Theory sounds right. Were you able to confirm this theory and fix your code to handle empty responses?
Thu, Oct 10
Tue, Oct 1
Wed, Sep 25
I think this was addressed by https://gitlab.com/libssh/libssh-mirror/merge_requests/63 for OpenSSL.
I think we should update the documentation so it is clear that the structure needs to be valid as long as the channel/session is valid.
Tue, Sep 24
Fri, Sep 20
Can you clarify on what platform you are experiencing these issues?
Thank you for the bug report. Indeed, the handling of the modulus and generator is wrong. These bignums are copied in the ssh_dh_set_parameters() into the keypair structures (when using openssl backend), but the calling function did not free them as expected. Also the handling of errors was wrong in case of some failures checking them.
Thu, Sep 19
The tests/pkd/pkd_util.c already parses the openssh version so it can be used with a bit of refactoring. Or you can adjust the tests/CMakeLists.txt to expose the already-collected versions also the pkd tests and allow these alogirhtms only on the newer openssh.
Wed, Sep 18
Thank you for testing. Whether and where to list your perl bindings in the main page, that is more up to @asn, but I do not think it should be a problem.
Tue, Sep 17
Thank you for the reproducer. I can confirm and reproduce with your attached code. This is the backtrace:
#0 0x00007f8df8576c10 in __poll_nocancel () from /lib64/libc.so.6 #1 0x00007f8df14d58c0 in ssh_poll () from /lib64/libssh.so.4 #2 0x00007f8df14d6033 in ssh_poll_ctx_dopoll () from /lib64/libssh.so.4 #3 0x00007f8df14d798c in ssh_handle_packets () from /lib64/libssh.so.4 #4 0x00007f8df14d7a93 in ssh_handle_packets_termination () from /lib64/libssh.so.4 #5 0x00007f8df14ae889 in channel_request () from /lib64/libssh.so.4 #6 0x00007f8df14afa6b in ssh_channel_request_exec () from /lib64/libssh.so.4 #7 0x00007f8df1740837 in XS_Libssh__Session_ssh_channel_request_exec () from /usr/lib64/perl5/vendor_perl/auto/Libssh/Session/Session.so #8 0x00007f8df989c41f in Perl_pp_entersub () from /usr/lib64/perl5/CORE/libperl.so #9 0x00007f8df9894b96 in Perl_runops_standard () from /usr/lib64/perl5/CORE/libperl.so #10 0x00007f8df9831985 in perl_run () from /usr/lib64/perl5/CORE/libperl.so #11 0x0000000000400ce9 in main ()
In the server, log the event of failed session opening is visible here:
Sep 17 14:59:42 jjelen-rhel-7.3 sshd: debug1: channel 10: new [server-session] Sep 17 14:59:42 jjelen-rhel-7.3 sshd: debug1: session_open: channel 10 Sep 17 14:59:42 jjelen-rhel-7.3 sshd: error: no more sessions Sep 17 14:59:42 jjelen-rhel-7.3 sshd: debug1: session open failed, free channel 10 Sep 17 14:59:42 jjelen-rhel-7.3 sshd: debug1: channel 10: free: server-session, nchannels 11 Sep 17 14:59:42 jjelen-rhel-7.3 sshd: debug3: channel 10: status: The following connections are open:\r\n ... Sep 17 14:59:42 jjelen-rhel-7.3 sshd: debug1: server_input_channel_open: failure session
Mon, Sep 16
I implemented the match exec. Can you try if it works for you as expected in your use case? There are several commits adjusting the tests and as well as I had to implement new token parsing function, but my basic tests looked good. This code also does not work on windows since I am not Windows developer, but if interested, I believe there will be somebody who could implement that.
The original issue should be resolved in master now.
Aris, can you open a new bug for this, ideally with more debug information as proposed by @ansasaki?
Sep 13 2019
Can you check the following patch if t addresses the issue for you?
I am getting exactly the same result as the OP when I remove the /usr/bin/nc, which is used in the respective failing test so I assume that this is the issue, but the error should be more properly reported and the test probably skipped in case the netcat is not in place. I will submit a patch.
Thank you for having a look into that. In that case, I am wondering why it did work for me and for the CI we run, but I think most of that is on Fedora, which might be a difference. Can you check whether the netcat (nc) is installed in your system?
I just tested the ssh-client from examples, which is using channel_open with sshd configured with MaxSessions 0 and it is correctly failing and not hanging for me. Can you test with current libssh master or 0.9 and provide more ellaborate reproducer?
The file doc/sftp.dox contains the following information (not sure whether it is rendered somewhere on the web):
This should be addressed by using the new API even inside of the deprecated functions without the change of functionality. Can you try the following patch?
The modification of the libssh.h is breaking other applications (for example the applications in example directory) using this header file not defining HAVE_INTTYPES_H and HAVE_UNISTD_H themselves. Can you clarify what problem are you solving by this and on what system you do not have these header files?
Did you try the latest libssh 0.9 . Not sure what went wrong with the old version, but the windows builds are part of CI and they should work in recent version
Aug 20 2019
I think the multi-criteria matches are generally supported, but I agree that they are not tested and that should be fixed.
Aug 8 2019
Aug 7 2019
Just for the record, the OpenSSH position to implementing this key exchange algorithm:
Aug 1 2019
Jul 30 2019
Jul 25 2019
You are probably right, Can you try with the following patch:
Jun 27 2019
Jun 26 2019
The current version should have improved memory handling. Can you retest whether the speed is better now?
Jun 24 2019
Thank you for the clarification and updated reproducer. I can reproduce it with the latest version installed by the package manager in Debian (0.8.7), but I can not reproduce it when I build the example against current master. I can not reproduce it even if I manually checkout the version 0.8.7 from git.
This is already available in master (e989c4afffa154d92fe8c4ae1716ecc6bb4c2fd5) and will be in 0.9. Unfortunately, this did not got updated in Fedora as we updated the default configuration file so I would propose to wait few days for the updated libssh or fill a fedora bug.
Jun 21 2019
Jun 19 2019
FYI, as a workaround, you can use ProxyCommand='ncat -vvv --proxy <proxy server:port> --proxy-type socks5 %h %p' -o'User <user> 2>/dev/null', which drops the stderr. But indeed, this is something to fix in libssh.
This was resolved recently by 4a67c191188601c8d98ec73b9d53ce8a2b6c6cf8 which implemented EtM modes of the MACs in libssh and which is available in latest libssh release.
Can you provide example of such key? Do you mean private keys in PEM format or in OpenSSH format? Or public keys?
I think this got resolved over the time and the PEM files are readable including the inline comments (probably because of the removal of the SSH1 protocol, which had special conditions for the old RSA1 keys).
Please, check the following commits in the master-fix branch:
Do you have a minimal reproducer that demonstrates this issue? The following parts do not look correct:
[2019/02/21 16:35:31.187831, 3] callback_receive_banner: ssh_protocol_version_exchange: Ncat: Version 7.70 ( https://nmap.org/ncat )
a bit later is what should be coming -- the SSH identification banner:
[2019/02/21 16:40:54.932576, 3] callback_receive_banner: ssh_protocol_version_exchange: libnsock nsock_trace_handler_callback(): Callback: READ SUCCESS for EID 10 [peer unspecified] (33 bytes): SSH-2.0-1.36_sshlib GlobalSCAPE..`
I am afraid that the ncat dumps some additional data to stderr, which is unintentionally picked up by the libssh. This is caused by the code in the src/socket.c:815, where both stdout and stderr is redirected to libssh
You ca try to drop the second line above, which should let the stderr to be mixed up with the libssh stderr or dropped (not sure from top of my head), but it should let you move on.
The same thing works for me with the current openssh-portable master that I just built in my Fedora VM. Could this be somehow related to the Ubuntu toolchains or packages, rather than to the 32 bits itsef?
I tried also the 0.8.7 branch as reported in the previous comment, but with the same result. Could it be an issue in valgrind or some other part of the toolchain?