Re: [PATCH 1/2] drivers: base: power: main: Use built-in RCU list checking

[joel@joelfernandes.org]

On March 1, 2020 3:12:53 PM EST, "Rafael J. Wysocki" <rafael@kernel.org> wrote:
>On Fri, Feb 28, 2020 at 6:47 PM <madhuparnabhowmik10@gmail.com> wrote:
>>
>> From: Madhuparna Bhowmik <madhuparnabhowmik10@gmail.com>
>>
>> This patch passes the cond argument to list_for_each_entry_rcu()
>> to fix the following false-positive lockdep warnings:
>>
>> [  330.302784] =============================
>> [  330.302789] WARNING: suspicious RCU usage
>> [  330.302796] 5.6.0-rc1+ #5 Not tainted
>> [  330.302801] -----------------------------
>> [  330.302808] drivers/base/power/main.c:326 RCU-list traversed in
>non-reader section!!
>>
>> [  330.303303] =============================
>> [  330.303307] WARNING: suspicious RCU usage
>> [  330.303311] 5.6.0-rc1+ #5 Not tainted
>> [  330.303315] -----------------------------
>> [  330.303319] drivers/base/power/main.c:1698 RCU-list traversed in
>non-reader section!!
>>
>> [  331.934969] =============================
>> [  331.934971] WARNING: suspicious RCU usage
>> [  331.934973] 5.6.0-rc1+ #5 Not tainted
>> [  331.934975] -----------------------------
>> [  331.934977] drivers/base/power/main.c:1238 RCU-list traversed in
>non-reader section!!
>>
>> [  332.467772] WARNING: suspicious RCU usage
>> [  332.467775] 5.6.0-rc1+ #5 Not tainted
>> [  332.467775] -----------------------------
>> [  332.467778] drivers/base/power/main.c:269 RCU-list traversed in
>non-reader section!!
>
>I don't see these warnings in the kernels run locally here.
>
>What do you do to get them?
>
>Joel, any comments here?

You have to enable lockdep in your config. Does your setup have that?

Thanks,

- Joel


>
>>
>> Signed-off-by: Madhuparna Bhowmik <madhuparnabhowmik10@gmail.com>
>> ---
>>  drivers/base/power/main.c | 12 ++++++++----
>>  1 file changed, 8 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
>> index 0e99a760aebd..742c05f3c1e7 100644
>> --- a/drivers/base/power/main.c
>> +++ b/drivers/base/power/main.c
>> @@ -266,7 +266,8 @@ static void dpm_wait_for_suppliers(struct device
>*dev, bool async)
>>          * callbacks freeing the link objects for the links in the
>list we're
>>          * walking.
>>          */
>> -       list_for_each_entry_rcu(link, &dev->links.suppliers, c_node)
>> +       list_for_each_entry_rcu(link, &dev->links.suppliers, c_node,
>> +                               device_links_read_lock_held())
>>                 if (READ_ONCE(link->status) != DL_STATE_DORMANT)
>>                         dpm_wait(link->supplier, async);
>>
>> @@ -323,7 +324,8 @@ static void dpm_wait_for_consumers(struct device
>*dev, bool async)
>>          * continue instead of trying to continue in parallel with
>its
>>          * unregistration).
>>          */
>> -       list_for_each_entry_rcu(link, &dev->links.consumers, s_node)
>> +       list_for_each_entry_rcu(link, &dev->links.consumers, s_node,
>> +                                device_links_read_lock_held())
>>                 if (READ_ONCE(link->status) != DL_STATE_DORMANT)
>>                         dpm_wait(link->consumer, async);
>>
>> @@ -1235,7 +1237,8 @@ static void dpm_superior_set_must_resume(struct
>device *dev)
>>
>>         idx = device_links_read_lock();
>>
>> -       list_for_each_entry_rcu(link, &dev->links.suppliers, c_node)
>> +       list_for_each_entry_rcu(link, &dev->links.suppliers, c_node,
>> +                                device_links_read_lock_held())
>>                 link->supplier->power.must_resume = true;
>>
>>         device_links_read_unlock(idx);
>> @@ -1695,7 +1698,8 @@ static void
>dpm_clear_superiors_direct_complete(struct device *dev)
>>
>>         idx = device_links_read_lock();
>>
>> -       list_for_each_entry_rcu(link, &dev->links.suppliers, c_node)
>{
>> +       list_for_each_entry_rcu(link, &dev->links.suppliers, c_node,
>> +                                device_links_read_lock_held()) {
>>                 spin_lock_irq(&link->supplier->power.lock);
>>                 link->supplier->power.direct_complete = false;
>>                 spin_unlock_irq(&link->supplier->power.lock);
>> --
>> 2.17.1
>>

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.