Marc Saegesser wrote:
>
> This is a truly fascinating thread of discussion. However, from reading the
> article _The "Double-Checked Locking is Broken" Declaration_
> (http://www.cs.umd.edu/~pugh/java/memoryModel/DoubleCheckedLocking.html)
>
> It seems to me that the following code is thread safe.
>
> if (_jspx_inited == false) {
> synchronized (this) {
> if (_jspx_inited == false) {
> _jspx_init();
> _jspx_inited = true;
> }
> }
> }
>
> The case described in the article is the following
>
> class Foo {
> private Helper helper = null;
> public Helper getHelper() {
> if (helper == null)
> synchronized(this) {
> if (helper == null)
> helper = new Helper();
> }
> return helper;
> }
> // other functions and members...
> }
> }
>
> The problem is that helper may be assigned a value before the Helper
> constructor executes. Thus another thread may come along and notice a
> non-null value for helper and attempt to use un-initialized values.
>
> In the _jspx_inited case above the only requirement is that compiler can't
> rewrite the code into the equivalent of
>
> if (_jspx_inited == false) {
> synchronized (this) {
> if (_jspx_inited == false) {
> _jspx_inited = true;
> _jspx_init();
> }
> }
> }
>
> Such a re-ordering seems illegal to me (what if jspx_init() uses the value
> of _jspx_inited?). If, however, it is legal reordering then the example of
> a "correct" double-check mechanism for 32-bit primitive values should work
> here. It would look something like
>
> boolean tmpInited = _jspx_inited;
> if (tmpInited == false) {
> synchronized (this) {
> if (tmpInited == false) {
> tmpInited = _jspx_init(); // NOTE: _jspx_init() needs to
> return true
> _jspx_inited = tmpInited;
> }
> }
> }
The issue as I understand it is that the constructor might
have been inlined and then the inlined instructions may have been
re-ordered. If _jspx_init() can be inlined then it might exhibit
the same problem. My question is what the spec says about inlining
virtual methods... I'm pretty sure that _jspx_init() is only a
candidate for inlining through runtime compilation by Hotspot. But
I'm not even sure of that.
Otherwise, if it can never be inlined then your original
assertion is correct.
-Paul Speed
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, email: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
