auditor

An aarch64 constant-time memory access auditing tool.
git clone git://git.ppad.tech/auditor.git
Log | Files | Refs | README | LICENSE

ARCH8.md (1345B)

      1 # ARCH8: Inter-Procedural Performance Caches
      2 
      3 ## Goal
      4 
      5 Eliminate avoidable recomputation in inter-procedural analysis by
      6 caching call-graph and function block ranges, reducing worklist
      7 iteration cost on large inputs.
      8 
      9 ## Scope
     10 
     11 - Precompute call graph once per run.
     12 - Precompute per-function block ranges/indices once per run.
     13 - Keep analysis semantics unchanged.
     14 
     15 ## Hotspots
     16 
     17 - `runInterProc` rebuilds call graph and rescans functions each
     18   iteration.
     19 - `functionBlocks` linearly scans blocks every call.
     20 
     21 ## Architecture
     22 
     23 ### Call Graph Cache
     24 
     25 - Build `callGraph :: Map Text [Text]` once.
     26 - Build `callerMap :: Map Text [Text]` once by inverting the call graph.
     27 - Inter-proc loop uses `callerMap` for worklist expansion.
     28 
     29 ### Function Block Cache
     30 
     31 - Build `funcBlocks :: Map Text [Int]` once using a single pass through
     32   block labels.
     33 - Optionally store `(startIdx, endIdx)` to avoid per-block membership
     34   sets where possible.
     35 
     36 ## Integration
     37 
     38 - Add helper(s) in `CFG` or `Taint` to compute caches.
     39 - Thread caches into `runInterProc` and config-aware variants.
     40 - Preserve existing APIs where possible; add internal variants if
     41   needed.
     42 
     43 ## Risks
     44 
     45 - Incorrect function boundary detection could change analysis coverage.
     46   Use existing `isFunctionLabel` semantics to derive ranges.
     47 - Memory overhead is minimal compared to analysis state.