harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexei Fedotov (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HARMONY-3862) [drlvm][verifier] A subroutine splits execution into several ret instructions
Date Tue, 19 Jun 2007 16:36:26 GMT

    [ https://issues.apache.org/jira/browse/HARMONY-3862?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12506240

Alexei Fedotov commented on HARMONY-3862:

This is an interesting example. My current belief is that a compiler generates non-verifiable
code. Consider two ret instructions: 

  44: ret 3 ; subroutine 1
  53: ret 1 ; subroutine 2

Spec says: "When executing the ret instruction, which implements a return from a subroutine,
there must be only one possible subroutine from which the instruction can be returning." Now,
let's ask the question about instruction 41: invokestatic - which subroutine does it belong

Since it is followed by instruction 44:, it belongs to the first subroutine. Since it belongs
to the range [20:; 52:] it may transfer control to an exception handler 52: pop 53: ret 1
and end in 53: ret 1 instruction. 

Any ideas how to consistently explain this contradiction of generated code to spec? 

> [drlvm][verifier] A subroutine splits execution into several ret instructions
> -----------------------------------------------------------------------------
>                 Key: HARMONY-3862
>                 URL: https://issues.apache.org/jira/browse/HARMONY-3862
>             Project: Harmony
>          Issue Type: Bug
>          Components: App-Oriented Bug Reports, DRLVM
>         Environment: Harmony 5.0 M1
> Windows XP SP 2
> Eclipse 3.2
> OpenSSH_4.5p1
>            Reporter: James Roper
> I am using Eclipse with an external SSH client (the one bundled with Cygwin) to connect
to a CVS server (using the "ext" method in Eclipse).  It works fine when running Eclipse under
the Sun HotSpot JVM, however, with Harmony, Eclipse reports the following error:
> An internal error occured during: "Updating Synchronize view for CVS Workspace.".
> (class: org/eclipse/team/internal/core/subscribers/ContentComparator, method: compare(Ljava/lang/Object;Lorg/eclipse/core/runtime/IProgressMonitor;)Z)
A subroutine splits execution into several ret instructions
> If you need any more information about my environment, let me know.  Off topic, are these
the sort of bugs I should be reporting in JIRA?  I couldn't find any bugs that looked like
they were the same as this, I'm just a user trying out Milestone 1 because I want to help
the Harmony project.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message