]> git.ipfire.org Git - thirdparty/openssh-portable.git/log
thirdparty/openssh-portable.git
5 years agodepend V_8_3 anongit/V_8_3 github-selfhosted/V_8_3 github/V_8_3 V_8_3_P1
Damien Miller [Wed, 27 May 2020 00:38:00 +0000 (10:38 +1000)] 
depend

5 years agoupstream: avoid possible NULL deref; from Pedro Martelletto
djm@openbsd.org [Mon, 18 May 2020 04:29:35 +0000 (04:29 +0000)] 
upstream: avoid possible NULL deref; from Pedro Martelletto

OpenBSD-Commit-ID: e6099c3fbb70aa67eb106e84d8b43f1fa919b721

5 years agoprefer ln to cp for temporary copy of sshd
Damien Miller [Thu, 14 May 2020 02:22:09 +0000 (12:22 +1000)] 
prefer ln to cp for temporary copy of sshd

I saw failures on the reexec fallback test on Darwin 19.4 where
fork()ed children of a process that had it's executable removed
would instantly fail. Using ln to preserve the inode avoids this.

5 years agoActually skip pty tests when needed.
Darren Tucker [Wed, 13 May 2020 05:24:51 +0000 (15:24 +1000)] 
Actually skip pty tests when needed.

5 years agoSkip building sk-dummy library if no SK support.
Darren Tucker [Wed, 13 May 2020 03:56:45 +0000 (13:56 +1000)] 
Skip building sk-dummy library if no SK support.

5 years agoexplicitly manage .depend and .depend.bak
Damien Miller [Wed, 13 May 2020 02:08:34 +0000 (12:08 +1000)] 
explicitly manage .depend and .depend.bak

Bring back removal of .depend to give the file a known state before
running makedepend, but manually move aside the current .depend file
and restore it as .depend.bak afterwards so the stale .depend check
works as expected.

5 years agomake depend
Damien Miller [Wed, 13 May 2020 02:03:42 +0000 (12:03 +1000)] 
make depend

5 years agorevert removal of .depend before makedepend
Damien Miller [Wed, 13 May 2020 02:01:10 +0000 (12:01 +1000)] 
revert removal of .depend before makedepend

Commit 83657eac4 started removing .depend before running makedepend
to reset the contents of .depend to a known state. Unfortunately
this broke the depend-check step as now .depend.bak would only ever
be created as an empty file.

ok dtucker

5 years agoprepare for 8.3 release
Damien Miller [Tue, 12 May 2020 02:58:46 +0000 (12:58 +1000)] 
prepare for 8.3 release

5 years agoEnsure SA_SIGNAL test only signals itself.
Darren Tucker [Fri, 8 May 2020 11:50:43 +0000 (21:50 +1000)] 
Ensure SA_SIGNAL test only signals itself.

When the test's child signals its parent and it exits the result of
getppid changes.  On Ubuntu 20.04 this results in the ppid being that
of the GDM session, causing it to exit.  Analysis and testing from pedro
at ambientworks.net

5 years agosync config.guess/config.sub with latest versions
Damien Miller [Fri, 8 May 2020 03:31:53 +0000 (13:31 +1000)] 
sync config.guess/config.sub with latest versions

ok dtucker@

5 years agoupstream: openssh-8.3; ok deraadt@
djm@openbsd.org [Wed, 6 May 2020 20:58:01 +0000 (20:58 +0000)] 
upstream: openssh-8.3; ok deraadt@

OpenBSD-Commit-ID: c8831ec88b9c750f5816aed9051031fb535d22c1

5 years agoupstream: another case where a utimes() failure could make scp send
djm@openbsd.org [Wed, 6 May 2020 20:57:38 +0000 (20:57 +0000)] 
upstream: another case where a utimes() failure could make scp send

a desynchronising error; reminded by Aymeric Vincent ok deraadt markus

OpenBSD-Commit-ID: 2ea611d34d8ff6d703a7a8bf858aa5dbfbfa7381

5 years agoCheck if -D_REENTRANT is needed for localtime_r.
Darren Tucker [Thu, 7 May 2020 05:34:12 +0000 (15:34 +1000)] 
Check if -D_REENTRANT is needed for localtime_r.

On at least HP-UX 11.11, the localtime_r declararation is behind
ifdef _REENTRANT.  Check for and add if needed.

5 years agoSkip security key tests if ENABLE_SK not set.
Darren Tucker [Tue, 5 May 2020 01:32:43 +0000 (11:32 +1000)] 
Skip security key tests if ENABLE_SK not set.

5 years agoupstream: sure enough, some of the test data that we though were in
djm@openbsd.org [Fri, 1 May 2020 04:03:14 +0000 (04:03 +0000)] 
upstream: sure enough, some of the test data that we though were in

new format were actually in the old format; fix from Michael Forney

OpenBSD-Regress-ID: a41a5c43a61b0f0b1691994dbf16dfb88e8af933

5 years agoupstream: make mktestdata.sh generate old/new format keys that we
djm@openbsd.org [Fri, 1 May 2020 04:00:29 +0000 (04:00 +0000)] 
upstream: make mktestdata.sh generate old/new format keys that we

expect. This script was written before OpenSSH switched to new-format private
keys by default and was never updated to the change (until now) From Michael
Forney

OpenBSD-Regress-ID: 38cf354715c96852e5b71c2393fb6e7ad28b7ca7

5 years agoupstream: portability fix for sed that always emil a newline even
djm@openbsd.org [Fri, 1 May 2020 03:58:02 +0000 (03:58 +0000)] 
upstream: portability fix for sed that always emil a newline even

if the input does not contain one; from Michael Forney

OpenBSD-Regress-ID: 9190c3ddf0d2562ccc02c4a95fce0e392196bfc7

5 years agoupstream: remove obsolete RSA1 test keys; spotted by Michael Forney
djm@openbsd.org [Fri, 1 May 2020 03:36:25 +0000 (03:36 +0000)] 
upstream: remove obsolete RSA1 test keys; spotted by Michael Forney

OpenBSD-Regress-ID: 6384ba889594e217d166908ed8253718ab0866da

5 years agoUpdate .depend.
Darren Tucker [Sat, 2 May 2020 08:34:47 +0000 (18:34 +1000)] 
Update .depend.

5 years agoRemove use of tail for 'make depend'.
Darren Tucker [Sat, 2 May 2020 08:29:40 +0000 (18:29 +1000)] 
Remove use of tail for 'make depend'.

Not every tail supports +N and we can do with out it so just remove it.
Prompted by mforney at mforney.org.

5 years agoupstream: we have a sshkey_save_public() function to save public keys;
djm@openbsd.org [Sat, 2 May 2020 07:19:43 +0000 (07:19 +0000)] 
upstream: we have a sshkey_save_public() function to save public keys;

use it and save a bunch of redundant code.

Patch from loic AT venez.fr; ok markus@ djm@

OpenBSD-Commit-ID: f93e030a0ebcd0fd9054ab30db501ec63454ea5f

5 years agoUse LONG_LONG_MAX and friends if available.
Darren Tucker [Fri, 1 May 2020 08:32:25 +0000 (18:32 +1000)] 
Use LONG_LONG_MAX and friends if available.

If we don't have LLONG_{MIN,MAX} but do have LONG_LONG_{MIN,MAX}
then use those instead.  We do calculate these values in configure,
but it turns out that at least one compiler (old HP ANSI C) can't
parse "-9223372036854775808LL" without mangling it. (It can parse
"-9223372036854775807LL" which is presumably why its limits.h defines
LONG_LONG_MIN as the latter minus 1.)

Fixes rekey test when compiled with the aforementioned compiler.

5 years agoupstream: when receving a file in sink(), be careful to send at
djm@openbsd.org [Fri, 1 May 2020 06:31:42 +0000 (06:31 +0000)] 
upstream: when receving a file in sink(), be careful to send at

most a single error response after the file has been opened. Otherwise the
source() and sink() can become desyncronised. Reported by Daniel Goujot,
Georges-Axel Jaloyan, Ryan Lahfa, and David Naccache.

ok deraadt@ markus@

OpenBSD-Commit-ID: 6c14d233c97349cb811a8f7921ded3ae7d9e0035

5 years agoupstream: expose vasnmprintf(); ok (as part of other commit) markus
djm@openbsd.org [Fri, 1 May 2020 06:28:52 +0000 (06:28 +0000)] 
upstream: expose vasnmprintf(); ok (as part of other commit) markus

deraadt

OpenBSD-Commit-ID: 2e80cea441c599631a870fd40307d2ade5a7f9b5

5 years agoupstream: avoid NULL dereference when attempting to convert invalid
djm@openbsd.org [Fri, 1 May 2020 04:23:11 +0000 (04:23 +0000)] 
upstream: avoid NULL dereference when attempting to convert invalid

ssh.com private keys using "ssh-keygen -i"; spotted by Michael Forney

OpenBSD-Commit-ID: 2e56e6d26973967d11d13f56ea67145f435bf298

5 years agoSee if SA_RESTART signals will interrupt select().
Darren Tucker [Fri, 1 May 2020 05:09:26 +0000 (15:09 +1000)] 
See if SA_RESTART signals will interrupt select().

On some platforms (at least older HP-UXes such as 11.11, possibly others)
setting SA_RESTART on signal handers will cause it to not interrupt
select(), at least for calls that do not specify a timeout.  Try to
detect this and if found, don't use SA_RESTART.

POSIX says "If SA_RESTART has been set for the interrupting signal, it
is implementation-dependent whether select() restarts or returns with
[EINTR]" so this behaviour is within spec.

5 years agofix reversed test
Damien Miller [Fri, 1 May 2020 03:55:03 +0000 (13:55 +1000)] 
fix reversed test

5 years agowrap sha2.h inclusion in #ifdef HAVE_SHA2_H
Damien Miller [Fri, 1 May 2020 03:29:16 +0000 (13:29 +1000)] 
wrap sha2.h inclusion in #ifdef HAVE_SHA2_H

5 years agoupstream: adapt dummy FIDO middleware to API change; ok markus@
djm@openbsd.org [Tue, 28 Apr 2020 04:59:29 +0000 (04:59 +0000)] 
upstream: adapt dummy FIDO middleware to API change; ok markus@

OpenBSD-Regress-ID: 8bb84ee500c2eaa5616044314dd0247709a1790f

5 years agoupstream: tweak previous; ok markus
jmc@openbsd.org [Thu, 30 Apr 2020 18:28:37 +0000 (18:28 +0000)] 
upstream: tweak previous; ok markus

OpenBSD-Commit-ID: 41895450ce2294ec44a5713134491cc31f0c09fd

5 years agoupstream: bring back debug() removed in rev 1.74; noted by pradeep
markus@openbsd.org [Thu, 30 Apr 2020 17:12:20 +0000 (17:12 +0000)] 
upstream: bring back debug() removed in rev 1.74; noted by pradeep

kumar

OpenBSD-Commit-ID: 8d134d22ab25979078a3b48d058557d49c402e65

5 years agoupstream: run the 2nd ssh with BatchMode for scp -3
markus@openbsd.org [Thu, 30 Apr 2020 17:07:10 +0000 (17:07 +0000)] 
upstream: run the 2nd ssh with BatchMode for scp -3

OpenBSD-Commit-ID: 77994fc8c7ca02d88e6d0d06d0f0fe842a935748

5 years agoupstream: when signing a challenge using a FIDO toke, perform the
djm@openbsd.org [Tue, 28 Apr 2020 04:02:29 +0000 (04:02 +0000)] 
upstream: when signing a challenge using a FIDO toke, perform the

hashing in the middleware layer rather than in ssh code. This allows
middlewares that call APIs that perform the hashing implicitly (including
Microsoft's AFAIK). ok markus@

OpenBSD-Commit-ID: c9fc8630aba26c75d5016884932f08a5a237f37d

5 years agoupstream: Fix comment typo. Patch from mforney at mforney.org.
dtucker@openbsd.org [Sun, 26 Apr 2020 09:38:14 +0000 (09:38 +0000)] 
upstream: Fix comment typo. Patch from mforney at mforney.org.

OpenBSD-Commit-ID: 3565f056003707a5e678e60e03f7a3efd0464a2b

5 years agoupstream: We've standardized on memset over bzero, replace a couple
dtucker@openbsd.org [Sat, 25 Apr 2020 06:59:36 +0000 (06:59 +0000)] 
upstream: We've standardized on memset over bzero, replace a couple

that had slipped in.  ok deraadt markus djm.

OpenBSD-Commit-ID: f5be055554ee93e6cc66b0053b590bef3728dbd6

5 years agoInclude sys/byteorder.h for htons and friends.
Darren Tucker [Fri, 1 May 2020 02:21:58 +0000 (12:21 +1000)] 
Include sys/byteorder.h for htons and friends.

These are usually in netinet/in.h but on HP-UX they are not defined if
_XOPEN_SOURCE_EXTENDED is set.  Only needed for netcat in the regression
tests.

5 years agoFix conditional for openssl-based chacha20.
Darren Tucker [Thu, 30 Apr 2020 23:21:52 +0000 (09:21 +1000)] 
Fix conditional for openssl-based chacha20.

Fixes warnings or link errors when building against older OpenSSLs.
ok djm

5 years agoError out if given RDomain if unsupported.
Darren Tucker [Fri, 24 Apr 2020 05:07:55 +0000 (15:07 +1000)] 
Error out if given RDomain if unsupported.

If the config contained 'RDomain %D' on a platform that did not support
it, the error would not be detected until runtime resulting in a broken
sshd.  Detect this earlier and error out if found.  bz#3126, based on a
patch from jjelen at redhat.com, tweaks and ok djm@

5 years agoupstream: Fix incorrect error message for "too many known hosts files."
dtucker@openbsd.org [Fri, 24 Apr 2020 03:33:21 +0000 (03:33 +0000)] 
upstream: Fix incorrect error message for "too many known hosts files."

bz#3149, patch from jjelen at redhat.com.

OpenBSD-Commit-ID: e0fcb07ed5cf7fd54ce340471a747c24454235e5

5 years agoupstream: Remove leave_non_blocking() which is now dead code
dtucker@openbsd.org [Fri, 24 Apr 2020 02:19:40 +0000 (02:19 +0000)] 
upstream: Remove leave_non_blocking() which is now dead code

because nothing sets in_non_blocking_mode any more. Patch from
michaael.meeks at collabora.com, ok djm@

OpenBSD-Commit-ID: c403cefe97a5a99eca816e19cc849cdf926bd09c

5 years agoupstream: ce examples of "Ar arg Ar arg" with "Ar arg arg" and
jmc@openbsd.org [Thu, 23 Apr 2020 21:28:09 +0000 (21:28 +0000)] 
upstream: ce examples of "Ar arg Ar arg" with "Ar arg arg" and

stop the spread;

OpenBSD-Commit-ID: af0e952ea0f5e2019c2ce953ed1796eca47f0705

5 years agoUpdate .depend.
Darren Tucker [Fri, 24 Apr 2020 01:10:18 +0000 (11:10 +1000)] 
Update .depend.

5 years agoMailing list is now closed to non-subscribers.
Darren Tucker [Wed, 22 Apr 2020 04:07:00 +0000 (14:07 +1000)] 
Mailing list is now closed to non-subscribers.

While there, add a reference to the bugzilla.  ok djm@

5 years agoPut the values from env vars back.
Darren Tucker [Wed, 22 Apr 2020 02:09:40 +0000 (12:09 +1000)] 
Put the values from env vars back.

This merges the values from the recently removed environment into make's
command line arguments since we actually need those.

5 years agoPass configure's egrep through to test-exec.sh.
Darren Tucker [Wed, 22 Apr 2020 01:33:15 +0000 (11:33 +1000)] 
Pass configure's egrep through to test-exec.sh.

Use it to create a wrapper function to call it from tests.  Fixes the
keygen-comment test on platforms with impoverished default egrep (eg
Solaris).

5 years agoRemove unneeded env vars from t-exec invocation.
Darren Tucker [Wed, 22 Apr 2020 00:56:44 +0000 (10:56 +1000)] 
Remove unneeded env vars from t-exec invocation.

5 years agoupstream: Backslash '$' at then end of string. Prevents warning on
dtucker@openbsd.org [Tue, 21 Apr 2020 23:14:58 +0000 (23:14 +0000)] 
upstream: Backslash '$' at then end of string. Prevents warning on

some shells.

OpenBSD-Regress-ID: 5dc27ab624c09d34078fd326b10e38c1ce9c741f

5 years agoSync rev 1.49.
Darren Tucker [Tue, 21 Apr 2020 08:27:23 +0000 (18:27 +1000)] 
Sync rev 1.49.

Prevent infinite for loop since i went from ssize_t to size_t.  Patch from
eagleoflqj via OpenSSH github PR#178, ok djm@, feedback & ok millert@

5 years agoupstream: regression test for printing of private key fingerprints and
djm@openbsd.org [Mon, 20 Apr 2020 04:44:47 +0000 (04:44 +0000)] 
upstream: regression test for printing of private key fingerprints and

key comments, mostly by loic AT venez.fr (slightly tweaked for portability)
ok dtucker@

OpenBSD-Regress-ID: 8dc6c4feaf4fe58b6d634cd89afac9a13fd19004

5 years agoupstream: fix a bug I introduced in r1.406: when printing private key
djm@openbsd.org [Mon, 20 Apr 2020 04:43:57 +0000 (04:43 +0000)] 
upstream: fix a bug I introduced in r1.406: when printing private key

fingerprint of old-format key, key comments were not being displayed. Spotted
by loic AT venez.fr, ok dtucker

OpenBSD-Commit-ID: 2d98e4f9eb168eea733d17e141e1ead9fe26e533

5 years agoupstream: repair private key fingerprint printing to also print
djm@openbsd.org [Fri, 17 Apr 2020 07:16:07 +0000 (07:16 +0000)] 
upstream: repair private key fingerprint printing to also print

comment after regression caused by my recent pubkey loading refactor.
Reported by loic AT venez.fr, ok dtucker@

OpenBSD-Commit-ID: f8db49acbee6a6ccb2a4259135693b3cceedb89e

5 years agoupstream: refactor out some duplicate private key loading code;
djm@openbsd.org [Fri, 17 Apr 2020 07:15:11 +0000 (07:15 +0000)] 
upstream: refactor out some duplicate private key loading code;

based on patch from loic AT venez.fr, ok dtucker@

OpenBSD-Commit-ID: 5eff2476b0d8d0614924c55e350fb7bb9c84f45e

5 years agoupstream: add space beteen macro arg and punctuation;
jmc@openbsd.org [Fri, 17 Apr 2020 06:12:41 +0000 (06:12 +0000)] 
upstream: add space beteen macro arg and punctuation;

OpenBSD-Commit-ID: c93a6cbb4bf9468fc4c13e64bc1fd4efee201a44

5 years agoupstream: auth2-pubkey r1.89 changed the order of operations to
djm@openbsd.org [Fri, 17 Apr 2020 04:27:03 +0000 (04:27 +0000)] 
upstream: auth2-pubkey r1.89 changed the order of operations to

checking AuthorizedKeysFile first and falling back to AuthorizedKeysCommand
if no key was found in a file. Document this order here; bz3134

OpenBSD-Commit-ID: afce0872cbfcfc1d4910ad7722e50f792a1dce12

5 years agosys/sysctl.h is only used on OpenBSD
Damien Miller [Fri, 17 Apr 2020 04:07:15 +0000 (14:07 +1000)] 
sys/sysctl.h is only used on OpenBSD

so change the preprocessor test used to include it to check
__OpenBSD__, matching the code that uses the symbols it declares.

5 years agoupstream: fix reversed test that caused IdentitiesOnly=yes to not
djm@openbsd.org [Fri, 17 Apr 2020 03:38:47 +0000 (03:38 +0000)] 
upstream: fix reversed test that caused IdentitiesOnly=yes to not

apply to keys loaded from a PKCS11Provider; bz3141, ok dtucker@

OpenBSD-Commit-ID: e3dd6424b94685671fe84c9b9dbe352fb659f677

5 years agoupstream: mention that /etc/hosts.equiv and /etc/shosts.equiv are
djm@openbsd.org [Fri, 17 Apr 2020 03:34:42 +0000 (03:34 +0000)] 
upstream: mention that /etc/hosts.equiv and /etc/shosts.equiv are

not considered for HostbasedAuthentication when the target user is root;
bz3148

OpenBSD-Commit-ID: fe4c1256929e53f23af17068fbef47852f4bd752

5 years agoupstream: make IgnoreRhosts a tri-state option: "yes" ignore
djm@openbsd.org [Fri, 17 Apr 2020 03:30:05 +0000 (03:30 +0000)] 
upstream: make IgnoreRhosts a tri-state option: "yes" ignore

rhosts/shosts, "no" allow rhosts/shosts or (new) "shosts-only" to allow
.shosts files but not .rhosts. ok dtucker@

OpenBSD-Commit-ID: d08d6930ed06377a80cf53923c1955e9589342e9

5 years agoupstream: allow the IgnoreRhosts directive to appear anywhere in a
djm@openbsd.org [Fri, 17 Apr 2020 03:23:13 +0000 (03:23 +0000)] 
upstream: allow the IgnoreRhosts directive to appear anywhere in a

sshd_config, not just before any Match blocks; bz3148, ok dtucker@

OpenBSD-Commit-ID: e042467d703bce640b1f42c5d1a62bf3825736e8

5 years agoupstream: add space between macro arg and punctuation;
jmc@openbsd.org [Sat, 11 Apr 2020 20:20:09 +0000 (20:20 +0000)] 
upstream: add space between macro arg and punctuation;

OpenBSD-Commit-ID: e579e4d95eef13059c30931ea1f09ed8296b819c

5 years agoAdd sys/syscall.h for syscall numbers.
Darren Tucker [Wed, 15 Apr 2020 00:58:02 +0000 (10:58 +1000)] 
Add sys/syscall.h for syscall numbers.

In some architecture/libc configurations we need to explicitly include
sys/syscall.h for the syscall number (__NR_xxx) definitions.  bz#3085,
patch from blowfist at xroutine.net.

5 years agoupstream: Refactor private key parsing. Eliminates a fair bit of
djm@openbsd.org [Sat, 11 Apr 2020 10:16:11 +0000 (10:16 +0000)] 
upstream: Refactor private key parsing. Eliminates a fair bit of

duplicated code and fixes oss-fuzz#20074 (NULL deref) caused by a missing key
type check in the ECDSA_CERT parsing path.

feedback and ok markus@

OpenBSD-Commit-ID: 4711981d88afb7196d228f7baad9be1d3b20f9c9

5 years agoupstream: Add tests for TOKEN expansion of LocalForward and
dtucker@openbsd.org [Fri, 10 Apr 2020 00:54:03 +0000 (00:54 +0000)] 
upstream: Add tests for TOKEN expansion of LocalForward and

RemoteForward.

OpenBSD-Regress-ID: 90fcbc60d510eb114a2b6eaf4a06ff87ecd80a89

5 years agoupstream: Add utf8.c for asmprintf used by krl.c
dtucker@openbsd.org [Mon, 6 Apr 2020 09:43:55 +0000 (09:43 +0000)] 
upstream: Add utf8.c for asmprintf used by krl.c

OpenBSD-Regress-ID: 433708d11165afdb189fe635151d21659dd37a37

5 years agoupstream: Add TOKEN percent expansion to LocalFoward and RemoteForward
dtucker@openbsd.org [Fri, 10 Apr 2020 00:52:07 +0000 (00:52 +0000)] 
upstream: Add TOKEN percent expansion to LocalFoward and RemoteForward

when used for Unix domain socket forwarding.  Factor out the code for the
config keywords that use the most common subset of TOKENS into its own
function. bz#3014, ok jmc@ (man page bits) djm@

OpenBSD-Commit-ID: bffc9f7e7b5cf420309a057408bef55171fd0b97

5 years agoupstream: let sshkey_try_load_public() load public keys from the
djm@openbsd.org [Wed, 8 Apr 2020 00:10:37 +0000 (00:10 +0000)] 
upstream: let sshkey_try_load_public() load public keys from the

unencrypted envelope of private key files if not sidecar public key file is
present.

ok markus@

OpenBSD-Commit-ID: 252a0a580e10b9a6311632530d63b5ac76592040

5 years agoupstream: simplify sshkey_try_load_public()
djm@openbsd.org [Wed, 8 Apr 2020 00:09:24 +0000 (00:09 +0000)] 
upstream: simplify sshkey_try_load_public()

ok markus@

OpenBSD-Commit-ID: 05a5d46562aafcd70736c792208b1856064f40ad

5 years agoupstream: add sshkey_parse_pubkey_from_private_fileblob_type()
djm@openbsd.org [Wed, 8 Apr 2020 00:08:46 +0000 (00:08 +0000)] 
upstream: add sshkey_parse_pubkey_from_private_fileblob_type()

Extracts a public key from the unencrypted envelope of a new-style
OpenSSH private key.

ok markus@

OpenBSD-Commit-ID: 44d7ab446e5e8c686aee96d5897b26b3939939aa

5 years agoupstream: simplify sshkey_parse_private_fileblob_type()
djm@openbsd.org [Wed, 8 Apr 2020 00:07:19 +0000 (00:07 +0000)] 
upstream: simplify sshkey_parse_private_fileblob_type()

Try new format parser for all key types first, fall back to PEM
parser only for invalid format errors.

ok markus@

OpenBSD-Commit-ID: 0173bbb3a5cface77b0679d4dca0e15eb5600b77

5 years agoupstream: check private key type against requested key type in
djm@openbsd.org [Wed, 8 Apr 2020 00:05:59 +0000 (00:05 +0000)] 
upstream: check private key type against requested key type in

new-style private decoding; ok markus@

OpenBSD-Commit-ID: 04d44b3a34ce12ce5187fb6f6e441a88c8c51662

5 years agoupstream: check that pubkey in private key envelope matches actual
djm@openbsd.org [Wed, 8 Apr 2020 00:04:32 +0000 (00:04 +0000)] 
upstream: check that pubkey in private key envelope matches actual

private key

(this public key is currently unusued)

ok markus@

OpenBSD-Commit-ID: 634a60b5e135d75f48249ccdf042f3555112049c

5 years agoupstream: refactor private key parsing a little
djm@openbsd.org [Wed, 8 Apr 2020 00:01:52 +0000 (00:01 +0000)] 
upstream: refactor private key parsing a little

Split out the base64 decoding and private section decryption steps in
to separate functions. This will make the decryption step easier to fuzz
as well as making it easier to write a "load public key from new-format
private key" function.

ok markus@

OpenBSD-Commit-ID: 7de31d80fb9062aa01901ddf040c286b64ff904e

5 years agoInclude openssl-compat.h before checking ifdefs.
Darren Tucker [Mon, 6 Apr 2020 10:54:34 +0000 (20:54 +1000)] 
Include openssl-compat.h before checking ifdefs.

Fixes problem where unsuitable chacha20 code in libressl would be used
unintentionally.

5 years agofix inverted test for LibreSSL version
Damien Miller [Mon, 6 Apr 2020 00:04:56 +0000 (10:04 +1000)] 
fix inverted test for LibreSSL version

5 years agoupstream: Indicate if we're using a cached key in trace output.
dtucker@openbsd.org [Sat, 4 Apr 2020 23:04:41 +0000 (23:04 +0000)] 
upstream: Indicate if we're using a cached key in trace output.

OpenBSD-Regress-ID: 409a7b0e59d1272890fda507651c0c3d2d3c0d89

5 years agoUse /usr/bin/xp4g/id if necessary.
Darren Tucker [Sat, 4 Apr 2020 22:43:57 +0000 (08:43 +1000)] 
Use /usr/bin/xp4g/id if necessary.

Solaris' native "id" doesn't support the options we use but the one
in /usr/bin/xp4g does, so use that instead.

5 years agoupstream: Some platforms don't have "hostname -s", so use cut to trim
dtucker@openbsd.org [Sat, 4 Apr 2020 22:14:26 +0000 (22:14 +0000)] 
upstream: Some platforms don't have "hostname -s", so use cut to trim

short hostname instead.

OpenBSD-Regress-ID: ebcf36a6fdf287c9336b0d4f6fc9f793c05307a7

5 years agoupstream: Compute hash locally and re-enable %C tests.
dtucker@openbsd.org [Fri, 3 Apr 2020 07:53:10 +0000 (07:53 +0000)] 
upstream: Compute hash locally and re-enable %C tests.

OpenBSD-Regress-ID: 94d1366e8105274858b88a1f9ad2e62801e49770

5 years agoprefer libcrypto chacha20-poly1305 where possible
Damien Miller [Fri, 3 Apr 2020 06:25:46 +0000 (17:25 +1100)] 
prefer libcrypto chacha20-poly1305 where possible

5 years agoupstream: Temporarily remove tests for '%C' since the hash contains the
dtucker@openbsd.org [Fri, 3 Apr 2020 05:43:11 +0000 (05:43 +0000)] 
upstream: Temporarily remove tests for '%C' since the hash contains the

local hostname and it doesn't work on any machine except mine... spotted by
djm@

OpenBSD-Regress-ID: 2d4c3585b9fcbbff14f4a5a5fde51dbd0d690401

5 years agoupstream: r1.522 deleted one too many lines; repair
djm@openbsd.org [Fri, 3 Apr 2020 06:07:57 +0000 (06:07 +0000)] 
upstream: r1.522 deleted one too many lines; repair

OpenBSD-Commit-ID: 1af8851fd7a99e4a887b19aa8f4c41a6b3d25477

5 years agoupstream: sort -N and add it to usage();
jmc@openbsd.org [Fri, 3 Apr 2020 05:53:52 +0000 (05:53 +0000)] 
upstream: sort -N and add it to usage();

OpenBSD-Commit-ID: 5b00e8db37c2b0a54c7831fed9e5f4db53ada332

5 years agoupstream: avoid another compiler warning spotted in -portable
djm@openbsd.org [Fri, 3 Apr 2020 05:48:57 +0000 (05:48 +0000)] 
upstream: avoid another compiler warning spotted in -portable

OpenBSD-Commit-ID: 1d29c51ac844b287c4c8bcaf04c63c7d9ba3b8c7

5 years agoupstream: this needs utf8.c too
djm@openbsd.org [Fri, 3 Apr 2020 04:07:48 +0000 (04:07 +0000)] 
upstream: this needs utf8.c too

OpenBSD-Regress-ID: 445040036cec714d28069a20da25553a04a28451

5 years agoupstream: Add percent_expand test for 'Match Exec'.
dtucker@openbsd.org [Fri, 3 Apr 2020 03:14:03 +0000 (03:14 +0000)] 
upstream: Add percent_expand test for 'Match Exec'.

OpenBSD-Regress-ID: a41c14fd6a0b54d66aa1e9eebfb9ec962b41232f

5 years agoupstream: fix format string (use %llu for uint64, not %lld). spotted by
djm@openbsd.org [Fri, 3 Apr 2020 04:43:24 +0000 (04:43 +0000)] 
upstream: fix format string (use %llu for uint64, not %lld). spotted by

Darren and his tinderbox tests

OpenBSD-Commit-ID: 3b4587c3d9d46a7be9bdf028704201943fba96c2

5 years agoupstream: Add a flag to re-enable verbose output when in batch
djm@openbsd.org [Fri, 3 Apr 2020 04:34:15 +0000 (04:34 +0000)] 
upstream: Add a flag to re-enable verbose output when in batch

mode; requested in bz3135; ok dtucker

OpenBSD-Commit-ID: 5ad2ed0e6440562ba9c84b666a5bbddc1afe2e2b

5 years agoupstream: chacha20-poly1305 AEAD using libcrypto EVP_chacha20
djm@openbsd.org [Fri, 3 Apr 2020 04:32:21 +0000 (04:32 +0000)] 
upstream: chacha20-poly1305 AEAD using libcrypto EVP_chacha20

Based on patch from Yuriy M. Kaminskiy. ok + lots of assistance along the
way at a2k20 tb@

OpenBSD-Commit-ID: 5e08754c13d31258bae6c5e318cc96219d6b10f0

5 years agoupstream: make Chacha20-POLY1305 context struct opaque; ok tb@ as
djm@openbsd.org [Fri, 3 Apr 2020 04:27:03 +0000 (04:27 +0000)] 
upstream: make Chacha20-POLY1305 context struct opaque; ok tb@ as

part of a larger diff at a2k20

OpenBSD-Commit-ID: a4609b7263284f95c9417ef60ed7cdbb7bf52cfd

5 years agoupstream: fix debug statement
djm@openbsd.org [Fri, 3 Apr 2020 04:06:26 +0000 (04:06 +0000)] 
upstream: fix debug statement

OpenBSD-Commit-ID: 42c6edeeda5ce88b51a20d88c93be3729ce6b916

5 years agoupstream: the tunnel-forwarding vs ExitOnForwardFailure fix that I
djm@openbsd.org [Fri, 3 Apr 2020 04:03:51 +0000 (04:03 +0000)] 
upstream: the tunnel-forwarding vs ExitOnForwardFailure fix that I

committed earlier had an off-by-one. Fix this and add some debugging that
would have made it apparent sooner.

OpenBSD-Commit-ID: 082f8f72b1423bd81bbdad750925b906e5ac6910

5 years agoupstream: %C expansion just added to Match Exec should include
dtucker@openbsd.org [Fri, 3 Apr 2020 03:12:11 +0000 (03:12 +0000)] 
upstream: %C expansion just added to Match Exec should include

remote user not local user.

OpenBSD-Commit-ID: 80f1d976938f2a55ee350c11d8b796836c8397e2

5 years agoupstream: Add regression test for percent expansions where possible.
dtucker@openbsd.org [Fri, 3 Apr 2020 02:33:31 +0000 (02:33 +0000)] 
upstream: Add regression test for percent expansions where possible.

OpenBSD-Regress-ID: 7283be8b2733ac1cbefea3048a23d02594485288

5 years agoupstream: make failures when establishing "Tunnel" forwarding terminate
djm@openbsd.org [Fri, 3 Apr 2020 02:40:32 +0000 (02:40 +0000)] 
upstream: make failures when establishing "Tunnel" forwarding terminate

the connection when ExitOnForwardFailure is enabled; bz3116; ok dtucker

OpenBSD-Commit-ID: ef4b4808de0a419c17579b1081da768625c1d735

5 years agoupstream: Make with config keywords support which
dtucker@openbsd.org [Fri, 3 Apr 2020 02:27:12 +0000 (02:27 +0000)] 
upstream: Make with config keywords support which

percent_expansions more consistent.  - %C is moved into its own function and
added to Match Exec.  - move the common (global) options into a macro.  This
is ugly but it's    the least-ugly way I could come up with.  - move
IdentityAgent and ForwardAgent percent expansion to before the    config dump
to make it regression-testable.  - document all of the above

ok jmc@ for man page bits, "makes things less terrible" djm@ for the rest.

OpenBSD-Commit-ID: 4b65664bd6d8ae2a9afaf1a2438ddd1b614b1d75

5 years agoupstream: give ssh-keygen the ability to dump the contents of a
djm@openbsd.org [Fri, 3 Apr 2020 02:26:56 +0000 (02:26 +0000)] 
upstream: give ssh-keygen the ability to dump the contents of a

binary key revocation list: ssh-keygen -lQf /path bz#3132; ok dtucker

OpenBSD-Commit-ID: b76afc4e3b74ab735dbde4e5f0cfa1f02356033b

5 years agoupstream: add allocating variant of the safe utf8 printer; ok
djm@openbsd.org [Fri, 3 Apr 2020 02:25:21 +0000 (02:25 +0000)] 
upstream: add allocating variant of the safe utf8 printer; ok

dtucker as part of a larger diff

OpenBSD-Commit-ID: 037e2965bd50eacc2ffb49889ecae41552744fa0

5 years agoupstream: Cast lifetime to u_long for comparison to prevent unsigned
dtucker@openbsd.org [Mon, 16 Mar 2020 02:17:02 +0000 (02:17 +0000)] 
upstream: Cast lifetime to u_long for comparison to prevent unsigned

comparison warning on 32bit arches.  Spotted by deraadt, ok djm.

OpenBSD-Commit-ID: 7a75b2540bff5ab4fa00b4d595db1df13bb0515a

5 years agoInclude fido.h when checking for fido/credman.h.
Darren Tucker [Sat, 14 Mar 2020 09:58:46 +0000 (20:58 +1100)] 
Include fido.h when checking for fido/credman.h.

It's required for fido_dev_t, otherwise configure fails with
when given --with-security-key-builtin.