Just to say, I did retry the reproducer test and also re-ran the adt
tests to double check that this no longer fails on 5.4.0-12.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
AppleTalk is disabled on focal s390x 5.4.0-12 kernels so this bug cannot
be tripped. Marking this as fixed released even though it's not a direct
fix, it does stop the issue.
** Changed in: linux (Ubuntu)
Status: Triaged => Fix Released
--
You received this bug notification because you
** Changed in: linux (Ubuntu)
Importance: High => Low
** Changed in: linux (Ubuntu)
Status: Incomplete => Triaged
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
stress-ng
See: https://lkml.org/lkml/2019/12/5/476
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
stress-ng sctp stressor breaks 5.4.0.7-8 on s390x
To manage notifications about this bug go
This goes right back to 4.6.x:
.6.7 crash (see below)
4.7.10 crash in xfrm6_dst_ifdown
4.8.17 crash in xfrm6_dst_ifdown
4.12.14 crash (see below)
4.13.16 reports "unregister_netdevice: waiting for eth0 to become free. Usage
count = 2"
4.14.157 reports "unregister_netdevice: waiting for eth0 to
Ah, fails on 5.2.0-15-generic, 5.3.0-18 generic too. Appears that the
regression test was enabled quite recently:
commit b5b9181c2403025b2c7ae7ea44333fd8fe6dbb54 (between 5.4-rc3 and 5.4-rc4)
Author: David Ahern
Date: Mon Oct 21 19:02:43 2019 -0600
selftests: Make l2tp.sh executable
Occurs between 5.3 and 5.4-rc1
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
stress-ng sctp stressor breaks 5.4.0.7-8 on s390x
To manage notifications about this bug go to:
Easy steps to reproduce this issue:
sudo modprobe l2tp_core
sudo ./linux-5.4.0/tools/testing/selftests/net/l2tp.sh
./close
where close is compiled from:
#include
#include
#include
#include
int main()
{
int fd;
printf("calling socket..\n");
fd = socket(AF_APPLETALK,
The unregister_netdevice issue occurs running the kernel self test in
testing/selftests/net/l2tp.sh after modprobing the l2tp driver. A hang
can be the produced by running the stress-ng close stressor, this is
just expediting an eventual hang caused by this test.
--
You received this bug
I added a background task to dump out new dmesg messages and I now see
messages such as the following *before* any stress-ng tests run. I think
we can therefore assume the damage to the kernel occurred in prior ADT
tests.
11:02:46 DEBUG| [stdout] [ 3093.210307] unregister_netdevice: waiting
for
** Changed in: linux (Ubuntu)
Assignee: (unassigned) => Colin Ian King (colin-king)
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
stress-ng sctp stressor breaks 5.4.0.7-8 on
And on an arm64 platform we have something similar:
15:55:45 DEBUG| [stdout] Number of CPUs: 4
15:55:45 DEBUG| [stdout] Number of CPUs Online: 4
15:55:45 DEBUG| [stdout]
15:55:45 DEBUG| [stdout] access STARTING
15:55:49 DEBUG| [stdout] [ 7016.776865] unregister_netdevice: waiting for eth0
to
This makes sense as the af-alg stressor now exercises a far wider set of
crypto engines.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
stress-ng sctp stressor breaks 5.4.0.7-8 on
The unregister_netdevice message appears after the af-alg stressor
starts, so it maybe a crypto algo that is the root cause:
14:34:33 DEBUG| [stdout] af-alg STARTING
14:34:35 DEBUG| [stdout] [ 2895.954700] unregister_netdevice: waiting for lo to
become free. Usage count = 1
--
You received
net/core/dev.c netdev_wait_allrefs() states:
**
* netdev_wait_allrefs - wait until all references are gone.
* @dev: target net_device
*
* This is called when unregistering network devices.
*
* Any protocol or device that holds a reference should register
* for netdevice notification, and
This stress test has not changed much lately, so I'm assuming this is a
racy kernel regression.
Last stress-sctp changes in stress-ng were:
commit 27b045a498b360ccbc761c3b62e3dd38dd744f09
Author: Colin Ian King
Date: Sat Aug 10 13:25:34 2019 +0100
stress-sctp: voidify unused return
I can't easily reproduce this on a s390 VM instance.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1854968
Title:
stress-ng sctp stressor breaks 5.4.0.7-8 on s390x
To manage notifications about
17 matches
Mail list logo