www-builds mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Martin Desruisseaux <martin.desruisse...@geomatys.fr>
Subject Re: Apache build infrastructure or Oracle JVM problem: crash in native JDK code
Date Wed, 06 Feb 2013 11:12:49 GMT
Hello all, and thanks for the reply.

Le 06/02/13 02:49, Jesse Glick a écrit :
> On 02/05/2013 07:30 PM, sebb wrote:
>> a pure Java application should not be able to cause a JVM crash
> Depends entirely on what you mean by “pure Java application”. When you 
> have filesystem and Process access, you can do whatever the user shell 
> account can do, which certainly includes triggering native errors.

But at least in the Process case, the crash would occur outside the JVM, 
isn't it?


> At any rate I personally agree that using java.util.zip in this way 
> should not cause JVM crashes—but it does, and the JDK team has 
> explicitly decided that this is “not a bug” and closed discussion. The 
> workaround, or depending on your perspective the fix, is to ensure 
> that you do not clobber an open ZIP file (which would also prevent a 
> lock error on Windows).

In our case, we are not handling the ZIP files ourself. The ZIP file is 
opened by the Java ClassLoader, and the JAR file were created by the 
standard Maven plugin. We are not doing parallel compilation. 
Consequently, it is not sure that the crash is caused by clobbering an 
open ZIP file. Furthermore, the approach that we wanted to apply has 
been used for more than 5 years on Windows, Solaris, Linux Gentoo, Linux 
Ubuntu, MacOS, and other Hudson/Jenkins servers without any JVM crash.

Anyway, if the JVM were "allowed" to crash when clobbering an open ZIP 
file, then any user or external application touching a ZIP file could 
crash a running JVM...

     Martin


Mime
View raw message