Re #41, interesting that unity-panel-service *is* installed on Touch
image though :)
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1235649
Title:
uevent spam causes session upstart to c
Right - recall that I am testing on i386 as I don't have mako/maguro :)
However, note that on Touch, the upstart code in unity8
(plugins/Unity/Indicators/indicatorsmanager.cpp) is almost identical to
the upstart code in unity-panel-service (unity/services/panel-
service.c).
Is it possible to some
Still unclear why memory is being cached in Upstart address space, but
I'm starting to think this might be a libdbus issue.
What is triggering the memory rise though is unity-panel-service:
stopping this from running results in no memory increase within the
Session Init.
** Also affects: unity (U
Persistent D-Bus connections are as expected (1 / bridge and 1 for
unity-panel-service).
Ephemeral D-Bus connections are being handled correctly and removed from the
control_conns list.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to lin
To be clear, adding a call to dbus_connection_close() in
control_disconnected() did not resolve the problem.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1235649
Title:
uevent spam cau
init/control.c:control_disconnected() could do with a call to
dbus_connection_close(), but each connection is being unref'd correctly
fwics.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1
The working theory at this stage is that
init/control.c:control_disconnected() is not cleaning up the dbus
connections created by the upstart-udev-bridge for each event it
requests upstart emit.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed
The problem is triggered by init/control.c:control_emit_event_emitted().
It looks like Upstart itself is clearing up memory, but either dbus
isn't, or upstart is missing a dbus unref call.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to li
Valgrinding the session init shows that most of the memory is correctly
freed at quiesce time.
** Attachment added: "valgrind session init consuming ~250MB (with
suppressions)"
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1235649/+attachment/3866119/+files/valgrind-250M.log.gz
--
Yo
Valgrind run as:
valgrind -v -v -v -v --show-reachable=yes --track-origins=yes --leak-
check=full init --debug --user
** Attachment added: "valgrind session init consuming ~500MB (no suppressions)"
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1235649/+attachment/3866123/+files/valgri
** Changed in: upstart (Ubuntu Saucy)
Assignee: (unassigned) => James Hunt (jamesodhunt)
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1235649
Title:
session upstart leaks mass
Attempting to run the script in #30 against a session init as below
*outside* of a the standard gui environment with no .conf files does not
trigger the issue:
$ mkdir -p /tmp/conf /tmp/log
$ init --debug --no-startup-event --confdir /tmp/conf --logdir /tmp/log --user
Same outcome running like th
I can no recreate the issue with memory not being freed by running:
http://paste.ubuntu.com/6205552/.
This is recreatable on i386 desktop images so although problem (1) in
#27 is h/w specific, problem (2) is not.
--
You received this bug notification because you are a member of Kernel
Packages,
Could someone run the "make check" tests for the libnih package on a
mako/maguro device?
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1235649
Title:
session upstart leaks massive amoun
To summarise: there are 2 problems here:
1) The kernel spamming userspace with udev messages on mako+maguro
(grouper doesn't appear to be affected but not independently verified
yet).
2) Upstart not releasing memory once the kernel has finished spamming
userspace (artificially forced by running "
>From info provided by jibel...
Forcing a log flush (#25) doesn't resolve the high memory usage (as we'd
expect, but worth ruling out).
The state file in #24 looks sane to me; forcing a re-exec of the Session
Init does cause the memory to fall to reasonable levels. This almost
feels like a compil
Can someone try running "/sbin/initctl notify-disk-writeable" and see if
the session inits memory consumption drops? Alternatively, run the
session init as "init --user --debug --no-log".
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to lin
@jibel - that get_state.out appears to have either been run as root, or
atleast outside the session init environment. Please could you re-run as
user phablet?
Also, please could you attach /var/log/udev ?
--
You received this bug notification because you are a member of Kernel
Packages, which is
Apparently, the high cpu load stop when the upstart-event-bridge is
stopped. This lends weight to the high cpu issue being caused the a
kernel udev event issue (like bug 1234743).
However, what is not explained is why the Session Init retains the
memory it consumes once the upstart-event-bridge is
Also, please could you run
http://people.canonical.com/~jhunt/upstart/scripts/get_state.sh as user
phablet and attach to this bug.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1235649
Ti
jibel - thanks. It does indeed look like bug 1234743 as the trace shows
a lot of the following events:
:sys:cpu-device-online
:sys:cpu-device-offline
:sys:power_supply-device-changed
This would explain the Session consuming a lot of CPU since those events
are generated by the kernel, picked up by
Thanks Colin. All the platform-device-changed events upstart is emitting
are from omapfb.ko:
KERNEL=omapfb
DEVPATH=/devices/platform/omapfb
SUBSYSTEM=platform
ACTION=change
DRIVER=omapfb
MODALIAS=platform:omapfb
SEQNUM=13545
USEC_INITIALIZED=1146951
VSYNC=12601146402237
So certainly looks like a
Hi Colin,
This issue looks similar to bug 1201865 - Upstart is being spammed by
the upstart-udev-bridge which itself is being spammed by all the
SUBSYSTEM=platform udev events from the kernel.
I cannot recreate the issue you are seeing on my nexus 7, so that
coupled with the fact that the event a
When did this problem first start?
If pressing the caps-lock key does not result in the caps-lock light
coming on, this sounds like it might be a kernel issue. Just after
starting a reboot, try pressing the escape key to view shutdown
messages.
** Also affects: linux (Ubuntu)
Importance: Unde
Public bug reported:
Upstart now manages the Touch session via an Upstart Session Init. To
allow that sub-init running as a non-priv user to receive all SIGCHLD's
for processes it starts, it attempts to use PR_SET_CHILD_SUBREAPER.
However, that prctl was added in kernel 3.4 and the kernel on my Ne
Public bug reported:
network operations spam console with repeated messages:
[ 3554.436766] brcmsmac bcma0:0: brcms_c_ampdu_dotxstatus_complete: ampdu tx
phy error (0x1)
[ 3554.437842] brcmsmac bcma0:0: phyerr 0x1, rate 0xa0a
[ 3554.437852] brcmsmac bcma0:0: brcms_c_ampdu_dotxstatus_complete: am
Re #27, removing -nographic still causes the same lockup for me.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1208455
Title:
general protection fault running apt-get inside double nest
'qemu -serial file:/tmp/kvm.log' Looks like an alternative to 'qemu
-nographic >/tmp/kvm.log 2>&1'.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1208455
Title:
general protection fault
The latest advice for running on Canonistack (kvm guest instance where
nested=Y is to):
- Only run on a 64-bit Canonistack guest instance.
- Modify bin/prepare-testbed and bin/run-adt-test to start the autopkgtest VM
using 'qemu-system-x86_64 -enable-kvm'.
- Have the DEP8 test running inside the
Latest kernel panic (from 2nd-level guest) using following kernel
versions:
host: 3.11.0-1-generic (-2 causes problems for my graphics hardware atm)
1st-level guest: 3.11.0-2-generic
2nd-level guest: 3.11.0-2-generic
** Attachment added: "kvm.log"
https://bugs.launchpad.net/ubuntu/+source/li
To recreate:
1) Follow instructions on internal wiki (page
"InformationInfrastructure/IS/CanonicalOpenstack") to create a keypair.
2) Create an guest instance:
$ keyname="${USER}_lcy01"
$ image=$(euca-describe-images|grep saucy|head -1|awk '{print $2}')
$ euca-run-instances -k $keyname -t m1.med
The kernel BUG: above happens when running "bin/prepare-testbed -r saucy -d
i386" from lp:auto-package-testing in canonistack:
the kvm instance spins at ~200% CPU again.
I modified bin/prepare-testbed slightly to log the output of the vm and
disable byoubu. See attached.
** Patch added: "auto-p
Now that 3.11 is out of proposed, we're seeing a different kernel
failure, starting with
run-parts: executing /etc/kernel/postinst.d/zz-update-grub 3.11.0-2-generic
/boot/vmlinuz-3.11.0-2-generic
Generating grub.cfg ...
Found linux image: /boot/vmlinuz-3.11.0-2-generic
Found initrd image: /boot/i
I got the panic again last night. This was running a 3.11 kernel for the
host, but of course the 1st and 2nd level guests were still running 3.10
as 3.11 is still in -proposed. So, although it's possible to get a test
run with (3.11, 3.10, 3.10) I really need to try running with (3.11,
3.11, 3.11)
Testing with linux-image-3.11.0-0-generic from -proposed with nested=Y
in all 3 environments does appear to solve this issue.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1208455
Title:
Fresh kernel panic from 2nd-level nested VM running with host, 1st level
and 2nd level all running with nested=1.
** Attachment added: "kvm.log"
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1208455/+attachment/3763650/+files/kvm.log
--
You received this bug notification because you a
What we know so far:
- the issue is not related to cloud-init.
- the very high CPU load was atleast partly caused by byobu bug 1208853. To
disable byobu...
=== modified file 'bin/prepare-testbed'
--- bin/prepare-testbed 2013-08-06 08:05:16 +
+++ bin/prepare-testbed 2013-08-06 14:01:42 +
High CPU may be cloud-init. Using todays server image, running 'prepare-
testbed' spins the CPU at 200%. The end of the log shows:
Aug 6 10:13:15 autopkgtest [CLOUDINIT] helpers.py[DEBUG]: Running
config-package-update-upgrade-install using lock ()
Aug 6 10:13:15 autopkgtest [CLOUDINIT] util.py
** Changed in: linux (Ubuntu)
Status: Incomplete => Confirmed
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1208455
Title:
general protection fault running apt-get inside nested
Note that #4 was taken from a local nested kvm environment (not
OpenStack).
** Summary changed:
- general protection fault running apt-get inside nested kvm VM on OpenStack
+ general protection fault running apt-get inside nested kvm VM
--
You received this bug notification because you are a me
*** This bug is a duplicate of bug 1208455 ***
https://bugs.launchpad.net/bugs/1208455
Public bug reported:
Fighting with apport (bug 1208508). The attachments should really be
added to bug 1208455.
ProblemType: Bug
DistroRelease: Ubuntu 13.10
Package: linux-image-3.10.0-6-generic 3.10.0-6.1
Kernel logs can be found on bug 1208509.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1208455
Title:
general protection fault running apt-get inside nested kvm VM on
OpenStack
Statu
** Attachment added: "Another general protection fault panic"
https://bugs.launchpad.net/ubuntu/+source/qemu-kvm/+bug/1208455/+attachment/3761555/+files/kvm.log
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://b
I'm suspecting an issue with nested=1... the nested VM does eventually
boot but it takes many minutes to do so. At this point, the cpu usage of
qemu-kvm drops to reasonable values. But if you ssh into the vm, the
usage jumps up to ~200% again.
To recreate:
$ sudo apt-get install -y genisoimage li
Attempting to repeat the dep8 test run in the same environment has so
far resulted in the qemu-kvm binary consuming 200% CPU (recreated on
multiple nodes).
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launc
Public bug reported:
Attempting to run apt-get in a nested vm hosted in OpenStack results in
a general protection fault in the kernel in the nested VM.
** Affects: linux (Ubuntu)
Importance: Undecided
Status: New
** Affects: qemu-kvm (Ubuntu)
Importance: Undecided
St
Looks like this issue may in fact be bug 1203211.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1203751
Title:
ibm fan control broken and iwlwifi no longer accepts 11n_disable
option
Public bug reported:
The 3.10.0-4-generic update broke a few things for me:
1) IBM thinkpad acpi fan-control no longer works:
$ sudo modprobe thinkpad_acpi fan_control=1
ERROR: could not insert 'thinkpad_acpi': Unknown symbol in module, or unknown
parameter (see dmesg)
$ dmesg|grep -i fan
101 - 148 of 148 matches
Mail list logo