This is a simple extension that allows the server to clearly
communicate transfer limits it is imposing so the client doesn't
have to guess, or force the user to manually tune. This is
particularly useful when an attempt to use too large of a value
causes the server to abort the connection.
Damien Miller [Thu, 18 Feb 2021 00:25:38 +0000 (11:25 +1100)]
prefer login_getpwclass() to login_getclass()
FreeBSD has login_getpwclass() that does some special magic for
UID=0. Prefer this to login_getclass() as its easier to emulate
the former with the latter.
Based on FreeBSD PR 37416 via Ed Maste; ok dtucker@
Darren Tucker [Fri, 15 Jan 2021 03:11:43 +0000 (14:11 +1100)]
Add self-hosted runners for VMs of other platforms.
Github only hosts a limited number of platforms, and the runner code
is only supported on slightly wider range of platforms. To increase
our test coverage beyond that, we run the runner natively on a VM host,
where it runs a jobs that boot VMs of other platforms, waits for them
to come up then runs the build and test by ssh'ing into the guest.
This means that the minimum dependencies for the guests are quite low
(basically just sshd, a compiler and make).
The interface to the VM host is fairly simple (basically 3 scripts:
vmstartup, vmrun and vmshutdown), but those are specific to the VM host
so are not in the public repo. We also mount the working directory on the
host via sshfs, so things like artifact upload by the runner also work.
As part of this we are moving the per-test-target configs into a single
place (.github/configs) where there will be referenced by a single short
"config" key. I plan to make the github-hosted runners use this too.
The self-hosted runners are run off a private repo on github since that
prevents third parties from accessing them[0], and since runner quota is
limited on private repos, we avoid running the tests we run on the public
repo.
Darren Tucker [Fri, 5 Feb 2021 22:37:01 +0000 (09:37 +1100)]
Add a hostname function for systems that don't have it.
Some systems don't have a hostname command (it's not required by POSIX).
The do have uname -n (which is), but as found by tim@ some others (eg
UnixWare) do not report the FQDN from uname -n.
upstream: Roll back the hostname->uname change in rev 1.10. It turns
out uname -n doesn't do what we need for some platforms in portable, so we'll
fix the original problem (that some other platforms don't have hostname at
all) by providing wrapper function to implement it.
djm@openbsd.org [Sun, 31 Jan 2021 22:55:29 +0000 (22:55 +0000)]
upstream: more strictly enforce KEX state-machine by banning packet
types once they are received. Fixes memleak caused by duplicate
SSH2_MSG_KEX_DH_GEX_REQUEST (spotted by portable OpenSSH kex_fuzz via
oss-fuzz #30078).
HostbasedAcceptedKeyTypes (sshd) to HostbasedAcceptedAlgorithms, which more
accurately reflects its effect. This matches a previous change to
PubkeyAcceptedAlgorithms. The previous names are retained as aliases. ok
djm@
Darren Tucker [Tue, 26 Jan 2021 03:48:07 +0000 (14:48 +1100)]
Disable sntrup761 if compiler doesn't support VLAs.
The sntrup761 code sourced from supercop uses variable length
arrays. Although widely supported, they are not part of the ANSI
C89 spec so if the compiler does not support VLAs, disable the
sntrup761x25519-sha512@openssh.com KEX method by replacing the kex
functions with no-op ones similar to what we do in kexecdh.c.
This should allow OpenSSH to build with a plain C89 compiler again.
Spotted by tim@, ok djm@.
djm@openbsd.org [Tue, 26 Jan 2021 00:54:49 +0000 (00:54 +0000)]
upstream: refactor key constraint parsing in ssh-agent
Key constraints parsing code previously existed in both the "add regular
key" and "add smartcard key" path. This unifies them but also introduces
more consistency checking: duplicated constraints and constraints that
are nonsensical for a particular situation (e.g. FIDO provider for a
smartcard key) are now banned.
djm@openbsd.org [Tue, 26 Jan 2021 00:46:17 +0000 (00:46 +0000)]
upstream: factor out common code in the agent client
Add a ssh_request_reply_decode() function that sends a message to
the agent, reads and parses a success/failure reply.
Use it for all requests that only expect success/failure
djm@openbsd.org [Mon, 25 Jan 2021 06:00:17 +0000 (06:00 +0000)]
upstream: make ssh hostbased authentication send the signature
algorithm in its SSH2_MSG_USERAUTH_REQUEST packets instead of the key type.
This make HostbasedAcceptedAlgorithms do what it is supposed to - filter on
signature algorithm and not key type.
upstream: Rename PubkeyAcceptedKeyTypes keyword to
PubkeyAcceptedAlgorithms. While the two were originally equivalent, this
actually specifies the signature algorithms that are accepted. Some key
types (eg RSA) can be used by multiple algorithms (eg ssh-rsa, rsa-sha2-512)
so the old name is becoming increasingly misleading. The old name is
retained as an alias. Prompted by bz#3253, help & ok djm@, man page help jmc@
upstream: In waitfd(), when poll returns early we are subtracting
the elapsed time from the timeout each loop, so we only want to measure the
elapsed time the poll() in that loop, not since the start of the function.
Spotted by chris.xj.zhu at gmail.com, ok djm@
upstream: Change convtime() from returning long to returning int.
On platforms where sizeof(int) != sizeof(long), convtime could accept values
>MAX_INT which subsequently truncate when stored in an int during config
parsing. bz#3250, ok djm@
upstream: Move address handling functions out into their own file
in order to reuse them for per-source maxstartups limiting. Supplement with
some additional functions from djm's flowtools that we'll also need. ok djm@
(as part of a larger diff).
djm@openbsd.org [Fri, 8 Jan 2021 02:57:24 +0000 (02:57 +0000)]
upstream: If a signature operation on a FIDO key fails with a
"incorrect PIN" reason and no PIN was initially requested from the user, then
request a PIN and retry the operation.
This smoothes over a few corner cases including FIDO devices that
require PINs for all hosted credentials, biometric FIDO devices that
fall back to requiring PIN when reading the biometric failed, devices
that don't implement reading credProtect status for downloaded keys
and probably a few more cases that I haven't though of yet.
upstream: Update the sntrup761 creation script and generated code:
- remove unneeded header files and typedefs and rely on crypto_api.h - add
defines to map types used to the crypto_api ones instead of typedefs. This
prevents typedef name collisions in -portable. - remove CRYPTO_NAMESPACE
entirely instead of making it a no-op - delete unused functions and make the
remaining ones that aren't exported static.
projects / thirdparty / openssh-portable.git / log
