Wrong offset in hs_err_*.log

[Zhengyu Gu]

That explains. When there is not .pdb file, it decodes to the nearest public (exported) symbol from corresponding DLL file.

Thanks,

-Zhengyu


On Jun 17, 2013, at 10:40 AM, Artem Ananiev wrote:

> Hi, Mikael,
> 
> On 6/14/2013 8:17 PM, Mikael Gerdin wrote:
>> Artem,
>> 
> 
> [skip]
> 
>> 
>> Do you have a .pdb file for glass.dll?
> 
> I personally have never observed crashes like this, all of them are reported by external people. They run promoted JDK/JavaFX builds, which don't contain any .pdb files.
> 
>> There is some logic to set up paths to find .pdb files but it may need
>> some tuning.
>> 
>> That's the only way we can give you a sensible stacktrace for
>> non-exported functions.
>> 
>> I think that the reason you get _getKeyCodeForChar() is that it's the
>> closest exported symbol to where you crash.
> 
> It is quite expected: we only export Java native methods in release builds. In such case, I would expect the log file to contain something like this:
> 
>> # Problematic frame:
>> # J  com.sun.glass.ui.win.WinWindow._close(J)Z
> 
> Thanks,
> 
> Artem
> 
>> /Mikael