markus [Thu, 15 Jan 2015 21:37:14 +0000 (21:37 +0000)]
switch to sshbuf
guenther [Thu, 15 Jan 2015 21:19:22 +0000 (21:19 +0000)]
UVM_INH_* and UVM_ADV_* are gone; use MAP_INHERIT_* and MADV_*
miod [Thu, 15 Jan 2015 21:17:54 +0000 (21:17 +0000)]
Rework the disk label handling once again. Disk label is read at attach time
because it teaches us the drive geometry, but from then on there is no reason
not to implement disklabel spoof support and a proper DIOCGPDINFO ioctl.
As a result, this makes disklabel -A now work on SMD disks.
The drawback of this, is that, since native labels do not have a pcylinders
value, the code will now always assume pcylinders == ncylinders + acylinders.
(this ought to be the case by default for labels produced by SunOS format(8).)
miod [Thu, 15 Jan 2015 20:48:51 +0000 (20:48 +0000)]
Be sure to print time_t with the appropriate format specifier. Repairs
dump of eeprom variables on sun4.
guenther [Thu, 15 Jan 2015 20:40:59 +0000 (20:40 +0000)]
Test shmat() with a non-NULL address, both failure and success
schwarze [Thu, 15 Jan 2015 20:37:36 +0000 (20:37 +0000)]
Use .An in the AUTHORS section where appropriate;
from Kaspars at Bankovskis dot net
millert [Thu, 15 Jan 2015 20:36:17 +0000 (20:36 +0000)]
The flags variable in shmat was not actually used. We need UVM_FLAG_*
flags, not mmap-style flags for UVM_MAPFLAG(). Remove the nonsensical
MAP_ANON|MAP_SHARED value and convert MAP_FIXED to UVM_FLAG_FIXED.
OK guenther@
tedu [Thu, 15 Jan 2015 19:55:58 +0000 (19:55 +0000)]
we don't have an alloca() symbol anywhere in libc, don't prototype it.
instead, #define to __builtin_alloca. if that doesn't work, you're hosed.
from an old discussion about the perils of alloca. (don't use alloca!)
schwarze [Thu, 15 Jan 2015 19:26:27 +0000 (19:26 +0000)]
use .In where appropriate; from Kaspars at Bankovskis dot net
schwarze [Thu, 15 Jan 2015 19:20:59 +0000 (19:20 +0000)]
typo; from Kaspars at Bankovskis dot net
espie [Thu, 15 Jan 2015 19:16:17 +0000 (19:16 +0000)]
oops, forgotten back-pointer to state for error messages
problem noticed by jca@
schwarze [Thu, 15 Jan 2015 19:06:31 +0000 (19:06 +0000)]
remove .Tn; from Jan Stary <hans at stare dot cz>
naddy [Thu, 15 Jan 2015 18:32:54 +0000 (18:32 +0000)]
handle UMAC128 initialization like UMAC; ok djm@ markus@
miod [Thu, 15 Jan 2015 17:54:14 +0000 (17:54 +0000)]
Fix subtle breakage for polled commands introduced during iopool conversion.
chl [Thu, 15 Jan 2015 17:34:15 +0000 (17:34 +0000)]
remove unused variable
ok tedu@
chl [Thu, 15 Jan 2015 17:32:43 +0000 (17:32 +0000)]
remove unused variable
ok tedu@
deraadt [Thu, 15 Jan 2015 17:14:04 +0000 (17:14 +0000)]
all modern systems can do this SUSPEND thing
deraadt [Thu, 15 Jan 2015 17:13:37 +0000 (17:13 +0000)]
NR_OF_EOFS is a festering boil, lance & drain.
ok millert
kettenis [Thu, 15 Jan 2015 16:31:19 +0000 (16:31 +0000)]
Map the sigcode page with MAP_INHERIT_COPY to make sure it isn't shared ater
a fork.
ok deraadt@
sf [Thu, 15 Jan 2015 15:30:17 +0000 (15:30 +0000)]
Don't include i82489 headers in cpu.h
Only pmap.c was not including them explicitly. Fix that.
"The direction is good" deraadt@
sf [Thu, 15 Jan 2015 13:58:55 +0000 (13:58 +0000)]
Define and print HV cpuid flag on i386.
This is set by many hypervisors, including kvm, vmware, hyper-v.
nicm [Thu, 15 Jan 2015 13:43:13 +0000 (13:43 +0000)]
Remove an unnecessary variable and shorten a line.
nicm [Thu, 15 Jan 2015 13:35:13 +0000 (13:35 +0000)]
Shorten some long lines.
bluhm [Thu, 15 Jan 2015 13:15:17 +0000 (13:15 +0000)]
Add a test for syslogd that reconnects to a TCP syslog server after
an established connection has been closed.
millert [Thu, 15 Jan 2015 13:05:59 +0000 (13:05 +0000)]
Use explicit_bzero instead of memset in hash Final and End functions.
OK deraadt@ djm@
sobrado [Thu, 15 Jan 2015 11:54:31 +0000 (11:54 +0000)]
tell the truth about DES.
joint work with djm@ and jsing@
ok djm@
bluhm [Thu, 15 Jan 2015 11:49:59 +0000 (11:49 +0000)]
When the connection to a TCP syslog server fails or the TCP connection
terminates, try to reconnect after an increasing timeout.
test and OK jasper@
djm [Thu, 15 Jan 2015 11:04:36 +0000 (11:04 +0000)]
fix regression reported by brad@ for passworded keys without
agent present
kettenis [Thu, 15 Jan 2015 10:04:26 +0000 (10:04 +0000)]
Map the sigcode page copy-on-write. This allows userland to put breakpoints
in the signal trampoline. Solves some long-standing issues with debugging
signal handlers in GDB.
ok deraadt@
djm [Thu, 15 Jan 2015 09:40:00 +0000 (09:40 +0000)]
sync ssh-keysign, ssh-keygen and some dependencies to the new
buffer/key API; mostly mechanical, ok markus@
ajacoutot [Thu, 15 Jan 2015 09:20:37 +0000 (09:20 +0000)]
Extend the description of _timeout.
gilles [Thu, 15 Jan 2015 09:05:37 +0000 (09:05 +0000)]
when enqueueing offline mails from within the daemon session, we should not
rely on getlogin() otherwise mail will end up enqueued as coming from user
who started smtpd.
bug spotted by deraadt@, diff ok todd@
djm [Thu, 15 Jan 2015 07:57:08 +0000 (07:57 +0000)]
remove commented-out test code now that it has moved to a
proper unit test
djm [Thu, 15 Jan 2015 07:36:28 +0000 (07:36 +0000)]
unit tests for KRL bitmap
deraadt [Thu, 15 Jan 2015 06:57:18 +0000 (06:57 +0000)]
back in september I did the large abstraction refactoring to allow these
other systems to fit into the same mold, so add copyright
schwarze [Thu, 15 Jan 2015 04:26:06 +0000 (04:26 +0000)]
Fatal errors no longer exist.
If a file can be opened, mandoc will produce some output;
at worst, the output may be almost empty.
Simplifies error handling and frees a message type for future use.
millert [Thu, 15 Jan 2015 03:54:28 +0000 (03:54 +0000)]
Sync with libc version.
millert [Thu, 15 Jan 2015 03:54:12 +0000 (03:54 +0000)]
Make strlcpy/strlcat slightly easier to read.
millert [Thu, 15 Jan 2015 03:19:43 +0000 (03:19 +0000)]
POSIX does not require an explicit inclusion of sys/types.h to get
uid_t or gid_t (and neither do we). OK deraadt@
schwarze [Thu, 15 Jan 2015 02:29:07 +0000 (02:29 +0000)]
downgrade .so failure from FATAL to ERROR
jsg [Thu, 15 Jan 2015 01:19:28 +0000 (01:19 +0000)]
Define a new wscons mouse type for Synaptics clickpad devices that lack
physical buttons. This will be used if the acpi pnpid for the mouse
matches a list provided by Synaptics found in Linux.
Combined with the xenocara changes this will remove the need for an xorg
config file for the x240/t440/t540 generation of thinkpads.
ok matthieu@ shadchin@ kettenis@
deraadt [Thu, 15 Jan 2015 00:48:10 +0000 (00:48 +0000)]
I bet the reference to <sys/ttycom.h> near the ldisc list in tty(4)
led this program and the associated manual pages to including that
file directly, rather than the canonical include path of <sys/ioctl.h>
discussed with millert
deraadt [Thu, 15 Jan 2015 00:45:20 +0000 (00:45 +0000)]
do not be specific regarding which ioctl backend .h file; the list
is good enough, and we do not want to mislead. the file to include
is at the top of the file.
deraadt [Thu, 15 Jan 2015 00:40:23 +0000 (00:40 +0000)]
not supposed to include sys/ttycom.h (these may have rpcgen history)
deraadt [Thu, 15 Jan 2015 00:32:36 +0000 (00:32 +0000)]
wrong place to mention nmea. cua is different, that hints at how it
works to many.
deraadt [Thu, 15 Jan 2015 00:25:00 +0000 (00:25 +0000)]
delete overly-specific details that cause espie angst.
ok millert tedu
deraadt [Wed, 14 Jan 2015 23:36:15 +0000 (23:36 +0000)]
change _yp_domain[] to HOST_NAME_MAX+1, and re-arrange those who include
it to get limits.h early enough
deraadt [Wed, 14 Jan 2015 23:03:50 +0000 (23:03 +0000)]
these never needed sys/param.h
schwarze [Wed, 14 Jan 2015 22:57:57 +0000 (22:57 +0000)]
downgrade ".so with absolute path" from FATAL to ERROR;
allows to get rid of ROFF_ERR
schwarze [Wed, 14 Jan 2015 22:44:51 +0000 (22:44 +0000)]
simplify by getting rid of ROFF_ERR in tbl(7) parsing; no functional change
schwarze [Wed, 14 Jan 2015 22:02:00 +0000 (22:02 +0000)]
To get rid of SYSERR entries in enum mandocerr, downgrade problems with
missing and unreadable files from SYSERR to ERROR.
Needed for upcoming work.
As a bonus, this minimally simplifies code and documentation.
miod [Wed, 14 Jan 2015 21:35:43 +0000 (21:35 +0000)]
Don't forget to register event counter for VME interrupts.
schwarze [Wed, 14 Jan 2015 21:27:01 +0000 (21:27 +0000)]
if earlier files set a non-zero exit status,
do not allow later files to reset it to zero
deraadt [Wed, 14 Jan 2015 21:17:52 +0000 (21:17 +0000)]
This does not need <sys/param.h>. If we change some other stuff around,
it might not need <sys/types.h> either. Even further changes could
eliminate it. Anyone up for the exercise?
miod [Wed, 14 Jan 2015 21:17:09 +0000 (21:17 +0000)]
When a read or write operation fails on a sector found on the bad block table,
be sure to adjust the IOPB data pointer before redirecting the I/O for this
particular sector to the replacement location.
Otherwise, the data pointer still points to the first sector of the I/O, which
may not necessarily be the one which failed.
(This is yet another 19 years old bug, making your filesystems self-destruct
even faster than intended)
deraadt [Wed, 14 Jan 2015 21:15:36 +0000 (21:15 +0000)]
This does not need <sys/param.h>. <sys/types.h> is enough.
discussed with ratchov and sthen
miod [Wed, 14 Jan 2015 21:14:49 +0000 (21:14 +0000)]
When the outcome of the watchdog processing is a reset, do not bother checking
further pending requests, since they will be forcefully failed by the reset
routine.
naddy [Wed, 14 Jan 2015 21:14:27 +0000 (21:14 +0000)]
Switch drift file format to ppm for compatibility with the ntp.org
daemon.
Old drift files will be interpreted as a minuscule adjustment and
ntpd will proceed to rediscover the drift, like starting from zero
on a newly installed machine.
ok deraadt@
miod [Wed, 14 Jan 2015 21:13:46 +0000 (21:13 +0000)]
Make the special buffer used for the few sector I/O at attach time a per-softc
member, rather than a global. No functional change.
djm [Wed, 14 Jan 2015 20:54:29 +0000 (20:54 +0000)]
whitespace
djm [Wed, 14 Jan 2015 20:05:27 +0000 (20:05 +0000)]
move authfd.c and its tentacles to the new buffer/key API;
ok markus@
djm [Wed, 14 Jan 2015 19:33:41 +0000 (19:33 +0000)]
fix small regression: ssh-agent would return a success message
but an empty signature if asked to sign using an unknown key;
ok markus@
deraadt [Wed, 14 Jan 2015 19:15:01 +0000 (19:15 +0000)]
pathconf(1) never happened.
ok guenther
miod [Wed, 14 Jan 2015 19:02:59 +0000 (19:02 +0000)]
Make x[dy]c_e2str return a const char *.
Remove obsolete comments in x[dy]dump().
miod [Wed, 14 Jan 2015 19:01:00 +0000 (19:01 +0000)]
Remove pass-through ioctl for SMD disks (DIOSXDCMD). Nothing uses this, because
noone wrote an equivalent of SunOS format(8) and noone ever will.
millert [Wed, 14 Jan 2015 18:28:15 +0000 (18:28 +0000)]
Use reallocarray() instead of calloc() when making a copy of the
environment. We already explicitly NULL terminate the array.
schwarze [Wed, 14 Jan 2015 17:45:25 +0000 (17:45 +0000)]
Simplify handling of system errors: just exit(3).
We already do the same for malloc(3) failure.
The is no virtue in trying to survive failure of fork(2) and the like.
millert [Wed, 14 Jan 2015 17:30:53 +0000 (17:30 +0000)]
Replace MAXHOSTNAMELEN with HOST_NAME_MAX+1 and MAXNAMLEN with NAME_MAX
and use limits.h instead of sys/param.h.
millert [Wed, 14 Jan 2015 17:27:51 +0000 (17:27 +0000)]
Use HAVE_FOO for BSD-specific features instead of relying on the
BSD macro from sys/param.h.
millert [Wed, 14 Jan 2015 17:27:29 +0000 (17:27 +0000)]
Use standard types for wait, readdir, signals and pids.
millert [Wed, 14 Jan 2015 17:27:13 +0000 (17:27 +0000)]
Remove compatibility defines for ancient systems. We assume a
reasonable level of POSIX compliance.
djm [Wed, 14 Jan 2015 15:02:39 +0000 (15:02 +0000)]
avoid BIGNUM in KRL code by using a simple bitmap;
feedback and ok markus
djm [Wed, 14 Jan 2015 13:54:13 +0000 (13:54 +0000)]
update sftp client and server to new buffer API.
pretty much just mechanical changes; with & ok markus
markus [Wed, 14 Jan 2015 13:09:09 +0000 (13:09 +0000)]
switch to sshbuf/sshkey; with & ok djm@
millert [Wed, 14 Jan 2015 11:59:10 +0000 (11:59 +0000)]
Replace "errx(1, "malloc failed")" with "err(1, NULL)" to get
standard "Cannot allocate memory" message.
djm [Wed, 14 Jan 2015 10:46:28 +0000 (10:46 +0000)]
avoid an warning for the !OPENSSL case
markus [Wed, 14 Jan 2015 10:30:34 +0000 (10:30 +0000)]
swith auth-options to new sshbuf/sshkey; ok djm@
djm [Wed, 14 Jan 2015 10:29:45 +0000 (10:29 +0000)]
make non-OpenSSL aes-ctr work on sshd w/ privsep; ok markus@
markus [Wed, 14 Jan 2015 10:24:42 +0000 (10:24 +0000)]
remove unneeded includes, sync my copyright across files & whitespace; ok djm@
markus [Wed, 14 Jan 2015 09:58:21 +0000 (09:58 +0000)]
re-add comment about full path
markus [Wed, 14 Jan 2015 09:54:38 +0000 (09:54 +0000)]
don't reset to the installed sshd; connect before reconfigure, too
gilles [Wed, 14 Jan 2015 09:07:51 +0000 (09:07 +0000)]
recipient and sender lists now support the user+TAG notation, allowing
among other things a secondary MX to filter recipients to be relayed to a
primary MX even if they are using tags. there are other nice things to do
with that feature, use your imagination.
tested and ok florian@, tested by several users for a few days too
gilles [Wed, 14 Jan 2015 08:50:32 +0000 (08:50 +0000)]
bring back reverted commits, the crash was unrelated
bluhm [Wed, 14 Jan 2015 00:37:19 +0000 (00:37 +0000)]
It is sufficient to protect in6_ifattach() with splsoftnet().
OK mpi@
millert [Tue, 13 Jan 2015 23:22:33 +0000 (23:22 +0000)]
Move debug printf for duplicate af to correct location.
stsp [Tue, 13 Jan 2015 23:16:59 +0000 (23:16 +0000)]
In IBSS mode, stop calling the driver's newassoc() each time a probe
response is received from a node. Only call it once when the node is
initially discovered, as used to be the case before r1.4 of this file.
Asking the driver to set up its per-node private state once is enough.
Also remove an outdated comment.
ok sthen deraadt jsg
schwarze [Tue, 13 Jan 2015 23:16:12 +0000 (23:16 +0000)]
Be developer-friendly, 'cause OpenBSD devs like to:
cd /usr/src/share/man/man4; vi newdev.4 Makefile; make install; man newdev
When a manual is missing from an outdated database, let man(1)
show it anyway, using a KISS file system lookup as a fallback.
Requested by deraadt@.
87 new lines of code doesn't seem too much bloat to me.
Of course, keeping your mandoc.db(5) files up to date with makewhatis(8)
or weekly(8) is still required for apropos(1) to find your new pages.
millert [Tue, 13 Jan 2015 21:49:36 +0000 (21:49 +0000)]
Remove bogus getopt.h include.
millert [Tue, 13 Jan 2015 21:45:53 +0000 (21:45 +0000)]
Fix printing of number of whitelist entries in debug mode.
millert [Tue, 13 Jan 2015 21:42:59 +0000 (21:42 +0000)]
Use address-family specific lists of addr/mask entries instead of
a union that can store either ipv4 or ipv6. The old method used
4x as much memory as was really needed for ipv4. The spamd-setup
protocol has changed from: tag;message;a/m;a/m;a/m...\n
to :tag;message;af;count;a/m;a/m;a/m...[af;count;a/m;a/m;a/m]\n
OK phessler@ "nice" beck@
miod [Tue, 13 Jan 2015 20:40:11 +0000 (20:40 +0000)]
Initialize d_type to DTYPE_SMD in the default label to avoid disklabel(8)
asking for the device type upon labeling a freshly formatted disk.
kettenis [Tue, 13 Jan 2015 20:05:43 +0000 (20:05 +0000)]
Generate a PT_PHDR entry for static PIE binaries. Modern GDB needs this to
do some sanity checking while it determines the load base.
ok kurt@
kettenis [Tue, 13 Jan 2015 20:05:01 +0000 (20:05 +0000)]
Generate a PT_PHDR entry for static PIE binaries. Modern GDB needs this to
do some sanity checking while it determines the load base.
ok kurt@
markus [Tue, 13 Jan 2015 19:31:40 +0000 (19:31 +0000)]
adapt mac.c to ssherr.h return codes (de-fatal) and simplify dependencies
ok djm@
djm [Tue, 13 Jan 2015 19:06:49 +0000 (19:06 +0000)]
sync changes from libopenssh; prepared by markus@
mostly debug output tweaks, a couple of error return value changes
and some other minor stuff
djm [Tue, 13 Jan 2015 19:04:35 +0000 (19:04 +0000)]
sync changes from libopenssh; prepared by markus@
mostly debug output tweaks, a couple of error return value changes
and some other minor stuff
kettenis [Tue, 13 Jan 2015 18:51:27 +0000 (18:51 +0000)]
Many architectures call initmsgbuf() really really early, before uvm is
initialized. Calling malloc(9) at that point is not a good idea. So
initialize consbuf later.
Fixes dmesg -s on sparc64 (and probably a few other architectures).
ok miod@, deraadt@
espie [Tue, 13 Jan 2015 18:30:15 +0000 (18:30 +0000)]
zap really odd code that's not actually in-use.
okay millert@
bluhm [Tue, 13 Jan 2015 17:35:35 +0000 (17:35 +0000)]
For non-blocking sockets tls_connect_fds() could fail with EAGAIN.
Use the same logic from the read, write, accept functions to inform
the caller wether a readable or writable socket is needed. After
that event, the connect function must be called again. All the
checks before connecting are done only once.
OK tedu@
bcallah [Tue, 13 Jan 2015 17:02:28 +0000 (17:02 +0000)]
Remove a variable that's not really being used.
ok deraadt@ jasper@