[Soot-list] Bug in traps translation

[Bodden, Eric]

Hi Martin.

> Which means that block4 goes into an infinite loop if I'm not mistaking, right? 

Well, it means that it _might_ go into an infinite loop - which it _will_ if the unlocking fails at runtime. Again it seems like the compiler is at error here. Soot only translates 1:1 what's stated in the bytecode.

> That causes a lot of headache in my analysis... is there an easy way I can fix that? Btw., I'm still on soot-2.5.0 because of the line number problem. Someone mentioned I can fix that by using coffi instead of ASM but I have no idea what that means :(

With 2.5.0 you don't need to worry. You are using Coffi anyway. But as I wrote, this is not the issue. The bytecode already looks odd. So for Soot there is not much it could do. You could just write a Soot pass that finds such situations and somehow modifies the Jimple to make them go away (diverging from the semantics of the original bytecode). It would be better to figure out why the compiler is doing what it's doing.

Best wishes,
Eric
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP using GPGMail
Url : http://mailman.CS.McGill.CA/pipermail/soot-list/attachments/20140910/44b20e7f/attachment.bin