From: Will Deacon <will@kernel.org>
To: linux-kernel@vger.kernel.org
Cc: kernel-team@android.com, Will Deacon <will@kernel.org>,
Peter Zijlstra <peterz@infradead.org>,
Waiman Long <longman@redhat.com>,
Zefan Li <lizefan.x@bytedance.com>, Tejun Heo <tj@kernel.org>,
Johannes Weiner <hannes@cmpxchg.org>,
cgroups@vger.kernel.org
Subject: [PATCH 1/2] cpuset: Fix cpuset_cpus_allowed() to not filter offline CPUs
Date: Tue, 31 Jan 2023 22:17:18 +0000 [thread overview]
Message-ID: <20230131221719.3176-2-will@kernel.org> (raw)
In-Reply-To: <20230131221719.3176-1-will@kernel.org>
From: Peter Zijlstra <peterz@infradead.org>
There is a difference in behaviour between CPUSET={y,n} that is now
wrecking havoc with {relax,force}_compatible_cpus_allowed_ptr().
Specifically, since commit 8f9ea86fdf99 ("sched: Always preserve the
user requested cpumask") relax_compatible_cpus_allowed_ptr() is
calling __sched_setaffinity() unconditionally.
But the underlying problem goes back a lot further, possibly to
commit: ae1c802382f7 ("cpuset: apply cs->effective_{cpus,mems}") which
switched cpuset_cpus_allowed() from cs->cpus_allowed to
cs->effective_cpus.
The problem is that for CPUSET=y cpuset_cpus_allowed() will filter out
all offline CPUs. For tasks that are part of a (!root) cpuset this is
then later fixed up by the cpuset hotplug notifiers that re-evaluate
and re-apply cs->effective_cpus, but for (normal) tasks in the root
cpuset this does not happen and they will forever after be excluded
from CPUs onlined later.
As such, rewrite cpuset_cpus_allowed() to return a wider mask,
including the offline CPUs.
Fixes: 8f9ea86fdf99 ("sched: Always preserve the user requested cpumask")
Reported-by: Will Deacon <will@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20230117160825.GA17756@willie-the-truck
Signed-off-by: Will Deacon <will@kernel.org>
---
kernel/cgroup/cpuset.c | 39 ++++++++++++++++++++++++++++++++++-----
1 file changed, 34 insertions(+), 5 deletions(-)
diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index a29c0b13706b..8552cc2c586a 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -3683,23 +3683,52 @@ void __init cpuset_init_smp(void)
BUG_ON(!cpuset_migrate_mm_wq);
}
+static const struct cpumask *__cs_cpus_allowed(struct cpuset *cs)
+{
+ const struct cpumask *cs_mask = cs->cpus_allowed;
+ if (!parent_cs(cs))
+ cs_mask = cpu_possible_mask;
+ return cs_mask;
+}
+
+static void cs_cpus_allowed(struct cpuset *cs, struct cpumask *pmask)
+{
+ do {
+ cpumask_and(pmask, pmask, __cs_cpus_allowed(cs));
+ cs = parent_cs(cs);
+ } while (cs);
+}
+
/**
* cpuset_cpus_allowed - return cpus_allowed mask from a tasks cpuset.
* @tsk: pointer to task_struct from which to obtain cpuset->cpus_allowed.
* @pmask: pointer to struct cpumask variable to receive cpus_allowed set.
*
- * Description: Returns the cpumask_var_t cpus_allowed of the cpuset
- * attached to the specified @tsk. Guaranteed to return some non-empty
- * subset of cpu_online_mask, even if this means going outside the
- * tasks cpuset.
+ * Description: Returns the cpumask_var_t cpus_allowed of the cpuset attached
+ * to the specified @tsk. Guaranteed to return some non-empty intersection
+ * with cpu_online_mask, even if this means going outside the tasks cpuset.
**/
void cpuset_cpus_allowed(struct task_struct *tsk, struct cpumask *pmask)
{
unsigned long flags;
+ struct cpuset *cs;
spin_lock_irqsave(&callback_lock, flags);
- guarantee_online_cpus(tsk, pmask);
+ rcu_read_lock();
+
+ cs = task_cs(tsk);
+ do {
+ cpumask_copy(pmask, task_cpu_possible_mask(tsk));
+ cs_cpus_allowed(cs, pmask);
+
+ if (cpumask_intersects(pmask, cpu_online_mask))
+ break;
+
+ cs = parent_cs(cs);
+ } while (cs);
+
+ rcu_read_unlock();
spin_unlock_irqrestore(&callback_lock, flags);
}
--
2.39.1.456.gfc5497dd1b-goog
next prev parent reply other threads:[~2023-01-31 22:17 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-31 22:17 [PATCH 0/2] Fix broken cpuset affinity handling on heterogeneous systems Will Deacon
2023-01-31 22:17 ` Will Deacon [this message]
2023-02-01 4:14 ` [PATCH 1/2] cpuset: Fix cpuset_cpus_allowed() to not filter offline CPUs Waiman Long
2023-02-01 9:14 ` Peter Zijlstra
2023-02-01 15:16 ` Waiman Long
2023-02-01 18:46 ` Waiman Long
2023-02-01 19:14 ` Waiman Long
2023-02-01 19:17 ` Waiman Long
2023-02-01 21:10 ` Peter Zijlstra
2023-02-02 3:34 ` Waiman Long
2023-02-03 11:50 ` Will Deacon
2023-02-03 15:13 ` Waiman Long
2023-02-03 15:26 ` Peter Zijlstra
2023-02-03 15:35 ` Waiman Long
2023-02-02 8:34 ` Peter Zijlstra
2023-02-02 16:06 ` Waiman Long
2023-02-02 19:42 ` Peter Zijlstra
2023-02-02 20:46 ` Waiman Long
2023-02-02 20:48 ` Tejun Heo
2023-02-02 20:53 ` Waiman Long
2023-02-02 21:05 ` Waiman Long
2023-02-02 21:50 ` Tejun Heo
2023-02-03 0:54 ` Waiman Long
2023-02-03 16:31 ` Will Deacon
2023-01-31 22:17 ` [PATCH 2/2] cpuset: Call set_cpus_allowed_ptr() with appropriate mask for task Will Deacon
2023-02-01 2:22 ` Waiman Long
2023-02-01 9:15 ` Peter Zijlstra
2023-02-01 15:03 ` Waiman Long
2023-02-01 9:27 ` Peter Zijlstra
2023-02-03 17:55 ` Waiman Long
2023-02-06 20:21 ` Tejun Heo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230131221719.3176-2-will@kernel.org \
--to=will@kernel.org \
--cc=cgroups@vger.kernel.org \
--cc=hannes@cmpxchg.org \
--cc=kernel-team@android.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lizefan.x@bytedance.com \
--cc=longman@redhat.com \
--cc=peterz@infradead.org \
--cc=tj@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).