review request for JDK-8013461 There is a symbol AsyncGetCallTrace in libjvm.symbols that does not exist in minimal/libjvm.a when DEBUG_LEVEL == release

[Kelly O'Hair]

On May 21, 2013, at 12:19 PM, JOSEPH PROVINO wrote:

>> 
>> Let's ask the Solaris Studio guys directly.
>> I'm adding Oleg to the mailing list.
>> 
>> Oleg,
>> 
>> Could you, please, share your view on this problem?
> 
> In particular what will the Sun Studio Profiler collector do if it gets the error
> 
> trace->num_frames = ticks_no_class_load; // -1
> 
> Thanks.
> 
> joe

I am pretty sure you want to keep the symbol in the library and have it return an error code, rather
than remove the symbol entirely.
Most tools will not necessarily know what kind of jdk/jre is being used, and getting a runtime linker
error means the tool can't get off the ground, unless the tool library is doing a dlsym() to see if the
symbol exists or not first, and I kind of doubt they would do that in all cases, but maybe. Kind of
depends on what JDK implementations that they might expect their tool to run on. Hard to tell.

The tools will have many different features, and maybe the need for this symbol is only part of one
feature, and the other features might work fine in a minimal environment.

Granted, I'm speaking with very limited knowledge if this detailed discussion, but removing symbols
from shared libraries can be a very disruptive thing to tool vendors. Just a warning.

-kto
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/serviceability-dev/attachments/20130521/82863069/attachment.html