deraadt [Wed, 7 Jan 2015 00:40:57 +0000 (00:40 +0000)]
sync
bluhm [Tue, 6 Jan 2015 23:11:23 +0000 (23:11 +0000)]
Backout revision 1.37. Setting ev->ev_pncalls to NULL results in
a use after free if the callback has freed the ev. With F in
malloc.conf both tmux and the regression tests triggered a segmentation
fault.
OK nicm@
deraadt [Tue, 6 Jan 2015 22:19:36 +0000 (22:19 +0000)]
mips64 architectures moving to static PIE
deraadt [Tue, 6 Jan 2015 22:14:53 +0000 (22:14 +0000)]
add 5.8 base key
sthen [Tue, 6 Jan 2015 22:05:51 +0000 (22:05 +0000)]
firmware key for 5.8
sthen [Tue, 6 Jan 2015 21:34:27 +0000 (21:34 +0000)]
add Medion Mobile S4222; Ingo Feinerer
stsp [Tue, 6 Jan 2015 21:26:46 +0000 (21:26 +0000)]
Remove the NOINET6 interface flag, a left-over from the times when IPv6
was enabled by default. Add AFATTACH/AFDETACH ioctls which enable/disable
an address family for an interface (currently used for IPv6 only).
New kernel needs new ifconfig for IPv6 configuration (address assignment
still works with old ifconfig making this easy to cross over).
Committing on behalf of henning@ who is currently lebensmittelvergiftet.
ok stsp, benno, mpi
nicm [Tue, 6 Jan 2015 21:14:42 +0000 (21:14 +0000)]
Revert to r1.16 since this is still clearly broken and I can't see how
right now.
bcook [Tue, 6 Jan 2015 21:08:11 +0000 (21:08 +0000)]
add initial HP-UX getentropy/arc4random support.
patch from Kinichiro Inoguchi, tested on HP-UX 11.31
ok deraadt@
stsp [Tue, 6 Jan 2015 17:55:28 +0000 (17:55 +0000)]
Make httpd return "404 not found" if an intermediate component of a requested
file path does not exist rather than returning "500 internal server error".
ok reyk
reyk [Tue, 6 Jan 2015 17:48:04 +0000 (17:48 +0000)]
I missed one goto abort instead of free(line).
Found by Fabian Raetz at gmail
armani [Tue, 6 Jan 2015 17:27:58 +0000 (17:27 +0000)]
We do not support freeing memory using reqbufs with a zero size so
return EINVAL in this case.
Also change an easily triggerable panic by a printf and return EINVAL.
Reminded by brad@, one typo spotted by sthen@ and ok mpi@
ajacoutot [Tue, 6 Jan 2015 14:48:59 +0000 (14:48 +0000)]
Simplify special vars argument checking; invalidating some bogus constructs
in the process.
reyk [Tue, 6 Jan 2015 14:07:48 +0000 (14:07 +0000)]
Only open a socket once for each unique "listen on" statement. This
prevents running out of file descriptors when loading a configuration
with many aliases.
OK florian@
reyk [Tue, 6 Jan 2015 13:48:15 +0000 (13:48 +0000)]
Instead of calling free(line) in each error case, call it once in fail:.
From Fabian Raetz at gmail
reyk [Tue, 6 Jan 2015 13:38:59 +0000 (13:38 +0000)]
Return "400 Bad Request" instead of "500 Internal Server Error" for
unknown/invalid HTTP requests.
From Fabian Raetz at gmail
dlg [Tue, 6 Jan 2015 12:50:47 +0000 (12:50 +0000)]
when we're entering an interrupt handler, record its ipl so splassert
can check if we're entering code we think should only be used from
lower ipls.
modelled a bit on how sparc64 does things.
with help from and ok kettenis@
ajacoutot [Tue, 6 Jan 2015 12:49:55 +0000 (12:49 +0000)]
While on death-row, 'rcctl status' still is still allowed for now -- so
make its output match the older one.
ajacoutot [Tue, 6 Jan 2015 12:28:56 +0000 (12:28 +0000)]
Sync with new rcctl usage. Will get some much needed love within the next
few days.
ajacoutot [Tue, 6 Jan 2015 12:27:42 +0000 (12:27 +0000)]
Rename some internal variables (flag and flags is way too error-prone):
flag -> var
flags -> args
deraadt [Tue, 6 Jan 2015 11:58:29 +0000 (11:58 +0000)]
sync
ajacoutot [Tue, 6 Jan 2015 11:47:50 +0000 (11:47 +0000)]
Major rcctl(8) rewrite to simplify it and add new features. It can now
configure "user" and "timeout". Committing now because there's been no
release yet including rcctl(8) so it's easier to modify its usage.
Usage extended (*retaining full backward compatibility for now*) to:
rcctl set|get|getdef foobar [flags|user|timeout|status]
The followings will be dropped soon but not right now to give people
time to adjust:
rcctl enable sshd flags # 'enable' takes no flags, use 'rcctl set ...'
rcctl status sshd # use 'rcctl get sshd [flags|status|timeout|user]'
rcctl default sshd # use 'rcctl getdef sshd [flags|status|timeout|user]'
rcctl status # use 'rcctl getall'
Man page changes will come soon. Make sure you have an up-to-date rc.subr.
discussed with schwarze@ robert@ jasper@ sthen@
"love the idea" jasper@
ok schwarze@
bluhm [Tue, 6 Jan 2015 11:42:37 +0000 (11:42 +0000)]
Apply commit
e0e6958aa074a7714cd7c4aa779a1dfede3a03b1 from upstream.
- Avoid deadlock when activating signals. Fixes bug
3048812.
Based on patch by Nicholas Marriott.
The deadlock was ultimately fixed in a different way (by disabling
reinit - see event.c r1.25). Add it now for consistency but without
the Windows compatibility code. Convert the fnctl() calls to
SOCK_CLOEXEC | SOCK_NONBLOCK to simplify the code.
OK nicm@
bluhm [Tue, 6 Jan 2015 11:27:35 +0000 (11:27 +0000)]
Apply commit
2d8cf0b720cdd5f9f292f174a10ff74e62a380ec from upstream.
- Defensive programming to prevent (hopefully impossible) stack-stomping
OK nicm@
mpi [Tue, 6 Jan 2015 10:42:16 +0000 (10:42 +0000)]
Support for MEDION S4222 which needs another funky umass quirk. From
Ingo Feinerer.
mpi [Tue, 6 Jan 2015 10:39:06 +0000 (10:39 +0000)]
regen
mpi [Tue, 6 Jan 2015 10:38:36 +0000 (10:38 +0000)]
MediaTek Inc. from Ingo Feinerer.
gilles [Tue, 6 Jan 2015 10:31:09 +0000 (10:31 +0000)]
move the message parser init earlier to avoid a possible use after free in
an unlikely error code path
gilles [Tue, 6 Jan 2015 09:32:13 +0000 (09:32 +0000)]
fix whitespace and indentation, by Kyle Milz
nicm [Tue, 6 Jan 2015 09:12:53 +0000 (09:12 +0000)]
Unzoom before kill, from Thomas Adam.
nicm [Tue, 6 Jan 2015 09:12:02 +0000 (09:12 +0000)]
Add missing default bindings, from Theo Buehler.
brad [Tue, 6 Jan 2015 04:05:43 +0000 (04:05 +0000)]
Always enable the receive filter. It was being enabled via sis_init()
but not if calling sis_iff() via sis_ioctl().
jsg [Tue, 6 Jan 2015 01:49:45 +0000 (01:49 +0000)]
init a potentially uninitialised var in pf_postprocess_addr
ok mikeb@ henning@
dlg [Tue, 6 Jan 2015 00:38:32 +0000 (00:38 +0000)]
implement atomic_swap_{uint,ulong,ptr) and some md variants. use these
to replace x86_atomic_testset_{u32,u64}.
help from guenther@ kettenis@
ok kettenis@
dlg [Mon, 5 Jan 2015 23:54:18 +0000 (23:54 +0000)]
splassert on some archs (or just sparc64) check that you're not in
an interrupt handler at an ipl level higher than what you're
splasserting you should be at. if you think code should be protected
by IPL_BIO and its entered from an interrupt handler established
at IPL_NET, you have a bug.
add some asserts to gets and puts so we can pick those cases up.
dlg [Mon, 5 Jan 2015 23:18:36 +0000 (23:18 +0000)]
there's already three different types of chips in this family of
controllers. the flags used in sgls on the first gen (thunderbolt)
are different to the ones used on the second and third gens (fury
and invader).
this creates an mfii_iop struct to store differences between these
chips, and uses them to set the flags on the sgls we generate for
the chip.
this solves lockups caused by stuck io on the following chips:
mfii0 at pci1 dev 0 function 0 "Symbios Logic MegaRAID SAS3108" rev 0x02: msi
mfii0: "PERC H730 Mini", firmware 25.2.1.0037, 1024MB cache
and
mfii0 at pci1 dev 0 function 0 "Symbios Logic MegaRAID SAS3008" rev 0x02: msi
mfii0: "PERC H330 Mini", firmware 25.2.1.0037
ive also tested this diff on:
mfii0 at pci10 dev 0 function 0 "Symbios Logic MegaRAID SAS2208" rev 0x05: msi
mfii0: "PERC H810 Adapter", firmware 21.2.0-0007, 1024MB cache
and
mfii0 at pci1 dev 0 function 0 "Symbios Logic MegaRAID SAS2208" rev 0x05: msi
mfii0: "PERC H710 Mini", firmware 21.3.0-0009, 512MB cache
Hrvoje Popovski reported the bug and verified the fix on his hardware.
bluhm [Mon, 5 Jan 2015 23:14:36 +0000 (23:14 +0000)]
Fix CVE-2014-6272 in Libevent 1.4 from upstream:
- https://github.com/libevent/libevent/commit/
7b21c4eabf1f3946d3f63cce1319c490caab8ecf
- For this fix, we need to make sure that passing too-large inputs to
the evbuffer functions can't make us do bad things with the heap.
On top of that do:
- Update libevent version to 1.4.15-stable.
- Use SIZE_MAX from limits.h instead of a private define.
- Do not declare 'size_t need' twice to avoid a compiler warning.
OK sthen
bluhm [Mon, 5 Jan 2015 22:41:37 +0000 (22:41 +0000)]
Fix relayd http regression tests. If http response is not ok, the
response's content length may be an unexpected value as an error
page is shown.
millert [Mon, 5 Jan 2015 21:58:52 +0000 (21:58 +0000)]
Remove #ifdef notdef bits--we are not going to change the fgetln()
API to NUL-terminate the buffer.
lum [Mon, 5 Jan 2015 21:45:10 +0000 (21:45 +0000)]
As pointed out by Kaspars Bankovskis on tech@, there is no ifdef for
STARTUP.
tedu [Mon, 5 Jan 2015 21:04:04 +0000 (21:04 +0000)]
rename kern enter/exit macros to malloc enter/leave to better reflect
what's going on.
gilles [Mon, 5 Jan 2015 21:00:36 +0000 (21:00 +0000)]
some third-party mda require a prepended From separator line, so remove the
prepending from delivery_filename backend and prepend conditionally earlier
in the code path if delivery method is through mda or filename.
ok eric@
millert [Mon, 5 Jan 2015 20:09:25 +0000 (20:09 +0000)]
Add SORTPROG environment variable to test a different sort program.
Replace -y with -S (supported by GNU and FreeBSD sorts).
uaa [Mon, 5 Jan 2015 19:59:40 +0000 (19:59 +0000)]
fix .Nd line
ok by deraadt@
bluhm [Mon, 5 Jan 2015 18:48:31 +0000 (18:48 +0000)]
Make syslogd TCP tests more reliable. Increase timeouts, shutdown
write to trigger connection close, allow errors EPIPE or ECONNRESET.
espie [Mon, 5 Jan 2015 16:32:28 +0000 (16:32 +0000)]
a few tweaks to match the old script better (for now)
- progress meter set like the old script
- path to firmware printed
- update both needed and installed drivers
- coda when we don't find anything to install/delete
- error out if @ARGV is not a possible driver
jsing [Mon, 5 Jan 2015 15:25:39 +0000 (15:25 +0000)]
Convert openssl(1) passwd to new option handling.
ok doug@
tedu [Mon, 5 Jan 2015 14:07:12 +0000 (14:07 +0000)]
encrypt can use the scrypt scaling code in libc now via crypt_newhash
tedu [Mon, 5 Jan 2015 13:52:47 +0000 (13:52 +0000)]
use the correct read count when determining how swabbing is necessary.
bug report from Martin Wolters
sthen [Mon, 5 Jan 2015 13:29:34 +0000 (13:29 +0000)]
zap file which is unused and already removed upstream
sthen [Mon, 5 Jan 2015 13:28:44 +0000 (13:28 +0000)]
use new ub_conf_dir to set the directory for keys and certificates
ok brad@, committed upstream (r3297)
sthen [Mon, 5 Jan 2015 13:27:33 +0000 (13:27 +0000)]
regen
sthen [Mon, 5 Jan 2015 13:27:28 +0000 (13:27 +0000)]
provide a new ub_conf_dir variable, to be used in unbound-control-setup
ok brad@, committed upstream (r3297)
sthen [Mon, 5 Jan 2015 13:18:31 +0000 (13:18 +0000)]
When building the package path for the new fw_update, only use
/firmware/snapshots/ for releases tagged with "-current", otherwise
use the version number e.g. /firmware/5.7/ for 5.7-beta / 5.7.
tedu [Mon, 5 Jan 2015 13:14:24 +0000 (13:14 +0000)]
simplify realloc loop by not using realloc. we don't care about old data,
so don't make realloc copy it. just allocate what we need.
ok deraadt guenther
tedu [Mon, 5 Jan 2015 13:10:10 +0000 (13:10 +0000)]
convert clock() to clock_gettime() for improved precision (and accuracy?)
guenther suggested using thread time, which actually may improve accuracy
if somebody puts this in a threaded program.
mpi [Mon, 5 Jan 2015 12:38:16 +0000 (12:38 +0000)]
Prevent a race condition upon resume by adding a supplementary delay.
This is a workaround needed at least by Renesas controllers. I didn't
find any documentation about this issue and I guess other open source
xHCI implementations do not see this race because they do much more work
upon resume.
Thanks to Remi Locherer for reporting this issue on bugs@.
reyk [Mon, 5 Jan 2015 11:03:57 +0000 (11:03 +0000)]
Be more specific: path is a component of the URI/URL, so use "path"
instead of "URI" or "URL" when referring to it.
mpi [Mon, 5 Jan 2015 10:21:58 +0000 (10:21 +0000)]
Explicitly check for a non NULL ifp in in_control(). Only MROUTING
ioctl(2)s are an exception to this rule. This makes clear that ifa
cannot be used initialized as found by jsg@ with gcc/clang.
ok krw@, bluhm@, jca@
lteo [Mon, 5 Jan 2015 03:54:55 +0000 (03:54 +0000)]
Zap a reference to .rnd, which is likely the last RANDFILE remnant on
this man page.
ok jsing@
jmc [Sun, 4 Jan 2015 22:38:31 +0000 (22:38 +0000)]
fix SEE ALSO;
chrisz [Sun, 4 Jan 2015 22:23:58 +0000 (22:23 +0000)]
add new url stripping option:
strip number
Strip number path components from the beginning of the
request URI before looking up the stripped-down URI at
the document root.
reviewed with much patience and OK by reyk@
uaa [Sun, 4 Jan 2015 21:11:16 +0000 (21:11 +0000)]
add uslhcom(4) description to usb(4)
overhaul SYNOPSYS of uhidev(4)
ok by deraadt@
uaa [Sun, 4 Jan 2015 20:55:38 +0000 (20:55 +0000)]
add new driver uslhcom(4) USB HID based UART.
uaa [Sun, 4 Jan 2015 20:29:09 +0000 (20:29 +0000)]
add uslhcom(4) to SYNOPSYS
ok by jmc@
mpi [Sun, 4 Jan 2015 20:10:08 +0000 (20:10 +0000)]
Only set the status of a completed xfer just before giving it back to
the stack. This will allow stricter checks when aborting transfers.
While here update a comment about short transfer and multi-TRB TD since
bulk transfers can also use a chain now.
miod [Sun, 4 Jan 2015 18:52:46 +0000 (18:52 +0000)]
EV68CB pass 2.4 is numeric revision 5, as reported by SRM.
tedu [Sun, 4 Jan 2015 18:44:12 +0000 (18:44 +0000)]
back to using ramdisk encrypt again. theo found and fixed the (my) bug.
daniel [Sun, 4 Jan 2015 16:12:42 +0000 (16:12 +0000)]
remove code for ancient gcc.
ok millert@, kettenis@
espie [Sun, 4 Jan 2015 14:55:47 +0000 (14:55 +0000)]
remove extra scaffolding that's actually NOT in the committed copy of
the old fw_update
espie [Sun, 4 Jan 2015 14:54:12 +0000 (14:54 +0000)]
same semantics as old program
espie [Sun, 4 Jan 2015 14:48:19 +0000 (14:48 +0000)]
cleaner logic, fix things so that fw_update -da makes sense
(deletes drivers not corresponding to that box, and updates/installs what's
needed)
espie [Sun, 4 Jan 2015 14:20:04 +0000 (14:20 +0000)]
tweak the ntogo messages slightly so that they work better in all situations
espie [Sun, 4 Jan 2015 14:10:20 +0000 (14:10 +0000)]
too many todo. Use unambiguous names, since they're not the same operation
at all
reyk [Sun, 4 Jan 2015 13:19:50 +0000 (13:19 +0000)]
The PRG airport has been renamed in honor of Vaclav Havel quite some
time ago.
From Jan Stary.
mpi [Sun, 4 Jan 2015 13:01:42 +0000 (13:01 +0000)]
Implement splassert(9) for powerpc.
This changes the logic to prevent a recursion when processing soft
interrupts. Previously a per-CPU flag was set before re-enabling
interrupts. Now the IPL level is raised to SOFTTTY which makes
splsoftassert() happy, greatly inspired by mips64.
As a side effect, the ppc_intr_{disable,enable}() dance is now done
only once instead of twice per splx(9).
While here, make use of dosoftint() instead of having 3 different
functions for dispatching soft interrupts.
Tested by deraadt@ on G4 smp and by myself G5 smp, G3, G4 and socppc.
No objection from the usual (and over busy) suspects.
mpi [Sun, 4 Jan 2015 12:30:39 +0000 (12:30 +0000)]
Sort wireless nodes by signal strength, from Simon Nicolussi.
ok stsp@, deraadt@
miod [Sun, 4 Jan 2015 09:07:44 +0000 (09:07 +0000)]
Use __buffer__ instead of __string__ as the __bounded type. The former causes
extra warnings with gcc 3, due to the way we use siphash in the filesystem
code.
With dlg@
dlg [Sun, 4 Jan 2015 08:54:01 +0000 (08:54 +0000)]
back out r1.173, aka the "* 8" diff. it tickles a problem on some
landisk machines. we've been unable to figure out due to a lack of
hardware (on my part) or time.
discussed with and ok miod@
jsg [Sun, 4 Jan 2015 08:42:04 +0000 (08:42 +0000)]
return 0 for success in ucycom_param() instead of uninitialised memory
brad [Sun, 4 Jan 2015 07:14:41 +0000 (07:14 +0000)]
Fix error handling for re_encap() within re_start() and some tidying up.
jsg [Sun, 4 Jan 2015 02:53:19 +0000 (02:53 +0000)]
avoid the use of an uninitialised variable in one of the codepaths in
pool_setlowat()
ok dlg@ tedu@
deraadt [Sun, 4 Jan 2015 02:28:26 +0000 (02:28 +0000)]
fix -b a mode, spotted by rpe
tedu [Sun, 4 Jan 2015 02:26:29 +0000 (02:26 +0000)]
go back to using /mnt encrypt. it seems something isn't working yet.
bcook [Sun, 4 Jan 2015 01:48:49 +0000 (01:48 +0000)]
use MAXIMUM as the canonical local MAX macro.
suggested by deraadt@ re: more general MIN/MAX cleanups
bcook [Sun, 4 Jan 2015 01:24:43 +0000 (01:24 +0000)]
rename sockaddr_un variables from 'sun' to the more common 'sa'.
This avoids a namespace conflict with Solaris build environments.
discussed with deraadt@ and kettenis@
bcook [Sun, 4 Jan 2015 01:19:46 +0000 (01:19 +0000)]
don't rely on sys/param.h having a MAX macro.
Switch to local definitions where MAX is needed.
discussed with deraadt@
bcook [Sun, 4 Jan 2015 01:16:10 +0000 (01:16 +0000)]
Remove the unused dns_pid variable.
Nothing is done with the return value from ntp_dns, and it already calls
fatal() on failure.
ok deraadt@
bcook [Sun, 4 Jan 2015 01:13:10 +0000 (01:13 +0000)]
removed unused libevent header and structures in ntpd.h
ok deraadt@ phessler@
bcook [Sun, 4 Jan 2015 01:11:24 +0000 (01:11 +0000)]
Add a missing include for time.h to get the definitions of clock_gettime and strftime.
ok jsing@ phessler@ naddy@
reyk [Sat, 3 Jan 2015 23:54:25 +0000 (23:54 +0000)]
Reset tls key and cert to NULL when duplicating a server - avoids a
possible double free in the error path of the parser.
Found by + OK doug@
kettenis [Sat, 3 Jan 2015 23:14:33 +0000 (23:14 +0000)]
Implement BMAC support; ports 2 & 3 work now as well.
uaa [Sat, 3 Jan 2015 21:43:42 +0000 (21:43 +0000)]
add new driver uslhcom(4) USB HID based UART.
uaa [Sat, 3 Jan 2015 21:24:19 +0000 (21:24 +0000)]
add new driver uslhcom(4) USB HID based UART.
uaa [Sat, 3 Jan 2015 21:20:26 +0000 (21:20 +0000)]
add new driver uslhcom(4) USB HID based UART.
miod [Sat, 3 Jan 2015 20:44:23 +0000 (20:44 +0000)]
Some bowels of the bus_dma code assume it is ok to perform bitwise operations
with the base address of the window, which only works if its size is smaller
than the address it begins at.
Until these assumptions are carefully fixed, revert r1.9, and go back to using
only one direct map window.
uaa [Sat, 3 Jan 2015 20:39:11 +0000 (20:39 +0000)]
add new driver uslhcom(4) USB HID based UART.
uaa [Sat, 3 Jan 2015 20:22:56 +0000 (20:22 +0000)]
(new) device driver for Silicon Labs CP2110 USB HID based UART.
ok by deraadt@
kettenis [Sat, 3 Jan 2015 19:08:40 +0000 (19:08 +0000)]
Make port 1 on the 4x1G card work. Port 2 & 3 still need some work.
doug [Sat, 3 Jan 2015 18:07:29 +0000 (18:07 +0000)]
Fix incorrect OPENSSL_assert() usage.
Instead of asserting, return an error code for I/O errors. This is based
on OpenSSL commit
2521fcd8527008ceb3e4748f95b0ed4e2d70cfef. Added checks
for two calloc()s while I'm here.
ok miod@