[PATCH] rcu: wrong stall log when only one task blocked within RCU read-side

From: yanfei . xu
Date: Sat May 15 2021 - 12:45:41 EST


From: Yanfei Xu <yanfei.xu@xxxxxxxxxxxxx>

The for loop in rcu_print_task_stall() always omits the first blocked
task in RCU read-ride critical section. It will cause status of the
blocked task can't be printed and lead a wrong ndetected variable.
With the above, the wired stall log comes, All QSes seen, but qsmask
still is not zero:

rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
rcu: Tasks blocked on level-1 rcu_node (CPUs 12-23):
(detected by 15, t=6504 jiffies, g=164777, q=9011209)
rcu: All QSes seen, last rcu_preempt kthread activity 1
(4295252379-4295252378), jiffies_till_next_fqs=1, root ->qsmask 0x2
BUG: sleeping function called from invalid context at
include/linux/uaccess.h:156
in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 70613, name:
msgstress04
INFO: lockdep is turned off.
Preemption disabled at:
[<ffff8000104031a4>] create_object.isra.0+0x204/0x4b0
CPU: 15 PID: 70613 Comm: msgstress04 Kdump: loaded Not tainted
5.12.2-yoctodev-standard #1
Hardware name: Marvell OcteonTX CN96XX board (DT)
Call trace:
dump_backtrace+0x0/0x2cc
show_stack+0x24/0x30
dump_stack+0x110/0x188
___might_sleep+0x214/0x2d0
__might_sleep+0x7c/0xe0

Fixes: c583bcb8f5ed ("rcu: Don't invoke try_invoke_on_locked_down_task() with irqs disabled")
Signed-off-by: Yanfei Xu <yanfei.xu@xxxxxxxxxxxxx>
---
kernel/rcu/tree_stall.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h
index 59b95cc5cbdf..b72311d24a9f 100644
--- a/kernel/rcu/tree_stall.h
+++ b/kernel/rcu/tree_stall.h
@@ -280,8 +280,8 @@ static int rcu_print_task_stall(struct rcu_node *rnp, unsigned long flags)
break;
}
raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
- for (i--; i; i--) {
- t = ts[i];
+ while (i) {
+ t = ts[--i];
if (!try_invoke_on_locked_down_task(t, check_slow_task, &rscr))
pr_cont(" P%d", t->pid);
else
--
2.27.0