andr3/scyther
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

- Debug of optimization.

Browse Source 
- Put down skeleton of new idea, related to a lemma from Niek Palm's
  work. This needs more investigating though.
This commit is contained in:
ccremers 2004-07-13 15:24:47 +00:00
parent b6806f6aaf
commit 39a2b4878c
1 changed files with 55 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
src/modelchecker.c
View File

@ -208,26 +208,25 @@ explorify (const System sys, const int run)
flag = 0;
/*
* Special check: if agents have been instantiated in such a way that no more claims need to be evaluated, then we can skip
* further traversal.
* Two cases: internal read or first read of a run; both are the first event of a run.
*
* Efficiency of the next check heavily relies on lazy L-R evaluation
* Special checks after (implicit) choose events; always first in run reads.
*/
if (rd == sys->runs[run].start && rd->type == READ)
{
int rid;
if (inTermlist (sys->untrusted, agentOfRun (sys, run)))
{
/* this run is executed by an untrusted agent, do not explore */
return 0;
}
else
{
/* executed by trusted agent, but is there a claim left to explore? */
/* executed by trusted agent */
/* Special check 1: if agents have been instantiated in such a way that no more claims need to be evaluated, then we can skip
* further traversal.
*/
if (sys->secrets == NULL)
{ /* there are no remaining secrecy claims to be checked */
Roledef rdscan;
int rid;
int validclaim;
rid = 0;
@ -261,8 +260,25 @@ explorify (const System sys, const int run)
return 0;
}
}
/* Special check 2: if all agents in each run send only encrypted stuff, and all agents are trusted,
* there is no way for the intruder to learn anything else than encrypted terms, so secrecy claims will not
* be violated anymore if they contain no terms that are encrypted with such keys */
//!@todo For now, there is no check that the runs only send publicly encrypted stuff! Just an assumption to be made true using static analysis.
/*
rid = 0;
while (rid < sys->maxruns)
{
if (!untrustedAgent (sys, sys->runs[rid].agents))
{
}
rid++;
}
*/
}
/* Apparently, all is well, and we can explore further */
if (executeStep (sys, run))
{
/* traverse the system after the step */