[Mono-bugs] [Bug 651895] New: Reproducable crash associated with SGen

[bugzilla_noreply at novell.com]

https://bugzilla.novell.com/show_bug.cgi?id=651895

https://bugzilla.novell.com/show_bug.cgi?id=651895#c0


           Summary: Reproducable crash associated with SGen
    Classification: Mono
           Product: Mono: Runtime
           Version: 2.8.x
          Platform: i686
        OS/Version: Linux
            Status: NEW
          Severity: Normal
          Priority: P5 - None
         Component: GC
        AssignedTo: lupus at novell.com
        ReportedBy: stefanor at cox.net
         QAContact: mono-bugs at lists.ximian.com
          Found By: ---
           Blocker: ---


User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.13)
Gecko/20100916 Iceweasel/3.5.13 (like Firefox/3.5.13)

When running the testcase at
http://feather.perl6.nl/~sorear/Bug-mono-20101105.zip with mono-sgen 2.8.0,
there is a crash in the collector.  With a test vector size of 3, the crash
occurs during domain cleanup.

If the test vector size is increased (say, to 1000), then a spurious exception
will be generated during the execution of the test itself; the segfault then
happens in ves_icall_System_Environment_Exit.

The attached executable is the result of an in-development compiler, so I'm at
a loss for how to effectively reduce it.  The compiler generates C# rather than
directly emitting CIL, and the .exe passes mono --verify-all, so I doubt that
invalid or memory-unsafe IL is involved.

Reproducible: Always

Steps to Reproduce:
See run.sh in the linked zip.
Actual Results:  
SEGV in native code; see run.log in the zip

Expected Results:  
Same as expected results up to the OK, but without the crash.  With the 1000
element test vector, the expected results are the same but much longer.

This copy of Mono 2.8 was installed on top of a 2.6.7 install.

-- 
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.