Before the `scheduler` plugin, `tuna` was used for CPU isolation.
There is a leftover use in the verification of the `realtime` profile.
Removing that, the dependency on the `tuna` package, and now-empty
scripts from the realtime profiles.
closes#594
Signed-off-by: Adriaan Schmidt <adriaan.schmidt@siemens.com>
Just explain what isolated_cores= is for and recommend to only have one
isolated_cores= line to avoid wrong system configuration.
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
The interruptions avoided by the rtentsk logic:
"Add plugin to open a socket with SOF_TIMESTAMPING_OPT_TX_SWHW.
This is for avoiding interruptions due to static key IPIs due
to opening socket with timestamping enabled (by opening a
socket ourselves the static key is kept enabled)."
Are also desired on systems that use cpu-partitioning.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Commit 49cc817871
("realtime: Set tsc as 'reliable'") reads:
"We already disable the clocksource watchdog, it's there to double-check
whether clock results are coherent between reads. But a new reliability
test was introduced in recent kernels[1], tracking tsc drift between
CPUs[2]. It works by programming timers on all CPUs, including isolated
ones, and checks whether the tsc was adjusted in between runs. This
introduces unwarranted latency on real time systems. So let's promote
tsc to 'reliable', which disables both checks."
The timer is also undesired interruption to systems that use cpu-partitioning.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
The following sysctl settings are the same between cpu-partitioning and
realtime profiles:
kernel.hung_task_timeout_secs = 600
kernel.nmi_watchdog = 0
vm.stat_interval = 10
kernel.timer_migration = 0
Unify them in the network-latency profile, parent of both profiles.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
The calc_isolated_cores function expands to the list of cores to
isolate. It accepts optional argument which specifies how many cores
from each socket reserve for housekeeping. If not specified, one core
from each socket is reserved for housekeeping.
Example:
Machine with 2 sockets, each 4 cores, using the following user variable
configuration file, e.g. /etc/tuned/realtime-variables.conf:
isolated_cores=${f:calc_isolated_cores:2}
It will expand to:
isolated_cores=2, 3, 6, 7
I.e. cores 0, 1 and 4, 5 will be used for housekeeping.
Also updated the TuneD profiles to use this builtin function for
isolated_cores preset.
Resolves: rhbz#2093847
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
We already disable the clocksource watchdog, it's there to double-check
whether clock results are coherent between reads. But a new reliability
test was introduced in recent kernels[1], tracking tsc drift between
CPUs[2]. It works by programming timers on all CPUs, including isolated
ones, and checks whether the tsc was adjusted in between runs. This
introduces unwarranted latency on real time systems. So let's promote
tsc to 'reliable', which disables both checks.
There shouldn't be any impact on old nor new setups. First, this test
didn't exist in the past, so any eventual drift was left uncorrected.
Second, most telco and HPC use-cases will make heavy use of tsc in
user-space, and its reliability is already assumed in the system's
design.
[1] c7719e793478 x86/tsc: Add a timer to make sure TSC_adjust is always
checked
[2] Generally caused by rogue BIOSes adjusting the tsc on a CPU and not
doing so on the rest
Signed-off-by: Nicolas Saenz Julienne <nsaenzju@redhat.com>
Enable 'isolate_managed_irq' parameter by default. It is important
for -realtime systems. When it is enabled, kernel directs interrupts
to the housekeeping CPUs and thus reduces latencies for the isolated
ones.
Fixes: RHBZ#1820626
tuned(8) realtime-virtual-{host|guest} profiles set Kernel and
KVM module parameters via script.sh plugin. These parameters are
to be verified in a verify() function, invoked by tuned-adm verify
command.
This patch updates verify() functions to validate KVM module
parameters.
It moves kernel parameters to the tuned.conf file under the
[sysfs] & [sysctl] plugin sections. And removes call to
disable_ksm function, no longer required.
Fixes: RHBZ#1947858
This allows the user to adjust the queue count for each
network device.
Also add check_net_queue_count built-in function.
The functions check if the user has specified any
queue count value or not. If no value is passed the function
returns the number of housekeeping CPUs as the queue count.
Resolves: rhbz#1846767
Signed-off-by: Nitesh Narayan Lal <nitesh@redhat.com>
Use the new irqbalance plugin instead of a script to set up
irqbalance.
I'm placing the 'irqbalance' instance before the 'script' instance so
that it's executed at the same time as the script, however I don't
think the 'irqbalance' instance needs to be ordered in any specific
way and could be placed almost anywhere.
Resolves: rhbz#1784645
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
Also added regex_search_ternary built-in function.
It takes arguments in the following form:
STR1, REGEX, STR2, STR3
If REGEX matches STR1 (re.search is used), STR2 is returned,
if it doesn't match STR3 is returned.
Example:
[variables]
foo=Y
bar=${f:regex_search_ternary:${foo}:\b[y,Y,1,t,T]\b:foo:bar}
It will result in the 'foo' string stored in the '${bar}' variable.
Resolves: rhbz#1797025
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
Make sure the isolated_cores and no_balance_cores variables are
defined before any of the variables that use them are defined. This
enforces a certain ordering of variable expansions so that child
profiles can set the variables directly in the profile (tuned.conf),
e.g.:
[main]
include=cpu-partitioning
[variables]
isolated_cores=3
Resolves: rhbz#1781664
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
This option disables the TSC clocksource watchdog
for isolated CPUs (which avoids interruptions
by the per-CPU clocksource watchdog timer).
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Realtime profiles require to restrict irqbalance too. Apply them to
the realtime profile so that realtime-virtual-{host|guest} can benefit
from that too.
Signed-off-by: Peter Xu <peterx@redhat.com>
In the cpu-partitioning profile, IRQ affinity change is already done
as part of the 'isolated_cores' option of the scheduler plugin, so
calling defirqaffinity is, at best, redundant. So let's remove the call.
In the realtime* profiles, it's essentially the same, except that tuna
was used instead of isolated_cores. So let's use built-in functionality
instead of tuna and drop calls to defirqaffinity.
Resolves: rhbz#1590937
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
Also removed the parameter from the realtime profile, because it already
includes the network-latency profile.
Resolves: rhbz#1451073
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
By default the timer tick will fire at the same time on all
CPUs. This causes the jiffies_lock to get contended in the
real-time kernel, as it's a raw spinlock. skew_tick=1 causes
the timer tick to fire at different times on all CPUs,
fixing the contention.
Note skew_tick=1 usage is going to be propagated to the
NFV profiles as well.
Resolves: rhbz#1447938
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Error message about no isolated cores set wasn't displayed,
because the current plugin_script code only logs messages written
to STDERR.
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
kernel.sched_rt_runtime_us - a global limit on how much time realtime
scheduling may use. A run time of -1 specifies runtime == period, ie. no limit.
Resolves: rhbz#1346715
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
It was found during code review that the kernel always try to
migrate timers away from idle CPUs to busy CPUs. Real-time
CPUs running dpdk polling threads for example, are busy so
they are a great candidate to receive timers from idle CPUs
(see get_nohz_timer_target() in the kernel sources).
It is not clear how this problem can affect real-time
applications and KVM-RT. However, for real-time and
high-performance the best thing is for a timer to always
fire in the CPU where it was created.
Resolves: rhbz#1323283
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Added new tuned-adm functionality 'profile_info'. It has optional parameter
'profile_name'. If called without parameter, active profile is used.
It outputs name of the profile it is querying, profile summary, and profile
description. If there is no profile summary or description available it
outputs empty string(s).
Extended tuned-adm 'list' functionality. It now displays profile summary
in the list (if available).
Added summary to profiles.
Extended D-Bus API to support the new feature and for better Cockpit support:
- Added 'profile_info' method. It gets profile_name argument and it queries
the requested profile for information. If profile_name is empty string it
queries the active/selected profile. It returns tuple of the following
format (status, profile_name, summary, description), where status is
boolean. If it is False, the query mechanism failed and the other fields
don't contain valid data
- Added 'profiles2' method. It's extension of 'profiles' method. It returns
list of tuples. Tuples have following format:
(profile_name, profile_summary)
Related: rhbz#1228356
Fixes: #46
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
The latest RHEL and RT kernels got support for setting the cpumask of unbound
workqueues in sysfs (https://bugzilla.redhat.com/show_bug.cgi?id=1176155).
This patch sets the unbound workqueues cpumask to the list of non-isolated
cores for the real-time profiles.
Patch from Luiz Capitulino <lcapitulino@redhat.com>
Resolves: rhbz#1259043
Signed-off-by: Jaroslav Škarvada <jskarvad@redhat.com>
