openbsd
6 years agoAdd kernel support for the VFP FPU/SIMD unit. Based on a diff by drahn@.
kettenis [Fri, 26 Jan 2018 22:25:23 +0000 (22:25 +0000)]
Add kernel support for the VFP FPU/SIMD unit.  Based on a diff by drahn@.
This allows us to use floating-pointer and vector instructions in userland
code.  The current implementation assumes all 32 VFP registers are present.
This should be the case on all armv7 hardware currently supported by
OpenBSD.

ok patrick@

6 years ago- do not junk pages returned by free_bytes(), all freed chunks are already
otto [Fri, 26 Jan 2018 19:14:51 +0000 (19:14 +0000)]
- do not junk pages returned by free_bytes(), all freed chunks are already
  junked
- freezero(): only clear requested size

6 years agoHandle the transfer size option when faking up /etc/random.seed.
naddy [Fri, 26 Jan 2018 16:40:14 +0000 (16:40 +0000)]
Handle the transfer size option when faking up /etc/random.seed.
Without this, clients that queried the size would receive 0 as response.
ok jca@

6 years agoAdd kernel support for the VFP FPU/SIMD unit. Based on a diff by drahn@.
kettenis [Fri, 26 Jan 2018 16:22:19 +0000 (16:22 +0000)]
Add kernel support for the VFP FPU/SIMD unit.  Based on a diff by drahn@.
This allows us to use floating-pointer and vector instructions in userland
code.  The current implementation assumes all 32 VFP registers are present.
This should be the case on all armv7 hardware currently supported by
OpenBSD.

ok patrick@

6 years agoCleanup VFP code.
kettenis [Fri, 26 Jan 2018 16:15:26 +0000 (16:15 +0000)]
Cleanup VFP code.

ok patrick@

6 years agoDocument MODCMAKE_DEBUG
jca [Fri, 26 Jan 2018 13:12:08 +0000 (13:12 +0000)]
Document MODCMAKE_DEBUG

ok landry@

6 years agoallow unpriviledged user to run "encrypt" and "spf walk"
eric [Fri, 26 Jan 2018 08:00:54 +0000 (08:00 +0000)]
allow unpriviledged user to run "encrypt" and "spf walk"

ok todd@, gilles@, sunil@

6 years agoFix Failure on Start
ccardenas [Fri, 26 Jan 2018 05:34:41 +0000 (05:34 +0000)]
Fix Failure on Start

When attempting to start a vm from vm.conf that fails due to inadequate
resources and such, do not remote vm entry from vm list.

Reported by mpi@

ok mpi@

6 years agoAdd DPRINTF() and use it to replace log_debug() statements inside
krw [Thu, 25 Jan 2018 15:43:51 +0000 (15:43 +0000)]
Add DPRINTF() and use it to replace log_debug() statements inside
#ifdef DEBUG/#endif blocks.

Suggestions on correct idiom (C99 vs gcc) from millert@

ok tom@

6 years agoRemove mutex implementations that now live in MI code.
mpi [Thu, 25 Jan 2018 15:19:33 +0000 (15:19 +0000)]
Remove mutex implementations that now live in MI code.

6 years agoMove common mutex implementations to a MI place.
mpi [Thu, 25 Jan 2018 15:06:29 +0000 (15:06 +0000)]
Move common mutex implementations to a MI place.

Archs not yet converted can to the jump by defining __USE_MI_MUTEX.

ok visa@

6 years agoUse a workaround for detached parent in carp_proto_input_c().
mpi [Thu, 25 Jan 2018 14:47:35 +0000 (14:47 +0000)]
Use a workaround for detached parent in carp_proto_input_c().

A NULL dereference can happen since processing protocol layer is
deffered to a second task.  In other words the NET_LOCK() is released
then regrabbed between ip_input() and carp_proto_input().

The same workaround is already in use in carp_output() due to deffered
processing in case of IPsec.

The real fix is to make carp(4) MP-safe and use if_get(9) there, any
taker?

Found & fix tested by Hrvoje Popovski.

6 years agoUse a new LDFLAGS variable to pass "-melf_i386_obsd" on amd64.
mpi [Thu, 25 Jan 2018 14:09:52 +0000 (14:09 +0000)]
Use a new LDFLAGS variable to pass "-melf_i386_obsd" on amd64.

Makes the linker generate a correct i386 gap.o file.

ok kettenis@, deraadt@

6 years agoAssert that ifiq_destroy() is not called with the NET_LOCK() held.
mpi [Thu, 25 Jan 2018 14:04:36 +0000 (14:04 +0000)]
Assert that ifiq_destroy() is not called with the NET_LOCK() held.

Calling taskq_barrier() on a softnet thread while holding the lock
is clearly a deadlock.

ok visa@, dlg@, bluhm@

6 years agocertificate options are case-sensitive; fix case on one that had it wrong.
djm [Thu, 25 Jan 2018 03:34:43 +0000 (03:34 +0000)]
certificate options are case-sensitive; fix case on one that had it wrong.

move a badly-place sentence to a less bad place

6 years agoFix perl build not to install libperl* twice. This has caused an
yasuoka [Thu, 25 Jan 2018 02:47:00 +0000 (02:47 +0000)]
Fix perl build not to install libperl* twice.  This has caused an
actual error if "install" was done parallelly.

ok espie todd andrew1, input andrew1

6 years agoMake the NEON codepaths conditional on __STRICT_ALIGNMENT not being
kettenis [Wed, 24 Jan 2018 23:03:37 +0000 (23:03 +0000)]
Make the NEON codepaths conditional on __STRICT_ALIGNMENT not being
defined as they rely on unaligned access.

ok joel@

6 years agoDon't display the seconds until renewal. i.e. "em0: bound to 1.2.3.4
krw [Wed, 24 Jan 2018 19:12:49 +0000 (19:12 +0000)]
Don't display the seconds until renewal. i.e. "em0: bound to 1.2.3.4
-- renewal in 300000 seconds" becomes simply "em0: bound to 1.2.3.4"

While here avoid setting timeouts before the current time.

ok florian@

6 years agoImplement support for specifying multiple proposals. This means we can
patrick [Wed, 24 Jan 2018 17:01:52 +0000 (17:01 +0000)]
Implement support for specifying multiple proposals.  This means we can
have a higher flexibility in negotiating with other peers, or even ease
migration from one proposal to a more secure one.

ok sthen@

6 years agoadd the glue to generate future libset, which are currently 100% unused
espie [Wed, 24 Jan 2018 16:52:44 +0000 (16:52 +0000)]
add the glue to generate future libset, which are currently 100% unused

6 years agoPOSIX requires that awk support \v and \a escapes. I used '\007'
millert [Wed, 24 Jan 2018 16:28:25 +0000 (16:28 +0000)]
POSIX requires that awk support \v and \a escapes.  I used '\007'
for BEL since that is what lex.c uses, though we could safely use
'\a' there instead.  OK martijn@

6 years agoLog some more errors in the ca code, since there seems to be still a bug
claudio [Wed, 24 Jan 2018 13:51:36 +0000 (13:51 +0000)]
Log some more errors in the ca code, since there seems to be still a bug
hidden somewhere. Also return after a poll timeout, there is no reason to
wait longer than a second for the answer of the ca process.
OK jsing@

6 years agoUse closefrom(3) instead of manually closing all file descriptors
tb [Wed, 24 Jan 2018 13:25:25 +0000 (13:25 +0000)]
Use closefrom(3) instead of manually closing all file descriptors
between 3 and 19.

ok martijn, millert, jca

6 years agoDo not print that the firmware was successfully loaded in the USB backend
patrick [Wed, 24 Jan 2018 13:10:20 +0000 (13:10 +0000)]
Do not print that the firmware was successfully loaded in the USB backend
driver.  It is only useful to know if firmware loading has failed.  This
means fewer clutter in dmesg(8).

Spotted by and ok stsp@

6 years agoPut the default case, which prints information about an event we do not yet
patrick [Wed, 24 Jan 2018 13:07:31 +0000 (13:07 +0000)]
Put the default case, which prints information about an event we do not yet
handle, into a debug printf.  Also do not print a pointer to kernel buffers
so that we don't leak kernel addresses.

Spotted by and ok stsp@

6 years agoWe don't do booleans in the kernel, so replace a few occurrences of 'true'
patrick [Wed, 24 Jan 2018 13:04:22 +0000 (13:04 +0000)]
We don't do booleans in the kernel, so replace a few occurrences of 'true'
with nicer integer equivalents.

Spotted by and ok stsp@

6 years agodocument bpfsattach and bpfsdetach
dlg [Wed, 24 Jan 2018 00:33:45 +0000 (00:33 +0000)]
document bpfsattach and bpfsdetach

6 years agoadd support for bpf on "subsystems", not just network interfaces
dlg [Wed, 24 Jan 2018 00:25:17 +0000 (00:25 +0000)]
add support for bpf on "subsystems", not just network interfaces

bpf assumed that it was being unconditionally attached to network
interfaces, and maintained a pointer to a struct ifnet *. this was
mostly used to get at the name of the interface, which is how
userland asks to be attached to a particular interface. this diff
adds a pointer to the name and uses it instead of the interface
pointer for these lookups. this in turn allows bpf to be attached
to arbitrary subsystems in the kernel which just have to supply a
name rather than an interface pointer. for example, bpf could be
attached to pf_test so you can see what packets are about to be
filtered. mpi@ is using this to look at usb transfers.

bpf still uses the interface pointer for bpfwrite, and for enabling
and disabling promisc. however, these are nopped out for subsystems.

ok mpi@

6 years agoAdd a smarter test that opens a file, unlinks it, and remounts the
bluhm [Tue, 23 Jan 2018 22:30:34 +0000 (22:30 +0000)]
Add a smarter test that opens a file, unlinks it, and remounts the
file system read-only.  After unmounting this should result in a
dirty file system that is not marked clean.  Disable both tests
with dangling vnodes for now as they fail.

6 years agoThe IPv6 gateway address passed to rt6_flush() comes either from a
bluhm [Tue, 23 Jan 2018 22:06:42 +0000 (22:06 +0000)]
The IPv6 gateway address passed to rt6_flush() comes either from a
packet through ip6_input() or from the routing table.  In both cases
the KAME hack has added the embeded scope to the address, so it is
not necessary to fill the scope id again in rt6_flush().  Assert
that it is already there.
OK mpi@

6 years agoUpdate to tzdata2018c from ftp.iana.org
millert [Tue, 23 Jan 2018 21:54:04 +0000 (21:54 +0000)]
Update to tzdata2018c from ftp.iana.org

6 years agoThe TCP reaper timeout was still imlemented as soft timeout. So
bluhm [Tue, 23 Jan 2018 21:41:17 +0000 (21:41 +0000)]
The TCP reaper timeout was still imlemented as soft timeout.  So
it could run immediately and was not synchronized with the TCP
timeouts, although that was the intension when it was introduced
in revision 1.85.  Convert the reaper to an ordinary TCP timeout
so it is scheduled on the same timeout thread after all timeouts
have finished.  A net lock is not necessary as the process calling
tcp_close() will not access the tcpcb after arming the reaper
timeout.
OK mikeb@

6 years agoAlthough it is a dead store here, always reassign the tcpcb after
bluhm [Tue, 23 Jan 2018 21:06:47 +0000 (21:06 +0000)]
Although it is a dead store here, always reassign the tcpcb after
calling tcp_close(), tcp_drop(), and tcp_disconnect().  Then no
freed memory can be used after closing a TCP connection.
OK mikeb@ mpi@

6 years agoBring back the PRU_ATTACH constant and description. Then trpt(8)
bluhm [Tue, 23 Jan 2018 20:49:58 +0000 (20:49 +0000)]
Bring back the PRU_ATTACH constant and description.  Then trpt(8)
prints the ATTACH action for TCP debugging socketes correctly.
OK bluhm@

6 years agoFix printf(9) format strings so that a kernel with TCPDEBUG compiles.
bluhm [Tue, 23 Jan 2018 20:41:42 +0000 (20:41 +0000)]
Fix printf(9) format strings so that a kernel with TCPDEBUG compiles.
Mark the string variables with TCP debugging names as constant.
OK mpi@

6 years agoFix a logic bug in sshd_exchange_identification which prevented clients
stsp [Tue, 23 Jan 2018 20:00:58 +0000 (20:00 +0000)]
Fix a logic bug in sshd_exchange_identification which prevented clients
using major protocol version 2 from connecting to the server.
ok millert@

6 years agoAdd missing braces; fixes 'write: Socket is not connected' error in ssh.
stsp [Tue, 23 Jan 2018 18:33:49 +0000 (18:33 +0000)]
Add missing braces; fixes 'write: Socket is not connected' error in ssh.
ok deraadt@

6 years agouse END_STRONG. Yes, everything in the sh ecosystem is different..
deraadt [Tue, 23 Jan 2018 17:11:02 +0000 (17:11 +0000)]
use END_STRONG.  Yes, everything in the sh ecosystem is different..

6 years agoAlways lock the kernel lock with non-MP-safe interrupt handlers.
visa [Tue, 23 Jan 2018 14:47:21 +0000 (14:47 +0000)]
Always lock the kernel lock with non-MP-safe interrupt handlers.

Suggested by kettenis@

6 years agoRevise 'struct fpreg' such that it can actually represent the full VFPv3-D32
kettenis [Tue, 23 Jan 2018 12:38:14 +0000 (12:38 +0000)]
Revise 'struct fpreg' such that it can actually represent the full VFPv3-D32
state.

ok patrick@

6 years agoFix underline rotation on CCW (quarter counter-clockwise) rotated
fcambus [Tue, 23 Jan 2018 10:10:32 +0000 (10:10 +0000)]
Fix underline rotation on CCW (quarter counter-clockwise) rotated
screens.

The "underline" was actually drawn above text.

OK jcs@

6 years agoFix table wrt missing element and `, detailed analysis and diff from
otto [Tue, 23 Jan 2018 07:06:55 +0000 (07:06 +0000)]
Fix table wrt missing element and `, detailed analysis and diff from
David Fifield; ok tb@

6 years agoDrop compatibility hacks for some ancient SSH implementations, including
djm [Tue, 23 Jan 2018 05:27:21 +0000 (05:27 +0000)]
Drop compatibility hacks for some ancient SSH implementations, including
ssh.com <=2.* and OpenSSH <= 3.*.

These versions were all released in or before 2001 and predate the
final SSH RFCs. The hacks in question aren't necessary for RFC-
compliant SSH implementations.

ok markus@

6 years agotry harder to preserve errno during ssh_connect_direct() to make the
djm [Tue, 23 Jan 2018 05:17:04 +0000 (05:17 +0000)]
try harder to preserve errno during ssh_connect_direct() to make the
final error message possibly accurate; bz#2814, ok dtucker@

6 years agounbreak support for clients that advertise a protocol version
djm [Tue, 23 Jan 2018 05:12:12 +0000 (05:12 +0000)]
unbreak support for clients that advertise a protocol version
of "1.99" (indicating both v2 and v1 support). Busted by me during
SSHv1 purge in r1.358; bz2810, ok dtucker

6 years agodon't attempt to force hostnames that are addresses to lowercase, but
djm [Tue, 23 Jan 2018 05:06:25 +0000 (05:06 +0000)]
don't attempt to force hostnames that are addresses to lowercase, but
instead canonicalise them through getnameinfo/getaddrinfo to remove
ambiguities (e.g. ::0001 => ::1) before they are matched against
known_hosts; bz#2763, ok dtucker@

6 years agoavoid modifying pw->pw_passwd; let endpwent() clean up for us, but
djm [Tue, 23 Jan 2018 05:01:15 +0000 (05:01 +0000)]
avoid modifying pw->pw_passwd; let endpwent() clean up for us, but
keep a scrubbed copy; bz2777, ok dtucker@

6 years agoFix checks for error return from urtw_alloc_rx_data_list() and
kevlo [Tue, 23 Jan 2018 02:53:26 +0000 (02:53 +0000)]
Fix checks for error return from urtw_alloc_rx_data_list() and
urtw_alloc_tx_data_list() in 8187b init.

From James Jerkins, thanks.

6 years ago1. Add a field for configuring the size of edge areas on touchpads.
bru [Mon, 22 Jan 2018 22:14:11 +0000 (22:14 +0000)]
1. Add a field for configuring the size of edge areas on touchpads.
2. Remove code that is obsolete because all touchpad drivers configure
wsmouse for compat mode now.

6 years agoAdd support for 'a:' and 'exists:' mechanisms.
okan [Mon, 22 Jan 2018 20:56:47 +0000 (20:56 +0000)]
Add support for 'a:' and 'exists:' mechanisms.

ok gilles@

6 years agoBring back the detach messages in trpt(8) for tcp debugging. They
bluhm [Mon, 22 Jan 2018 20:27:28 +0000 (20:27 +0000)]
Bring back the detach messages in trpt(8) for tcp debugging.  They
got lost with the pr_detach conversion.  While there, remove some
dead code.
OK mpi@

6 years agoTriggered by Klemens Nanni patch, move OLD_WRKDIR_NAME into OBSOLETE VARIABLES
espie [Mon, 22 Jan 2018 13:18:37 +0000 (13:18 +0000)]
Triggered by Klemens Nanni patch, move OLD_WRKDIR_NAME into OBSOLETE VARIABLES
and deprecate WRKDIR_LINKNAME.

6 years agoUse the proper address space when using a kernel that supports
helg [Mon, 22 Jan 2018 13:16:48 +0000 (13:16 +0000)]
Use the proper address space when using a kernel that supports
Supervisor Mode Access Prevention (SMAP) compiled with FUSE_DEBUG.
While here, add some sanity check to the debug function fuse_dump_buff.

Thanks to Thomas Jeunet.

OK mpi@, helg@

6 years agoMore sizes for free() in wscons(4).
fcambus [Mon, 22 Jan 2018 12:20:54 +0000 (12:20 +0000)]
More sizes for free() in wscons(4).

OK visa@

6 years ago<machine/mutex.h> -> <sys/mutex.h>
mpi [Mon, 22 Jan 2018 10:04:46 +0000 (10:04 +0000)]
<machine/mutex.h> -> <sys/mutex.h>

6 years agoInclude <sys/mutex.h> instead of <machine/mutex.h>
mpi [Mon, 22 Jan 2018 09:40:45 +0000 (09:40 +0000)]
Include <sys/mutex.h> instead of <machine/mutex.h>

required by upcoming MI mutex change.

6 years agoInclude <sys/mutex.h> rather than <machine/mutex.h>
mpi [Mon, 22 Jan 2018 09:31:35 +0000 (09:31 +0000)]
Include <sys/mutex.h> rather than <machine/mutex.h>

Required by upcoming MI mutex change.

6 years agoAlways grab the KERNEL_LOCK() for !IPL_MPSAFE handlers.
mpi [Mon, 22 Jan 2018 09:08:43 +0000 (09:08 +0000)]
Always grab the KERNEL_LOCK() for !IPL_MPSAFE handlers.

Suggested by kettenis@, ok visa@

6 years agoUse ip_send() to not call ip_output() recursively in *start() routine.
mpi [Mon, 22 Jan 2018 09:06:22 +0000 (09:06 +0000)]
Use ip_send() to not call ip_output() recursively in *start() routine.

ok visa@

6 years agoDo not call ip_output() recursively in vxlan_start().
mpi [Mon, 22 Jan 2018 09:05:06 +0000 (09:05 +0000)]
Do not call ip_output() recursively in vxlan_start().

ip_output() currently needs the NET_LOCK() which is not held when an
interface is added to a pf(4) queue.

Fix an issue reported by Jason Tubnor, ok dlg@

6 years agoAdd SHT_SYMTAB_SHNDX
mpi [Mon, 22 Jan 2018 09:01:39 +0000 (09:01 +0000)]
Add SHT_SYMTAB_SHNDX

Required by Karel Gardas to unbreak GHC head.

6 years agoregen
jsg [Mon, 22 Jan 2018 03:14:33 +0000 (03:14 +0000)]
regen

6 years agoadd clang-local(1) to SEE ALSO
jsg [Mon, 22 Jan 2018 03:13:09 +0000 (03:13 +0000)]
add clang-local(1) to SEE ALSO
suggested by jmc@ ok patrick@

6 years agoMatch on Iris Pro Graphics 580 and other skylake GT4 ids and correct
jsg [Mon, 22 Jan 2018 02:28:09 +0000 (02:28 +0000)]
Match on Iris Pro Graphics 580 and other skylake GT4 ids and correct
the Iris Graphics P555 GT3 id.

Derived from the following linux commits:

drm/i915/skl: Add SKL GT4 PCI IDs
From Mika Kuoppala
15620206ae87ba9643ffa6f5ddb5471be7192006 in mainline linux

drm/i915: Update SKL SRV GT4 pci ids reference.
From Rodrigo Vivi
5390974f981907cd02d1cb48145b03471279877a in mainline linux

drm/i915/skl: Add missing SKL ID
From Michal Winiarski
ca7a45ba6fb9e7ceca56d10b91db29c2f3451a2e in mainline linux

Expanded version of a diff from ccardenas@.
Reported and tested by Bryan Vyhmeister.
Tested by and ok ccardenas@

6 years agoMove some code back from locore0.S to locore.S that we will need to bring up
kettenis [Sun, 21 Jan 2018 22:30:41 +0000 (22:30 +0000)]
Move some code back from locore0.S to locore.S that we will need to bring up
secondary CPUs.  This involves adjusting get_virt_delta() to account for the
fact that it may no longer live near the start of kernel code.  Heavily based
on a diff from drahn@.

ok patrick@, deraadt@

6 years agoImplement ffs(3) using the CLZ instructions which has been available ever
kettenis [Sun, 21 Jan 2018 21:56:02 +0000 (21:56 +0000)]
Implement ffs(3) using the CLZ instructions which has been available ever
since ARMv5.  Should be much faster but more importantly it removes the
data table from .text which could introduce unwanted ROP gadgets.

Based on changes in Android/Bionic by Elliott Hughes.

ok patrick@

6 years agoAdd missing OpenBSD CVS tag
patrick [Sun, 21 Jan 2018 21:37:01 +0000 (21:37 +0000)]
Add missing OpenBSD CVS tag

6 years agoImplement network boot support in efiboot(8). This changes efiboot(8)
patrick [Sun, 21 Jan 2018 21:35:34 +0000 (21:35 +0000)]
Implement network boot support in efiboot(8).  This changes efiboot(8)
to recognize if it has been booted via PXE.  The new TFTP file system
layer will then use the matching PXE base code protocol handle to load
the files.  Since this uses the PXE base code protocol for abstraction
instead of the raw Simple Network protocol this will at this point not
work on u-boot based machines.

ok kettenis@

6 years agoadd new '-c' option which runs a command instead of an interactive shell
jasper [Sun, 21 Jan 2018 20:18:20 +0000 (20:18 +0000)]
add new '-c' option which runs a command instead of an interactive shell

from Paul de Weerd
ok ian@ job@; manpage bits are ok jmc@

6 years agofix error handling while fetching sets
halex [Sun, 21 Jan 2018 19:58:37 +0000 (19:58 +0000)]
fix error handling while fetching sets

reported by naddy@ long ago
ok rpe@

6 years agoRelease interrupt source if the interrupt task is already scheduled
mikeb [Sun, 21 Jan 2018 18:54:46 +0000 (18:54 +0000)]
Release interrupt source if the interrupt task is already scheduled

Tested by Imre Oolberg, thanks!

6 years agoImprove panic message
mikeb [Sun, 21 Jan 2018 18:51:19 +0000 (18:51 +0000)]
Improve panic message

6 years agoadd arch to Dt;
jmc [Sun, 21 Jan 2018 16:00:50 +0000 (16:00 +0000)]
add arch to Dt;

6 years agoremove the ref to gcc-local. the alternative was to add clang-local,
jmc [Sun, 21 Jan 2018 10:28:37 +0000 (10:28 +0000)]
remove the ref to gcc-local. the alternative was to add clang-local,
but i think the ref to cc should suffice for this page;

6 years agouse virt instructions with clang instead of equivalent raw bytes
jsg [Sat, 20 Jan 2018 23:57:44 +0000 (23:57 +0000)]
use virt instructions with clang instead of equivalent raw bytes

6 years agoProvide a simpler example ifstated.conf.
mpf [Sat, 20 Jan 2018 23:07:58 +0000 (23:07 +0000)]
Provide a simpler example ifstated.conf.
The existing example was written over a decade ago, when carp(4) was
not supporting carpdemote and did not track it's interfaces link state itself.
OK jmc@, rob@

6 years agoImprove the config example.
mpf [Sat, 20 Jan 2018 22:57:11 +0000 (22:57 +0000)]
Improve the config example.
Suggest using carpdemote instead of tweaking the advskew setting.
Also add a check for a non-carp link state.
Prodded by jmc@. OK jmc@, rob@.

6 years agoDisable TCP and UDP transmit checksum offloading
mikeb [Sat, 20 Jan 2018 20:03:45 +0000 (20:03 +0000)]
Disable TCP and UDP transmit checksum offloading

In certain configurations, transmit checksum offloading doesn't
appear to work correctly, preventing correct TCP and UDP operation.
Some of these issues involving VLAN tagging are documented by the
Xen project.

Problem reported and fix tested by Imre Oolberg, thanks!

6 years agoFix uninitialized variable.
rob [Sat, 20 Jan 2018 18:49:12 +0000 (18:49 +0000)]
Fix uninitialized variable.

Ok jca@

6 years agoRe-enable the code that unmaps the startup code once more now that lld is
kettenis [Sat, 20 Jan 2018 18:35:41 +0000 (18:35 +0000)]
Re-enable the code that unmaps the startup code once more now that lld is
fixed.

6 years agosync
deraadt [Sat, 20 Jan 2018 18:11:31 +0000 (18:11 +0000)]
sync

6 years agoRemove code that has been disabled since 1999. From Michael W. Bombardieri who
anton [Sat, 20 Jan 2018 15:32:20 +0000 (15:32 +0000)]
Remove code that has been disabled since 1999. From Michael W. Bombardieri who
also took the time to provide an excellent explanation on why the code isn't
useful on tech@.

ok jca@

6 years agooops, no chomp, so get rid of the extra blank line by using print
espie [Sat, 20 Jan 2018 12:01:56 +0000 (12:01 +0000)]
oops, no chomp, so get rid of the extra blank line by using print

6 years agodwiic: cleanup function prototypes after ACPI/PCI split
jcs [Fri, 19 Jan 2018 18:20:38 +0000 (18:20 +0000)]
dwiic: cleanup function prototypes after ACPI/PCI split

6 years agosync
kettenis [Fri, 19 Jan 2018 16:55:46 +0000 (16:55 +0000)]
sync

6 years agoBuild clang on sparc64.
kettenis [Fri, 19 Jan 2018 16:54:54 +0000 (16:54 +0000)]
Build clang on sparc64.

ok deraadt@

6 years agoSwitch the default compiler on armv7 to clang.
kettenis [Fri, 19 Jan 2018 16:17:36 +0000 (16:17 +0000)]
Switch the default compiler on armv7 to clang.

ok patrick@

6 years agoUse static inline such that the functions are guaranteed to be present when
kettenis [Fri, 19 Jan 2018 16:16:09 +0000 (16:16 +0000)]
Use static inline such that the functions are guaranteed to be present when
compiled with clang.

ok visa@

6 years agoIn pfioctl() a pf unlock was missing in the error path.
bluhm [Fri, 19 Jan 2018 15:33:27 +0000 (15:33 +0000)]
In pfioctl() a pf unlock was missing in the error path.
OK lteo@ sashan@

6 years agoVMD: vioscsi - fix large ISO support in Linux
ccardenas [Fri, 19 Jan 2018 14:23:52 +0000 (14:23 +0000)]
VMD: vioscsi - fix large ISO support in Linux

Check if n_blocks is less than UINT32_MAX and not the image size
in various places (READ_CAPACITY_* and GET_CONFIGURATION).

ok mlarkin@ and dlg@

6 years agoUse the pf state key linking functions in two more places instead
bluhm [Fri, 19 Jan 2018 12:57:15 +0000 (12:57 +0000)]
Use the pf state key linking functions in two more places instead
of doing it manually.
OK sashan@ visa@

6 years agodisplay the actual redirect message, as that's informative.
espie [Fri, 19 Jan 2018 11:34:41 +0000 (11:34 +0000)]
display the actual redirect message, as that's informative.
won't flood the terminal, since the first redirect changes the host
discussed with sthen

6 years agoPartial revert of rev. 1.7, fixes build with clang
jca [Fri, 19 Jan 2018 11:10:43 +0000 (11:10 +0000)]
Partial revert of rev. 1.7, fixes build with clang

The error was:
--8<--
ffs.S:57:2: error: unsupported relocation on symbol
adrne r2, .L_ffs_table
^
-->8--

Minimal workaround suggested by deraadt@, ok guenther@ kettenis@

6 years ago/etc/snmpd.conf may contain passwords or other sensitive
gsoares [Fri, 19 Jan 2018 00:19:58 +0000 (00:19 +0000)]
/etc/snmpd.conf may contain passwords or other sensitive
data. diff provided by alf <a.schlichting at lemarit.com>
ok millert

6 years agoWow, it cannot even be in .data...
deraadt [Thu, 18 Jan 2018 23:42:13 +0000 (23:42 +0000)]
Wow, it cannot even be in .data...

6 years agoZap the rotor, it was a wrong idea. Cluebat applied by kshe who
otto [Thu, 18 Jan 2018 20:06:16 +0000 (20:06 +0000)]
Zap the rotor, it was a wrong idea. Cluebat applied by kshe who
came also up with this diff. Simple, no bias and benchmarks show the extra
random calls disappear in te measurement noise.

6 years agoSorry, the PIC macros cannot reach to .rodata....
deraadt [Thu, 18 Jan 2018 18:21:01 +0000 (18:21 +0000)]
Sorry, the PIC macros cannot reach to .rodata....

6 years agoRemove quirk matching all Yamaha devices, as this prevents non-MIDI
ratchov [Thu, 18 Jan 2018 18:09:38 +0000 (18:09 +0000)]
Remove quirk matching all Yamaha devices, as this prevents non-MIDI
devices from attaching (ex uaudio devices).

ok mpi

6 years agoWhile booting it does not make sense to wait for memory, there is
bluhm [Thu, 18 Jan 2018 18:08:51 +0000 (18:08 +0000)]
While booting it does not make sense to wait for memory, there is
no other process which could free it.  Better panic in malloc(9)
or pool_get(9) instead of sleeping forever.
tested by visa@ patrick@ Jan Klemkow
suggested by kettenis@;  OK deraadt@

6 years agoMake URL redirection sticky in the installer
job [Thu, 18 Jan 2018 17:23:15 +0000 (17:23 +0000)]
Make URL redirection sticky in the installer

Ensures that when URL redirection takes place, the rest of the install/upgrade
session will stick to that redirected location. This should help ensure that
the fetched set of files is coherent when a load balancer emits 301/302.

With feedback from rpe@

OK sthen@ rpe@