RFR(S): 8008357: [sampling] assert(sender_blob->is_runtime_stub() || sender_blob->is_nmethod()) failed: Impossible call chain
Rickard Bäckman
rickard.backman at oracle.com
Mon Mar 25 04:13:23 PDT 2013
Hi,
trying again, can I please have two reviews on this change?
Some small changes after offline discussion.
The webrev is available here:
http://cr.openjdk.java.net/~rbackman/8008357.u1/
Thanks
/R
On Mar 5, 2013, at 8:07 AM, Rickard Bäckman wrote:
> Anyone have time to look at this?
>
> Thanks
> /R
>
> On Mar 1, 2013, at 10:27 AM, Rickard Bäckman wrote:
>
>> Hi all,
>>
>> here comes another update to frame.safe_for_sender.
>> If the PC at a place where the stack doesn't match the _frame_size we sometimes read an invalid return PC.
>> In this case we read one that pointed into the Safepoint blob.
>>
>> We now pretty much guarded for all kind of blobs in safe_for_sender,
>> so I've changed the method to not assert in the end but to do it the same was as the frame_sparc.cpp did it. Everything that is not
>> a nmethod at the end of the method is not safe.
>>
>> I've also changed another check for a frame_size == 0 to frame_size <= 0 to make it somewhat more safer.
>>
>> The bug: http://bugs.sun.com/view_bug.do?bug_id=8008357
>> Webrev: http://cr.openjdk.java.net/~rbackman/8008357/
>>
>> This webrev is for HS24.
>>
>> Thanks
>> /R
>
More information about the hotspot-runtime-dev
mailing list