On Monday, December 30, 2013 4:40:28 PM UTC+4, Dmitry Vyukov wrote:
>
> However, keep in mind the following aspects: 
> 1. We would prefer to do checks before the memory access in order to 
> not corrupt memory. This rules out some optimizations based on 
> post-dominance. 
>

Good point. Obviously, the optimizer would need special handling of the 
checks.
 

> 2. We would like to have precise line number in reports. This rules 
> out some inter-basicblock optimizations. 
> There is still some room for aggregation of accesses to adjacent 
> locations and optimizations based on pre-dominance. 
>

You mean we could have an instance of the checking code shared between 
several pre-dominators so that associated source location info is a 
phi-function, correct? If so, we would need to segregate the location info 
from the checking code as well...

Another way to consider is to encode several source positions for the same 
instance of the checking code.

OK, it's something I would need to think about somewhat more. Thanks a lot 
for the comments!

-- 
You received this message because you are subscribed to the Google Groups 
"address-sanitizer" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to