Re: [PATCH] cpu/hotplug: Fix rollback during error-out in takedown_cpu()

From: Mukesh Ojha
Date: Wed Sep 05 2018 - 08:10:30 EST




On 9/5/2018 5:03 PM, Thomas Gleixner wrote:
On Tue, 4 Sep 2018, Neeraj Upadhyay wrote:
If takedown_cpu() fails during _cpu_down(), st->state is reset,
by calling cpuhp_reset_state(). This results in an additional
increment of st->state, which results in CPUHP_AP_SMPBOOT_THREADS
state being skipped during rollback. Fix this by not calling
cpuhp_reset_state() and doing the state reset directly in
_cpu_down().

Fixes: 4dddfb5faa61 ("smp/hotplug: Rewrite AP state machine core")
Signed-off-by: Neeraj Upadhyay <neeraju@xxxxxxxxxxxxxx>
---
kernel/cpu.c | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/kernel/cpu.c b/kernel/cpu.c
index aa7fe85..9f49edb 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -970,7 +970,14 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
*/
ret = cpuhp_down_callbacks(cpu, st, target);
if (ret && st->state > CPUHP_TEARDOWN_CPU && st->state < prev_state) {
- cpuhp_reset_state(st, prev_state);
+ /*
+ * As st->last is not set, cpuhp_reset_state() increments
+ * st->state, which results in CPUHP_AP_SMPBOOT_THREADS being
+ * skipped during rollback. So, don't use it here.
+ */
+ st->rollback = true;
+ st->target = prev_state;
+ st->bringup = !st->bringup;
No, this is just papering over the actual problem.

The state inconsistency happens in take_cpu_down() when it returns with a
failure from __cpu_disable() because that returns with state = TEARDOWN_CPU
and st->state is then incremented in undo_cpu_down().

That's the real issue and we need to analyze the whole cpu_down rollback
logic first.

Could this be done like below ?

diff --git a/kernel/cpu.c b/kernel/cpu.c
index aa7fe85..47bce90 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -802,17 +802,18 @@ static int take_cpu_down(void *_param)
ÂÂÂÂÂÂÂ int err, cpu = smp_processor_id();
ÂÂÂÂÂÂÂ int ret;

-ÂÂÂÂÂÂ /* Ensure this CPU doesn't handle any more interrupts. */
-ÂÂÂÂÂÂ err = __cpu_disable();
-ÂÂÂÂÂÂ if (err < 0)
-ÂÂÂÂÂÂÂÂÂÂÂÂÂÂ return err;
-
ÂÂÂÂÂÂÂ /*
ÂÂÂÂÂÂÂÂ * We get here while we are in CPUHP_TEARDOWN_CPU state and we must not
ÂÂÂÂÂÂÂÂ * do this step again.
ÂÂÂÂÂÂÂÂ */
ÂÂÂÂÂÂÂ WARN_ON(st->state != CPUHP_TEARDOWN_CPU);
ÂÂÂÂÂÂÂ st->state--;
+
+ÂÂÂÂÂÂ /* Ensure this CPU doesn't handle any more interrupts. */
+ÂÂÂÂÂÂ err = __cpu_disable();
+ÂÂÂÂÂÂ if (err < 0)
+ÂÂÂÂÂÂÂÂÂÂÂÂÂÂ return err;
+
ÂÂÂÂÂÂÂ /* Invoke the former CPU_DYING callbacks */

Thanks,
Mukesh