svadev AT lists.siebelschool.illinois.edu
Subject: Svadev mailing list
List archive
- From: leckie peng <pengjingjian1016 AT gmail.com>
- To: svadev AT cs.uiuc.edu
- Subject: [svadev] Redundent Check Elimination
- Date: Tue, 25 Sep 2012 18:26:00 +0800
- List-archive: <http://lists.cs.uiuc.edu/pipermail/svadev/>
- List-id: <svadev.cs.uiuc.edu>
Hi,
I am a MSc candidate and very interested in the open project that creates common infrastructure for memory safety tools. To finish my degree I've been working on it for a while and read some of the literature.
As for eliminating redundant checks, which is among the goals of this project, I was considering if we can employ data flow analysis to transfer the "safety" of the memory address along CFG. Namely, if one memory access has been checked, it's possible that accesses to the same location down the execution paths don't need to be checked if the "safety" of the location can be transferred there. This is a simple idea and can be implemented with classic bit vector data flow analysis. It seems to me that the optimizations implemented in ASan and those by Ott Tinn didn't use this approach, although they can eliminate a large part of redundant checks by now, by using much analysis. This data flow approach, however, seems to me more elegant and straightforward. I don't have much experience with Compiler Techniques, it would be great to have your opinions.
Best regards.
--Jian Peng
- [svadev] Redundent Check Elimination, leckie peng, 09/25/2012
- Re: [svadev] Redundent Check Elimination, John Criswell, 09/25/2012
- Re: [svadev] Redundent Check Elimination, leckie peng, 09/26/2012
- Re: [svadev] Redundent Check Elimination, John Criswell, 09/26/2012
- Re: [svadev] Redundent Check Elimination, leckie peng, 09/26/2012
- Re: [svadev] Redundent Check Elimination, John Criswell, 09/25/2012
Archive powered by MHonArc 2.6.16.