RFR (S) 8205608: Fix 'frames()' in ThreadReferenceImpl.c to prevent quadratic runtime behavior

[Schmelter, Ralf]

Hi Serguei,

Sorry for the late reply, but it seems the spam filter has removed your emails. I just saw them in the archives.

Regarding this code:

288     if (length != count) {

289         error = JVMTI_ERROR_INTERNAL;

290     }

count is the number of frames filled into the array (it is set in the GetStackTrace JVMTI call) and length is the number of frames requested to be filled in. Both are independent of the start index at this point. Note that I've reused the count variable (it was first initialized to hold the number of frames on the stack). Maybe it is clearer to use a new variable in this call?

The package will not be send if an error code is set on the output stream (see outStream_sendReply()). This should cover all cases (in both the new and the old code).

Best regards,
Ralf
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/serviceability-dev/attachments/20180713/f76c1d9a/attachment.html>