upstream: Remove references to ChallengeResponseAuthentication in
favour of KbdInteractiveAuthentication. The former is what was in SSHv1, the
latter is what is in SSHv2 (RFC4256) and they were treated as somewhat but
not entirely equivalent. We retain the old name as deprecated alias so
config files continue to work and a reference in the man page for people
looking for it.
Prompted by bz#3303 which pointed out the discrepancy between the two
when used with Match. Man page help & ok jmc@, with & ok djm@
upstream: Replace SIGCHLD/notify_pipe kludge with pselect.
Previously sshd's SIGCHLD handler would wake up select() by writing a
byte to notify_pipe. We can remove this by blocking SIGCHLD, checking
for child terminations then passing the original signal mask through
to pselect. This ensures that the pselect will immediately wake up if
a child terminates between wait()ing on them and the pselect.
In -portable, for platforms that do not have pselect the kludge is still
there but is hidden behind a pselect interface.
Darren Tucker [Fri, 25 Jun 2021 05:08:18 +0000 (15:08 +1000)]
Move closefrom() to before first malloc.
When built against tcmalloc, tcmalloc allocates a descriptor for its
internal use, so calling closefrom() afterward causes the descriptor
number to be reused resulting in a corrupted connection. Moving the
closefrom a little earlier should resolve this. From kircherlike at
outlook.com via bz#3321, ok djm@
Darren Tucker [Fri, 18 Jun 2021 10:41:45 +0000 (20:41 +1000)]
Put second -lssh in link line for sftp-server.
When building --without-openssl the recent port-prngd.c change adds
a dependency on atomicio, but since nothing else in sftp-server uses
it, the linker may not find it. Add a second -lssh similar to other
binaries.
returns -1, eg if it was interrupted by a signal. This should prevent
the hang discovered by sthen@ wherein sshd receives a SIGHUP while it has
an unauthenticated child and goes on to a blocking read on a notify_pipe.
feedback deraadt@, ok djm@
djm@openbsd.org [Tue, 8 Jun 2021 07:09:42 +0000 (07:09 +0000)]
upstream: switch sshd_config parsing to argv_split()
similar to the previous commit, this switches sshd_config parsing to
the newer tokeniser. Config parsing will be a little stricter wrt
quote correctness and directives appearing without arguments.
feedback and ok markus@
tested in snaps for the last five or so days - thanks Theo and those who
caught bugs
djm@openbsd.org [Tue, 8 Jun 2021 07:07:15 +0000 (07:07 +0000)]
upstream: Switch ssh_config parsing to use argv_split()
This fixes a couple of problems with the previous tokeniser,
strdelim()
1. strdelim() is permissive wrt accepting '=' characters. This is
intended to allow it to tokenise "Option=value" but because it
cannot keep state, it will incorrectly split "Opt=val=val2".
2. strdelim() has rudimentry handling of quoted strings, but it
is incomplete and inconsistent. E.g. it doesn't handle escaped
quotes inside a quoted string.
3. It has no support for stopping on a (unquoted) comment. Because
of this readconf.c r1.343 added chopping of lines at '#', but
this caused a regression because these characters may legitimately
appear inside quoted strings.
The new tokeniser is stricter is a number of cases, including #1 above
but previously it was also possible for some directives to appear
without arguments. AFAIK these were nonsensical in all cases, and the
new tokeniser refuses to accept them.
The new code handles quotes much better, permitting quoted space as
well as escaped closing quotes. Finally, comment handling should be
fixed - the tokeniser will terminate only on unquoted # characters.
feedback & ok markus@
tested in snaps for the last five or so days - thanks Theo and those who
caught bugs
djm@openbsd.org [Mon, 7 Jun 2021 03:38:38 +0000 (03:38 +0000)]
upstream: fix debug message when finding a private key to match a
certificate being attempted for user authentication. Previously it would
print the certificate's path, whereas it was supposed to be showing the
private key's path. Patch from Alex Sherwin via GHPR247
djm@openbsd.org [Sun, 6 Jun 2021 03:40:39 +0000 (03:40 +0000)]
upstream: Client-side workaround for a bug in OpenSSH 7.4: this release
allows RSA/SHA2 signatures for public key authentication but fails to
advertise this correctly via SSH2_MSG_EXT_INFO. This causes clients of these
server to incorrectly match PubkeyAcceptedAlgorithms and potentially refuse
to offer valid keys.
Reported by and based on patch from Gordon Messmer via bz3213, thanks
also for additional analysis by Jakub Jelen. ok dtucker
Darren Tucker [Fri, 4 Jun 2021 08:39:48 +0000 (18:39 +1000)]
Add pselect implementation for platforms without.
This is basically the existing notify_pipe kludge from serverloop.c
moved behind a pselect interface. It works by installing a signal
handler that writes to a pipe that the select is watching, then calls
the original handler.
The select call in serverloop will become pselect soon, at which point the
kludge will be removed from thereand will only exist in the compat layer.
Original code by markus, help from djm.
By design, 'partial' logins are successful logins, so initially with
authenticated set to 1, for which another authentication is required. As
a result, authenticated is always reset to 0 when partial is set to 1.
However, even if authenticated is 0, those are not failed login
attempts, similarly to attempts with authctxt->postponed set to 1.
upstream: Switch the listening select loop from select() to
pselect() and mask signals while checking signal flags, umasking for pselect
and restoring afterwards. Also restore signals before sighup_restart so they
don't remain blocked after restart.
This prevents a race where a SIGTERM or SIGHUP can arrive between
checking the flag and calling select (eg if sshd is processing a
new connection) resulting in sshd not shutting down until the next
time it receives a new connection. bz#2158, with & ok djm@
djm@openbsd.org [Fri, 4 Jun 2021 05:02:40 +0000 (05:02 +0000)]
upstream: allow ssh_config SetEnv to override $TERM, which is otherwise
handled specially by the protocol. Useful in ~/.ssh/config to set TERM to
something generic (e.g. "xterm" instead of "xterm-256color") for destinations
that lack terminfo entries. feedback and ok dtucker@
djm@openbsd.org [Mon, 31 May 2021 06:48:42 +0000 (06:48 +0000)]
upstream: Hash challenge supplied by client during FIDO key enrollment
prior to passing it to libfido2, which does expect a hash.
There is no effect for users who are simply generating FIDO keys using
ssh-keygen - by default we generate a random 256 bit challenge, but
people building attestation workflows around our tools should now have
a more consistent experience (esp. fewer failures when they fail to
guess the magic 32-byte challenge length requirement).
Darren Tucker [Tue, 25 May 2021 01:40:47 +0000 (11:40 +1000)]
Rename README.md to ci-status.md.
The original intent was to provide a status page for the CIs configured
in that directory, but it had the side effect of replacing the top-level
README.md.
djm@openbsd.org [Wed, 19 May 2021 01:24:05 +0000 (01:24 +0000)]
upstream: restore blocking status on stdio fds before close
ssh(1) needs to set file descriptors to non-blocking mode to operate
but it was not restoring the original state on exit. This could cause
problems with fds shared with other programs via the shell, e.g.
> $ cat > test.sh << _EOF
> #!/bin/sh
> {
> ssh -Fnone -oLogLevel=verbose ::1 hostname
> cat /usr/share/dict/words
> } | sleep 10
> _EOF
> $ ./test.sh
> Authenticated to ::1 ([::1]:22).
> Transferred: sent 2352, received 2928 bytes, in 0.1 seconds
> Bytes per second: sent 44338.9, received 55197.4
> cat: stdout: Resource temporarily unavailable
This restores the blocking status for fds 0,1,2 (stdio) before ssh(1)
abandons/closes them.
This was reported as bz3280 and GHPR246; ok dtucker@
upstream: Clarify language about moduli. While both ends of the
connection do need to use the same parameters (ie groups), the DH-GEX
protocol takes care of that and both ends do not need the same contents in
the moduli file, which is what the previous text suggested. ok djm@ jmc@
Darren Tucker [Mon, 26 Apr 2021 23:18:02 +0000 (09:18 +1000)]
Fix custom OpenSSL tests.
Check out specified OpenSSL version. Install custom libcrypto where
configure expects to find it. Remove unneeded OpenSSL config time
options. Older OpenSSL versions were not make -j safe so remove it.
projects / thirdparty / openssh-portable.git / log
