Re: RFR: 8238566: java.security.Provider$Service.supportsParameter() is racy

[Daniel Fuchs]

Hi Valerie,

Given that hasKeyAttributes is already decelared as volatile,
may I suggest the following change that uses double locking?
It will  avoid synchronizing in the happy case where `hasKeyAttributes`
has already been computed.

1924         private boolean hasKeyAttributes() {
1925             Boolean b = hasKeyAttributes;
                 if (b != null) return b;
                 synchronized (this) {
                     b = hasKeyAttributes;
1926                 if (b != null) return b;
1927                 String s;
1928                 s = getAttribute("SupportedKeyFormats");

                     ...

1948                 hasKeyAttributes = b;
1949             }
1950             return b;
1951         }


best regards,

-- daniel

On 11/03/2020 20:31, Valerie Peng wrote:

Anyone can help reviewing this?

I addressed this by applying the double-checked-locking pattern for lazy initialization of instance fields.

Existing code already have most of the code structured for the pattern but misses the second check.

Bug: https://bugs.openjdk.java.net/browse/JDK-8238566

Webrev: http://cr.openjdk.java.net/~valeriep/8238566/webrev.00/

Thanks,

Valerie