Merge pull request #5 from br0kej/dev

Enhancing Call graph generation - Func Metadata and Parallel Deduplication

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "bin2ml"
-version = "0.2.3"
+version = "0.2.4"
 edition = "2021"
 
 [dependencies]

data-examples/processed/graphs/test_bin_cg-1hop-callers-meta/entry0-1hop-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"entry0","functionFeatureSubset":{"name":"entry0","ninstrs":13,"edges":0,"indegree":0,"outdegree":1,"nlocals":0,"nargs":1,"signature":"entry0 (int64_t arg3);"}},{"id":1,"funcName":"unk.0x3fe0","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-1hop-callers-meta/main-1hop-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"main","functionFeatureSubset":{"name":"main","ninstrs":38,"edges":11,"indegree":3,"outdegree":6,"nlocals":2,"nargs":0,"signature":"int main (int argc, char **argv, char **envp);"}},{"id":1,"funcName":"sym.imp.printf","functionFeatureSubset":{"name":"sym.imp.printf","ninstrs":2,"edges":0,"indegree":6,"outdegree":0,"nlocals":0,"nargs":0,"signature":"int sym.imp.printf (const char *format);"}}]}

data-examples/processed/graphs/test_bin_cg-1hop-callers-meta/sym.__do_global_dtors_aux-1hop-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__do_global_dtors_aux","functionFeatureSubset":{"name":"sym.__do_global_dtors_aux","ninstrs":14,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":0,"signature":"sym.__do_global_dtors_aux ();"}},{"id":1,"funcName":"sym..plt.got","functionFeatureSubset":{"name":"sym..plt.got","ninstrs":2,"edges":0,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym..plt.got ();"}},{"id":2,"funcName":"sym.deregister_tm_clones","functionFeatureSubset":{"name":"sym.deregister_tm_clones","ninstrs":9,"edges":4,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym.deregister_tm_clones ();"}}]}

data-examples/processed/graphs/test_bin_cg-1hop-callers-meta/sym.__libc_csu_init-1hop-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[{"id":3,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__libc_csu_init","functionFeatureSubset":{"name":"sym.__libc_csu_init","ninstrs":34,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":3,"signature":"sym.__libc_csu_init (int64_t arg1, int64_t arg2, int64_t arg3);"}},{"id":1,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":2,"funcName":"rsp","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}},{"id":3,"funcName":"unk.0xffffffffffffffff","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-1hop-callers-meta/sym._init-1hop-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[],[{"id":0,"weight":0}]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":1,"funcName":"unk.0xffffffffffffffff","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}},{"id":2,"funcName":"sym.__libc_csu_init","functionFeatureSubset":{"name":"sym.__libc_csu_init","ninstrs":34,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":3,"signature":"sym.__libc_csu_init (int64_t arg1, int64_t arg2, int64_t arg3);"}}]}

data-examples/processed/graphs/test_bin_cg-1hop-meta/entry0-1hop-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"entry0","functionFeatureSubset":{"name":"entry0","ninstrs":13,"edges":0,"indegree":0,"outdegree":1,"nlocals":0,"nargs":1,"signature":"entry0 (int64_t arg3);"}},{"id":1,"funcName":"unk.0x3fe0","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-1hop-meta/main-1hop-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"main","functionFeatureSubset":{"name":"main","ninstrs":38,"edges":11,"indegree":3,"outdegree":6,"nlocals":2,"nargs":0,"signature":"int main (int argc, char **argv, char **envp);"}},{"id":1,"funcName":"sym.imp.printf","functionFeatureSubset":{"name":"sym.imp.printf","ninstrs":2,"edges":0,"indegree":6,"outdegree":0,"nlocals":0,"nargs":0,"signature":"int sym.imp.printf (const char *format);"}}]}

data-examples/processed/graphs/test_bin_cg-1hop-meta/sym.__do_global_dtors_aux-1hop-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__do_global_dtors_aux","functionFeatureSubset":{"name":"sym.__do_global_dtors_aux","ninstrs":14,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":0,"signature":"sym.__do_global_dtors_aux ();"}},{"id":1,"funcName":"sym..plt.got","functionFeatureSubset":{"name":"sym..plt.got","ninstrs":2,"edges":0,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym..plt.got ();"}},{"id":2,"funcName":"sym.deregister_tm_clones","functionFeatureSubset":{"name":"sym.deregister_tm_clones","ninstrs":9,"edges":4,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym.deregister_tm_clones ();"}}]}

data-examples/processed/graphs/test_bin_cg-1hop-meta/sym.__libc_csu_init-1hop-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[{"id":3,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__libc_csu_init","functionFeatureSubset":{"name":"sym.__libc_csu_init","ninstrs":34,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":3,"signature":"sym.__libc_csu_init (int64_t arg1, int64_t arg2, int64_t arg3);"}},{"id":1,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":2,"funcName":"rsp","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}},{"id":3,"funcName":"unk.0xffffffffffffffff","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-1hop-meta/sym._init-1hop-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":1,"funcName":"unk.0xffffffffffffffff","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-cg-callers-meta/entry0-cg-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"entry0","functionFeatureSubset":{"name":"entry0","ninstrs":13,"edges":0,"indegree":0,"outdegree":1,"nlocals":0,"nargs":1,"signature":"entry0 (int64_t arg3);"}},{"id":1,"funcName":"unk.0x3fe0","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-cg-callers-meta/main-cg-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"main","functionFeatureSubset":{"name":"main","ninstrs":38,"edges":11,"indegree":3,"outdegree":6,"nlocals":2,"nargs":0,"signature":"int main (int argc, char **argv, char **envp);"}},{"id":1,"funcName":"sym.imp.printf","functionFeatureSubset":{"name":"sym.imp.printf","ninstrs":2,"edges":0,"indegree":6,"outdegree":0,"nlocals":0,"nargs":0,"signature":"int sym.imp.printf (const char *format);"}}]}

data-examples/processed/graphs/test_bin_cg-cg-callers-meta/sym.__do_global_dtors_aux-cg-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__do_global_dtors_aux","functionFeatureSubset":{"name":"sym.__do_global_dtors_aux","ninstrs":14,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":0,"signature":"sym.__do_global_dtors_aux ();"}},{"id":1,"funcName":"sym..plt.got","functionFeatureSubset":{"name":"sym..plt.got","ninstrs":2,"edges":0,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym..plt.got ();"}},{"id":2,"funcName":"sym.deregister_tm_clones","functionFeatureSubset":{"name":"sym.deregister_tm_clones","ninstrs":9,"edges":4,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym.deregister_tm_clones ();"}}]}

data-examples/processed/graphs/test_bin_cg-cg-callers-meta/sym.__libc_csu_init-cg-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__libc_csu_init","functionFeatureSubset":{"name":"sym.__libc_csu_init","ninstrs":34,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":3,"signature":"sym.__libc_csu_init (int64_t arg1, int64_t arg2, int64_t arg3);"}},{"id":1,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":2,"funcName":"rsp","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-cg-callers-meta/sym._init-cg-callers-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[],[{"id":0,"weight":0}]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":1,"funcName":"unk.0xffffffffffffffff","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}},{"id":2,"funcName":"sym.__libc_csu_init","functionFeatureSubset":{"name":"sym.__libc_csu_init","ninstrs":34,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":3,"signature":"sym.__libc_csu_init (int64_t arg1, int64_t arg2, int64_t arg3);"}}]}

data-examples/processed/graphs/test_bin_cg-cg-callers/entry0-cg-callers.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"entry0"},{"id":1,"funcName":"unk.0x3fe0"}]}

data-examples/processed/graphs/test_bin_cg-cg-callers/main-cg-callers.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"main"},{"id":1,"funcName":"sym.imp.printf"}]}

data-examples/processed/graphs/test_bin_cg-cg-callers/sym.__do_global_dtors_aux-cg-callers.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__do_global_dtors_aux"},{"id":1,"funcName":"sym..plt.got"},{"id":2,"funcName":"sym.deregister_tm_clones"}]}

data-examples/processed/graphs/test_bin_cg-cg-callers/sym.__libc_csu_init-cg-callers.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__libc_csu_init"},{"id":1,"funcName":"sym._init"},{"id":2,"funcName":"rsp"}]}

data-examples/processed/graphs/test_bin_cg-cg-callers/sym._init-cg-callers.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[],[{"id":0,"weight":0}]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym._init"},{"id":1,"funcName":"unk.0xffffffffffffffff"},{"id":2,"funcName":"sym.__libc_csu_init"}]}

data-examples/processed/graphs/test_bin_cg-cg-meta/entry0-cg-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"entry0","functionFeatureSubset":{"name":"entry0","ninstrs":13,"edges":0,"indegree":0,"outdegree":1,"nlocals":0,"nargs":1,"signature":"entry0 (int64_t arg3);"}},{"id":1,"funcName":"unk.0x3fe0","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-cg-meta/main-cg-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"main","functionFeatureSubset":{"name":"main","ninstrs":38,"edges":11,"indegree":3,"outdegree":6,"nlocals":2,"nargs":0,"signature":"int main (int argc, char **argv, char **envp);"}},{"id":1,"funcName":"sym.imp.printf","functionFeatureSubset":{"name":"sym.imp.printf","ninstrs":2,"edges":0,"indegree":6,"outdegree":0,"nlocals":0,"nargs":0,"signature":"int sym.imp.printf (const char *format);"}}]}

data-examples/processed/graphs/test_bin_cg-cg-meta/sym.__do_global_dtors_aux-cg-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__do_global_dtors_aux","functionFeatureSubset":{"name":"sym.__do_global_dtors_aux","ninstrs":14,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":0,"signature":"sym.__do_global_dtors_aux ();"}},{"id":1,"funcName":"sym..plt.got","functionFeatureSubset":{"name":"sym..plt.got","ninstrs":2,"edges":0,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym..plt.got ();"}},{"id":2,"funcName":"sym.deregister_tm_clones","functionFeatureSubset":{"name":"sym.deregister_tm_clones","ninstrs":9,"edges":4,"indegree":1,"outdegree":0,"nlocals":0,"nargs":0,"signature":"sym.deregister_tm_clones ();"}}]}

data-examples/processed/graphs/test_bin_cg-cg-meta/sym.__libc_csu_init-cg-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__libc_csu_init","functionFeatureSubset":{"name":"sym.__libc_csu_init","ninstrs":34,"edges":5,"indegree":0,"outdegree":2,"nlocals":0,"nargs":3,"signature":"sym.__libc_csu_init (int64_t arg1, int64_t arg2, int64_t arg3);"}},{"id":1,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":2,"funcName":"rsp","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-cg-meta/sym._init-cg-meta.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym._init","functionFeatureSubset":{"name":"sym._init","ninstrs":8,"edges":3,"indegree":1,"outdegree":1,"nlocals":0,"nargs":0,"signature":"sym._init ();"}},{"id":1,"funcName":"unk.0xffffffffffffffff","functionFeatureSubset":{"name":"","ninstrs":0,"edges":0,"indegree":0,"outdegree":0,"nlocals":0,"nargs":0,"signature":""}}]}

data-examples/processed/graphs/test_bin_cg-cg/entry0-cg.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"entry0"},{"id":1,"funcName":"unk.0x3fe0"}]}

data-examples/processed/graphs/test_bin_cg-cg/sym.__do_global_dtors_aux-cg.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__do_global_dtors_aux"},{"id":1,"funcName":"sym..plt.got"},{"id":2,"funcName":"sym.deregister_tm_clones"}]}

data-examples/processed/graphs/test_bin_cg-cg/sym.__libc_csu_init-cg.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":2,"weight":0},{"id":1,"weight":0}],[],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym.__libc_csu_init"},{"id":1,"funcName":"sym._init"},{"id":2,"funcName":"rsp"}]}

data-examples/processed/graphs/test_bin_cg-cg/sym._init-cg.json

@@ -0,0 +1 @@
+{"adjacency":[[{"id":1,"weight":0}],[]],"directed":"True","graph":[],"multigraph":false,"nodes":[{"id":0,"funcName":"sym._init"},{"id":1,"funcName":"unk.0xffffffffffffffff"}]}

src/afij.rs

@@ -98,7 +98,7 @@ pub struct Regvar {
     pub ref_field: String,
 }
 
-#[derive(Default, Debug, Clone, PartialEq, Serialize, Deserialize)]
+#[derive(Default, Debug, Clone, PartialEq, Hash, Eq, Serialize, Deserialize)]
 #[serde(rename_all = "camelCase")]
 pub struct AFIJFeatureSubset {
     pub name: String,
@@ -115,8 +115,8 @@ impl From<&AFIJFunctionInfo> for AFIJFeatureSubset {
     fn from(src: &AFIJFunctionInfo) -> AFIJFeatureSubset {
         AFIJFeatureSubset {
             name: src.name.clone(),
-            ninstrs: src.ninstrs.clone(),
-            edges: src.edges.clone(),
+            ninstrs: src.ninstrs,
+            edges: src.edges,
             indegree: src.indegree.unwrap_or(0),
             outdegree: src.outdegree.unwrap_or(0),
             nlocals: src.nlocals.unwrap_or(0),