jca [Thu, 15 May 2014 05:03:24 +0000 (05:03 +0000)]
Deal with - actually, ignore - route info messages on the listening side.
Issue reported by consus at gmx dot com on bugs@.
ok bluhm@
guenther [Thu, 15 May 2014 04:43:25 +0000 (04:43 +0000)]
knote_processexit() needs the thread to pass down to FRELE(), so pass it
the exiting thread instead of assuming that that's ps_mainproc.
Also, panic no matter which thread of init takes it down.
ok tedu@
guenther [Thu, 15 May 2014 04:36:33 +0000 (04:36 +0000)]
Prefer prsignal() to send process signals
guenther [Thu, 15 May 2014 03:52:25 +0000 (03:52 +0000)]
Move from struct proc to process the reference-count-holding pointers
to the process's vmspace and filedescs. struct proc continues to
keep copies of the pointers, copying them on fork, clearing them
on exit, and (for vmspace) refreshing on exec.
Also, make uvm_swapout_threads() thread aware, eliminating p_swtime
in kernel.
particular testing by ajacoutot@ and sebastia@
tedu [Wed, 14 May 2014 21:54:20 +0000 (21:54 +0000)]
stop flushing streams in abort(). it's hackish and unsafe, and no longer
required. try to document this fact and some of the history.
with feedback from deraadt guenther millert
henning [Wed, 14 May 2014 21:48:50 +0000 (21:48 +0000)]
improve style - basically, what my fingers did at Marrakesh airport when
waiting for the flight home from m2k14. ok reyk
todd [Wed, 14 May 2014 21:33:50 +0000 (21:33 +0000)]
sync
claudio [Wed, 14 May 2014 19:10:21 +0000 (19:10 +0000)]
Adjust route outputs to the new lo(4) MTU which is now the same on all archs.
Makes this regress work again.
claudio [Wed, 14 May 2014 19:08:33 +0000 (19:08 +0000)]
There is no reason why one needs to have a lo(4) MTU that is arch specific.
Also making the size 32k + some extra is not very helpful since we don't
do something like zerocopy or sendfile. Just go back to 32768 and we can
bikeshed over the right value later on once we have features that benefit
from it. OK mpi@
shadchin [Wed, 14 May 2014 18:28:22 +0000 (18:28 +0000)]
Remove second arg in SetMouseSpeed(), it is always 1200.
ok miod@
shadchin [Wed, 14 May 2014 18:23:22 +0000 (18:23 +0000)]
Remove mouse.baudrate, it is always 1200.
ok miod@
shadchin [Wed, 14 May 2014 18:18:01 +0000 (18:18 +0000)]
Remove unused code.
ok miod@
todd [Wed, 14 May 2014 18:15:26 +0000 (18:15 +0000)]
sync
shadchin [Wed, 14 May 2014 18:14:09 +0000 (18:14 +0000)]
Remove unused variables.
ok miod@
shadchin [Wed, 14 May 2014 18:11:24 +0000 (18:11 +0000)]
wsmoused() called if cmd == WSDISPLAYIO_WSMOUSED only,
don't need check cmd twice.
ok miod@
tedu [Wed, 14 May 2014 16:02:34 +0000 (16:02 +0000)]
let cpp do some of the work
tedu [Wed, 14 May 2014 15:56:41 +0000 (15:56 +0000)]
spell out base64 in error messages
tedu [Wed, 14 May 2014 15:55:11 +0000 (15:55 +0000)]
save some memory by reducing buffer sizes
tedu [Wed, 14 May 2014 15:52:24 +0000 (15:52 +0000)]
recode base64 hashes if necessary
tedu [Wed, 14 May 2014 15:33:41 +0000 (15:33 +0000)]
realloc like a boss, so checksum verification doesn't take minutes.
jsing [Wed, 14 May 2014 14:46:35 +0000 (14:46 +0000)]
Add poly1305 to libcrypto utilising Andrew Moon's public domain
implementation.
ok miod@
mikeb [Wed, 14 May 2014 09:42:22 +0000 (09:42 +0000)]
Sync ktable code with bgpd to fetch, store and perform lookups in
multiple routing tables. Currently it doesn't do anything useful
but it's a prerequisite for any future work in this direction.
ok sthen, reyk, blambert
nicm [Wed, 14 May 2014 06:45:35 +0000 (06:45 +0000)]
Change key-table to mode-table to allow for some future work. From Keith
Amling.
nicm [Wed, 14 May 2014 06:39:58 +0000 (06:39 +0000)]
Simplify copy lines, from Keith Amling.
nicm [Wed, 14 May 2014 06:21:19 +0000 (06:21 +0000)]
Now that cmdlists are reference counted, there is no need for two-step
deletion via the dead_key_bindings tree. From Keith Amling.
nicm [Tue, 13 May 2014 22:54:18 +0000 (22:54 +0000)]
Don't allow multiple buffers with the same name, from Thomas Adam.
jmc [Tue, 13 May 2014 22:07:03 +0000 (22:07 +0000)]
sync some of the text of ENVIRONMENT with more(1), and cut some
verbiage;
ok millert
krw [Tue, 13 May 2014 22:06:06 +0000 (22:06 +0000)]
We don't need no stinkin' 'extern long dev_bsize;'. At least not since
fsck_ffs stopped providing it.
jmc [Tue, 13 May 2014 20:51:00 +0000 (20:51 +0000)]
for non-valid combinations with `#', instead of trying to list them all,
simply say all other combinations are behaviour undefined;
ok guenther, who reminded me to change these too;
jmc [Tue, 13 May 2014 20:37:48 +0000 (20:37 +0000)]
- document two escape sequences as extensions in STANDARDS
- document undefined behaviour for unlisted formats in
combination with `#'; guenther helped me organise this into a saner
piece of text
ok guenther
claudio [Tue, 13 May 2014 14:33:25 +0000 (14:33 +0000)]
While Rev 1.285 fixed a RB tree corruption it caused a TAILQ corruption
in the case where the rdomain was not switched. Make sure ifa_add() is
only called if ifa_del() was called previously. Hopefully we got all the
corruption fixed.
With and OK mpi@
markus [Tue, 13 May 2014 14:24:35 +0000 (14:24 +0000)]
pass SA initiator not the exchange initator to sa_address(); ok mikeb@
jmc [Tue, 13 May 2014 14:05:02 +0000 (14:05 +0000)]
various macro fixes;
jmc [Tue, 13 May 2014 14:01:41 +0000 (14:01 +0000)]
zap trailing whitespace;
jmc [Tue, 13 May 2014 14:00:59 +0000 (14:00 +0000)]
ohash stuff in libutil now;
krw [Tue, 13 May 2014 12:51:40 +0000 (12:51 +0000)]
Nuke another pointless dev_bsize variable always set to 512 and use
DEV_BSIZE. David Vasek says it still works the same on 4K-byte-sector
device. i.e. prints one value incorrectly. For which a fix should
be following soon.
eric [Tue, 13 May 2014 11:57:35 +0000 (11:57 +0000)]
skip incomplete entries in /etc/hosts and /etc/networks
fix null deref spotted by Seth Hanford, pinpointed by dtucker@
ok florian@
mpi [Tue, 13 May 2014 09:30:27 +0000 (09:30 +0000)]
Remove some m->m_pkthdr.rcvif dereference to help for upcoming
receiving pointer -> index conversion. No functional change.
ok claudio@, mikeb@, lteo@
nicm [Tue, 13 May 2014 08:08:32 +0000 (08:08 +0000)]
If multiple arguments are given to new-session, new-window,
split-window, respawn-window or respawn-pane, pass them directly to
execvp() to help avoid quoting problems. One argument still goes to "sh
-c" like before. Requested by many over the years. Patch from J Raynor.
nicm [Tue, 13 May 2014 07:54:20 +0000 (07:54 +0000)]
Add a copy mode key binding to copy to a named buffer. From J Raynor.
nicm [Tue, 13 May 2014 07:34:35 +0000 (07:34 +0000)]
Add support for named buffers. If you don't name a buffer, things work
much as before - buffers are automatically named "buffer0000",
"buffer0001" and so on and ordered as a stack. Buffers can be named
explicitly when creating ("loadb -b foo" etc) or renamed ("setb -b
buffer0000 -n foo"). If buffers are named explicitly, they are not
deleted when buffer-limit is reached. Diff from J Raynor.
guenther [Tue, 13 May 2014 05:50:24 +0000 (05:50 +0000)]
Since this is limited to ffs, use ufsinfo_t instead of ino_t.
Grow the array of cached inodes exponentially instead of arithmetically.
Prefer sizeof(*pointer) over sizeof(type) in mallocs.
Don't leak memory in searchdir().
Fix multiple bugs in the handling of indirect blocks, including reuse of
a static buffer in a recursive function and failure to track the remaining
size of the directory blocks to process
Tested with a directory *doubly* indirect block!
ok krw@
millert [Tue, 13 May 2014 02:34:58 +0000 (02:34 +0000)]
All callers of helpfile() are #ifdef SMALL so only define that
function when SMALL is not defined. OK guenther@
andre [Mon, 12 May 2014 21:30:42 +0000 (21:30 +0000)]
Refactors "dummyrun" flag as "dryrun" with an approach similar to the one used
for client/server processes checks through $self->{down} state control
variable, while here, do not instantiate a client when noclient is defined,
reflecting server object instantiation behaviour.
ok bluhm
krw [Mon, 12 May 2014 21:10:35 +0000 (21:10 +0000)]
Remove useless dev_bsize variable and use DEV_BSIZE as needed. Divide
sblock_try[] entries by DEV_BSIZE to get proper 512-byte-block address
for bread().
Confirmed to still work on 4k-byte sector devices by David Vasek.
miod [Mon, 12 May 2014 20:50:46 +0000 (20:50 +0000)]
Fix format strings involving time_t arguments, fixes `ntpd -d' on sparc (and
probably more 32-bit platforms).
Problem noticed by tobiasu@; ok tobiasu@ dtucker@ sthen@ benno@
kettenis [Mon, 12 May 2014 19:29:16 +0000 (19:29 +0000)]
Move GTT management for Sandy Bridge and up into inteldrm(4). This makes
it possible to use the non-mappable part of the GTT, prepares the way for
using the PPGTT and reduces the diffs with Linux.
ok jsg@
espie [Mon, 12 May 2014 19:27:24 +0000 (19:27 +0000)]
these files have moved to libutil
espie [Mon, 12 May 2014 19:25:16 +0000 (19:25 +0000)]
...better if I actually unplug it, Makefile doesn't do it
miod [Mon, 12 May 2014 19:19:55 +0000 (19:19 +0000)]
Remove AES_bi_ige_encrypt() from libcrypto. This routine is supposed to use
two keys and four IVs to do much magic, is specified as such with test vectors,
but the implementation actually always uses the first key, and the test vectors
were computed with it, so they are wrong.
Fixing the code to match the intended specification would break
interoperability with existing code (assuming such code would exist), so it is
better to remove this interface, which is obviously too complex for mere
mortals if even its author can not implement it correctly.
Riding on the libcrypto major bump.
todd [Mon, 12 May 2014 19:18:23 +0000 (19:18 +0000)]
sync
miod [Mon, 12 May 2014 19:16:35 +0000 (19:16 +0000)]
Bring back ASN1_seq_pack(), ASN1_seq_unpack(), ASN1_pack_string() and
ASN1_unpack_string() from the NO_ASN1_OLD purge; turns out that some Ruby
extensions still use them, unfortunately, so we need to give them some more
time to catch up on ``recent'' interfaces.
Riding on the libcrypto major bump.
miod [Mon, 12 May 2014 19:14:14 +0000 (19:14 +0000)]
Move the `pqueue' part of libcrypto, which is a glorified sorted linked list
of 64-bit data, and only used by DTLS, to libssl where it belongs.
Remove pqueue_print() which is a debugging interface and serves no useful
purpose, except for the regress test, which grows its own pqueue_print()
routine.
Bump libcrypto major and libssl minor.
WARNING: do not update your tree right now, more changes are coming, which
will ride the libcrypto major bump.
tedu [Mon, 12 May 2014 19:13:51 +0000 (19:13 +0000)]
no md5crypt
tedu [Mon, 12 May 2014 19:13:14 +0000 (19:13 +0000)]
remove md5crypt.
while changing things, add a crypt_checkpass wrapper that handles most of
the edge cases. (not quite ready for production, though.)
ok deraadt
espie [Mon, 12 May 2014 19:11:19 +0000 (19:11 +0000)]
adjust to ohash being in libutil now, and to the interface changes.
fix potential integer overflows in memory allocation (mostly for pedagogical
purposes, these are unlikely to overflow in practice)
move the rest of lst.lib stuff into its own directory.
espie [Mon, 12 May 2014 19:09:00 +0000 (19:09 +0000)]
move the ohash functions into libutil by popular demand.
It's not a standard interface, so it doesn't belong in libc.
I hate duplicating the code in client programs, so do beck@, kettenis@,
schwarze@, millert@, miod@... and they agree with libutil.
espie [Mon, 12 May 2014 19:07:37 +0000 (19:07 +0000)]
tweak interface to do the right thing, after useful comments from millert@
(mostly use a calloc function to avoid potential integer overflow)
tedu [Mon, 12 May 2014 19:02:20 +0000 (19:02 +0000)]
change to having four freelists per size, to reduce another source of
deterministic behavior. four selected because it's more than three, less
than five. i.e., no particular reason.
krw [Mon, 12 May 2014 18:50:02 +0000 (18:50 +0000)]
Don't exit on encountering RTM_NEWADDR or RTM_DELADDR routing messages
that lack the appropriate address info.
Simply ignore the message as is done for short or incorrectly
versioned messages
ok claudio@
mpi [Mon, 12 May 2014 17:03:28 +0000 (17:03 +0000)]
Keep libusbhid's HID parser in sync with the kernel one. This is mostly
a backport from FreeBSD's r205728, r224511 and 225839 without the new
functions to set/get a report ID nor the usbhidctl(1) & usbhidaction(1)
bits.
Committing now to open the "Monday's Bump Parade".
eric [Mon, 12 May 2014 14:37:32 +0000 (14:37 +0000)]
fix a possible double free when tls is required but not advertised by
the server.
ok gilles@
kettenis [Mon, 12 May 2014 14:35:56 +0000 (14:35 +0000)]
Fix format strings in debug code.
andre [Mon, 12 May 2014 14:28:22 +0000 (14:28 +0000)]
Fix a leak from a recent added mark/marked keywords check.
ok reyk
krw [Mon, 12 May 2014 13:12:41 +0000 (13:12 +0000)]
Since all static leases are in one file, /etc/dhclient.conf, there is no
'wrong interface' specification. So display that error message only for
non-static leases. i.e. when reading /var/db/dhclient.leases.<if>.
krw [Mon, 12 May 2014 12:16:53 +0000 (12:16 +0000)]
No need for a dev_bsize variable that is always set to 512,
a.k.a. DEV_BSIZE. So just use DEV_BSIZE.
Confirmed to still work on 4k-byte sector disks by David Vasek.
mpi [Mon, 12 May 2014 09:50:44 +0000 (09:50 +0000)]
Backport FreeBSD's HID parser in order to properly parse `moderm'
input device descriptors, notably if they include an Item with an
Usage array (Min-Max range) and others single Usage elements.
Tested by many, thanks! ok andre@
mpi [Mon, 12 May 2014 09:15:00 +0000 (09:15 +0000)]
Includes a router altert option (RAO) in IGMP packets. Without this
option, required by the RFC2236, some L3 switches do not examine the
packets.
Based on FreeBSD's r14622 via Florian Riehm on tech@. ok bluhm@, jca@
beck [Mon, 12 May 2014 08:47:37 +0000 (08:47 +0000)]
Make ifconfig do something intelligent based on the required length of
WEP keys rather then being silently dumb, so when using WEP:
1) If the key is a plausible size try to use it.
2) If they key would be a plausible size with '0x' in front of it, add that.
3) If the key is not a plausible size, emit a warning and do not try to use it.
ok sthen@
halex [Sun, 11 May 2014 21:25:07 +0000 (21:25 +0000)]
replace realloc(p, N * M) with reallocarray(p, N, M) and remove some
pointless cleanup if we're obviously going to die anyway
ok guenther@
guenther [Sun, 11 May 2014 21:14:03 +0000 (21:14 +0000)]
struct direct's d_ino is no longer the same size as an ino_t, so copy it
for the call to bsearch()
problem report and ok krw@
mpi [Sun, 11 May 2014 16:33:21 +0000 (16:33 +0000)]
The relation between uhci(4), ohci(4) and their root hub device is
delicate. HC drivers do not always accept to be left alone.
I don't know if it is a love crisis, but apparently receiving a root
hub status change interrupt before having an uhub(4) attached breaks
resume.
So make sure the root hub is re-attached before interrupt get enabled.
It is safe to do it during DVACT_RESUME since attaching root hubs do
not require any USB transfer.
Based on a diff from yasuoka@, fix a regression introduced in last
commit an reported by Abel Abraham Camarillo Ojeda on tech@ and Nils
R on bugs@.
krw [Sun, 11 May 2014 14:00:52 +0000 (14:00 +0000)]
Tweak DHCPREQUEST log message to omit 'port N', since the port is not
set at that point, and is always set to 67 in send_packet().
deraadt [Sun, 11 May 2014 12:59:31 +0000 (12:59 +0000)]
need syscall.h..
krw [Sun, 11 May 2014 12:40:37 +0000 (12:40 +0000)]
Rework/restore recorded lease handling. Actually use 'lease'
statements in dhclient.conf.
Recorded and static leases should now be bound to interfaces when
appropriate.
.... Ken
jsing [Sun, 11 May 2014 12:20:20 +0000 (12:20 +0000)]
More KNF.
jsing [Sun, 11 May 2014 12:05:27 +0000 (12:05 +0000)]
KNF.
jmc [Sun, 11 May 2014 11:00:55 +0000 (11:00 +0000)]
zap double full stop; ok schwarze
jasper [Sun, 11 May 2014 09:23:49 +0000 (09:23 +0000)]
hide two printfs behind an ifdef DEBUG; spotted by miod@
rpe [Sun, 11 May 2014 08:23:46 +0000 (08:23 +0000)]
Fix configuration of static IPv6 default gateway
- Append IPv4 default gw to /tmp/mygate instead of overwriting it.
- Delete /tmp/mygate at the beginning of configure_ifs() to reset
previous default gw config on installer restarts.
pointed out by todd@
OK halex@ krw@
deraadt [Sun, 11 May 2014 01:40:02 +0000 (01:40 +0000)]
sync
guenther [Sun, 11 May 2014 00:12:43 +0000 (00:12 +0000)]
Move the increment of uvmexp.softs back to the caller of mi_ast():
it needs to be done atomicly on some MP archs and we don't have
atomic_add_int() everywhere yet. Also, mi_ast() was meant to be inline.
noted by miod@
guenther [Sun, 11 May 2014 00:05:38 +0000 (00:05 +0000)]
Pull in <sys/user.h> before <sys/syscall*.h>
chl [Sun, 11 May 2014 00:03:14 +0000 (00:03 +0000)]
remove unused variable
ok tedu@
jmc [Sat, 10 May 2014 23:31:40 +0000 (23:31 +0000)]
catch up with the inet(3) rename; this commit missed in previous
ok guenther
jasper [Sat, 10 May 2014 22:37:12 +0000 (22:37 +0000)]
spello in comment and tweak DUMP_BOOT_DESC a tad
jasper [Sat, 10 May 2014 22:25:16 +0000 (22:25 +0000)]
various format string fixes and remove -Wno-format from octeon
feedback/ok miod@
deraadt [Sat, 10 May 2014 21:58:56 +0000 (21:58 +0000)]
sigh, another case of user.h before syscall_mi.h
chl [Sat, 10 May 2014 21:50:40 +0000 (21:50 +0000)]
fix typo in function prototype
ok gilles@
reyk [Sat, 10 May 2014 21:34:07 +0000 (21:34 +0000)]
Fix SSL breakage that I accidentally introduced with my previous commit.
chl [Sat, 10 May 2014 21:21:09 +0000 (21:21 +0000)]
remove warning by adding explicit parentheses around && within ||
ok sperreault@
chl [Sat, 10 May 2014 20:09:38 +0000 (20:09 +0000)]
remove unused variable
ok shadchin@
kettenis [Sat, 10 May 2014 19:07:43 +0000 (19:07 +0000)]
Remove -Wno-format.
chris [Sat, 10 May 2014 19:07:25 +0000 (19:07 +0000)]
Allow -p prefix to override user home directory for chroot() (Like
openbsd-apache ServerRoot behavior)
Work continues to unify nginx behavior (Principle of Least Astonishment)
ok henning@
guenther [Sat, 10 May 2014 18:59:29 +0000 (18:59 +0000)]
Disable speedstep instead of panicing if the high and low speeds are the same.
Problem noted by Benjamin Baier (programmer (at) netzbasis.de)
improvements and ok kettenis@
kettenis [Sat, 10 May 2014 18:46:20 +0000 (18:46 +0000)]
We print time_t by casting to "long long" even in the kernel.
ok deraadt@
kettenis [Sat, 10 May 2014 18:41:55 +0000 (18:41 +0000)]
Sidestep format string issue by sanitizing a bit of debug code.
ok deraadt@
deraadt [Sat, 10 May 2014 18:15:31 +0000 (18:15 +0000)]
please, if there is any chance you've ever encountered the same command
during install time, be more careful here.
otto [Sat, 10 May 2014 18:14:55 +0000 (18:14 +0000)]
fix MALLOC_STATS code that was broken in rev 1.159, not compiled in by default
chris [Sat, 10 May 2014 17:48:09 +0000 (17:48 +0000)]
Allow nginx to chroot to a directory other than /var/www
ok henning@