[#RVM-440] Poisoned tests failing in finalizer thread - jira.codehaus.org

Details

Type: Bug
Status: Closed
Priority: Major
Resolution: Fixed
Affects Version/s: 2.9.2
Fix Version/s: 3.0.1
Component/s: Runtime
Labels:
None

Number of attachments :
0

Description

In r14119 poisoning regression tests were enabled. These stress read/write barriers. We got our first results [1] and there seems to be a problem in the finalizer thread, we have died in the PepTest of both Spec JBB versions [2] [3].

[1] http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/3447/statistics_report
[2] http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/3447/FullAdaptivePoisoned/default/SPECjbb2000/SPECjbb2000
[3] http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/3447/FullAdaptivePoisoned/default/SPECjbb2005/SPECjbb2005/default/Output.txt

Issue Links

This issue relates to:
RVM-609	Race in setting finalizer method
RVM-633	Avoid the use of Reflection to invoke Object.finalize()

Activity

Ascending order - Click to sort in descending order

Hide

Permalink

Ian Rogers added a comment - 27/Apr/08 11:16 AM

Still present and effect DaCapo benchmarks too:

http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/3587/FullAdaptivePoisoned/default/dacapo/lusearch/default/Output.txt

Show

Ian Rogers added a comment - 27/Apr/08 11:16 AM Still present and effect DaCapo benchmarks too: http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/3587/FullAdaptivePoisoned/default/dacapo/lusearch/default/Output.txt

Hide

Permalink

Ian Rogers added a comment - 15/Aug/08 1:44 PM

~~RVM-633~~ might have an effect on this issue.

Show

Ian Rogers added a comment - 15/Aug/08 1:44 PM ~~RVM-633~~ might have an effect on this issue.

Hide

Permalink

Ian Rogers added a comment - 15/Aug/08 2:39 PM

A more recent example of this failure:
http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/4608/FullAdaptivePoisoned/default/SPECjbb2005/SPECjbb2005/default/Output.txt

Show

Ian Rogers added a comment - 15/Aug/08 2:39 PM A more recent example of this failure: http://jikesrvm.anu.edu.au/cattrack/results/rvmx86lnx32b.anu.edu.au/sanity/4608/FullAdaptivePoisoned/default/SPECjbb2005/SPECjbb2005/default/Output.txt

Hide

Permalink

Daniel Frampton added a comment - 16/Aug/08 8:07 AM

This should have been fixed with the new Finalization processing code.

The previous code was unsafe in that it did not use the correct barriers when accessing the ready for finalize table. Closing to be reopened if the error shows up again.

Show

Daniel Frampton added a comment - 16/Aug/08 8:07 AM This should have been fixed with the new Finalization processing code. The previous code was unsafe in that it did not use the correct barriers when accessing the ready for finalize table. Closing to be reopened if the error shows up again.

People

Assignee:

Daniel Frampton

Reporter:

Ian Rogers

Vote (0)

Watch (0)

Dates

Created:

15/Apr/08 10:30 AM

Updated:

16/Aug/08 8:07 AM

Resolved:

16/Aug/08 8:07 AM