Re: Patch: replace mutex with object synchronization

[Andrew Haley <aph at redhat dot com>]

Jeff Sturm writes:
 > On Tue, 20 Apr 2004, Per Bothner wrote:
 > > Jeff Sturm wrote:
 > >
 > > > When I ran the -fno-assume-compiled experiment I was simply trying to
 > > > link as small an executable as possible.  Efficiency wasn't important.  I
 > > > ran the application once printing the name of each class as it was loaded,
 > >
 > > Did you use any special tools or patch for this?
 > 
 > Not really.  I simply added a print statement to initializeClass().
 > 
 > > I'm thinking it might be useful to implement some property that
 > > if set will print out to a file each class that gets initialized.
 > > Optionally, it might print out a stack trace, in case one wants
 > > to figure out "why is this class gettin initialized"?
 > 
 > Definitely.  At the very least, something like the JRE's -verbose:class
 > option is absent from libgcj.

There's a rough cut of this on the gcj-abi-2-dev-branch.

 > (The ELFClassLoader I wrote about earlier has a similar feature, enabled
 > with setVerbose.)

Ah yes, the ELFClassLoader.

The glibc shared library loader goes to great lengths to prevent a DSO
from being loaded twice.  The only way I can see around this is to
copy the library: simply changing its name by making a link doesn't
work, as the shared library loader checks the inode number.  So, it
looks to me like the ELFClassLoader could turn out to be very useful
indeed.

Andrew.