RFR(XXS): 8144197: Possible use after free in Arguments::add_property function

[Lindenmaier, Goetz]

Hi Dmitry,

this looks good.  

How did you spot this, did you really get an oom there?

Best regards,
  Goetz.

> -----Original Message-----
> From: hotspot-runtime-dev [mailto:hotspot-runtime-dev-
> bounces at openjdk.java.net] On Behalf Of Dmitry Dmitriev
> Sent: Thursday, December 10, 2015 10:13 AM
> To: hotspot-runtime-dev at openjdk.java.net
> Subject: Re: RFR(XXS): 8144197: Possible use after free in
> Arguments::add_property function
> 
> Hello,
> 
> Please, can I one more "r" review for that fix? Thank you!
> 
> Dmitry
> 
> On 01.12.2015 14:38, Dmitry Dmitriev wrote:
> > Hello,
> >
> > Please, review this small fix. JDK-8132725 removes memory leak in
> > Arguments::add_property function in
> > hotspot/src/share/vm/runtime/arguments.cpp module, but introduce
> > theoretical possibility of use after free of _java_command, because
> > _java_command used in reporting OOM errors. This can happen if
> > sun.java.command property is defined more than once and JVM can not
> > allocate memory for new java command when processing second(or later)
> > "sun.java.command" property. Thus sun.java.command can point to the
> > freed memory when reporting OOM error, because memory for
> > sun.java.command is freed before allocating new sun.java.command. Fix
> > is simple - free old sun.java.command only after allocating memory for
> > the new one. The similar thing is done for java.vendor.url.bug
> > property in case if fatal error occurred during allocation of new
> > java.vendor.url.bug property.
> >
> > JBS: https://bugs.openjdk.java.net/browse/JDK-8144197
> > webrev.00: http://cr.openjdk.java.net/~ddmitriev/8144197/webrev.00/
> > <http://cr.openjdk.java.net/%7Eddmitriev/8144197/webrev.00/>
> > Testing: locally, JPRT
> >
> > Thanks,
> > Dmitry