On 10/2/12 11:22 AM, Alan Bateman wrote:
This test has been failing intermittently for me on 32-bit systems since jdk8/tl picked up the perm gen removal work a few days ago. The test uses a 1GB heap and it doesn't seem to be possible to reduce it too much as it needs a largish heap on 64-bit systems. Any objection if I change the test to run without CDS?
No objection from me.
That allows the VM to reserve the 1GB. This is just a workaround for now, the alternative is to exclude it by putting it on the ProblemList.
Also, parsing large CRLs are problematic in general so this is really a bigger problem that needs to be fixed. There is already a CR open for that : 6670894: CRL parsing implementation is extremely inefficient
--Sean
Thanks, -Alan diff --git a/test/sun/security/provider/X509Factory/BigCRL.java b/test/sun/security/provider/X509Factory/BigCRL.java --- a/test/sun/security/provider/X509Factory/BigCRL.java +++ b/test/sun/security/provider/X509Factory/BigCRL.java @@ -25,7 +25,7 @@ * @test * @bug 7099399 * @summary cannot deal with CRL file larger than 16MB - * @run main/othervm -Xmx1024m BigCRL + * @run main/othervm -Xshare:off -Xmx1024m BigCRL */ import java.io.FileInputStream;
