use CTLFLAG_RD on MIBs that supposed to be read-only.
o u_int32t -> uint32_t
o Remove unneeded locking from poll_switch().
cognet [Mon, 3 Oct 2005 14:13:50 +0000 (14:13 +0000)]
- Provide the kernel l1pt physical address, for userland.
- Use the new API for pmap_copy_page() and pmap_zero_page().
- Just write-back the pages in pmap_qenter(), and invalidate it in
pmap_qremove().
- Nuke the cache flushing in pmap_enter_quick(), it's not needed anymore.
cognet [Mon, 3 Oct 2005 14:09:36 +0000 (14:09 +0000)]
Import a small ELF trampoline, in which the kernel is embedded, and that
is able to load the kernel into memory, symbol table included. This is
needed to be able to access the symbol table from DDB without a boot
loader.
cognet [Mon, 3 Oct 2005 14:07:57 +0000 (14:07 +0000)]
*blush*
Don't try to dereference map if it's NULL.
While I'm there, increase the minimum value to write-back/invalidate the
whole dcache in bus_dmamap_sync().
cperciva [Mon, 3 Oct 2005 12:49:54 +0000 (12:49 +0000)]
If sufficiently bad things happen during a call to kern_execve(), it is
possible for do_execve() to call exit1() rather than returning. As a
result, the sequence "allocate memory; call kern_execve; free memory"
can end up leaking memory.
This commit documents this astonishing behaviour and adds a call to
exec_free_args() before the exit1() call in do_execve(). Since all
the users of kern_execve() in the tree use exec_free_args() to free
the command-line arguments after kern_execve() returns, this should
be safe, and it fixes the memory leak which can otherwise occur.
Submitted by: Peter Holm
MFC after: 3 days
Security: Local denial of service
rwatson [Mon, 3 Oct 2005 11:09:39 +0000 (11:09 +0000)]
Acquire Giant conditionally in in_addmulti() and in_delmulti() based on
whether the interface being accessed is IFF_NEEDSGIANT or not. This
avoids lock order reversals when calling into the interface ioctl
handler, which could potentially lead to deadlock.
The long term solution is to eliminate non-MPSAFE network drivers.
scottl [Mon, 3 Oct 2005 07:05:34 +0000 (07:05 +0000)]
Reintroduce the lmc T1/E1/T3 WAN driver. This version is locked, supports
interface polling, compiles on 64-bit platforms, and compiles on NetBSD,
OpenBSD, BSD/OS, and Linux. Woo! Thanks to David Boggs for providing this
driver.
Altq, sppp, netgraph, and bpf are required for this driver to operate.
Userland tools and man pages will be committed next.
yar [Mon, 3 Oct 2005 02:24:21 +0000 (02:24 +0000)]
Improve handling flags that must be propagated
to the parent interface, such as IFF_PROMISC and
IFF_ALLMULTI. In addition, vlan(4) gains ability
to migrate from one parent to another w/o losing
its own flags.
yar [Mon, 3 Oct 2005 02:14:51 +0000 (02:14 +0000)]
Clean up consistency checks in if_setflag():
. use KASSERT for all checks so that the source of an error can be detected;
. use __func__ instead of spelling function name each time;
. fix a typo.
yar [Mon, 3 Oct 2005 01:47:43 +0000 (01:47 +0000)]
Log a message about entering or leaving permanently promiscuous mode,
as it is done for usual promiscuous mode already. This info is important
because promiscuous mode in the hands of a malicious party can jeopardize
the whole network.
truckman [Sun, 2 Oct 2005 23:27:56 +0000 (23:27 +0000)]
Always wire the sysctl output buffer in sysctl_kern_proc() before
calling sysctl_out_proc(). -- fix from jhb
Move the code in fill_kinfo_thread() that gathers data from struct proc
into the new function fill_kinfo_proc_only().
Change all callers of fill_kinfo_thread() to call both
fill_kinfo_proc_only() and fill_kinfo() thread. When gathering
data from a multi-threaded process, fill_kinfo_proc_only() only needs
to be called once.
Grab sched_lock before accessing the process thread list or calling
fill_kinfo_thread().
cognet [Sun, 2 Oct 2005 22:57:31 +0000 (22:57 +0000)]
- Call db_setup_paging() for traceall.
- Make it so one can't call db_setup_paging() if it has already been called
before. traceall needs this, or else the db_setup_paging() call from
db_trace_thread() will reset the printed line number, and override its
argument.
This is not perfect for traceall, because even if one presses 'q' while in
the middle of printing a backtrace it will finish printing the backtrace
before exiting, as db_trace_thread() won't be notified it should stop, but
it is hard to do better without reworking the pager interface a lot more.
hrs [Sun, 2 Oct 2005 21:02:37 +0000 (21:02 +0000)]
New release notes:
gre(4) IPv6 over GRE support,
ipfw IPv6 support,
new sysctl net.inet6.ip6.stealth added,
g_label now supports Ext2FS and ReiserFS,
bsdiff(1) and bspatch(1) added,
ping(8) "sweeping ping" support,
wcsdup() function added,
rc.d/gbde_swap renamed to rc.d/encswap,
rc.d/geli and rc.d/geli2 added, and
portsnap(8) imported.
yar [Sun, 2 Oct 2005 19:12:42 +0000 (19:12 +0000)]
Record dependency on the newly introduced pfsync.
Start before routing for better system protection.
(pf used to start late during system boot, after
many a network daemon have started already, which
sucked from security POV.)
Remark: For maximum security, pf should start before
netif, but it would create a dependency loop because
pfsync has to start after netif, yet before pf.
yar [Sun, 2 Oct 2005 18:59:02 +0000 (18:59 +0000)]
Add an rc.d script to start pfsync at the right moment of the
system boot, and hook it up in the system.
The separate script is needed because in the presence of various
interface lists in rc.conf ($network_interfaces, $cloned_interfaces,
$sppp_interfaces, $gif_interfaces, more to come) it is hard to start
them orderly, so that pfsync is brought up after its syncdev, which
is required for the proper startup of pfsync.
netchild [Sun, 2 Oct 2005 15:56:36 +0000 (15:56 +0000)]
* Fixed rate operation for es1370 chip to solve conflicting
sampling rate between playback and recording. This can be
disabled / enabled via kernel hints
(hint.pcm.<unit>.fixed_rate=0/4000-48000) or sysctl
hw.snd.pcm<unit>.fixed_rate=0/4000-48000). Default to 48khz
fixed rate. [1]
* Basic cleanup. *_es1371x_* -> *_es137x_*.
* Some locking fixes. [2]
Submitted by: Ariff Abdullah <skywizard@MyBSD.org.my>
Discussed with: yongari [2]
See also: http://lists.freebsd.org/pipermail/freebsd-multimedia/2005-September/002758.html [1]
Reported by: Jos Backus <jos at catnook.com> [1]
yar [Sun, 2 Oct 2005 15:54:26 +0000 (15:54 +0000)]
Use rc.subr(8) appropriately:
- utilize default methods instead of rolling local ones;
- avoid to specify BEFORE conditions we don't really need
(pflog will be REQUIRE'd by pf);
- omit extra decoration from warning messages, warn() will
decorate them sufficiently.
netchild [Sun, 2 Oct 2005 15:43:57 +0000 (15:43 +0000)]
sys/dev/sound/pcm/sndstat.c:
* General spl* cleanup. It doesn't serve any purpose anymore.
* Nuke sndstat_busy(). Addition of sndstat_acquire() /
sndstat_release() for sndstat exclusive access. [1]
sys/dev/sound/pcm/sound.c:
* Remove duplicate SLIST_INIT()
* Use sndstat_acquire() / release() to lock / release the entire
sndstat during pcm_unregister(). This should fix LOR #159 [1]
sys/dev/sound/pcm/sound.h:
* Definition of SD_F_SOFTVOL (part of feeder volume)
* Nuke sndstat_busy(). Addition of sndstat_acquire() /
sndstat_release() for exclusive sndstat access. [1]
rwatson [Sun, 2 Oct 2005 11:41:12 +0000 (11:41 +0000)]
Add a DDB "traceall" function, which stack traces all known process
threads. This is quite useful if generating a debug log for post-mortem
by another developer, in which case the person at the console may not
know which threads are of interest. The output of this can be quite
long.
maxim [Sun, 2 Oct 2005 08:43:57 +0000 (08:43 +0000)]
o Teach sysctl_drop() how to deal with the sockets in TIME_WAIT state.
This is a special case because tcp_twstart() destroys a tcp control
block via tcp_discardcb() so we cannot call tcp_drop(struct *tcpcb) on
such connections. Use tcp_twclose() instead.
avatar [Sun, 2 Oct 2005 04:29:08 +0000 (04:29 +0000)]
Fixing WEP bustage in hostap mode since 5.2-RELEASE.
- WEP TX fix:
The original code called software crypto, ieee80211_crypto_encap(),
which never worked since IEEE80211_KEY_SWCRYPT was never flagged due to
ieee80211_crypto_newkey() assumes that wi always supports hardware based
crypto regardless of operational mode(by virtue of IEEE80211_C_WEP).
This fix works around that issue by adding wi_key_alloc() to force
the use of s/w crypto. Also if anyone ever decides to cleanup ioctl
handling where key changes wouldn't cause a call to wi_init() every time,
we'll need wi_key_alloc() to DTRT.
In addition to that, this fix also adds code to wi_write_wep() to force
existing keys to be switched between h/w and s/w crypto such that an
operation mode change(sta <-> hostap) will flag IEEE80211_KEY_SWCRYPT
properly.
- WEP RX fix:
Clear IEEE80211_F_DROPUNENC even in hostap mode. Quote from Sam:
"This is really gross but I don't see an easy way around it.
By doing it we lose the ability to independently drop unencode
frames (and support mixed wep/!wep use). We should really be
setting the EXCLUDE_UNENCRYPTED flag written in wi_write_wep
based on IEEE80211_F_DROPUNENC but with our clearing it we can't
depend on it being set properly."
Reported by: Holm Tiffe <holm at freibergnet dot de>
Submitted by: sam
MFC after: 3 days
yar [Sat, 1 Oct 2005 20:58:03 +0000 (20:58 +0000)]
Fix the usage of rc_usage. The rc_usage function takes
a list of possible keywords, not all them in a single argument.
This also fixes the issue of extra delimiter characters appearing
on the help line from rc.d scripts not setting $extra_commands.
glebius [Sat, 1 Oct 2005 19:04:22 +0000 (19:04 +0000)]
- Catch up with the fact that HZ is 1000 now by default.
- Remove description of poll in trap feature.
- Tell that polling should be turned on and off with ifconfig.
- Move description of kern.polling.enable to the end and say
that this a deprecated way of turning polling on.
- Remove note that idle poll has some problems in CURRENT. I failed
to find them, while Sam and Luigi failed to remember what the
problem actually were there.
glebius [Sat, 1 Oct 2005 18:56:19 +0000 (18:56 +0000)]
Big polling(4) cleanup.
o Axe poll in trap.
o Axe IFF_POLLING flag from if_flags.
o Rework revision 1.21 (Giant removal), in such a way that
poll_mtx is not dropped during call to polling handler.
This fixes problem with idle polling.
o Make registration and deregistration from polling in a
functional way, insted of next tick/interrupt.
o Obsolete kern.polling.enable. Polling is turned on/off
with ifconfig.
Detailed kern_poll.c changes:
- Remove polling handler flags, introduced in 1.21. The are not
needed now.
- Forget and do not check if_flags, if_capenable and if_drv_flags.
- Call all registered polling handlers unconditionally.
- Do not drop poll_mtx, when entering polling handlers.
- In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx.
- In netisr_poll() axe the block, where polling code asks drivers
to unregister.
- In netisr_poll() and ether_poll() do polling always, if any
handlers are present.
- In ether_poll_[de]register() remove a lot of error hiding code. Assert
that arguments are correct, instead.
- In ether_poll_[de]register() use standard return values in case of
error or success.
- Introduce poll_switch() that is a sysctl handler for kern.polling.enable.
poll_switch() goes through interface list and enabled/disables polling.
A message that kern.polling.enable is deprecated is printed.
Detailed driver changes:
- On attach driver announces IFCAP_POLLING in if_capabilities, but
not in if_capenable.
- On detach driver calls ether_poll_deregister() if polling is enabled.
- In polling handler driver obtains its lock and checks IFF_DRV_RUNNING
flag. If there is no, then unlocks and returns.
- In ioctl handler driver checks for IFCAP_POLLING flag requested to
be set or cleared. Driver first calls ether_poll_[de]register(), then
obtains driver lock and [dis/en]ables interrupts.
- In interrupt handler driver checks IFCAP_POLLING flag in if_capenable.
If present, then returns.This is important to protect from spurious
interrupts.
tjr [Sat, 1 Oct 2005 06:09:55 +0000 (06:09 +0000)]
Merge from old copyout.c rev 1.2: Output a zero rdev except for bdevs,
cdevs, fifos and sockets. Don't output a file if the major, minor or
totality of its rdev would be truncated.
tjr [Sat, 1 Oct 2005 05:54:39 +0000 (05:54 +0000)]
Merge from old copypass.c rev 1.3: If we must yell at the user than a
file cannot be linked into place when requested (not required) to do it,
reassure them that cpio is still intelligent enough that it will perform
a full copy instead.
The NWFS code in RELENG_6 is broken due to a typo in
sys/fs/nwfs/nwfs_vfsop= s.c, introduced with the conversion to
nmount with revision 1.38. This causes mount_nwfs to fail with
the error message:
mount_nwfs: mount error: /mnt/netware: syserr = No such file or directo=
ry
This is caused by a typo on line 178, which specifies "nwfw_args"
rather than "nwfs_args".
Move some devstat collection to below where large IO operations are chopped
up. This make iostat report operations passed down to the device driver
instead of operations passed down to GEOM disk. The transfer size limit
imposed by the device driver is no longer hidden, improving the correlation
between iostat output and device driver workload.
Add a more generic version of the mii_phy_match routine (mii_phy_match_gen)
which can be used for phy that want to piggy back other data with their
table.
MFi386: revision 1.33.
> Cause all flags passed by boot2 to set the respective loader(8)
> boot_* variable. The end effect is that all flags from boot2
> are now passed to the kernel.
Un-staticize runningbufwakeup() and staticize updateproc.
Add a new private thread flag to indicate that the thread should
not sleep if runningbufspace is too large.
Set this flag on the bufdaemon and syncer threads so that they skip
the waitrunningbufspace() call in bufwrite() rather than than
checking the proc pointer vs. the known proc pointers for these two
threads. A way of preventing these threads from being starved for
I/O but still placing limits on their outstanding I/O would be
desirable.
Set this flag in ffs_copyonwrite() to prevent bufwrite() calls from
blocking on the runningbufspace check while holding snaplk. This
prevents snaplk from being held for an arbitrarily long period of
time if runningbufspace is high and greatly reduces the contention
for snaplk. The disadvantage is that ffs_copyonwrite() can start
a large amount of I/O if there are a large number of snapshots,
which could cause a deadlock in other parts of the code.
Call runningbufwakeup() in ffs_copyonwrite() to decrement runningbufspace
before attempting to grab snaplk so that I/O requests waiting on
snaplk are not counted in runningbufspace as being in-progress.
Increment runningbufspace again before actually launching the
original I/O request.
Prior to the above two changes, the system could deadlock if enough
I/O requests were blocked by snaplk to prevent runningbufspace from
falling below lorunningspace and one of the bawrite() calls in
ffs_copyonwrite() blocked in waitrunningbufspace() while holding
snaplk.
For the sake of consistency and easier typing,
introduce "-tunnel" as an alias for "deletetunnel".
The latter is overly long and prone to typos, but
keep it for POLA since it costs nothing.
After a rmdir()ed directory has been truncated, force an update of
the directory's inode after queuing the dirrem that will decrement
the parent directory's link count. This will force the update of
the parent directory's actual link to actually be scheduled. Without
this change the parent directory's actual link count would not be
updated until ufs_inactive() cleared the inode of the newly removed
directory, which might be deferred indefinitely. ufs_inactive()
will not be called as long as any process holds a reference to the
removed directory, and ufs_inactive() will not clear the inode if
the link count is non-zero, which could be the result of an earlier
system crash.
If a background fsck is run before the update of the parent directory's
actual link count has been performed, or at least scheduled by
putting the dirrem on the leaf directory's inodedep id_bufwait list,
fsck will corrupt the file system by decrementing the parent
directory's effective link count, which was previously correct
because it already took the removal of the leaf directory into
account, and setting the actual link count to the same value as the
effective link count after the dangling, removed, leaf directory
has been removed. This happens because fsck acts based on the
actual link count, which will be too high when fsck creates the
file system snapshot that it references.
This change has the fortunate side effect of more quickly cleaning
up the large number dirrem structures that linger for an extended
time after the removal of a large directory tree. It also fixes a
potential problem with the shutdown of the syncer thread timing out
if the system is rebooted immediately after removing a large directory
tree.