Re: [net PATCH] net: sched: lock imbalance in hhf qdisc

[John Fastabend <john.fastabend@gmail.com>]

On 05/01/2014 08:30 AM, Eric Dumazet wrote:
> On Thu, 2014-05-01 at 08:15 -0700, John Fastabend wrote:
>> hhf_change() takes the sch_tree_lock and releases it but misses the error
>> cases. Fix the missed case here.
>>
>> Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
>> ---
>>   net/sched/sch_hhf.c |    4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/net/sched/sch_hhf.c b/net/sched/sch_hhf.c
>> index edee03d..1803f89 100644
>> --- a/net/sched/sch_hhf.c
>> +++ b/net/sched/sch_hhf.c
>> @@ -565,8 +565,10 @@ static int hhf_change(struct Qdisc *sch, struct nlattr *opt)
>>   		new_hhf_non_hh_weight = nla_get_u32(tb[TCA_HHF_NON_HH_WEIGHT]);
>>
>>   	non_hh_quantum = (u64)new_quantum * new_hhf_non_hh_weight;
>> -	if (non_hh_quantum > INT_MAX)
>> +	if (non_hh_quantum > INT_MAX) {
>> +		sch_tree_unlock(sch);
>>   		return -EINVAL;
>> +	}
>>   	q->quantum = new_quantum;
>>   	q->hhf_non_hh_weight = new_hhf_non_hh_weight;
>>
>
> Good catch, could you refactor a bit more to do all checks before any
> change ?

Sure. What you have below looks good. I'll submit an official v2 patch.

-- 
John Fastabend         Intel Corporation