On SEL guests, the default configuration does not permit the
Control-Program Identifiation (CPI) data to be sent by the cpi
service, resulting in a failure message in systemctl/log output
on start of the cpi service.
This is not a failure because it is expected behavior and the
cpi service is a one-shot service whose function is already
performed when the exit code is returned. Prevent the error log
by filtering the associated exit code in systemd.
Signed-off-by: Shalini Chellathurai Saroja <shalini@linux.ibm.com>
Acked-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The cpictl utility sends control-program identification data
from protected virtualization guests to hosts by default.
This behaviour leaks the below potentially sensitive
information to untrusted hosts.
- system_type
- system_level
- sysplex_name
- system_name
To prevent this behaviour, enhance the cpictl utility to stop
setting CPI information on protected virtualization guests by
default. If the user chooses to set the CPI information, it
could be set by one of the below options
- use the command line option --permit-cpi
- set the environment variable CPI_PERMIT_ON_PVGUEST to 1 to
control the CPI service behaviour during boot
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Shalini Chellathurai Saroja <shalini@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
cpacfstatsd can be run successfully on z/VM and KVM mashines, but the
service unit did not allow that. A user would have to start the
cpacfstats daemon by hand every time until now.
Acked-by: Steffen Eiden <seiden@linux.ibm.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: Finn Callies <fcallies@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The optics monitoring tool opticsmon implements the user-space portion
of reporting optics data to the SE. Its basic functionality is to
collect optical module information equivalent to "ethtool --module-info"
for PCI Physical Functions and forwards this data to the SE using the
new SCLP Write Event Data Action Qualifier 3.
For the part of finding all PFs we need to look at all PCI
functions and determine which ones are PFs and what netdevs they
correspond to. This is a generally useful functionality so this part as
well as the SCLP issuing code go into a new libzpci library which also
includes a standalone example for listing PCI functions and their s390x
specific attributes. Medium term we plan to add this functionality to
lszdev.
For the opticsmon tool itself there are 2 basic operating modes:
* One-shot Mode: Without parameters opticsmon collects optical module
data and prints a summary of the netdevice in JSON format. With
--module-data it also includes a base64 encoded raw dump equivalent to
ethtool --module-info <netdev> raw on.
* Monitor Mode: With the --monitor flag opticsmon runs continuously
usually started via a systemd unit and collects new optical module
data on a time interval (default 24h) or when the operational state
("/sys/class/net/<netdev/operstate") changes. The tool listens for
changes via netlink so no polling on sysfs is necessary
Note: Both modes will *NOT* issues SCLPs without adding the
--send-report flag but will output a JSON summary for each data
collection so can be tested without firmware impact.
Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Until recently dumpconf might have been installed in two different
locations, /etc/init.d/ and /lib/s390-tools/. As dumpconf is now part of
the helper scripts in the scripts directory, extend the installation
rule and add the %S390_TOOLS_VERSION% wildcard to the script along the
way for correct version output.
The installation rule in the systemd/Makefile is removed, bringing it in
line with cpictl.
Acked-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Modern distributions use systemd and a corresponding systemd unit file
is provided that handles the dumpconf tool. The tool is also the last
script left in etc/init.d. Move it to scripts and get rid of the init.d
directory altogether.
Acked-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Noticed on RHEL CoreOS that the CPI service failed to start with:
"Cannot access lock file: /var/lock/cpictl.lock"
This was a timing issue where /var was not created yet and the symlink
to /run/lock was not present. Add sysinit.target to fix this.
This fix has been tested with RHEL CoreOS.
Closes: https://github.com/ibm-s390-tools/s390-tools/pull/82
Signed-off-by: Prashanth Sundararaman <psundara@redhat.com>
Suggested-by: Colin Walters <walters@verbum.org>
Reviewed-by: Dan Horák <dan@danny.cz>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Modern distributions do not provide the symlink /sbin/udevadm any more.
Use /usr/bin/udevadm instead for the example init script / systemd units.
Note that those are meant for example purpose and would need to be adjusted
by distributions, so no extra effort is taken to determine the actual
location of udevadm.
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
This fixes the same issue as in commit 82c8148983 ("cpuplugd: Improve
systemctl start error handling") for mon_tools (mon_procd and mon_fsstatd).
Currently "systemctl start mon_procd/fsstatd" does not report any errors
in case the startup fails.
Example (with mon_procd):
(change interval in /etc/sysconfig/mon_procd to an invalid value "abc")
# systemctl start mon_procd
The reason is that for type=simple systemd forks/execs mon_procd and if
that is successful immediately returns. There is no way to find out if the
initial startup fails.
Fix this by using type=fork and running the process in the background. In
this case systemd waits until the initial process returns.
In addition use PIDFile and ensure that the pid file is already available
when the initial process returns. To achieve this, use startup
synchronization via pipe. Without that systemd would print the following
warning:
systemd[1]: mon_procd.service: PID file /var/run/mon_procd.pid not readable
(yet?) after start: No such file or directory
With this patch, an early startup error like in the example above, is now
reported correctly in "systemctl start":
# systemctl start mon_procd
Job for mon_procd.service failed because the control process exited...
See "systemctl status mon_procd.service" and "journalctl -xe" for ...
# journalctl -xe | grep mon_procd
mon_procd[3184]: Error: Invalid interval (needs to be greater than 0)
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Currently "systemctl start cpuplugd" does not report any errors in
case the startup fails.
Example:
# mv /etc/cpuplugd.conf /etc/cpuplugd.conf.xxx
# systemctl start cpuplugd
The reason is that for type=simple systemd forks/execs cpuplugd and if
that is successful immediately returns. There is no way to find out
if the initial startup fails.
Fix this by using type=fork and running cpuplugd in the background. In
this case systemd waits until the initial process returns.
In addition use PIDFile and ensure that the pid file is already available
when the initial cpuplugd process returns. To achieve this, replace the
daemon() function by our own implementation that introduces startup
synchronization via pipe. Without that systemd would print the following
warning:
systemd[1]: cpuplugd.service: PID file /var/run/cpuplugd.pid not readable
(yet?) after start: No such file or directory
With this patch, an early startup error like in the example above, is now
reported correctly in "systemctl start":
# systemctl start cpuplugd
Job for cpuplugd.service failed because the control process exited...
See "systemctl status cpuplugd.service" and "journalctl -xe" for ...
# journalctl -ex | grep cpuplugd
Nov 16 15:52:27 ... cpuplugd[5096]: Opening configuration file failed:
No such file or directory
Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Acked-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
The cpi service unit requires an Install section to tell systemd
when to start the cpi service. Add an install section and start
cpi for the multi-user target.
Reported-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Stefan Haberland <sth@linux.vnet.ibm.com>
Add a new udev rule that is triggered when a KVM VM is started
on the system the very first time. This rule reports over the
sysfs CPI kernel interface that the system is a KVM host.
Currently the cpi.service systemd unit already works with
the CPI sysfs interface. Introduce a new internal utility
script "cpictl" that is used by both the udev rule and
the systemd unit to guarantee coordinated access to the
kernel CPU interface.
Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>
Reviewed-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Reviewed-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
This commit is based on the s390-tools-1.39.0 version.
Changes on top of s390-tools-1.39.0:
- Add MIT license to all source files
- Add LICENSE file
- Transform REAMDE to README.md (markdown)
- Add AUTHORS.md file
- Add CONTRIBUTING.md file
- Move changelog from README to CHANGELOG.md file
Reviewed-by: Stefan Haberland <sth@linux.vnet.ibm.com>
Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
