Commit 879acba5 authored by thestig@chromium.org's avatar thestig@chromium.org

Overhaul base/process_util_linux.cc.

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/9584024

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@124839 0039d316-1c4b-4281-b951-d872f2087c98
parent 8f73f942
......@@ -4,15 +4,10 @@
#include "base/process_util.h"
#include <ctype.h>
#include <dirent.h>
#include <dlfcn.h>
#include <errno.h>
#include <fcntl.h>
#include <malloc.h>
#include <sys/time.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <time.h>
#include <unistd.h>
#include "base/file_util.h"
......@@ -32,26 +27,125 @@ enum ParsingState {
};
const char kProcDir[] = "/proc";
const char kStatFile[] = "stat";
// Returns a FilePath to "/proc/pid".
FilePath GetProcPidDir(pid_t pid) {
return FilePath(kProcDir).Append(base::Int64ToString(pid));
}
// Reads /proc/<pid>/stat and populates |proc_stats| with the values split by
// spaces. Returns true if successful.
bool GetProcStats(pid_t pid, std::vector<std::string>* proc_stats) {
// Fields from /proc/<pid>/stat, 0-based. See man 5 proc.
// If the ordering ever changes, carefully review functions that use these
// values.
enum ProcStatsFields {
VM_COMM = 1, // Filename of executable, without parentheses.
VM_STATE = 2, // Letter indicating the state of the process.
VM_PPID = 3, // PID of the parent.
VM_PGRP = 4, // Process group id.
VM_UTIME = 13, // Time scheduled in user mode in clock ticks.
VM_STIME = 14, // Time scheduled in kernel mode in clock ticks.
VM_VSIZE = 22, // Virtual memory size in bytes.
VM_RSS = 23, // Resident Set Size in pages.
};
// Reads /proc/<pid>/stat into |buffer|. Returns true if successful.
bool ReadProcStats(pid_t pid, std::string* buffer) {
// Synchronously reading files in /proc is safe.
base::ThreadRestrictions::ScopedAllowIO allow_io;
FilePath stat_file = GetProcPidDir(pid).Append("stat");
std::string mem_stats;
if (!file_util::ReadFileToString(stat_file, &mem_stats))
FilePath stat_file = GetProcPidDir(pid).Append(kStatFile);
if (!file_util::ReadFileToString(stat_file, buffer)) {
DLOG(WARNING) << "Failed to get process stats.";
return false;
}
return true;
}
// Takes |stats_data| and populates |proc_stats| with the values split by
// spaces. Taking into account the 2nd field may, in itself, contain spaces.
// Returns true if successful.
bool ParseProcStats(const std::string& stats_data,
std::vector<std::string>* proc_stats) {
// The stat file is formatted as:
// pid (process name) data1 data2 .... dataN
// Look for the closing paren by scanning backwards, to avoid being fooled by
// processes with ')' in the name.
size_t open_parens_idx = stats_data.find(" (");
size_t close_parens_idx = stats_data.rfind(") ");
if (open_parens_idx == std::string::npos ||
close_parens_idx == std::string::npos ||
open_parens_idx > close_parens_idx) {
NOTREACHED();
return false;
base::SplitString(mem_stats, ' ', proc_stats);
}
open_parens_idx++;
proc_stats->clear();
// PID.
proc_stats->push_back(stats_data.substr(0, open_parens_idx));
// Process name without parentheses.
proc_stats->push_back(
stats_data.substr(open_parens_idx + 1,
close_parens_idx - (open_parens_idx + 1)));
// Split the rest.
std::vector<std::string> other_stats;
base::SplitString(stats_data.substr(close_parens_idx + 2), ' ', &other_stats);
for (size_t i = 0; i < other_stats.size(); ++i)
proc_stats->push_back(other_stats[i]);
return true;
}
// Reads the |field_num|th field from |proc_stats|. Returns 0 on failure.
// This version does not handle the first 3 values, since the first value is
// simply |pid|, and the next two values are strings.
size_t GetProcStatsFieldAsInt(const std::vector<std::string>& proc_stats,
ProcStatsFields field_num) {
if (field_num < VM_PPID) {
NOTREACHED();
return 0;
}
if (proc_stats.size() > static_cast<size_t>(field_num)) {
size_t value;
if (base::StringToUint64(proc_stats[field_num], &value))
return value;
}
NOTREACHED();
return 0;
}
// Convenience wrapper around GetProcStatsFieldAsInt(), ParseProcStats() and
// ReadProcStats(). See GetProcStatsFieldAsInt() for details.
size_t ReadProcStatsAndGetFieldAsInt(pid_t pid, ProcStatsFields field_num) {
std::string stats_data;
if (!ReadProcStats(pid, &stats_data))
return 0;
std::vector<std::string> proc_stats;
if (!ParseProcStats(stats_data, &proc_stats))
return 0;
return GetProcStatsFieldAsInt(proc_stats, field_num);
}
// Reads the |field_num|th field from |proc_stats|.
// Returns an empty string on failure.
// This version only handles VM_COMM and VM_STATE, which are the only fields
// that are strings.
std::string GetProcStatsFieldAsString(
const std::vector<std::string>& proc_stats,
ProcStatsFields field_num) {
if (field_num < VM_COMM || field_num > VM_STATE) {
NOTREACHED();
return "";
}
if (proc_stats.size() > static_cast<size_t>(field_num))
return proc_stats[field_num];
NOTREACHED();
return 0;
}
// Reads /proc/<pid>/cmdline and populates |proc_cmd_line_args| with the command
// line arguments. Returns true if successful.
// Note: /proc/<pid>/cmdline contains command line arguments separated by single
......@@ -71,28 +165,53 @@ bool GetProcCmdline(pid_t pid, std::vector<std::string>* proc_cmd_line_args) {
return true;
}
// Take a /proc directory entry named |d_name|, and if it is the directory for
// a process, convert it to a pid_t.
// Returns 0 on failure.
// e.g. /proc/self/ will return 0, whereas /proc/1234 will return 1234.
pid_t ProcDirSlotToPid(const char* d_name) {
int i;
for (i = 0; i < NAME_MAX && d_name[i]; ++i) {
if (!IsAsciiDigit(d_name[i])) {
return 0;
}
}
if (i == NAME_MAX)
return 0;
// Read the process's command line.
pid_t pid;
std::string pid_string(d_name);
if (!base::StringToInt(pid_string, &pid)) {
NOTREACHED();
return 0;
}
return pid;
}
// Get the total CPU of a single process. Return value is number of jiffies
// on success or -1 on error.
int GetProcessCPU(pid_t pid) {
// Synchronously reading files in /proc is safe.
base::ThreadRestrictions::ScopedAllowIO allow_io;
// Use /proc/<pid>/task to find all threads and parse their /stat file.
FilePath path = GetProcPidDir(pid).Append("task");
FilePath task_path = GetProcPidDir(pid).Append("task");
DIR* dir = opendir(path.value().c_str());
DIR* dir = opendir(task_path.value().c_str());
if (!dir) {
DPLOG(ERROR) << "opendir(" << path.value() << ")";
DPLOG(ERROR) << "opendir(" << task_path.value() << ")";
return -1;
}
int total_cpu = 0;
while (struct dirent* ent = readdir(dir)) {
if (ent->d_name[0] == '.')
pid_t tid = ProcDirSlotToPid(ent->d_name);
if (!tid)
continue;
FilePath stat_path = path.AppendASCII(ent->d_name).AppendASCII("stat");
// Synchronously reading files in /proc is safe.
base::ThreadRestrictions::ScopedAllowIO allow_io;
std::string stat;
FilePath stat_path = task_path.Append(ent->d_name).Append(kStatFile);
if (file_util::ReadFileToString(stat_path, &stat)) {
int cpu = base::ParseProcStatCPU(stat);
if (cpu > 0)
......@@ -104,22 +223,21 @@ int GetProcessCPU(pid_t pid) {
return total_cpu;
}
} // namespace
namespace base {
ProcessId GetParentProcessId(ProcessHandle process) {
// Synchronously reading files in /proc is safe.
base::ThreadRestrictions::ScopedAllowIO allow_io;
FilePath stat_file = GetProcPidDir(process).Append("status");
// Read /proc/<pid>/status and returns the value for |field|, or 0 on failure.
// Only works for fields in the form of "Field: value kB".
size_t ReadProcStatusAndGetFieldAsInt(pid_t pid, const std::string& field) {
FilePath stat_file = GetProcPidDir(pid).Append("status");
std::string status;
if (!file_util::ReadFileToString(stat_file, &status))
return -1;
{
// Synchronously reading files in /proc is safe.
base::ThreadRestrictions::ScopedAllowIO allow_io;
if (!file_util::ReadFileToString(stat_file, &status))
return 0;
}
StringTokenizer tokenizer(status, ":\n");
ParsingState state = KEY_NAME;
StringPiece last_key_name;
base::StringPiece last_key_name;
while (tokenizer.GetNext()) {
switch (state) {
case KEY_NAME:
......@@ -128,16 +246,40 @@ ProcessId GetParentProcessId(ProcessHandle process) {
break;
case KEY_VALUE:
DCHECK(!last_key_name.empty());
if (last_key_name == "PPid") {
int ppid;
base::StringToInt(tokenizer.token_piece(), &ppid);
return ppid;
if (last_key_name == field) {
std::string value_str;
tokenizer.token_piece().CopyToString(&value_str);
std::string value_str_trimmed;
TrimWhitespaceASCII(value_str, TRIM_ALL, &value_str_trimmed);
std::vector<std::string> split_value_str;
base::SplitString(value_str_trimmed, ' ', &split_value_str);
if (split_value_str.size() != 2 || split_value_str[1] != "kB") {
NOTREACHED();
return 0;
}
size_t value;
if (!base::StringToUint64(split_value_str[0], &value)) {
NOTREACHED();
return 0;
}
return value;
}
state = KEY_NAME;
break;
}
}
NOTREACHED();
return 0;
}
} // namespace
namespace base {
ProcessId GetParentProcessId(ProcessHandle process) {
size_t pid = ReadProcStatsAndGetFieldAsInt(process, VM_PPID);
if (pid)
return pid;
return -1;
}
......@@ -166,66 +308,45 @@ ProcessIterator::~ProcessIterator() {
bool ProcessIterator::CheckForNextProcess() {
// TODO(port): skip processes owned by different UID
dirent* slot = 0;
const char* openparen;
const char* closeparen;
pid_t pid = kNullProcessId;
std::vector<std::string> cmd_line_args;
std::string stats_data;
std::vector<std::string> proc_stats;
// Arbitrarily guess that there will never be more than 200 non-process
// files in /proc. Hardy has 53.
// files in /proc. Hardy has 53 and Lucid has 61.
int skipped = 0;
const int kSkipLimit = 200;
while (skipped < kSkipLimit) {
slot = readdir(procfs_dir_);
dirent* slot = readdir(procfs_dir_);
// all done looking through /proc?
if (!slot)
return false;
// If not a process, keep looking for one.
bool notprocess = false;
int i;
for (i = 0; i < NAME_MAX && slot->d_name[i]; ++i) {
if (!isdigit(slot->d_name[i])) {
notprocess = true;
break;
}
}
if (i == NAME_MAX || notprocess) {
pid = ProcDirSlotToPid(slot->d_name);
if (!pid) {
skipped++;
continue;
}
// Read the process's command line.
std::string pid_string(slot->d_name);
int pid;
if (StringToInt(pid_string, &pid) && !GetProcCmdline(pid, &cmd_line_args))
if (!GetProcCmdline(pid, &cmd_line_args))
continue;
// Read the process's status.
char buf[NAME_MAX + 12];
sprintf(buf, "/proc/%s/stat", slot->d_name);
FILE* fp = fopen(buf, "r");
if (!fp)
if (!ReadProcStats(pid, &stats_data))
continue;
const char* result = fgets(buf, sizeof(buf), fp);
fclose(fp);
if (!result)
if (!ParseProcStats(stats_data, &proc_stats))
continue;
// Parse the status. It is formatted like this:
// %d (%s) %c %d %d ...
// pid (name) runstate ppid gid
// To avoid being fooled by names containing a closing paren, scan
// backwards.
openparen = strchr(buf, '(');
closeparen = strrchr(buf, ')');
if (!openparen || !closeparen)
std::string runstate = GetProcStatsFieldAsString(proc_stats, VM_STATE);
if (runstate.size() != 1) {
NOTREACHED();
continue;
char runstate = closeparen[2];
}
// Is the process in 'Zombie' state, i.e. dead but waiting to be reaped?
// Allowed values: D R S T Z
if (runstate != 'Z')
if (runstate[0] != 'Z')
break;
// Nope, it's a zombie; somebody isn't cleaning up after their children.
......@@ -237,17 +358,14 @@ bool ProcessIterator::CheckForNextProcess() {
return false;
}
// This seems fragile.
entry_.pid_ = atoi(slot->d_name);
entry_.ppid_ = atoi(closeparen + 3);
entry_.gid_ = atoi(strchr(closeparen + 4, ' '));
entry_.pid_ = pid;
entry_.ppid_ = GetProcStatsFieldAsInt(proc_stats, VM_PPID);
entry_.gid_ = GetProcStatsFieldAsInt(proc_stats, VM_PGRP);
entry_.cmd_line_args_.assign(cmd_line_args.begin(), cmd_line_args.end());
// TODO(port): read pid's commandline's $0, like killall does. Using the
// short name between openparen and closeparen won't work for long names!
int len = closeparen - openparen - 1;
entry_.exe_file_.assign(openparen + 1, len);
entry_.exe_file_ = GetProcStatsFieldAsString(proc_stats, VM_COMM);
return true;
}
......@@ -265,58 +383,22 @@ ProcessMetrics* ProcessMetrics::CreateProcessMetrics(ProcessHandle process) {
// On linux, we return vsize.
size_t ProcessMetrics::GetPagefileUsage() const {
std::vector<std::string> proc_stats;
if (!GetProcStats(process_, &proc_stats))
DLOG(WARNING) << "Failed to get process stats.";
const size_t kVmSize = 22;
if (proc_stats.size() > kVmSize) {
int vm_size;
base::StringToInt(proc_stats[kVmSize], &vm_size);
return static_cast<size_t>(vm_size);
}
return 0;
return ReadProcStatsAndGetFieldAsInt(process_, VM_VSIZE);
}
// On linux, we return the high water mark of vsize.
size_t ProcessMetrics::GetPeakPagefileUsage() const {
std::vector<std::string> proc_stats;
if (!GetProcStats(process_, &proc_stats))
DLOG(WARNING) << "Failed to get process stats.";
const size_t kVmPeak = 21;
if (proc_stats.size() > kVmPeak) {
int vm_peak;
if (base::StringToInt(proc_stats[kVmPeak], &vm_peak))
return vm_peak;
}
return 0;
return ReadProcStatusAndGetFieldAsInt(process_, "VmPeak") * 1024;
}
// On linux, we return RSS.
size_t ProcessMetrics::GetWorkingSetSize() const {
std::vector<std::string> proc_stats;
if (!GetProcStats(process_, &proc_stats))
DLOG(WARNING) << "Failed to get process stats.";
const size_t kVmRss = 23;
if (proc_stats.size() > kVmRss) {
int num_pages;
if (base::StringToInt(proc_stats[kVmRss], &num_pages))
return static_cast<size_t>(num_pages) * getpagesize();
}
return 0;
return ReadProcStatsAndGetFieldAsInt(process_, VM_RSS) * getpagesize();
}
// On linux, we return the high water mark of RSS.
size_t ProcessMetrics::GetPeakWorkingSetSize() const {
std::vector<std::string> proc_stats;
if (!GetProcStats(process_, &proc_stats))
DLOG(WARNING) << "Failed to get process stats.";
const size_t kVmHwm = 23;
if (proc_stats.size() > kVmHwm) {
int num_pages;
base::StringToInt(proc_stats[kVmHwm], &num_pages);
return static_cast<size_t>(num_pages) * getpagesize();
}
return 0;
return ReadProcStatusAndGetFieldAsInt(process_, "VmHWM") * 1024;
}
bool ProcessMetrics::GetMemoryBytes(size_t* private_bytes,
......@@ -326,7 +408,7 @@ bool ProcessMetrics::GetMemoryBytes(size_t* private_bytes,
return false;
if (private_bytes)
*private_bytes = ws_usage.priv << 10;
*private_bytes = ws_usage.priv * 1024;
if (shared_bytes)
*shared_bytes = ws_usage.shared * 1024;
......@@ -342,7 +424,7 @@ bool ProcessMetrics::GetWorkingSetKBytes(WorkingSetKBytes* ws_usage) const {
// First we need to get the page size, since everything is measured in pages.
// For details, see: man 5 proc.
const int page_size_kb = sysconf(_SC_PAGE_SIZE) / 1024;
const int page_size_kb = getpagesize() / 1024;
if (page_size_kb <= 0)
return false;
......@@ -473,24 +555,15 @@ ProcessMetrics::ProcessMetrics(ProcessHandle process)
// Exposed for testing.
int ParseProcStatCPU(const std::string& input) {
// /proc/<pid>/stat contains the process name in parens. In case the
// process name itself contains parens, skip past them.
std::string::size_type rparen = input.rfind(')');
if (rparen == std::string::npos)
std::vector<std::string> proc_stats;
if (!ParseProcStats(input, &proc_stats))
return -1;
// From here, we expect a bunch of space-separated fields, where the
// 0-indexed 11th and 12th are utime and stime. On two different machines
// I found 42 and 39 fields, so let's just expect the ones we need.
std::vector<std::string> fields;
base::SplitString(input.substr(rparen + 2), ' ', &fields);
if (fields.size() < 13)
return -1; // Output not in the format we expect.
int fields11, fields12;
base::StringToInt(fields[11], &fields11);
base::StringToInt(fields[12], &fields12);
return fields11 + fields12;
if (proc_stats.size() <= VM_STIME)
return -1;
size_t utime = GetProcStatsFieldAsInt(proc_stats, VM_UTIME);
size_t stime = GetProcStatsFieldAsInt(proc_stats, VM_STIME);
return utime + stime;
}
namespace {
......@@ -519,15 +592,15 @@ bool GetSystemMemoryInfo(SystemMemoryInfoKB* meminfo) {
FilePath meminfo_file("/proc/meminfo");
std::string meminfo_data;
if (!file_util::ReadFileToString(meminfo_file, &meminfo_data)) {
DLOG(WARNING) << "Failed to open /proc/meminfo.";
DLOG(WARNING) << "Failed to open " << meminfo_file.value();
return false;
}
std::vector<std::string> meminfo_fields;
SplitStringAlongWhitespace(meminfo_data, &meminfo_fields);
if (meminfo_fields.size() < kMemCachedIndex) {
DLOG(WARNING) << "Failed to parse /proc/meminfo. Only found " <<
meminfo_fields.size() << " fields.";
DLOG(WARNING) << "Failed to parse " << meminfo_file.value()
<< ". Only found " << meminfo_fields.size() << " fields.";
return false;
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment