> 2023年5月30日 08:30，Paul E. McKenney <paulmck@xxxxxxxxxx> 写道：
>
> On Sun, May 28, 2023 at 02:37:28PM +0000, Alan Huang wrote:
>> The cap here has not been published, plain access is enough.
>>
>> Signed-off-by: Alan Huang <mmpgouride@xxxxxxxxx>
>
> Good catch, thank you!
>
> Does line 27 of that same listing need a READ_ONCE(*ctrp)?
It does need! I have sent another patch.
Thanks,
Alan
>
> Thanx, Paul
>
>> ---
>> CodeSamples/count/count_end_rcu.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/CodeSamples/count/count_end_rcu.c b/CodeSamples/count/count_end_rcu.c
>> index 1171f0b6..8fc70611 100644
>> --- a/CodeSamples/count/count_end_rcu.c
>> +++ b/CodeSamples/count/count_end_rcu.c
>> @@ -89,7 +89,7 @@ void count_unregister_thread(int nthreadsexpected) //\lnlbl{unreg:b}
>> } //\lnlbl{unreg:alloc:e}
>> spin_lock(&final_mutex); //\lnlbl{unreg:acq}
>> *cap = *countarrayp; //\lnlbl{unreg:copy}
>> - WRITE_ONCE(cap->total, cap->total + counter); //\lnlbl{unreg:add}
>> + cap->total += counter; //\lnlbl{unreg:add}
>> cap->counterp[idx] = NULL; //\lnlbl{unreg:null}
>> capold = countarrayp; //\lnlbl{unreg:retain}
>> rcu_assign_pointer(countarrayp, cap); //\lnlbl{unreg:assign}
>> --
>> 2.34.1
>>
