Re: linux-next: manual merge of the paulmck tree with the rcu tree
From: Paul E. McKenney
Date: Fri Nov 20 2026 - 17:45:05 EST
On Wed, Nov 18, 2026 at 10:49:41AM -0500, Joel Fernandes wrote:
> On 11/17, "Paul E. McKenney" <paulmck@xxxxxxxxxx> wrote:
> > On Tue, Nov 17, 2026 at 10:44:28PM +0000, Mark Brown wrote:
> > > Hi all,
> > >
> > > Today's linux-next merge of the paulmck tree got a conflict in:
> > >
> > > tools/testing/selftests/rcutorture/bin/torture.sh
> > >
> > > between commit:
> > >
> > > 7ada0a57267ab ("torture: Use ${:+} to save a few lines in torture.sh's torture_one()")
> > >
> > > from the rcu tree and commit:
> > >
> > > 436f3c6b27de8 ("torture.sh: Add hazptr torturing")
> > >
> > > from the paulmck tree. AFAICT the change in the hazptr commit is adding
> > > quoting that is also covered by the other commit and should have been
> > > redundant anyway but you really want to double check here.
> >
> > I confess, I am confused. If you said that the commit from the rcu
> > tree was instead this one:
> >
> > ccabfbe01ca5 ("torture: Avoid duplicate identify_qemu_vcpus calls in torture.sh")
> >
> > I would be less confused.
>
> Maybe both patches conflicted? The one Mark posted resolved as follows:
>
> rcu/next tree:
> "$@" ${torture_bootargs:+--bootargs "$torture_bootargs"} --datestamp "$ds/results-$curflavor" > $T/$curflavor.out 2>&1
>
> Paul's tree:
> "$@" "${boottag}" "$cur_bootargs" --datestamp "$ds/results $curflavor" > $T/$curflavor.out 2>&1
>
> Mark's resolution:
> "$@" ${torture_bootargs:+--bootargs "$torture_bootargs"} --datestamp "$ds/results-$curflavor" > $T/$curflavor.out 2>&1
>
> Both versions look to be doing the same thing, which is adding the boot
> arguments if the `torture boot args` variable was defined.
What I have done is to pull the "torture.sh: Add hazptr torturing"
commit out of -next, avoiding this conflict.
I also rebased my stack on top of the rcu tree's "next" branch, which
showed some skew on this commit:
fd5121d99c86 ("rcu: Print CPU that preempted stalled reader wants to run on")
Joel, could you please fold the patch shown below into this commit?
Thanx, Paul
------------------------------------------------------------------------
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 09e05e2b6c9aa..bd586ea18fdac 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -2528,6 +2528,8 @@ extern void migrate_enable(void);
DEFINE_LOCK_GUARD_0(migrate, migrate_disable(), migrate_enable())
+struct task_struct *cpu_curr_snapshot_rcu(int cpu);
+
#endif
struct task_struct *cpu_curr_snapshot_rcu(int cpu);
diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h
index e0607f4235645..53564b780ec90 100644
--- a/kernel/rcu/tree_stall.h
+++ b/kernel/rcu/tree_stall.h
@@ -335,7 +335,7 @@ static int check_slow_task(struct task_struct *t, void *arg)
rscrp->rcu_rdr_cpu = cpu = task_cpu(t);
rscrp->rcu_rdr_cpu_offline = cpu_is_offline(cpu);
lockdep_assert_irqs_disabled();
- rscrp->rcu_rdr_cpu_task = th = cpu_curr_snapshot_rcu(cpu);
+ rscrp->rcu_rdr_cpu_task = cpu_curr_snapshot_rcu(cpu);
// There is no need for get_task_struct(t) because we are in
// an implicit RCU read-side critical section due to interrupts
// being disabled.