[jira] Updated: (HARMONY-2396) [drlvm][jit] Jitrino fails to run SpecJBB2005 in -Xem:opt mode

["Mikhail Fursov (JIRA)" <ji...@apache.org>]

     [ http://issues.apache.org/jira/browse/HARMONY-2396?page=all ]

Mikhail Fursov updated HARMONY-2396:
------------------------------------

    Attachment: tauedge.diff

the fix. 

> [drlvm][jit] Jitrino fails to run SpecJBB2005 in -Xem:opt mode
> --------------------------------------------------------------
>
>                 Key: HARMONY-2396
>                 URL: http://issues.apache.org/jira/browse/HARMONY-2396
>             Project: Harmony
>          Issue Type: Bug
>          Components: DRLVM
>            Reporter: Mikhail Fursov
>         Attachments: tauedge.diff
>
>
> SpecJBB2005 crashes in -Xem:opt mode.
> After analysis of IR I prepared this minimized testcase to reproduce the problem:
> public class Test {
>     public static void main(String[] args) {
>         foo3(new Test(), 10, new char[100]);
>     }
>     
>     int position;
>     static void foo3 (Test x, int i, char[] arr) {
>         int j=x.position;
>         if(j < i) {
>             System.arraycopy(arr, j, arr, i, 10);
>         }
>     }
> }
> The problem is in memopt: it produces code with tau opnds mixed with numerics at in the same insts:
> The part of IR before memopt:
>   I42:cgt:i4  t3, t9 -) t22:int32
>   I43:and     t18, t22 -) t23:int32
> The part of IR after memopt:
>   I125:copy     t55 -) t22:int32
>    I43:and      t18, t22 -) t23:int32
> Where t55 is
>   I118:tauedge() -) t55:tau
> I propose replacing I125 copy inst with ldconst #1 for tauedge insts in HVN pass to solve the problem.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira