Merge "gc: Find roots using libproc on Darwin" into main

src/libstore/local.mk

@@ -7,6 +7,8 @@ libstore_DIR := $(d)
 libstore_SOURCES := $(wildcard $(d)/*.cc $(d)/builtins/*.cc $(d)/build/*.cc)
 ifdef HOST_LINUX
 libstore_SOURCES += $(d)/platform/linux.cc
+else ifdef HOST_DARWIN
+libstore_SOURCES += $(d)/platform/darwin.cc
 else
 libstore_SOURCES += $(d)/platform/fallback.cc
 endif

src/libstore/meson.build

@@ -162,6 +162,9 @@ libstore_headers = files(
 if host_machine.system() == 'linux'
   libstore_sources += files('platform/linux.cc')
   libstore_headers += files('platform/linux.hh')
+elif host_machine.system() == 'darwin'
+  libstore_sources += files('platform/darwin.cc')
+  libstore_headers += files('platform/darwin.hh')
 else
   libstore_sources += files('platform/fallback.cc')
   libstore_headers += files('platform/fallback.hh')

src/libstore/platform.cc

@@ -2,6 +2,8 @@
 
 #if __linux__
 #include "platform/linux.hh"
+#elif __APPLE__
+#include "platform/darwin.hh"
 #else
 #include "platform/fallback.hh"
 #endif
@@ -11,6 +13,8 @@ std::shared_ptr<LocalStore> LocalStore::makeLocalStore(const Params & params)
 {
 #if __linux__
     return std::shared_ptr<LocalStore>(new LinuxLocalStore(params));
+#elif __APPLE__
+    return std::shared_ptr<LocalStore>(new DarwinLocalStore(params));
 #else
     return std::shared_ptr<LocalStore>(new FallbackLocalStore(params));
 #endif

src/libstore/platform/darwin.cc

@@ -0,0 +1,223 @@
+#include "gc-store.hh"
+#include "signals.hh"
+#include "platform/darwin.hh"
+#include "regex.hh"
+
+#include <sys/proc_info.h>
+#include <sys/sysctl.h>
+#include <libproc.h>
+
+#include <regex>
+
+namespace nix {
+
+void DarwinLocalStore::findPlatformRoots(UncheckedRoots & unchecked)
+{
+    auto storePathRegex = regex::storePathRegex(storeDir);
+
+    std::vector<int> pids;
+    int pidBufSize = 1;
+
+    while (pidBufSize > pids.size() * sizeof(int)) {
+        // Reserve some extra size so we don't fail too much
+        pids.resize((pidBufSize + pidBufSize / 8) / sizeof(int));
+        pidBufSize = proc_listpids(PROC_ALL_PIDS, 0, pids.data(), pids.size() * sizeof(int));
+
+        if (pidBufSize <= 0) {
+            throw SysError("Listing PIDs");
+        }
+    }
+
+    pids.resize(pidBufSize / sizeof(int));
+
+    for (auto pid : pids) {
+        // It doesn't make sense to ask about the kernel
+        if (pid == 0) {
+            continue;
+        }
+
+        try {
+            // Process cwd/root directory
+            struct proc_vnodepathinfo vnodeInfo;
+            if (proc_pidinfo(pid, PROC_PIDVNODEPATHINFO, 0, &vnodeInfo, sizeof(vnodeInfo)) <= 0) {
+                throw SysError("Getting pid %1% working directory", pid);
+            }
+
+            unchecked[std::string(vnodeInfo.pvi_cdir.vip_path)].emplace(fmt("{libproc/%d/cwd}", pid)
+            );
+            unchecked[std::string(vnodeInfo.pvi_rdir.vip_path)].emplace(
+                fmt("{libproc/%d/rootdir}", pid)
+            );
+
+            // File descriptors
+            std::vector<struct proc_fdinfo> fds;
+            int fdBufSize = 1;
+            while (fdBufSize > fds.size() * sizeof(struct proc_fdinfo)) {
+                // Reserve some extra size so we don't fail too much
+                fds.resize((fdBufSize + fdBufSize / 8) / sizeof(struct proc_fdinfo));
+                fdBufSize = proc_pidinfo(
+                    pid, PROC_PIDLISTFDS, 0, fds.data(), fds.size() * sizeof(struct proc_fdinfo)
+                );
+
+                if (fdBufSize <= 0) {
+                    throw SysError("Listing pid %1% file descriptors", pid);
+                }
+            }
+            fds.resize(fdBufSize / sizeof(struct proc_fdinfo));
+
+            for (auto fd : fds) {
+                // By definition, only a vnode is on the filesystem
+                if (fd.proc_fdtype != PROX_FDTYPE_VNODE) {
+                    continue;
+                }
+
+                struct vnode_fdinfowithpath fdInfo;
+                if (proc_pidfdinfo(
+                        pid, fd.proc_fd, PROC_PIDFDVNODEPATHINFO, &fdInfo, sizeof(fdInfo)
+                    )
+                    <= 0)
+                {
+                    // They probably just closed this fd, no need to cancel looking at ranges and
+                    // arguments
+                    if (errno == EBADF) {
+                        continue;
+                    }
+                    throw SysError("Getting pid %1% fd %2% path", pid, fd.proc_fd);
+                }
+
+                unchecked[std::string(fdInfo.pvip.vip_path)].emplace(
+                    fmt("{libproc/%d/fd/%d}", pid, fd.proc_fd)
+                );
+            }
+
+            // Regions (e.g. mmapped files, executables, shared libraries)
+            uint64_t nextAddr = 0;
+            while (true) {
+                // Seriously, what are you doing XNU?
+                // There's 3 flavors of PROC_PIDREGIONPATHINFO:
+                // * PROC_PIDREGIONPATHINFO includes all regions
+                // * PROC_PIDREGIONPATHINFO2 includes regions backed by a vnode
+                // * PROC_PIDREGIONPATHINFO3 includes regions backed by a vnode on a specified
+                // filesystem Only PROC_PIDREGIONPATHINFO is documented. Unfortunately, using it
+                // would make finding gcroots take about 100x as long and tests would fail from
+                // timeout. According to the Frida source code, PROC_PIDREGIONPATHINFO2 has been
+                // available since XNU 2782.1.97 in OS X 10.10
+                //
+                // 22 means PROC_PIDREGIONPATHINFO2
+                struct proc_regionwithpathinfo regionInfo;
+                if (proc_pidinfo(pid, 22, nextAddr, &regionInfo, sizeof(regionInfo)) <= 0) {
+                    // PROC_PIDREGIONPATHINFO signals we're done with an error,
+                    // so we're expected to hit this once per process
+                    if (errno == ESRCH || errno == EINVAL) {
+                        break;
+                    }
+                    throw SysError("Getting pid %1% region path", pid);
+                }
+
+                unchecked[std::string(regionInfo.prp_vip.vip_path)].emplace(
+                    fmt("{libproc/%d/region}", pid)
+                );
+
+                nextAddr = regionInfo.prp_prinfo.pri_address + regionInfo.prp_prinfo.pri_size;
+            }
+
+            // Arguments and environment variables
+            // We can't read environment variables of binaries with entitlements unless
+            // nix has the `com.apple.private.read-environment-variables` entitlement or SIP is off
+            // We can read arguments for all applications though.
+
+            // Yes, it's a sysctl, the proc_info and sysctl APIs are mostly similar,
+            // but both have exclusive capabilities
+            int sysctlName[3] = {CTL_KERN, KERN_PROCARGS2, pid};
+            size_t argsSize = 0;
+            if (sysctl(sysctlName, 3, nullptr, &argsSize, nullptr, 0) < 0) {
+                throw SysError("Reading pid %1% arguments", pid);
+            }
+
+            std::vector<char> args(argsSize);
+            if (sysctl(sysctlName, 3, args.data(), &argsSize, nullptr, 0) < 0) {
+                throw SysError("Reading pid %1% arguments", pid);
+            }
+
+            if (argsSize < args.size()) {
+                args.resize(argsSize);
+            }
+
+            // We have these perfectly nice arguments, but have to ignore them because
+            // otherwise we'd see arguments to nix-store commands and
+            // `nix-store --delete /nix/store/whatever` would always fail
+            // First 4 bytes are an int of argc.
+            if (args.size() < sizeof(int)) {
+                continue;
+            }
+            auto argc = reinterpret_cast<int *>(args.data())[0];
+
+            auto argsIter = args.begin();
+            std::advance(argsIter, sizeof(int));
+            // Executable then argc args, each separated by some number of null bytes
+            for (int i = 0; argsIter != args.end() && i < argc + 1; i++) {
+                argsIter = std::find(argsIter, args.end(), '\0');
+                argsIter = std::find_if(argsIter, args.end(), [](char ch) { return ch != '\0'; });
+            }
+
+            if (argsIter != args.end()) {
+                auto env_end = std::sregex_iterator{};
+                for (auto i = std::sregex_iterator{argsIter, args.end(), storePathRegex};
+                     i != env_end;
+                     ++i)
+                {
+                    unchecked[i->str()].emplace(fmt("{libproc/%d/environ}", pid));
+                }
+            };
+
+            // Per-thread working directories
+            struct proc_taskallinfo taskAllInfo;
+            if (proc_pidinfo(pid, PROC_PIDTASKALLINFO, 0, &taskAllInfo, sizeof(taskAllInfo)) <= 0) {
+                throw SysError("Reading pid %1% tasks", pid);
+            }
+
+            // If the process doesn't have the per-thread cwd flag then we already have the
+            // process-wide cwd from PROC_PIDVNODEPATHINFO
+            if (taskAllInfo.pbsd.pbi_flags & PROC_FLAG_THCWD) {
+                std::vector<uint64_t> tids(taskAllInfo.ptinfo.pti_threadnum);
+                int tidBufSize = proc_pidinfo(
+                    pid, PROC_PIDLISTTHREADS, 0, tids.data(), tids.size() * sizeof(uint64_t)
+                );
+                if (tidBufSize <= 0) {
+                    throw SysError("Listing pid %1% threads", pid);
+                }
+
+                for (auto tid : tids) {
+                    struct proc_threadwithpathinfo threadPathInfo;
+                    if (proc_pidinfo(
+                            pid,
+                            PROC_PIDTHREADPATHINFO,
+                            tid,
+                            &threadPathInfo,
+                            sizeof(threadPathInfo)
+                        )
+                        <= 0)
+                    {
+                        throw SysError("Reading pid %1% thread %2% cwd", pid, tid);
+                    }
+
+                    unchecked[std::string(threadPathInfo.pvip.vip_path)].emplace(
+                        fmt("{libproc/%d/thread/%d/cwd}", pid, tid)
+                    );
+                }
+            }
+        } catch (SysError & e) {
+            // ENOENT/ESRCH: Process no longer exists (proc_info)
+            // EINVAL: Process no longer exists (sysctl)
+            // EACCESS/EPERM: We don't have permission to read this field (proc_info)
+            // EIO: Kernel failed to read from target process memory during KERN_PROCARGS2 (sysctl)
+            if (errno == ENOENT || errno == ESRCH || errno == EINVAL || errno == EACCES
+                || errno == EPERM || errno == EIO)
+            {
+                continue;
+            }
+            throw;
+        }
+    }
+}
+}

src/libstore/platform/darwin.hh

@@ -0,0 +1,35 @@
+#pragma once
+///@file
+
+#include "gc-store.hh"
+#include "local-store.hh"
+
+namespace nix {
+
+/**
+ * Darwin-specific implementation of LocalStore
+ */
+class DarwinLocalStore : public LocalStore
+{
+public:
+    DarwinLocalStore(const Params & params)
+        : StoreConfig(params)
+        , LocalFSStoreConfig(params)
+        , LocalStoreConfig(params)
+        , Store(params)
+        , LocalFSStore(params)
+        , LocalStore(params)
+    {
+    }
+    DarwinLocalStore(const std::string scheme, std::string path, const Params & params)
+        : DarwinLocalStore(params)
+    {
+        throw UnimplementedError("DarwinLocalStore");
+    }
+
+private:
+
+    void findPlatformRoots(UncheckedRoots & unchecked) override;
+};
+
+}

src/libstore/platform/linux.cc

@@ -1,6 +1,7 @@
 #include "gc-store.hh"
 #include "signals.hh"
 #include "platform/linux.hh"
+#include "regex.hh"
 
 #include <regex>
 
@@ -26,12 +27,6 @@ static void readProcLink(const std::string & file, UncheckedRoots & roots)
     }
 }
 
-static std::string quoteRegexChars(const std::string & raw)
-{
-    static auto specialRegex = std::regex(R"([.^$\\*+?()\[\]{}|])");
-    return std::regex_replace(raw, specialRegex, R"(\$&)");
-}
-
 static void readFileRoots(const char * path, UncheckedRoots & roots)
 {
     try {
@@ -50,8 +45,7 @@ void LinuxLocalStore::findPlatformRoots(UncheckedRoots & unchecked)
         struct dirent * ent;
         auto digitsRegex = std::regex(R"(^\d+$)");
         auto mapRegex = std::regex(R"(^\s*\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+(/\S+)\s*$)");
-        auto storePathRegex =
-            std::regex(quoteRegexChars(storeDir) + R"(/[0-9a-z]+[0-9a-zA-Z\+\-\._\?=]*)");
+        auto storePathRegex = regex::storePathRegex(storeDir);
         while (errno = 0, ent = readdir(procDir.get())) {
             checkInterrupt();
             if (std::regex_match(ent->d_name, digitsRegex)) {

src/libutil/meson.build

@@ -22,6 +22,7 @@ libutil_sources = files(
   'position.cc',
   'print-elided.cc',
   'references.cc',
+  'regex.cc',
   'serialise.cc',
   'shlex.cc',
   'signals.cc',
@@ -77,6 +78,7 @@ libutil_headers = files(
   'ref.hh',
   'references.hh',
   'regex-combinators.hh',
+  'regex.hh',
   'repair-flag.hh',
   'serialise.hh',
   'shlex.hh',

src/libutil/regex.cc

@@ -0,0 +1,16 @@
+#include <string>
+#include <regex>
+
+namespace nix::regex {
+std::string quoteRegexChars(const std::string & raw)
+{
+    static auto specialRegex = std::regex(R"([.^$\\*+?()\[\]{}|])");
+    return std::regex_replace(raw, specialRegex, R"(\$&)");
+}
+
+std::regex storePathRegex(const std::string & storeDir)
+{
+    return std::regex(quoteRegexChars(storeDir) + R"(/[0-9a-z]+[0-9a-zA-Z\+\-\._\?=]*)");
+}
+
+}

src/libutil/regex.hh

@@ -0,0 +1,11 @@
+#pragma once
+///@file
+
+#include <string>
+#include <regex>
+
+namespace nix::regex {
+std::string quoteRegexChars(const std::string & raw);
+
+std::regex storePathRegex(const std::string & storeDir);
+}

tests/functional/common/vars-and-functions.sh.in

@@ -24,7 +24,6 @@ if [[ -n $NIX_STORE ]]; then
     export _NIX_TEST_NO_SANDBOX=1
 fi
 export _NIX_IN_TEST=$TEST_ROOT/shared
-export _NIX_TEST_NO_LSOF=1
 export NIX_REMOTE=${NIX_REMOTE_-}
 unset NIX_PATH
 export TEST_HOME=$TEST_ROOT/test-home

tests/functional/gc-runtime.nix

@@ -1,17 +1,29 @@
 with import ./config.nix;
 
-mkDerivation {
-  name = "gc-runtime";
-  builder =
-    # Test inline source file definitions.
-    builtins.toFile "builder.sh" ''
-      mkdir $out
+{
+  environ = mkDerivation {
+    name = "gc-runtime-environ";
+    buildCommand = "mkdir $out; echo environ > $out/environ";
+  };
 
-      cat > $out/program <<EOF
-      #! ${shell}
-      sleep 10000
-      EOF
+  open = mkDerivation {
+    name = "gc-runtime-open";
+    buildCommand = "mkdir $out; echo open > $out/open";
+  };
 
-      chmod +x $out/program
-    '';
+  program = mkDerivation {
+    name = "gc-runtime-program";
+    builder =
+      # Test inline source file definitions.
+      builtins.toFile "builder.sh" ''
+        mkdir $out
+
+        cat > $out/program <<EOF
+        #! ${shell}
+        sleep 10000 < \$1
+        EOF
+
+        chmod +x $out/program
+      '';
+  };
 }

tests/functional/gc-runtime.sh

@@ -1,38 +1,44 @@
 source common.sh
 
-case $system in
-    *linux*)
-        ;;
-    *)
-        skipTest "Not running Linux";
-esac
-
 set -m # enable job control, needed for kill
 
 profiles="$NIX_STATE_DIR"/profiles
 rm -rf $profiles
 
-nix-env -p $profiles/test -f ./gc-runtime.nix -i gc-runtime
+nix-env -p $profiles/test -f ./gc-runtime.nix -i gc-runtime-{program,environ,open}
 
-outPath=$(nix-env -p $profiles/test -q --no-name --out-path gc-runtime)
-echo $outPath
+programPath=$(nix-env -p $profiles/test -q --no-name --out-path gc-runtime-program)
+environPath=$(nix-env -p $profiles/test -q --no-name --out-path gc-runtime-environ)
+openPath=$(nix-env -p $profiles/test -q --no-name --out-path gc-runtime-open)
+echo $programPath $environPath $openPath
 
 echo "backgrounding program..."
-$profiles/test/program &
+export environPath
+$profiles/test/program $openPath/open &
 sleep 2 # hack - wait for the program to get started
 child=$!
 echo PID=$child
 
-nix-env -p $profiles/test -e gc-runtime
+nix-env -p $profiles/test -e gc-runtime-{program,environ,open}
 nix-env -p $profiles/test --delete-generations old
 
 nix-store --gc
 
 kill -- -$child
 
-if ! test -e $outPath; then
+if ! test -e $programPath; then
     echo "running program was garbage collected!"
     exit 1
 fi
 
+if ! test -e $environPath; then
+    echo "file in environment variable was garbage collected!"
+    exit 1
+fi
+
+if ! test -e $openPath; then
+    echo "opened file was garbage collected!"
+    exit 1
+fi
+
 exit 0