libssh dep8 fails on ppc64el/s390x

Bug #2024064 reported by Sergio Durigan Junior
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
libssh (Ubuntu)
In Progress
High
Sebastien Bacher

Bug Description

Reporting this bug for documentation purposes. libssh's dep8 test is failing on ppc64el/s390x.

I tried reproducing it in a ppc64el canonistack box, but the test always passes there. migration-reference/0 didn't help.

It's interesting to notice that the test succeeds on debci, which leads me to believe that this is something really particular to our autopkgtest infra.

Revision history for this message
Sebastien Bacher (seb128) wrote :

Trying on a canonistack s390x instance the tests fail after installing 0.10.5-2

the journal has that error

s390x kernel: User process fault: interruption code 0010 ilc:3 in mock-sshd[2aa2a880000+6000]
s390x kernel: Failing address: 000003ffce5ff000 TEID: 000003ffce5ff800
s390x kernel: Fault in primary space mode while using user ASCE.
s390x kernel: AS:000000004921c1c7 R3:0000000013d18007 S:0000000000000020
s390x kernel: CPU: 0 PID: 1002265 Comm: mock-sshd Not tainted 5.15.0-53-generic #59-Ubuntu
s390x kernel: Hardware name: IBM 2964 N63 400 (KVM/Linux)

Revision history for this message
Sebastien Bacher (seb128) wrote :

The kernel msg is just a sign of the program hitting an error

Doing
$ ./mock-sshd --user=bug --port 1234 -v

attaching gdb

and then issuing the client cmd
$ ssh -i test_rsa -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -p 1234 bug@localhost cat /etc/os-release

leads to this server segfault

Program received signal SIGSEGV, Segmentation fault.
0x000002aa02e02e84 in fd_data (fd=<error reading variable: Cannot access memory at address 0x3ffce57fe94>,
    revents=<error reading variable: Cannot access memory at address 0x3ffce57fe90>,
    user_data=<error reading variable: Cannot access memory at address 0x3ffce57fe88>) at mock-sshd.c:164
164 {
(gdb) bt
#0 0x000002aa02e02e84 in fd_data (fd=<error reading variable: Cannot access memory at address 0x3ffce57fe94>,
    revents=<error reading variable: Cannot access memory at address 0x3ffce57fe90>,
    user_data=<error reading variable: Cannot access memory at address 0x3ffce57fe88>) at mock-sshd.c:164
#1 0x000003ffa9b32128 in ssh_poll_ctx_dopoll (ctx=ctx@entry=0x2aa04d545a0, timeout=timeout@entry=-1)
    at /usr/src/libssh-0.10.5-2/src/poll.c:736
#2 0x000003ffa9b33732 in ssh_handle_packets (session=session@entry=0x2aa04d54ae0, timeout=timeout@entry=-1)
    at /usr/src/libssh-0.10.5-2/src/session.c:682
#3 0x000003ffa9b338e4 in ssh_handle_packets_termination (session=session@entry=0x2aa04d54ae0, timeout=<optimized out>,
    fct=fct@entry=0x3ffa9b33d30 <ssh_flush_termination>, user=user@entry=0x2aa04d54ae0)
    at /usr/src/libssh-0.10.5-2/src/session.c:751
#4 0x000003ffa9b33dc6 in ssh_blocking_flush (session=0x2aa04d54ae0, timeout=<optimized out>)
    at /usr/src/libssh-0.10.5-2/src/session.c:550
#5 0x000003ffa9b1a0dc in ssh_channel_flush (channel=0x2aa04db9bd0) at /usr/src/libssh-0.10.5-2/src/channels.c:1433
#6 ssh_channel_send_eof (channel=0x2aa04db9bd0) at /usr/src/libssh-0.10.5-2/src/channels.c:1316
...

Revision history for this message
Sebastien Bacher (seb128) wrote :

Bisecting it seems the issue started in https://gitlab.com/libssh/libssh-mirror/-/commit/bb5bdac3

Revision history for this message
Sebastien Bacher (seb128) wrote :

I've reported the issue and my findings upstream on https://gitlab.com/libssh/libssh-mirror/-/issues/202

Changed in libssh (Ubuntu):
importance: Undecided → High
assignee: nobody → Sebastien Bacher (seb128)
status: New → In Progress
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.