Re: Runtime Link-Loader Behavior and ODR (Is this a Bug?)

On Sat, Oct 23, 2010 at 10:33 PM, Ian Lance Taylor <[EMAIL PROTECTED]> wrote:
> Jeffrey Walton <[EMAIL PROTECTED]> writes:
>
>> (1) Since behavior for construction and destruction differ, (I expect
>> one behavior or another), I claim there is a bug in the runtime
>> link-loader.
>
> In your example, unloading Plugin/DSO 1 should not unload the Crypto++
> shared library.
In my test cases under the debugger, 'i sh' shows the Crypto++ [shared
object] is still loaded. Unless of course, gdb defers unloading.

> If that library is being unloaded, then it does seem
> that the dynamic linker has a bug.  If the library is not unloaded, then
> the global destructors should not be run, so then there is a different
> bug.
OK. This might be a 'different bug'.

> The dynamic linker is part of glibc, not gcc.  I recommend that you try
> to reproduce this bug with a complete standalone test case.  Then report
> that testcase at http://sourceware.org/bugzilla .