Hi,
On Fri, 25 Apr 2025, Sheng Liu wrote:
>
> If an RT process needs to run in an interrupt-free environment on an
> isolated CPU, how can it determine if the nohz_full conditions are met,
> similar to how the kernel uses can_stop_full_tick()?
>
> Would it be feasible to add a variable in procfs, which can be read by
> other programs and notify the RT process throught a shared memory flag.
>
This is overly-engineeded with no extra benefit. A simple cgroup setup
and some basic /sys/fs/cgroup/ file reads should do the job just fine.
Assuming a recent kernel, let's say you want to isolate the CPUs 0-1,
right now, on your own laptop. Open a shell and do:
CGROUP=rtisolated
echo "+cpuset" > /sys/fs/cgroup/cgroup.subtree_control
mkdir /sys/fs/cgroup/${CGROUP}.slice
pushd /sys/fs/cgroup/${CGROUP}.slice
echo 0-1 > cpuset.cpus
echo isolated > cpuset.cpus.partition
echo 0 > cpuset.mems
cat cpuset.cpus.partition # should print "isolated"
popd
On the same shell, spawn a /bin/bash process in that 'rtisolated' cgroup:
systemd-run \
--slice=${CGROUP} \
--unit=isolated-shell \
--pty \
/bin/bash
Running as unit: isolated-shell.service
Press ^] three times within 1s to disconnect TTY.
Now, from that isolated shell, do:
echo $$ # Bash own PID
302713
taskset --cpu-list --pid $$ # should print CPUs 0 and 1
pid 302713's current affinity list: 0,1
Now, also from that isolated shell, do:
cat /proc/$$/cgroup
cat /proc/self/cgroup
Both "cat"s should print something like:
0::/rtisolated.slice/isolated-shell.service
which can be easily done and verified from any C code as well, before
starting the RT loop (where multiple initiatory steps need to be done
anyway).
Running systemd-cgls, from outside that isolated shell, should also print
something like:
CGroup /:
-.slice
├─user.slice
│ └─ ...
├─init.scope
│ └─1 /sbin/init
├─system.slice
│ └─ ...
└─rtisolated.slice
└─isolated-shell.service
└─302713 /bin/bash
^^ ^^
Hope this helps a bit.
Good luck,
--
Ahmed S. Darwish
Linutronix GmbH
