libstdc++/5037: Multithreaded read access to strings not threadsafe on Solaris/Sparc

[Nathan Myers]

On Thu, Dec 06, 2001 at 07:40:26PM -0600, Loren James Rittle wrote:
>  ... I am afraid that this is an
>  ABI change for all sparc ports.  Assume that others might prefer
>  another final solution to fix this on the 3.0.X branch with a
>  different performance tradeoff.]

Yes, a more minimal change for 3.0.x seems advisable.
 
> In the interest of getting a fix in the mainline tree ASAP, here is
> the obvious solution that requires no change to the abstraction layer
> and I think meets any contention-efficiency goals.  
> ...   
> ! struct _Atomic_word
> ! {
> !   unsigned char __lock;
> !   int __word;
> !   operator int() const { return __word; }
> !   operator = (const int x) { return x == __word; }
> !   _Atomic_word () : __lock (0), __word (0) {}
> ! };

Unfortunately this makes a basic_string<>::_Rep substantially larger.
Also, it cannot be put in places that need a POD.  Probably if we do
something like this in 3.1 a change to the abstraction layer is 
appropriate.

Do we really expect to see much contention on this lock?  It's held
for such a short time. 

Nathan Myers
ncm at cantrip dot org