On 4/5/20 1:04 AM,
[email protected] wrote:
I already reviewed it, so we have two reviewers now.Hi Egor,
The fix looks good to me.
Thank you for taking care about this optimization!
The call to validateVM() is not really needed but does not harm as it is an empty method.
I'll sponsor the push for you.
Will submit a mach5 test run as soon as this gets reviewed.
Chris
Thanks,
Serguei
On 4/3/20 02:16, Egor Ushakov wrote:
Hi all, please review the fix
The com.sun.tools.jdi.ClassLoaderReferenceImpl#findType may be slow because ClassLoaderReferenceImpl#visibleClasses does not populate signature and we check it for every class in the loop just after, so requesting all unavailable signatures one by one.
The fix adds extra check in classesBySignature before comparing signatures of all visible classes.
bugid https://bugs.openjdk.java.net/browse/JDK-8241958
cr http://cr.openjdk.java.net/~eushakov/8241958/webrev.00/
Thanks!
