On 8/6/24 23:15, Miaohe Lin wrote:
On 2024/8/7 0:41, Waiman Long wrote:
The memory_failure_cpu structure is a per-cpu structure. Access to its
content requires the use of get_cpu_var() to lock in the current CPU
and disable preemption. The use of a regular spinlock_t for locking
purpose is fine for a non-RT kernel.
Since the integration of RT spinlock support into the v5.15 kernel,
a spinlock_t in a RT kernel becomes a sleeping lock and taking a
sleeping lock in a preemption disabled context is illegal resulting in
the following kind of warning.
[12135.732244] BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:48
[12135.732248] in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 270076, name: kworker/0:0
[12135.732252] preempt_count: 1, expected: 0
[12135.732255] RCU nest depth: 2, expected: 2
:
[12135.732420] Hardware name: Dell Inc. PowerEdge R640/0HG0J8, BIOS 2.10.2 02/24/2021
[12135.732423] Workqueue: kacpi_notify acpi_os_execute_deferred
[12135.732433] Call Trace:
[12135.732436] <TASK>
[12135.732450] dump_stack_lvl+0x57/0x81
[12135.732461] __might_resched.cold+0xf4/0x12f
[12135.732479] rt_spin_lock+0x4c/0x100
[12135.732491] memory_failure_queue+0x40/0xe0
[12135.732503] ghes_do_memory_failure+0x53/0x390
[12135.732516] ghes_do_proc.constprop.0+0x229/0x3e0
[12135.732575] ghes_proc+0xf9/0x1a0
[12135.732591] ghes_notify_hed+0x6a/0x150
[12135.732602] notifier_call_chain+0x43/0xb0
[12135.732626] blocking_notifier_call_chain+0x43/0x60
[12135.732637] acpi_ev_notify_dispatch+0x47/0x70
[12135.732648] acpi_os_execute_deferred+0x13/0x20
[12135.732654] process_one_work+0x41f/0x500
[12135.732695] worker_thread+0x192/0x360
[12135.732715] kthread+0x111/0x140
[12135.732733] ret_from_fork+0x29/0x50
[12135.732779] </TASK>
Fix it by using a raw_spinlock_t for locking instead. Also move the
pr_err() out of the lock critical section to avoid indeterminate latency
of this call.
Fixes: ea8f5fb8a71f ("HWPoison: add memory_failure_queue()")
We shouldn't have this problem before RT spinlock is supported? If so, this Fixes tag might be wrong.
OK, I can take out the Fixes tag. It is hard to pinpoint a particular RT
related commit.
Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
---
mm/memory-failure.c | 18 ++++++++++--------
1 file changed, 10 insertions(+), 8 deletions(-)
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 581d3e5c9117..7aeb5198c2a0 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -2417,7 +2417,7 @@ struct memory_failure_entry {
struct memory_failure_cpu {
DECLARE_KFIFO(fifo, struct memory_failure_entry,
MEMORY_FAILURE_FIFO_SIZE);
- spinlock_t lock;
+ raw_spinlock_t lock;
struct work_struct work;
};
@@ -2443,19 +2443,21 @@ void memory_failure_queue(unsigned long pfn, int flags)
{
struct memory_failure_cpu *mf_cpu;
unsigned long proc_flags;
+ bool buffer_overflow;
struct memory_failure_entry entry = {
.pfn = pfn,
.flags = flags,
};
mf_cpu = &get_cpu_var(memory_failure_cpu);
- spin_lock_irqsave(&mf_cpu->lock, proc_flags);
- if (kfifo_put(&mf_cpu->fifo, entry))
+ raw_spin_lock_irqsave(&mf_cpu->lock, proc_flags);
+ buffer_overflow = !kfifo_put(&mf_cpu->fifo, entry);
+ if (!buffer_overflow)
schedule_work_on(smp_processor_id(), &mf_cpu->work);
- else
+ raw_spin_unlock_irqrestore(&mf_cpu->lock, proc_flags);
+ if (buffer_overflow)
pr_err("buffer overflow when queuing memory failure at %#lx\n",
pfn);
Should we put pr_err() further under put_cpu_var()?
Yes, we should probably enable preemption first before calling pr_err().
Will make the change in v2.
Thanks,
Longman
- spin_unlock_irqrestore(&mf_cpu->lock, proc_flags)
put_cpu_var(memory_failure_cpu);
}
Will below diff be more straightforward?
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index b68953dc9fad..be172cbc6ca9 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -2553,20 +2553,23 @@ void memory_failure_queue(unsigned long pfn, int flags)
{
struct memory_failure_cpu *mf_cpu;
unsigned long proc_flags;
+ bool buffer_overflow = false;
struct memory_failure_entry entry = {
.pfn = pfn,
.flags = flags,
};
mf_cpu = &get_cpu_var(memory_failure_cpu);
- spin_lock_irqsave(&mf_cpu->lock, proc_flags);
+ raw_spin_lock_irqsave(&mf_cpu->lock, proc_flags);
if (kfifo_put(&mf_cpu->fifo, entry))
schedule_work_on(smp_processor_id(), &mf_cpu->work);
else
+ buffer_overflow = true;
+ raw_spin_unlock_irqrestore(&mf_cpu->lock, proc_flags);
+ put_cpu_var(memory_failure_cpu);
+ if (buffer_overflow)
pr_err("buffer overflow when queuing memory failure at %#lx\n",
pfn);
- spin_unlock_irqrestore(&mf_cpu->lock, proc_flags);
- put_cpu_var(memory_failure_cpu);
}
EXPORT_SYMBOL_GPL(memory_failure_queue);
But no strong opinion.
Thanks.
.
