djm [Fri, 12 Feb 2021 03:49:09 +0000 (03:49 +0000)]
sftp: add missing lsetstat@openssh.com documentation
patch from Mike Frysinger
djm [Fri, 12 Feb 2021 03:14:18 +0000 (03:14 +0000)]
factor SSH_AGENT_CONSTRAIN_EXTENSION parsing into its own function
and remove an unused variable; ok dtucker@
patrick [Thu, 11 Feb 2021 23:55:48 +0000 (23:55 +0000)]
Call exuart(4) early attach on arm64.
ok kettenis@
patrick [Thu, 11 Feb 2021 23:54:40 +0000 (23:54 +0000)]
Enable exuart(4).
ok kettenis@
patrick [Thu, 11 Feb 2021 23:53:42 +0000 (23:53 +0000)]
Don't hardcode com(4)'s major number in exuart(4).
ok kettenis@
tobhe [Thu, 11 Feb 2021 22:02:41 +0000 (22:02 +0000)]
Explicitly unset IKED_REQ_CERTVALID before sending cert to ca process.
ok markus@
bluhm [Thu, 11 Feb 2021 21:09:56 +0000 (21:09 +0000)]
More route-to fallout in pfctl regress.
mvs [Thu, 11 Feb 2021 20:28:57 +0000 (20:28 +0000)]
sbdrop(): use NULL instead of 0 in pointer assignment
ok bluhm@
mvs [Thu, 11 Feb 2021 20:28:01 +0000 (20:28 +0000)]
We link `ifp' to `if_list' before we perform if_attachsetup(). It is not
fully initialized because we initialize `if_groups' after linking. It's
not triggered because if_attach() and if_unit(9) are serialized by
kernel lock and `ifp' is often filled by nulls. Move `if_groups'
initialization to if_attach_common() to prevent this.
ok bluhm@ claudio@ deraadt@
bluhm [Thu, 11 Feb 2021 19:41:05 +0000 (19:41 +0000)]
Link isakmpd dynamically. Mount /usr on NFS via IPsec does not
work anyway. Dynamic binaries help building errata, reduce disk
usage and make ROP harder. Also remove an unused bsd.subdir.mk
include.
OK sthen@ mvs@ deraadt@ tobhe@ patrick@
deraadt [Thu, 11 Feb 2021 17:02:39 +0000 (17:02 +0000)]
When clang was changed to -fcommon, perl's P_hash_{seed,state} variables
moved into BSS in the .o, with padding rules following the types -- they
are both char[]. Since P_hash_seed is (system-dependent) not a multiple of 8,
P_hash_state gets layed out misaligned, which sucks because the hash functions
demand 64-bit alignment for both variables. There is the possibility of using
misalignment macros, but this is not cheap. Could also use kernel-trap fault
repair, but the performance would really suck for something so crucial.
The correct fix would be for upstream to declare these types as uint64[],
we have requested that in https://github.com/Perl/perl5/issues/18555
In the meantime, carry a diff to roundup P_hash_seed to 64-bit alignment so that
P_hash_state will land aligned.
ok afresh1
stsp [Thu, 11 Feb 2021 16:22:06 +0000 (16:22 +0000)]
Add missing break in switch statement of rge_activate().
CID
1501716
ok kevlo@
and mestre@ had the same diff
visa [Thu, 11 Feb 2021 14:44:13 +0000 (14:44 +0000)]
Leave out gp initialization from kernel entry on mips64
On OpenBSD/mips64, the kernel is compiled with -mno-abicalls. This
disables gp-relative addressing and essentially makes gp a spare
register in the kernel. Hence it is unnecessary to initialize gp when
entering the kernel. The _gp symbol is not needed either.
Suggested by miod@
otto [Thu, 11 Feb 2021 13:40:28 +0000 (13:40 +0000)]
"proc: table is full" actually means thread table is full; ok mpi@ sthen@
claudio [Thu, 11 Feb 2021 12:08:21 +0000 (12:08 +0000)]
In the various open functions reduce the fdplock() to only span over the
function which need the lock (falloc, fdinsert, fdremove). In most cases
it is not correct to hold the lock while calling VFS functions or e.g.
closef since those aquire or release long lived VFS locks.
OK visa@ mvs@
mestre [Thu, 11 Feb 2021 11:57:32 +0000 (11:57 +0000)]
Initialize var since it's used in a condition a little bit afterwards.
CID
1501713
ok jmatthew@
anton [Thu, 11 Feb 2021 11:03:57 +0000 (11:03 +0000)]
Initialize the stack local device id variable correctly.
CID
1501705
patrick [Thu, 11 Feb 2021 10:41:19 +0000 (10:41 +0000)]
Swap faddr/laddr and fport/lport arguments in call to stoeplitz_ipXport().
Technically the whole point of the stoeplitz API is that it's symmetric,
meaning that the order of addresses and ports doesn't matter and will produce
the same hash value.
Coverity CID
1501717
ok dlg@
nicm [Thu, 11 Feb 2021 09:39:29 +0000 (09:39 +0000)]
Add a couple of helper functions, and flush imsgs on exit.
nicm [Thu, 11 Feb 2021 09:03:38 +0000 (09:03 +0000)]
O_TRUNC is needed in case file exists.
nicm [Thu, 11 Feb 2021 08:28:45 +0000 (08:28 +0000)]
Move file handling protocol stuff all into file.c so it can be reused
more easily.
anton [Thu, 11 Feb 2021 07:26:03 +0000 (07:26 +0000)]
Make room for handling of HID++ 1.0 devices. No functional change.
anton [Thu, 11 Feb 2021 07:24:50 +0000 (07:24 +0000)]
Use idx suffix consistently.
anton [Thu, 11 Feb 2021 07:23:48 +0000 (07:23 +0000)]
Remove unused software id macro.
anton [Thu, 11 Feb 2021 07:22:21 +0000 (07:22 +0000)]
Fold long line.
anton [Thu, 11 Feb 2021 06:56:49 +0000 (06:56 +0000)]
Stop uhidpp from claiming all report ids, instead only claim the
necessary ones. Solves a regression introduced with the arrival of
uhidpp causing some Logitech HID devices from attaching to its
appropriate driver.
Thanks to <naszy at poczta dot fm> and Peter Kane <pwkane at gmail dot com>
for reporting and trying out diffs.
ok mglocker@
anton [Thu, 11 Feb 2021 06:55:10 +0000 (06:55 +0000)]
Add uhidev_unset_report_dev(), doing the opposite of
uhidev_set_report_dev(). Needed by some upcoming changes to uhidpp.
ok mglocker@
anton [Thu, 11 Feb 2021 06:53:44 +0000 (06:53 +0000)]
If uhidev_set_report_dev() already have been invoked for the given
report id, there's no point in trying to find a matching sub device.
ok mglocker@
tb [Thu, 11 Feb 2021 04:56:43 +0000 (04:56 +0000)]
KNF
deraadt [Thu, 11 Feb 2021 04:08:17 +0000 (04:08 +0000)]
sync
tobhe [Wed, 10 Feb 2021 22:25:54 +0000 (22:25 +0000)]
Delay deletion of IKE SAs on rekey when stickyaddress is enabled to make
sure peers can keep their previously assigned addresses.
ok patrick@
espie [Wed, 10 Feb 2021 22:04:14 +0000 (22:04 +0000)]
rephrase example in a more consistent way
kettenis [Wed, 10 Feb 2021 20:51:27 +0000 (20:51 +0000)]
Add a instruction barrier between writing CCSELR_EL1 and reading CCSIDR_EL1
to guarantee that we read the cache parameters of the cache we just selected.
The required ISB instruction is present in the examples in the ARM ARM.
Fixes the the report on the cores in Apple's M1 SoC.
ok patrick@
bluhm [Wed, 10 Feb 2021 18:28:06 +0000 (18:28 +0000)]
If pf changes the routing table when sending packets, the kernel
could get stuck in an endless recursion during TCP path MTU discovery.
Create a dynamic host route in ip_output() that can be used by
tcp_mtudisc() to store the MTU.
Reported by Peter Mueller and Sebastian Sturm
OK claudio@
claudio [Wed, 10 Feb 2021 16:37:29 +0000 (16:37 +0000)]
Add med test, this no longer fails in -current
bluhm [Wed, 10 Feb 2021 14:45:27 +0000 (14:45 +0000)]
Use the same check in kernel and ifconfig for group names. ifconfig
delete group does not need name sanitation. The kernel will just
report that it does not exist.
OK deraadt@ gnezdo@ anton@ mvs@ claudio@
bluhm [Wed, 10 Feb 2021 14:41:53 +0000 (14:41 +0000)]
Interface group names must fit into IFNAMSIZ and be unique. But
the kernel made the unique check before trunkating with strlcpy().
So there could be two interface groups with the same name. The kif
is created by a name lookup. The trunkated names are equal, so
there was only one kif owned by both groups. When the groups got
destroyed, the single kif was removed twice from the RB tree.
Check length of group name before doing the unique check.
The empty group name was allowed and is now invalid.
Reported-by: syzbot+f47e8296ebd559f9bbff@syzkaller.appspotmail.com
OK deraadt@ gnezdo@ anton@ mvs@ claudio@
mvs [Wed, 10 Feb 2021 13:38:46 +0000 (13:38 +0000)]
Remove `sc_dead' logic from pppac(4). It is used to prevent
pppac_ioctl() be called on dying pppac(4) interface. But now if_detach()
makes dying `ifp' inaccessible and waits for references which are in-use
in ioctl(2) path. This logic is not required anymore. Also if_detach()
was moved before klist_invalidate() to prevent the case while
pppac_qstart() bump `sc_rsel'.
ok yasuoka@
espie [Wed, 10 Feb 2021 12:44:13 +0000 (12:44 +0000)]
as usual, stuff got removed without updating the documentation
GC www/drupal7 description
mvs [Wed, 10 Feb 2021 08:20:09 +0000 (08:20 +0000)]
Move UNIX domain sockets out of kernel lock. The new `unp_lock' rwlock(9)
used as solock()'s backend to protect the whole layer.
With feedback from mpi@.
ok bluhm@ claudio@
nicm [Wed, 10 Feb 2021 07:17:07 +0000 (07:17 +0000)]
Use ~/.tmux.conf as an example rather than /etc/passwd, suggested by
deraadt@.
jmc [Wed, 10 Feb 2021 06:52:05 +0000 (06:52 +0000)]
some spacing/grammar fixes from dave voutila;
deraadt [Wed, 10 Feb 2021 02:53:43 +0000 (02:53 +0000)]
only amd64 & arm64 lldb work at the moment
deraadt [Wed, 10 Feb 2021 00:34:57 +0000 (00:34 +0000)]
cast large to reduce warning on 32-bit machines (an ELF type is printed with %ll)
kettenis [Tue, 9 Feb 2021 23:58:33 +0000 (23:58 +0000)]
Make sure that switching the console from serial to framebuffer works
for framebuffer nodes under / and /chosen.
ok patrick@
patrick [Tue, 9 Feb 2021 23:37:54 +0000 (23:37 +0000)]
pfsync_state_import() must not be called with the pf state lock held,
since the actual modification of the state table is done by a call to
pf_state_insert(), which takes the pf state lock itself. Other calls
to pfsync_state_import() also only have the pf lock.
Reported-by: syzbot+d6ea8620b43dc69ecbc6@syzkaller.appspotmail.com
ok bluhm@
patrick [Tue, 9 Feb 2021 21:58:46 +0000 (21:58 +0000)]
sync
patrick [Tue, 9 Feb 2021 21:57:25 +0000 (21:57 +0000)]
Build and install lldb.
Discussed with deraadt@
naddy [Tue, 9 Feb 2021 21:42:04 +0000 (21:42 +0000)]
rc: ensure that vfs.mounts.nfs check works without NFS
If NFS isn't compiled into the kernel, sysctl -n vfs.mounts.nfs
will produce no numerical output. Make sure that we always have
a valid arithmetic expression.
Reported by and ok patrick@
tobhe [Tue, 9 Feb 2021 21:35:48 +0000 (21:35 +0000)]
Add optional 'group none' transform for child SAs and fix handling of
'group none'. We currently send no transform of type DH by default,
which should be equivalent to explicitly sending a single DH transform
of type 'none'. However, the proposal matching logic had a bug where
these two would not match, effectively breaking the ability to negotiate
optional PFS. This commit fixes the bug but continues to send
no DH proposal by default to remain backwards compatible with older
versions.
ok patrick@
mortimer [Tue, 9 Feb 2021 21:35:45 +0000 (21:35 +0000)]
Fix lldb.
Map deliberately invalid signal to zero when passing to PT_STEP and P_CONTINUE.
Also clean up getting Environment so setting LLDB_DEBUGSERVER env vars works again.
ok patrick@
claudio [Tue, 9 Feb 2021 17:00:30 +0000 (17:00 +0000)]
These regress tests expect coredumps to be written so run them with
ulimit -c unlimited. Also simplify the logic a bit as requested by bluhm@.
OK bluhm@ deraadt@
claudio [Tue, 9 Feb 2021 16:58:00 +0000 (16:58 +0000)]
These regress test expect coredumps as an effect of the test so run
the tests with ulimit -c unlimited to make sure coredumps are written.
OK bluhm@ deraadt@
claudio [Tue, 9 Feb 2021 16:55:51 +0000 (16:55 +0000)]
Walk over all results from getaddrinfo() instead of giving up after the
first entry. This way ocspcheck will try all returned IPs to contact
the OCSP server. Found by the regress test and a resolv.conf file with
'family inet6 inet4'.
OK kn@ deraadt@
deraadt [Tue, 9 Feb 2021 14:49:22 +0000 (14:49 +0000)]
sync
jcs [Tue, 9 Feb 2021 14:37:13 +0000 (14:37 +0000)]
ddb: when a new wsdisplay console attaches, resize ddb cols/rows to it
ok visa
nicm [Tue, 9 Feb 2021 14:25:40 +0000 (14:25 +0000)]
Do not expand times and #() inside #().
patrick [Tue, 9 Feb 2021 14:06:19 +0000 (14:06 +0000)]
Activate use of PF_LOCK() by removing the WITH_PF_LOCK ifdefs.
Silence from the network group
ok sashan@
deraadt [Tue, 9 Feb 2021 07:12:20 +0000 (07:12 +0000)]
sync
jmatthew [Tue, 9 Feb 2021 01:50:10 +0000 (01:50 +0000)]
Add a barrier between reading the cqe flags and the command ID, which
should ensure that we don't read a stale command ID and complete the
wrong scsi io. powerpc64 base builds were crashing like this fairly
regularly.
ok deraadt@ dlg@
sthen [Mon, 8 Feb 2021 22:18:21 +0000 (22:18 +0000)]
7.0 firmware key
naddy [Mon, 8 Feb 2021 22:09:57 +0000 (22:09 +0000)]
7.0 packages key
jmc [Mon, 8 Feb 2021 20:32:07 +0000 (20:32 +0000)]
correct return type for compressBound();
from pedro martelletto
stsp [Mon, 8 Feb 2021 20:07:04 +0000 (20:07 +0000)]
change discipline name from "RAID1C" to "RAID 1C" to match the man pages
stsp [Mon, 8 Feb 2021 20:05:20 +0000 (20:05 +0000)]
add RAID 1C to the list of supported softraid(4) disciplines
jan [Mon, 8 Feb 2021 19:37:15 +0000 (19:37 +0000)]
Remove maxburst feature from tcp_output
OK bluhm@, claudio@, deraadt@
deraadt [Mon, 8 Feb 2021 19:09:05 +0000 (19:09 +0000)]
sync
stsp [Mon, 8 Feb 2021 19:05:05 +0000 (19:05 +0000)]
Make bioctl properly verify raidlevels specified via the -c option.
Trailing characters in the option argument were ignored, such that
-cC1 (typo of -c1C) was interpreted as -cC instead of being rejected.
ok jsing@
jsing [Mon, 8 Feb 2021 17:21:50 +0000 (17:21 +0000)]
Update DTLS client hello due to ECC changes.
jsing [Mon, 8 Feb 2021 17:20:47 +0000 (17:20 +0000)]
Remove bogus DTLS checks to disable ECC and OCSP.
ECC and OCSP can be used with DTLS, so remove bogus checks that currently
prevent it. These are long lasting remnants from the original OpenSSL code.
ok tb@
jsing [Mon, 8 Feb 2021 17:18:39 +0000 (17:18 +0000)]
Enforce read ahead with DTLS.
DTLS is largely broken/useless without read ahead being enabled, so enforce
it for DTLS. This behaviour matches both our documentation and OpenSSL.
ok tb@
jsing [Mon, 8 Feb 2021 17:17:02 +0000 (17:17 +0000)]
Use dtls1_retrieve_buffered_record() to load buffered application data.
Replace the current copy of dtls1_retrieve_buffered_record() with a call
to it instead.
ok tb@
deraadt [Mon, 8 Feb 2021 16:15:06 +0000 (16:15 +0000)]
add future 7.0 base key
tobhe [Mon, 8 Feb 2021 16:13:58 +0000 (16:13 +0000)]
Clean up kernel IPsec flows and security associations on shutdown.
Discussed with sthen@
ok patrick@
nicm [Mon, 8 Feb 2021 14:46:53 +0000 (14:46 +0000)]
Add "pipe" variants of the "copy-pipe" commands which do not copy, from
Christian Zangl.
bluhm [Mon, 8 Feb 2021 12:30:10 +0000 (12:30 +0000)]
Start refcounting interface groups with 1. if_creategroup() returns
a new object that is already refcounted, so carp attach does not
reach into internal structures. Add kasserts to detect counter
overflow or underflow.
OK mvs@
stsp [Mon, 8 Feb 2021 11:21:53 +0000 (11:21 +0000)]
Add a RAID1C (raid1 + crypto) softraid(8) discipline.
The RAID1C discipline encrypts data like the CRYPTO discipline, and accepts
multiple chunks during creation and assembly like the RAID1 discipline.
To deal with failing disks a RAID1C volume may be assembled with a smaller
number of chunks than the volume was created with. The volume will then come
up in degraded state. If the volume is now detached and assembled again with
the correct number of chunks, any re-added chunks will require a rebuild.
Consequently, assembling RAID1C volumes requires careful attention to the
chunks passed via 'bioctl -l'. If a chunk is accidentally omitted from the
command line during volume assembly, then this chunk will need to be rebuilt.
At least one known-good chunk is required in order to assemble the volume.
Like CRYPTO, RAID1C supports passphrase and key-disk authentication.
Key-disk based volumes are assembled automatically if the key disk is present
while the system is booting up.
Unlike CRYPTO and RAID1, there is no boot support for RAID1C yet.
RAID1C largely reuses existing code of RAID1 and CRYPTO disciplines.
At present RAID1C's discipline-specific data structure is shared with that
of the CRYPTO discipline to allow re-use of existing CRYPTO code. A custom
RAID1C data structure would require CRYPTO code to access struct sr_crypto
via a pointer instead of via a member field of struct sr_discipline.
ok jsing@
stsp [Mon, 8 Feb 2021 11:20:03 +0000 (11:20 +0000)]
Add a RAID1C (raid1 + crypto) softraid(8) discipline.
The RAID1C discipline encrypts data like the CRYPTO discipline, and accepts
multiple chunks during creation and assembly like the RAID1 discipline.
To deal with failing disks a RAID1C volume may be assembled with a smaller
number of chunks than the volume was created with. The volume will then come
up in degraded state. If the volume is now detached and assembled again with
the correct number of chunks, any re-added chunks will require a rebuild.
Consequently, assembling RAID1C volumes requires careful attention to the
chunks passed via 'bioctl -l'. If a chunk is accidentally omitted from the
command line during volume assembly, then this chunk will need to be rebuilt.
At least one known-good chunk is required in order to assemble the volume.
Like CRYPTO, RAID1C supports passphrase and key-disk authentication.
Key-disk based volumes are assembled automatically if the key disk is present
while the system is booting up.
Unlike CRYPTO and RAID1, there is no boot support for RAID1C yet.
RAID1C largely reuses existing code of RAID1 and CRYPTO disciplines.
At present RAID1C's discipline-specific data structure is shared with that
of the CRYPTO discipline to allow re-use of existing CRYPTO code. A custom
RAID1C data structure would require CRYPTO code to access struct sr_crypto
via a pointer instead of via a member field of struct sr_discipline.
ok jsing@
mpi [Mon, 8 Feb 2021 10:51:01 +0000 (10:51 +0000)]
Revert the convertion of per-process thread into a SMR_TAILQ.
We did not reach a consensus about using SMR to unlock single_thread_set()
so there's no point in keeping this change.
mpi [Mon, 8 Feb 2021 09:46:45 +0000 (09:46 +0000)]
Extend binary operators support, required for more filter features.
Improve debugging of filters and print operator names in debug output.
claudio [Mon, 8 Feb 2021 09:28:58 +0000 (09:28 +0000)]
Also print the recently added caRepository URI alongside rpkiManifest
and rpkiNotify SIA.
claudio [Mon, 8 Feb 2021 09:22:53 +0000 (09:22 +0000)]
Extract the 1.3.6.1.5.5.7.48.5 (caRepository) SIA from the certificate.
Make sure that the caRepository matches the start of the URI of rpkiManifest.
The MFT must be a file inside the caRepository.
OK tb@
claudio [Mon, 8 Feb 2021 09:18:30 +0000 (09:18 +0000)]
Do not hold onto the fdplock longer then needed. Release the lock after
the initial falloc() calls and then regrab it for the fdinsert() or
fdremove() calls respectiviely. Also move closef() outside of the lock.
This replaces the previously reverted lock order change that was reverted.
OK mvs@ visa@
nicm [Mon, 8 Feb 2021 08:33:54 +0000 (08:33 +0000)]
Include "focused" in client flags, from Dan Aloni in GitHub issue 2558.
mpi [Mon, 8 Feb 2021 08:18:45 +0000 (08:18 +0000)]
Simplify sleep_setup API to two operations in preparation for splitting
the SCHED_LOCK().
Putting a thread on a sleep queue is reduce to the following:
sleep_setup();
/* check condition or release lock */
sleep_finish();
Previous version ok cheloha@, jmatthew@, ok claudio@
dlg [Sun, 7 Feb 2021 23:53:46 +0000 (23:53 +0000)]
route-to rules take ips, not interfaces with optional ips.
another gentle reminder by bluhm@ who pointed out i missed some
dlg [Sun, 7 Feb 2021 23:53:00 +0000 (23:53 +0000)]
i missed a dup-to when tweaking route-to things to takes ips, not ifaces.
gentle reminded by bluhm@
jsing [Sun, 7 Feb 2021 15:26:32 +0000 (15:26 +0000)]
Absorb ssl3_get_algorithm2() into ssl_get_handshake_evp_md().
The mess that is ssl_get_algorithm2() only exists to upgrade the handshake
MAC of a pre-TLSv1.2 cipher suite to SHA256 when used with TLSv1.2. We can
readily do this in ssl_get_handshake_evp_md(), which is far more readable.
ok tb@
jsing [Sun, 7 Feb 2021 15:12:52 +0000 (15:12 +0000)]
Correct handshake MAC/PRF for various TLSv1.2 cipher suites.
For some reason various TLSv1.2 cipher suites were added with the default
handshake MAC and PRF, rather than the SHA256 handshake MAC and PRF. This
gets patched up in ssl3_get_algorithm2(), hence goes unnoticed.
ok tb@
jsing [Sun, 7 Feb 2021 15:04:10 +0000 (15:04 +0000)]
Factor out the legacy stack version checks.
Also check for explicit version numbers, rather than just the major version
value.
ok tb@
jsing [Sun, 7 Feb 2021 14:52:17 +0000 (14:52 +0000)]
Enable auto DHE and disable session tickets for some tests.
This allows us to drop the server messages that we intend on dropping.
florian [Sun, 7 Feb 2021 13:35:41 +0000 (13:35 +0000)]
Only probe for DNS64 presence when we know that we can talk to the slaacd
provided nameservers, i.e. the stub resolver check succeeded.
Previously we would only probe DNS64 on network change but would not
reschedule when it failed. Sometimes (most of the time?) this failes
because our address is still tentative or a default route has
not yet been installed.
OK phessler
deraadt [Sun, 7 Feb 2021 01:24:39 +0000 (01:24 +0000)]
sync
tobhe [Sun, 7 Feb 2021 00:56:06 +0000 (00:56 +0000)]
Free X509_STOREs in ca_shutdown().
tobhe [Sun, 7 Feb 2021 00:51:53 +0000 (00:51 +0000)]
Fix address leaks in expand_flows().
ok patrick@
mestre [Sat, 6 Feb 2021 21:42:30 +0000 (21:42 +0000)]
while learning (teachgammon(1)) you might want to save your game so "{w,c}path"
pledge(2) permissions are required
naddy [Sat, 6 Feb 2021 21:40:59 +0000 (21:40 +0000)]
switch clang to -fno-common by default
Common variables will be flagged as errors by default. This can
be overridden with -fcommon.
Cherry-pick this change that will be part of LLVM11. We bring it
forward so we can finish fixing the fallout in the ports tree.
ok deraadt@ sthen@ mglocker@
deraadt [Sat, 6 Feb 2021 21:26:19 +0000 (21:26 +0000)]
6.9-beta
florian [Sat, 6 Feb 2021 18:01:02 +0000 (18:01 +0000)]
Revert delayed opening of trust anchor file. The code was somewhat
ugly and the underlying problem (dhclient and unwind playing well
together) should be solved differently.
Final straw was jca reporting that it breaks his setup.
bluhm [Sat, 6 Feb 2021 13:15:37 +0000 (13:15 +0000)]
Simplex interface sends packet back without hardware checksum
offloading. The checksum must be calculated in software. Use the
same condition in ether_resolve() to send the broadcast packet back
to the stack and in in_ifcap_cksum() to force software checksumming.
This fixes regress/sys/kern/sosplice/loop.
OK procter@