openbsd
9 years agoKill outdated comment.
jca [Mon, 25 May 2015 21:59:37 +0000 (21:59 +0000)]
Kill outdated comment.

ok eric@

9 years agoMake SSL_CIPHER_get_bits() report ChaCha20-Poly1305 ciphers as using
guenther [Mon, 25 May 2015 21:35:35 +0000 (21:35 +0000)]
Make SSL_CIPHER_get_bits() report ChaCha20-Poly1305 ciphers as using
256bit keys

problem noted by Tim Kuijsten (info (at) netsend.nl)
ok deraadt@ miod@ bcook@

9 years agoSkip search domains iteration if RES_DNSRCH and/or RES_DEFNAMES is unset.
eric [Mon, 25 May 2015 19:30:25 +0000 (19:30 +0000)]
Skip search domains iteration if RES_DNSRCH and/or RES_DEFNAMES is unset.

prodded by Brad
ok jca@

9 years agobump up the default Diffie-Hellman group to modp3072; ok mikeb@ djm@
naddy [Mon, 25 May 2015 19:29:36 +0000 (19:29 +0000)]
bump up the default Diffie-Hellman group to modp3072; ok mikeb@ djm@

9 years agogetnameinfo(3) doesn't need to initialize the resolver when it's only used
eric [Mon, 25 May 2015 19:16:08 +0000 (19:16 +0000)]
getnameinfo(3) doesn't need to initialize the resolver when it's only used
for address/port formatting (e.g. NI_NUMERICHOST).

ok deraadt@ jca@

9 years agofix a panic in import_identities() in case the ID isnt loaded
benno [Mon, 25 May 2015 18:48:17 +0000 (18:48 +0000)]
fix a panic in import_identities() in case the ID isnt loaded
(triggered by bgpd).
ok marku@s, mikeb@

9 years agoInitialize ipa_ndrq in isascan() too...
miod [Mon, 25 May 2015 15:19:22 +0000 (15:19 +0000)]
Initialize ipa_ndrq in isascan() too...

9 years agoMatch newer elantech v4 touchpads, logic taken from Linux.
mpi [Mon, 25 May 2015 15:04:26 +0000 (15:04 +0000)]
Match newer elantech v4 touchpads, logic taken from Linux.

Based on a submission from and ok jcs@.

9 years agoonly scan sensors if they are configured
deraadt [Mon, 25 May 2015 14:58:34 +0000 (14:58 +0000)]
only scan sensors if they are configured
ok bcook

9 years agoPort the ELF m88k work to binutils 2.17. Good enough to build a booting
miod [Mon, 25 May 2015 14:56:26 +0000 (14:56 +0000)]
Port the ELF m88k work to binutils 2.17. Good enough to build a booting
kernel, and hopefully userland as well.

9 years agosort
deraadt [Mon, 25 May 2015 14:50:10 +0000 (14:50 +0000)]
sort

9 years agoMake this build when using the __STRICT_ALIGNMENT version of USETW.
jsg [Mon, 25 May 2015 12:53:12 +0000 (12:53 +0000)]
Make this build when using the __STRICT_ALIGNMENT version of USETW.

9 years agovax ELF bits for binutils 2.17.
miod [Mon, 25 May 2015 12:42:54 +0000 (12:42 +0000)]
vax ELF bits for binutils 2.17.

9 years agoConvert from ether_input() with separate mbuf data and Ethernet header, to
miod [Mon, 25 May 2015 12:40:04 +0000 (12:40 +0000)]
Convert from ether_input() with separate mbuf data and Ethernet header, to
if_input(). Based upon an initial diff from mpi@, and then painfully made
STRICT_ALIGNMENT-compliant. Tested on 4/260.
ok mpi@

9 years agoChange ENTRY to __start to match binutils 2.15, needed for static PIE
miod [Mon, 25 May 2015 12:12:42 +0000 (12:12 +0000)]
Change ENTRY to __start to match binutils 2.15, needed for static PIE

9 years agoPrevent a use after free in by closing all open endpoints upon detach.
mpi [Mon, 25 May 2015 11:52:15 +0000 (11:52 +0000)]
Prevent a use after free in by closing all open endpoints upon detach.

Fix a panic reported by landry@ with Android's ADB.

Tested and ok ajacoutot@

9 years agoallow pkg_add as nonroot to soft-fail when outside of local base.
espie [Mon, 25 May 2015 07:20:31 +0000 (07:20 +0000)]
allow pkg_add as nonroot to soft-fail when outside of local base.

9 years agoa dreaded whitespace; Kyle Milz
deraadt [Mon, 25 May 2015 03:07:49 +0000 (03:07 +0000)]
a dreaded whitespace; Kyle Milz

9 years agomissing word in comment; Kyle Milz
deraadt [Mon, 25 May 2015 03:07:07 +0000 (03:07 +0000)]
missing word in comment; Kyle Milz

9 years agoTeach binutils the {rd,wr}{fs,gs}base instructions.
guenther [Mon, 25 May 2015 00:12:59 +0000 (00:12 +0000)]
Teach binutils the {rd,wr}{fs,gs}base instructions.

Flag bits worked out with kettenis@
ok mlarkin@

9 years agoadd missing 'c' option to getopt(), case statement was already
djm [Sun, 24 May 2015 23:39:16 +0000 (23:39 +0000)]
add missing 'c' option to getopt(), case statement was already
there; from Felix Bolte

9 years agoMaximilian dot Fillinger at uni-duesseldorf dot de
schwarze [Sun, 24 May 2015 15:44:52 +0000 (15:44 +0000)]
Maximilian dot Fillinger at uni-duesseldorf dot de
starts helping with the pod2mdoc(1)-based conversion
of LibreSSL crypto manuals from perlpod(1) to mdoc(7).
Here comes the first file, slightly tweaked by me.

9 years agoInitialize ipa_nirq in isascan(). Gets rid of spurious irq locators being
miod [Sun, 24 May 2015 15:25:34 +0000 (15:25 +0000)]
Initialize ipa_nirq in isascan(). Gets rid of spurious irq locators being
printed for isadma(4).

9 years agoadd the chromebook board id the exynos code uses
jsg [Sun, 24 May 2015 11:06:16 +0000 (11:06 +0000)]
add the chromebook board id the exynos code uses

9 years agoimx_board_devs -> exynos_board_devs
jsg [Sun, 24 May 2015 11:01:48 +0000 (11:01 +0000)]
imx_board_devs -> exynos_board_devs

9 years agoFollow the recent pckbc@isa changes and always establish all the necessary
miod [Sun, 24 May 2015 10:57:47 +0000 (10:57 +0000)]
Follow the recent pckbc@isa changes and always establish all the necessary
interrupts at pckbc attach time, and get rid of the `intr_establish'
pckbc callback.

Tested on hppa (gsckbc) and sgi (pckbc@hpc); not tested on sparc64 (pckbc@ebus)
but this attachment was already behaving this way and its intr_establish
callback was an empty function.

9 years agoAdd udl(4) and uvideo(4) to armv7 GENERIC. Tested on my sabre lite (imx).
matthieu [Sun, 24 May 2015 08:01:46 +0000 (08:01 +0000)]
Add udl(4) and uvideo(4) to armv7 GENERIC. Tested on my sabre lite (imx).
enable udl firmware and COMPAT_RAW_KBD to make udl useable with X.
ok jsg@

9 years agopass subst to the installer state, so that -Dunsigned would work
espie [Sun, 24 May 2015 07:53:14 +0000 (07:53 +0000)]
pass subst to the installer state, so that -Dunsigned would work

9 years agoTreat primary cpu like others and put pointer to its GDT in cpu_info.ci_gdt
guenther [Sun, 24 May 2015 01:01:49 +0000 (01:01 +0000)]
Treat primary cpu like others and put pointer to its GDT in cpu_info.ci_gdt

requested by and ok mlarkin@

9 years agobump to version 2.2
bcook [Sat, 23 May 2015 21:09:46 +0000 (21:09 +0000)]
bump to version 2.2

ok deraadt@

9 years agofix a memory leak in an error path
jsg [Sat, 23 May 2015 14:28:37 +0000 (14:28 +0000)]
fix a memory leak in an error path
ok markus@ dtucker@

9 years agofix a memory leak in an error path
jsg [Sat, 23 May 2015 14:26:06 +0000 (14:26 +0000)]
fix a memory leak in an error path

9 years agoPACKET_TAG_IPSEC_PENDING_TDB is gone, too.
markus [Sat, 23 May 2015 12:57:09 +0000 (12:57 +0000)]
PACKET_TAG_IPSEC_PENDING_TDB is gone, too.

9 years agoremove PACKET_TAG_IPSEC_PENDING_TDB, it is never set; ok mikeb@
markus [Sat, 23 May 2015 12:52:59 +0000 (12:52 +0000)]
remove PACKET_TAG_IPSEC_PENDING_TDB, it is never set; ok mikeb@

9 years agointroduce ipsec-id bundles and use them for ipsecflowinfo,
markus [Sat, 23 May 2015 12:38:53 +0000 (12:38 +0000)]
introduce ipsec-id bundles and use them for ipsecflowinfo,
fixes rekeying for l2tp/ipsec against multiple windows clients
and saves memory (for many SAs to same peers); feedback and ok mikeb@

9 years agouse & not && when testing lcr bits
jsg [Sat, 23 May 2015 12:08:14 +0000 (12:08 +0000)]
use & not && when testing lcr bits
tested by jmatthew
ok pirofti@ jmatthew@ jasper@

9 years agoCall if_input() instead of setting the "rcvif" pointer ourself.
mpi [Sat, 23 May 2015 08:32:12 +0000 (08:32 +0000)]
Call if_input() instead of setting the "rcvif" pointer ourself.

ok dlg@

9 years agoPass output packets to bpf(4). This is helpful when debugging stack
mpi [Sat, 23 May 2015 08:31:05 +0000 (08:31 +0000)]
Pass output packets to bpf(4).  This is helpful when debugging stack
issues.

ok reyk@

9 years agoCanonicalize all devices to DUIDs in order to make -w and -W output consistent.
guenther [Sat, 23 May 2015 05:17:20 +0000 (05:17 +0000)]
Canonicalize all devices to DUIDs in order to make -w and -W output consistent.

Based on diff from Manuel Giraud (manuel (at) ledu-giraud.fr)  Thanks!

9 years agodefine BYTEORDER so the endian tests will work
jsg [Sat, 23 May 2015 00:53:25 +0000 (00:53 +0000)]
define BYTEORDER so the endian tests will work
ok deraadt@ miod@

9 years agoAdd tests for relayd TLS inspection with plain SSL and HTTPS.
bluhm [Fri, 22 May 2015 19:09:18 +0000 (19:09 +0000)]
Add tests for relayd TLS inspection with plain SSL and HTTPS.

9 years agosync
deraadt [Fri, 22 May 2015 15:10:13 +0000 (15:10 +0000)]
sync

9 years agoCut down on if statements around pf_icmp_state_lookup
mikeb [Fri, 22 May 2015 14:18:55 +0000 (14:18 +0000)]
Cut down on if statements around pf_icmp_state_lookup

Checked with blambert@, OK millert, henning

9 years agoCleanup leftover PF_ICMP_MULTI_* code that is not needed anymore.
mikeb [Fri, 22 May 2015 14:16:09 +0000 (14:16 +0000)]
Cleanup leftover PF_ICMP_MULTI_* code that is not needed anymore.

ok henning

9 years agobe pedantic with sizeof use
jsg [Fri, 22 May 2015 13:48:25 +0000 (13:48 +0000)]
be pedantic with sizeof use
no change in behaviour as sizeof(char **) is the same as sizeof(char *)
ok otto@ guenther@

9 years agoDon't use an uninitialised softc pointer in midiread/midiwrite.
jsg [Fri, 22 May 2015 12:52:00 +0000 (12:52 +0000)]
Don't use an uninitialised softc pointer in midiread/midiwrite.
ok ratchov@

9 years agoLITTE_ENDIAN -> LITTLE_ENDIAN
jsg [Fri, 22 May 2015 12:46:38 +0000 (12:46 +0000)]
LITTE_ENDIAN -> LITTLE_ENDIAN
ok ratchov@

9 years agoLimit the number of dma segments used for transmitting packets to
kettenis [Fri, 22 May 2015 06:50:54 +0000 (06:50 +0000)]
Limit the number of dma segments used for transmitting packets to
IWM_NUM_OF_TBS - 2.  We have IWM_NUM_OF_TBS slots, but use two of those
for sending commands to the firmware.  Hopefully fixes the

  iwm0: hardware error, stopping device

errors I've seen somewhat regularly.

ok claudio@, deraadt@

9 years agomention ssh-keygen -E for comparing legacy MD5 fingerprints; bz#2332
djm [Fri, 22 May 2015 05:28:45 +0000 (05:28 +0000)]
mention ssh-keygen -E for comparing legacy MD5 fingerprints; bz#2332

9 years agoReorder EscapeChar option parsing to avoid a single-byte out-
djm [Fri, 22 May 2015 04:45:52 +0000 (04:45 +0000)]
Reorder EscapeChar option parsing to avoid a single-byte out-
of-bounds read. bz#2396 from Jaak Ristioja; ok dtucker@

9 years agoadd knob to relax GSSAPI host credential check for multihomed hosts
djm [Fri, 22 May 2015 03:50:02 +0000 (03:50 +0000)]
add knob to relax GSSAPI host credential check for multihomed hosts
bz#928, patch by Simon Wilkinson; ok dtucker
(kerberos/GSSAPI is not compiled by default on OpenBSD)

9 years agoUpdate DH groups
dtucker [Fri, 22 May 2015 02:45:42 +0000 (02:45 +0000)]
Update DH groups

9 years agoRemove 6k and 8k bit moduli fragments since they are now kept in
dtucker [Fri, 22 May 2015 02:43:59 +0000 (02:43 +0000)]
Remove 6k and 8k bit moduli fragments since they are now kept in
usr.bin/ssh/moduli-gen.

9 years agoUpdate DH groups
dtucker [Fri, 22 May 2015 02:34:53 +0000 (02:34 +0000)]
Update DH groups

9 years agosync
deraadt [Fri, 22 May 2015 01:48:21 +0000 (01:48 +0000)]
sync

9 years agosync
deraadt [Fri, 22 May 2015 01:46:31 +0000 (01:46 +0000)]
sync

9 years agofix a non safe use of TAILQ_FOREACH with TAILQ_REMOVE
jsg [Fri, 22 May 2015 01:34:13 +0000 (01:34 +0000)]
fix a non safe use of TAILQ_FOREACH with TAILQ_REMOVE
ok reyk@

9 years agofix a non safe use of LIST_FOREACH with LIST_REMOVE
jsg [Fri, 22 May 2015 01:30:27 +0000 (01:30 +0000)]
fix a non safe use of LIST_FOREACH with LIST_REMOVE
ok claudio@ kettenis@ reyk@

9 years agoUse m_defrag(9) instead of rolling our own version of it.
kettenis [Thu, 21 May 2015 22:13:55 +0000 (22:13 +0000)]
Use m_defrag(9) instead of rolling our own version of it.

ok jca@

9 years agosync
deraadt [Thu, 21 May 2015 22:00:36 +0000 (22:00 +0000)]
sync

9 years agoEstablish interrupts for both keyboard and mouse slots at pckbc attach time,
miod [Thu, 21 May 2015 19:32:29 +0000 (19:32 +0000)]
Establish interrupts for both keyboard and mouse slots at pckbc attach time,
rather than lazily from pckbc when slots are discovered. This is consistent
with what other isa devices (and pckbc on non-isa busses) do, and as a side
effect, this makes the dmesg output shorter.

This will also let us get rid of pckbc's intr_establish() callback in a
later diff.

Prompted by krw@ noticing ugly kernel output in a configuration with the
mouse slot left empty. ok krw@ mpi@

9 years agoReport all valid interrupt locators in isaprint() - although config(8) stanzas
miod [Thu, 21 May 2015 19:29:31 +0000 (19:29 +0000)]
Report all valid interrupt locators in isaprint() - although config(8) stanzas
only allow one irq for isa devices, there is actually support for more since
we got isapnp(4) support, and upcoming changes will actually have regular isa(4)
devices claim more than one irq in their indirect match function.

9 years agono such thing as mips64le
miod [Thu, 21 May 2015 19:26:34 +0000 (19:26 +0000)]
no such thing as mips64le

9 years agoSwitch amd64, hppa, mips64, mips64le and powerpc to binutils 2.17.
kettenis [Thu, 21 May 2015 19:13:59 +0000 (19:13 +0000)]
Switch amd64, hppa, mips64, mips64le and powerpc to binutils 2.17.

ok deraadt@

9 years agoNo need to call tzset() and log_init() in the forked constraint
reyk [Thu, 21 May 2015 14:24:43 +0000 (14:24 +0000)]
No need to call tzset() and log_init() in the forked constraint
handler.  It is run in a chroot, so tzset() wouldn't even succeed to
open the zone file.  Found with tame.

OK deraadt@

9 years agoRename caddr_t p to cp in an inner block to avoid aliasing the outer
nicm [Thu, 21 May 2015 13:35:15 +0000 (13:35 +0000)]
Rename caddr_t p to cp in an inner block to avoid aliasing the outer
struct proc *p, ok deraadt

9 years agoSupport "ssh-keygen -lF hostname" to find search known_hosts and
djm [Thu, 21 May 2015 12:01:19 +0000 (12:01 +0000)]
Support "ssh-keygen -lF hostname" to find search known_hosts and
print key hashes. Already advertised by ssh-keygen(1), but not
delivered by code; ok dtucker@

9 years agosome fixes from pjanzen;
jmc [Thu, 21 May 2015 10:42:30 +0000 (10:42 +0000)]
some fixes from pjanzen;

9 years agoNo need for ifp since we do not set "rcvif".
mpi [Thu, 21 May 2015 09:44:32 +0000 (09:44 +0000)]
No need for ifp since we do not set "rcvif".

9 years agoNo need to set "rcvif", if_input() does it for you.
mpi [Thu, 21 May 2015 09:36:20 +0000 (09:36 +0000)]
No need to set "rcvif", if_input() does it for you.

9 years agotedu commented out xl_testpacket(), remove one of the IFQ_ENQUEUE()
mpi [Thu, 21 May 2015 09:25:18 +0000 (09:25 +0000)]
tedu commented out xl_testpacket(), remove one of the IFQ_ENQUEUE()
in the tree.

9 years agoConvert to if_output().
mpi [Thu, 21 May 2015 09:22:39 +0000 (09:22 +0000)]
Convert to if_output().

9 years agoCorrectly state the link state to INVALID when creating a carp interface.
mpi [Thu, 21 May 2015 09:17:53 +0000 (09:17 +0000)]
Correctly state the link state to INVALID when creating a carp interface.

Since vhe are allocated with M_ZERO and INIT is also defined to be 0,
carp_set_state() would result in a no-op because of the state check.

So explicitly initialize the state of a vhe to INIT and move the state
check in carp_set_state_all() to prevent similar issues in the future.

Problem and initial diff from Johan Ymerson, thanks!

ok henning@

9 years agoAccess to uninitialized variable fixed.
gerhard [Thu, 21 May 2015 07:39:52 +0000 (07:39 +0000)]
Access to uninitialized variable fixed.

ok mikeb@

9 years agoregress test for AuthorizedPrincipalsCommand
djm [Thu, 21 May 2015 06:44:25 +0000 (06:44 +0000)]
regress test for AuthorizedPrincipalsCommand

9 years agoadd AuthorizedPrincipalsCommand that allows getting authorized_principals
djm [Thu, 21 May 2015 06:43:30 +0000 (06:43 +0000)]
add AuthorizedPrincipalsCommand that allows getting authorized_principals
from a subprocess rather than a file, which is quite useful in
deployments with large userbases

feedback and ok markus@

9 years agoregress test for AuthorizedKeysCommand arguments
djm [Thu, 21 May 2015 06:40:02 +0000 (06:40 +0000)]
regress test for AuthorizedKeysCommand arguments

9 years agosupport arguments to AuthorizedKeysCommand
djm [Thu, 21 May 2015 06:38:35 +0000 (06:38 +0000)]
support arguments to AuthorizedKeysCommand

bz#2081 loosely based on patch by Sami Hartikainen
feedback and ok markus@

9 years agorefactor: split base64 encoding of pubkey into its own
djm [Thu, 21 May 2015 04:55:51 +0000 (04:55 +0000)]
refactor: split base64 encoding of pubkey into its own
sshkey_to_base64() function and out of sshkey_write();
ok markus@

9 years agoRe-remove extra perl utils, patch lost in 5.20.2 update
afresh1 [Thu, 21 May 2015 03:58:09 +0000 (03:58 +0000)]
Re-remove extra perl utils, patch lost in 5.20.2 update

pointed out by miod@

9 years agoRemove clauses 3 and 4 from Christos Zoulas' BSD license.
schwarze [Wed, 20 May 2015 23:39:55 +0000 (23:39 +0000)]
Remove clauses 3 and 4 from Christos Zoulas' BSD license.
This is safe because Christos did that himself in NetBSD in 2008.
No code change.

9 years agoRemove function argument name from posix_spawnattr_getsigmask()
millert [Wed, 20 May 2015 22:50:07 +0000 (22:50 +0000)]
Remove function argument name from posix_spawnattr_getsigmask()
prototype to match other prototypes in the file.  OK guenther@ deraadt@

9 years agoFix sign compare bug introduced when rnum() was redefined to use
millert [Wed, 20 May 2015 20:26:00 +0000 (20:26 +0000)]
Fix sign compare bug introduced when rnum() was redefined to use
arc4random_uniform().  From pjanzen@, OK deraadt@

9 years agoMerge the get_drive() function with install_disk(), which is the
rpe [Wed, 20 May 2015 19:14:35 +0000 (19:14 +0000)]
Merge the get_drive() function with install_disk(), which is the
only remaining consumer.

OK krw@

9 years agoSigned types are bad array indicies - let it panic instead.
pelikan [Wed, 20 May 2015 15:21:57 +0000 (15:21 +0000)]
Signed types are bad array indicies - let it panic instead.

ok deraadt krw millert

9 years agoscrap unused ixgbe_get_link_capabilities_X540
mikeb [Wed, 20 May 2015 14:34:27 +0000 (14:34 +0000)]
scrap unused ixgbe_get_link_capabilities_X540

9 years agoRemove hotplug(4) sensor support: the code has been disabled by
reyk [Wed, 20 May 2015 13:32:39 +0000 (13:32 +0000)]
Remove hotplug(4) sensor support: the code has been disabled by
henning@ 9 years ago because of an issue with the /dev/hotplug device
- it does not support multiple readers opening it.  Nobody ever cared
enough to fix it so it is time to sent the dead code to the Attic.

OK henning@ (feeling sad about it), mpi@ and others

9 years agoUse off_t instead of size_t to pass file size and print it using %lld when
kettenis [Wed, 20 May 2015 09:28:47 +0000 (09:28 +0000)]
Use off_t instead of size_t to pass file size and print it using %lld when
constructing the Content-Length header field.  Should fix some, but probably
not all, problems with serving files bigger than 2G on 32-bit architectures.

ok reyk@, florian@

9 years agoKeep track of the ifih corresponding to a vlan instance to ease its
mpi [Wed, 20 May 2015 08:54:37 +0000 (08:54 +0000)]
Keep track of the ifih corresponding to a vlan instance to ease its
removal.

As soon as carp(4) will be converted to the new if_input() API it
will be possible to add multiple vlan(4) and carp(4) pseudo-ifps on
top of the same parent interface.  When such thing happens we can no
longer assume that the first pseudo-ifp to be destroyed will be the
last configured.

ok dlg@

9 years agoDo not increment if_opackets in if_output(). It might make sense to do
mpi [Wed, 20 May 2015 08:28:54 +0000 (08:28 +0000)]
Do not increment if_opackets in if_output().  It might make sense to do
that later but all drivers should be adapated.

Should fix a double output packet accounting, reported by Hrvoje Popovski.

9 years agoReturn empty string if format is empty rather than attempting to
nicm [Wed, 20 May 2015 06:39:02 +0000 (06:39 +0000)]
Return empty string if format is empty rather than attempting to
allocate zero bytes.

9 years agoNo need to check the return value of memcpy() if you actually checked this
miod [Wed, 20 May 2015 04:33:35 +0000 (04:33 +0000)]
No need to check the return value of memcpy() if you actually checked this
pointer for NULL the line above; ok doug@

9 years agoRemove cubieboard specific gpio led setting.
jsg [Wed, 20 May 2015 03:49:23 +0000 (03:49 +0000)]
Remove cubieboard specific gpio led setting.
From Artturi Alm in bitrig.

9 years agoNow all the socs use the same va entry point and don't have any
jsg [Wed, 20 May 2015 01:44:20 +0000 (01:44 +0000)]
Now all the socs use the same va entry point and don't have any
conflicting symbols we can combine the configs.

Multiple umg files are still required however.  The bsd.umg target in
the kernel is replaced by targets for bsd.IMX.umg, bsd.OMAP.umg and
bsd.SUNXI.umg.

9 years agoARM L2C driver is only relevant on Cortex-A9 machines.
jsg [Wed, 20 May 2015 00:39:16 +0000 (00:39 +0000)]
ARM L2C driver is only relevant on Cortex-A9 machines.
From Patrick Wildt in bitrig.

9 years agoadd per soc match functions instead of using armv7_match
jsg [Wed, 20 May 2015 00:14:55 +0000 (00:14 +0000)]
add per soc match functions instead of using armv7_match

9 years agoInstead of testing for __ELF__ and/or vax, leave out the bits for interfacing
guenther [Tue, 19 May 2015 20:50:06 +0000 (20:50 +0000)]
Instead of testing for __ELF__ and/or vax, leave out the bits for interfacing
with ld.so locking whenever building NOPIC

pointless use of __ELF__ noted by brad@
ok miod@

9 years agoOnly attempt to load /etc/random.seed from the boot device after the kernel
miod [Tue, 19 May 2015 20:42:11 +0000 (20:42 +0000)]
Only attempt to load /etc/random.seed from the boot device after the kernel
image has been succesfully loaded (with the recent loadfile changes allowing
us to know where the randomness needs to be loaded). While there, don't
bother doing this when booting from tape.

This works around the sun4e PROM 1.6, which gets confused by PROM open() -
close() sequences without any I/O happening in between.

Crank boot blocks version to 2.11.

9 years agoExtend the libsa loadfile(9) granularity to tell apart randomness from the rest
miod [Tue, 19 May 2015 20:39:12 +0000 (20:39 +0000)]
Extend the libsa loadfile(9) granularity to tell apart randomness from the rest
of the kernel, and extend the array filled by loadfile to report the location
of the randomness area.

This doesn't introduce any change for bootblocks (save for a slightly larger
stack usage due to the larger array), for the new {LOAD,COUNT}_RANDOM bits
are included in the {LOAD,COUNT}_ALL masks everything uses or computes from.

9 years agoMove acquisition of the kernel lock deeper in the interrupt path, and make
miod [Tue, 19 May 2015 20:28:14 +0000 (20:28 +0000)]
Move acquisition of the kernel lock deeper in the interrupt path, and make
sure clock interrupts do not attempt to acquire it.
This will also eventually allow for IPL_MPSAFE interrupts on alpha.

Tested by dlg@ and I.