On 12/21/2017 07:39 PM, Stephen Boyd wrote:
On 12/20, Stephen Boyd wrote:
On 12/20, David Lechner wrote:
On 12/20/2017 02:33 PM, David Lechner wrote:
So, the question I have is: what is the actual "correct" behavior of
spin_trylock_irqsave()? Is it really supposed to always return true
when CONFIG_DEBUG_SPINLOCK=n and CONFIG_SMP=n or is this a bug?
Thanks for doing the analysis in this thread.
When CONFIG_DEBUG_SPINLOCK=n and CONFIG_SMP=n, spinlocks are
compiler barriers, that's it. So even if it is a bug to always
return true, I fail to see how we can detect that a spinlock is
already held in this configuration and return true or false.
I suppose the best option is to make clk_enable_lock() and
clk_enable_unlock() into nops or pure owner/refcount/barrier
updates when CONFIG_SMP=n. We pretty much just need the barrier
semantics when there's only a single CPU.
How about this patch? It should make the trylock go away on UP
configs and then we keep everything else for refcount and
ownership. We would test enable_owner outside of any
irqs/preemption disabled section though. That needs a think.
---8<----
diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 3526bc068f30..b6f61367aa8d 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -143,7 +143,8 @@ static unsigned long clk_enable_lock(void)
{
unsigned long flags;
- if (!spin_trylock_irqsave(&enable_lock, flags)) {
+ if (!IS_ENABLED(CONFIG_SMP) ||
+ !spin_trylock_irqsave(&enable_lock, flags)) {
if (enable_owner == current) {
enable_refcnt++;
__acquire(enable_lock);
I came up with the exact same patch earlier today, but did not have a
chance to send it. I've tested it and it fixes the problem for me.
I'm afraid I don't know enough about how preemption works yet to be of
much help to say what or if something else is needed to protect
enable_owner/enable_refcnt.
