Tracking Benchmark Changes for Thesis

[michael-schwarz]

The branch I used to benchmark for my thesis diverges at d8be117.

This draft MR is not intended to be merged but should allow me to keep an overview of the changes and see what should be maybe back-ported.

PR for master exists:

• Fixes for Self-Validation Issues #1641 addressing Apron: Invariants for variables of type unsigned long wrong #1542 and YAML: Relational invariants about escaped variables are incorrect (due to Apron->CIL conversion not considering escaping) #1547
• Make meet in AddressDomain more precise #1468
• More useful stats output by privPrecCompare #1642
• Relational MM: Cluster LMust #1643

To consider:

• Way to supply suffix to name of privPrecCompare output
• Do not colorize privPrecCompare output to stderr (in case it is saved to somewhere)
• exp.priv-prec-dump-proj option to project precision of dumps to enums, intervals, and def_exc for comparison between different activated domains
• Warn instead of crash if an argument to a string function does not evaluate to an address.
• 22fdef6 (adapt traces and traces-rel configurations to be more similar)
• 1b180d4 (Add bot_in_blob_leq_bot so bot and Blob(bot) compare equal in privPrecCompare)
• Fix to force refinement to only happen when dumping precision comparisons (to avoid Int domain refinement causes fixpoint error #1005)
• Merged Make update_offset idempotent for blobs #1559

---

Done

Back-ported:

• Typos (c28ca49, e192c8e, 5073759, c5d2b36, 3ad6b39)
• Traces: Enable def_exc_widen_by_join for reproducibility for SAS '21 results 3e13e04
• 3a30b0c (check whether cast to long long is injective before emitting invariants)

PR landed on master concurrently with MS proceeding on this branch (nothing to do):

• BaseAnalysis: For non-definite AD, join over **all** components not just cpa in set #1458
• Fix catch-all exception handlers #1435
• Union: meet in Field: Do not raise Uncomparable when one arg is Top #1471
• Fix global_init for escaped globals in mutex-meet-tid cluster configuration #1481
• Fix relation analysis ignoring extern inits #1444
• Fix mutex-meet for malloc after thread creation #1492
• Refactor must-locksets to use definite mvals instead of addresses #1430
• Fix privatization unsoundness due to non-definite and unknown mutexes #1500
• Make privatization sync Join more precise when threadflag is path-sensitive #1475
• Introduce sync reason JoinCall & Limit sync at function start nodes #1508

Hotfixes:

• Hotfix for Printing of Apron values very slow  #1513 (should be solved by addressing Pretty.text very slow and uses a lot of memory cil#169 instead)

PR with principled fix exists:

• Quick fix for Crash on pthread_mutex_lock with complicated argument #1421