harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Pavel Afremov (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HARMONY-5309) [EUT] 274 NPE in checkAndTagAsMalformed on Linux x86_64
Date Tue, 18 Dec 2007 20:15:43 GMT

    [ https://issues.apache.org/jira/browse/HARMONY-5309?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12552882
] 

Pavel Afremov commented on HARMONY-5309:
----------------------------------------

Evaluating of Jited code in the place of NullPointer Exception throwing, shows that bug is
in JET. JET generate incorrect offset for switch table in the case when switch value is large
and fits in 32 bits, but offset in the table (value*8) doesn't fit in 32 bits.

In Our case following code is generated:
0x00002aaac347153e:     mov    0xfffffffffffffd88(%rbp),%r11d
0x00002aaac3471545:     cmp    $0x60000106,%r11d
0x00002aaac347154c:     jg     0x2aaac3471585
0x00002aaac3471552:     cmp    $0x600000cc,%r11d
0x00002aaac3471559:     jl     0x2aaac3471585
0x00002aaac347155f:     mov    $0x2aaac3078ad0,%rax
0x00002aaac3471569:     movslq %r11d,%r11
0x00002aaac347156c:     mov    0xfffffffffffff9a0(%rax,%r11,8),%rax
0x00002aaac3471574:     jmpq   *%rax

Where :
(gdb) i r r11
r11            0x600000f0       1610612976

0x600000f0 * 8 generates too large offset 

So SigSegv happened:
Program received signal SIGSEGV, Segmentation fault.
0x00002aaac347156c in ?? ()
0x2aaac347156c: mov    0xfffffffffffff9a0(%rax,%r11,8),%rax


> [EUT] 274 NPE in checkAndTagAsMalformed on Linux x86_64
> -------------------------------------------------------
>
>                 Key: HARMONY-5309
>                 URL: https://issues.apache.org/jira/browse/HARMONY-5309
>             Project: Harmony
>          Issue Type: Bug
>          Components: App-Oriented Bug Reports
>         Environment: Linux / x86_64
>            Reporter: Vladimir Beliaev
>
> 274 tests from EUT suites failed with NPE in the same method on Linux x86_64:
>       java.lang.NullPointerException
>      at org.eclipse.jdt.core.dom.ASTSyntaxErrorPropagator.checkAndTagAsMalformed(ASTSyntaxErrorPropagator.java:34)
>      at org.eclipse.jdt.core.dom.ASTSyntaxErrorPropagator.visit(ASTSyntaxErrorPropagator.java:117)
>      at org.eclipse.jdt.core.dom.CompilationUnit.accept0(CompilationUnit.java:208)
>      <skipped>
> This is not a regressions, still a sensetive difference of EUT behaviour on Linux x86_64
from x86 platforms, so it better be fixed.

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


Mime
View raw message