[LLVMdev] A working garbage collector - finally :)

[Rafael Avila de Espindola]

On 11-03-09 02:34 PM, Reid Kleckner wrote:
> Where's the problem?  A pointer to v's alloca escapes to llvm.gcroot,
> so the optimizers should know that foo could modify the value it
> holds.  foo() might also read v[i] through the escaped pointer, so the
> store will have to happen before the call.

It is fine, sorry. I had a bug in the example IL I wrote, I had added 
the call after the load of 'v' but before the getelement pointer. If I 
put it in the right place I get a load of 'v' in the correct side of the 
call.

> Reid

Thanks,
Rafael