Merge pull request #275 from plasma-umass/filter-rust-stdlib

Filter Rust stdlib/dependency paths from profiling by default

Author: emeryberger

libcoz/inspect.cpp

@@ -33,6 +33,7 @@
 #include "util.h"
 
 #include "ccutil/log.h"
+#include "path_filter.h"
 
 using namespace std;
 
@@ -221,38 +222,6 @@ bool in_scope(const string& name, const unordered_set<string>& scope) {
   return in_scope_normalized(normalized, scope);
 }
 
-static bool path_has_prefix(const string& path, const string& prefix) {
-  if(prefix.empty())
-    return false;
-  if(prefix.size() > path.size())
-    return false;
-  if(path.compare(0, prefix.size(), prefix) != 0)
-    return false;
-  return path.size() == prefix.size() || path[prefix.size()] == '/';
-}
-
-static bool is_system_path(const string& normalized) {
-  static const vector<string> prefixes = {
-    "/usr/include",
-    "/usr/lib",
-    "/usr/local/include",
-    "/usr/local/lib",
-    "/lib",
-    "/lib64"
-  };
-  for(const auto& prefix : prefixes) {
-    if(path_has_prefix(normalized, prefix))
-      return true;
-  }
-  return false;
-}
-
-static bool is_coz_header(const string& path) {
-  // Never profile coz's own instrumentation header
-  const string suffix = "/coz.h";
-  return path.size() >= suffix.size() &&
-         path.compare(path.size() - suffix.size(), suffix.size(), suffix) == 0;
-}
 
 static bool file_matches_scope(const string& name,
                                const unordered_set<string>& scope,
@@ -266,6 +235,10 @@ static bool file_matches_scope(const string& name,
     return false;
   if(scope.empty())
     return true;
+  // Filter Rust toolchain/dependency paths unless user specified an explicit source scope
+  bool default_scope = (scope.size() == 1 && scope.count("%") == 1);
+  if(default_scope && is_rust_path(normalized))
+    return false;
   return in_scope_normalized(normalized, scope);
 }
 

libcoz/path_filter.h

@@ -0,0 +1,65 @@
+/*
+ * Copyright (c) 2015, Charlie Curtsinger and Emery Berger,
+ *                     University of Massachusetts Amherst
+ * This file is part of the Coz project. See LICENSE.md file at the top-level
+ * directory of this distribution and at http://github.com/plasma-umass/coz.
+ */
+
+#ifndef COZ_PATH_FILTER_H
+#define COZ_PATH_FILTER_H
+
+#include <cstdlib>
+#include <string>
+#include <vector>
+
+inline bool path_has_prefix(const std::string& path, const std::string& prefix) {
+  if(prefix.empty())
+    return false;
+  if(prefix.size() > path.size())
+    return false;
+  if(path.compare(0, prefix.size(), prefix) != 0)
+    return false;
+  return path.size() == prefix.size() || path[prefix.size()] == '/';
+}
+
+inline bool is_system_path(const std::string& normalized) {
+  static const std::vector<std::string> prefixes = {
+    "/usr/include",
+    "/usr/lib",
+    "/usr/local/include",
+    "/usr/local/lib",
+    "/lib",
+    "/lib64"
+  };
+  for(const auto& prefix : prefixes) {
+    if(path_has_prefix(normalized, prefix))
+      return true;
+  }
+  return false;
+}
+
+inline bool is_rust_path(const std::string& normalized) {
+  // /rustc/<hash>/... paths are synthetic, always start with /rustc
+  if(path_has_prefix(normalized, "/rustc"))
+    return true;
+  // .rustup and .cargo are only Rust toolchain paths when under $HOME
+  const char* home = std::getenv("HOME");
+  if(home && home[0] != '\0') {
+    std::string h(home);
+    if(path_has_prefix(normalized, h + "/.rustup"))
+      return true;
+    if(path_has_prefix(normalized, h + "/.cargo/registry"))
+      return true;
+    if(path_has_prefix(normalized, h + "/.cargo/git"))
+      return true;
+  }
+  return false;
+}
+
+inline bool is_coz_header(const std::string& path) {
+  const std::string suffix = "/coz.h";
+  return path.size() >= suffix.size() &&
+         path.compare(path.size() - suffix.size(), suffix.size(), suffix) == 0;
+}
+
+#endif  // COZ_PATH_FILTER_H

rust/src/lib.rs

@@ -209,12 +209,11 @@ type GetCounterFn = unsafe extern "C" fn(libc::c_int, *const libc::c_char) -> *m
 /// `typedef void (*coz_add_delays_t)(void);`
 type AddDelaysFn = unsafe extern "C" fn();
 
-#[cfg(target_os = "linux")]
 fn coz_get_counter(ty: libc::c_int, name: &CStr) -> Option<*mut coz_counter_t> {
     static GET_COUNTER: OnceCell<Option<GetCounterFn>> = OnceCell::new();
     let func = GET_COUNTER.get_or_init(|| {
         let name = CStr::from_bytes_with_nul(b"_coz_get_counter\0").unwrap();
-        // SAFETY: We are calling an external function that does exist in Linux.
+        // SAFETY: dlsym is available on all POSIX platforms (Linux, macOS, etc.).
         // No specific invariants that we must uphold have been defined.
         let func = unsafe { libc::dlsym(libc::RTLD_DEFAULT, name.as_ptr()) };
         if func.is_null() {
@@ -237,7 +236,6 @@ fn coz_get_counter(ty: libc::c_int, name: &CStr) -> Option<*mut coz_counter_t> {
 /// This must be called after every counter increment to allow the profiler to
 /// inject virtual delays for causal profiling experiments. Without this call,
 /// the profiler cannot detect progress points and will report 0 experiments.
-#[cfg(target_os = "linux")]
 fn coz_add_delays() {
     static ADD_DELAYS: OnceCell<Option<AddDelaysFn>> = OnceCell::new();
     let func = ADD_DELAYS.get_or_init(|| {
@@ -255,11 +253,3 @@ fn coz_add_delays() {
         unsafe { f() };
     }
 }
-
-#[cfg(not(target_os = "linux"))]
-fn coz_get_counter(_ty: libc::c_int, _name: &CStr) -> Option<*mut coz_counter_t> {
-    None
-}
-
-#[cfg(not(target_os = "linux"))]
-fn coz_add_delays() {}

tests/CMakeLists.txt

@@ -1,3 +1,12 @@
+add_executable(path_filter_test
+  ${CMAKE_SOURCE_DIR}/tests/path_filter/path_filter_test.cpp)
+target_include_directories(path_filter_test PRIVATE
+  ${CMAKE_SOURCE_DIR}/libcoz)
+target_compile_features(path_filter_test PRIVATE cxx_std_11)
+
+add_test(NAME path_filter
+  COMMAND path_filter_test)
+
 add_executable(dwarf_scope_test
   ${CMAKE_SOURCE_DIR}/tests/dwarf/dwarf_scope_test.cpp)
 target_include_directories(dwarf_scope_test PRIVATE
@@ -21,3 +30,12 @@ add_test(NAME dwarf_scope_filter
           ${TEST_OUTPUT_DIR})
 set_tests_properties(dwarf_scope_filter PROPERTIES
   ENVIRONMENT "PYTHONUNBUFFERED=1")
+
+set(RUST_FILTER_OUTPUT_DIR "${CMAKE_BINARY_DIR}/tests/rust_filter")
+
+add_test(NAME rust_source_filter
+  COMMAND ${CMAKE_SOURCE_DIR}/tests/run_rust_filter_test.sh
+          $<TARGET_FILE:coz>
+          ${RUST_FILTER_OUTPUT_DIR})
+set_tests_properties(rust_source_filter PROPERTIES
+  ENVIRONMENT "PYTHONUNBUFFERED=1")