1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 04:02:41 +01:00
llvm-mirror/lib/Support/Unix/Program.inc
Oleg Ranevskyy 2d3f6608fd [Clang/Support/Windows/Unix] Command lines created by clang may exceed the command length limit set by the OS
Summary:
Hi Rafael,

Would you be able to review this patch, please?

(Clang part of the patch is D15832).

When clang runs an external tool, e.g. a linker, it may create a command line that exceeds the length limit.

Clang uses the llvm::sys::argumentsFitWithinSystemLimits function to check if command line length fits the OS 

limitation. There are two problems in this function that may cause exceeding of the limit:

1. It ignores the length of the program path in its calculations. On the other hand, clang adds the program 

path to the command line when it runs the program.

2. It assumes no space character is inserted after the last argument, which is not true for Windows. The flattenArgs function adds the trailing space for *each* argument. The result of this is that the terminating NULL character is not counted and may be placed beyond the length limit if the command line is exactly 32768 characters long. The WinAPI's CreateProcess does not find the NULL character and fails.

Reviewers: rafael, ygao, probinson

Subscribers: asl, llvm-commits

Differential Revision: http://reviews.llvm.org/D15831

llvm-svn: 256866
2016-01-05 19:56:12 +00:00

470 lines
14 KiB
C++

//===- llvm/Support/Unix/Program.cpp -----------------------------*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements the Unix specific portion of the Program class.
//
//===----------------------------------------------------------------------===//
//===----------------------------------------------------------------------===//
//=== WARNING: Implementation here must contain only generic UNIX code that
//=== is guaranteed to work on *all* UNIX variants.
//===----------------------------------------------------------------------===//
#include "Unix.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/Config/config.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/FileSystem.h"
#include "llvm/Support/Path.h"
#include "llvm/Support/raw_ostream.h"
#if HAVE_SYS_STAT_H
#include <sys/stat.h>
#endif
#if HAVE_SYS_RESOURCE_H
#include <sys/resource.h>
#endif
#if HAVE_SIGNAL_H
#include <signal.h>
#endif
#if HAVE_FCNTL_H
#include <fcntl.h>
#endif
#if HAVE_UNISTD_H
#include <unistd.h>
#endif
#ifdef HAVE_POSIX_SPAWN
#ifdef __sun__
#define _RESTRICT_KYWD
#endif
#include <spawn.h>
#if defined(__APPLE__)
#include <TargetConditionals.h>
#endif
#if defined(__APPLE__) && !(defined(TARGET_OS_IPHONE) && TARGET_OS_IPHONE)
#define USE_NSGETENVIRON 1
#else
#define USE_NSGETENVIRON 0
#endif
#if !USE_NSGETENVIRON
extern char **environ;
#else
#include <crt_externs.h> // _NSGetEnviron
#endif
#endif
namespace llvm {
using namespace sys;
ProcessInfo::ProcessInfo() : Pid(0), ReturnCode(0) {}
ErrorOr<std::string> sys::findProgramByName(StringRef Name,
ArrayRef<StringRef> Paths) {
assert(!Name.empty() && "Must have a name!");
// Use the given path verbatim if it contains any slashes; this matches
// the behavior of sh(1) and friends.
if (Name.find('/') != StringRef::npos)
return std::string(Name);
SmallVector<StringRef, 16> EnvironmentPaths;
if (Paths.empty())
if (const char *PathEnv = std::getenv("PATH")) {
SplitString(PathEnv, EnvironmentPaths, ":");
Paths = EnvironmentPaths;
}
for (auto Path : Paths) {
if (Path.empty())
continue;
// Check to see if this first directory contains the executable...
SmallString<128> FilePath(Path);
sys::path::append(FilePath, Name);
if (sys::fs::can_execute(FilePath.c_str()))
return std::string(FilePath.str()); // Found the executable!
}
return errc::no_such_file_or_directory;
}
static bool RedirectIO(const StringRef *Path, int FD, std::string* ErrMsg) {
if (!Path) // Noop
return false;
std::string File;
if (Path->empty())
// Redirect empty paths to /dev/null
File = "/dev/null";
else
File = *Path;
// Open the file
int InFD = open(File.c_str(), FD == 0 ? O_RDONLY : O_WRONLY|O_CREAT, 0666);
if (InFD == -1) {
MakeErrMsg(ErrMsg, "Cannot open file '" + File + "' for "
+ (FD == 0 ? "input" : "output"));
return true;
}
// Install it as the requested FD
if (dup2(InFD, FD) == -1) {
MakeErrMsg(ErrMsg, "Cannot dup2");
close(InFD);
return true;
}
close(InFD); // Close the original FD
return false;
}
#ifdef HAVE_POSIX_SPAWN
static bool RedirectIO_PS(const std::string *Path, int FD, std::string *ErrMsg,
posix_spawn_file_actions_t *FileActions) {
if (!Path) // Noop
return false;
const char *File;
if (Path->empty())
// Redirect empty paths to /dev/null
File = "/dev/null";
else
File = Path->c_str();
if (int Err = posix_spawn_file_actions_addopen(
FileActions, FD, File,
FD == 0 ? O_RDONLY : O_WRONLY | O_CREAT, 0666))
return MakeErrMsg(ErrMsg, "Cannot dup2", Err);
return false;
}
#endif
static void TimeOutHandler(int Sig) {
}
static void SetMemoryLimits (unsigned size)
{
#if HAVE_SYS_RESOURCE_H && HAVE_GETRLIMIT && HAVE_SETRLIMIT
struct rlimit r;
__typeof__ (r.rlim_cur) limit = (__typeof__ (r.rlim_cur)) (size) * 1048576;
// Heap size
getrlimit (RLIMIT_DATA, &r);
r.rlim_cur = limit;
setrlimit (RLIMIT_DATA, &r);
#ifdef RLIMIT_RSS
// Resident set size.
getrlimit (RLIMIT_RSS, &r);
r.rlim_cur = limit;
setrlimit (RLIMIT_RSS, &r);
#endif
#ifdef RLIMIT_AS // e.g. NetBSD doesn't have it.
// Don't set virtual memory limit if built with any Sanitizer. They need 80Tb
// of virtual memory for shadow memory mapping.
#if !LLVM_MEMORY_SANITIZER_BUILD && !LLVM_ADDRESS_SANITIZER_BUILD
// Virtual memory.
getrlimit (RLIMIT_AS, &r);
r.rlim_cur = limit;
setrlimit (RLIMIT_AS, &r);
#endif
#endif
#endif
}
}
static bool Execute(ProcessInfo &PI, StringRef Program, const char **args,
const char **envp, const StringRef **redirects,
unsigned memoryLimit, std::string *ErrMsg) {
if (!llvm::sys::fs::exists(Program)) {
if (ErrMsg)
*ErrMsg = std::string("Executable \"") + Program.str() +
std::string("\" doesn't exist!");
return false;
}
// If this OS has posix_spawn and there is no memory limit being implied, use
// posix_spawn. It is more efficient than fork/exec.
#ifdef HAVE_POSIX_SPAWN
if (memoryLimit == 0) {
posix_spawn_file_actions_t FileActionsStore;
posix_spawn_file_actions_t *FileActions = nullptr;
// If we call posix_spawn_file_actions_addopen we have to make sure the
// c strings we pass to it stay alive until the call to posix_spawn,
// so we copy any StringRefs into this variable.
std::string RedirectsStorage[3];
if (redirects) {
std::string *RedirectsStr[3] = {nullptr, nullptr, nullptr};
for (int I = 0; I < 3; ++I) {
if (redirects[I]) {
RedirectsStorage[I] = *redirects[I];
RedirectsStr[I] = &RedirectsStorage[I];
}
}
FileActions = &FileActionsStore;
posix_spawn_file_actions_init(FileActions);
// Redirect stdin/stdout.
if (RedirectIO_PS(RedirectsStr[0], 0, ErrMsg, FileActions) ||
RedirectIO_PS(RedirectsStr[1], 1, ErrMsg, FileActions))
return false;
if (redirects[1] == nullptr || redirects[2] == nullptr ||
*redirects[1] != *redirects[2]) {
// Just redirect stderr
if (RedirectIO_PS(RedirectsStr[2], 2, ErrMsg, FileActions))
return false;
} else {
// If stdout and stderr should go to the same place, redirect stderr
// to the FD already open for stdout.
if (int Err = posix_spawn_file_actions_adddup2(FileActions, 1, 2))
return !MakeErrMsg(ErrMsg, "Can't redirect stderr to stdout", Err);
}
}
if (!envp)
#if !USE_NSGETENVIRON
envp = const_cast<const char **>(environ);
#else
// environ is missing in dylibs.
envp = const_cast<const char **>(*_NSGetEnviron());
#endif
// Explicitly initialized to prevent what appears to be a valgrind false
// positive.
pid_t PID = 0;
int Err = posix_spawn(&PID, Program.str().c_str(), FileActions,
/*attrp*/nullptr, const_cast<char **>(args),
const_cast<char **>(envp));
if (FileActions)
posix_spawn_file_actions_destroy(FileActions);
if (Err)
return !MakeErrMsg(ErrMsg, "posix_spawn failed", Err);
PI.Pid = PID;
return true;
}
#endif
// Create a child process.
int child = fork();
switch (child) {
// An error occurred: Return to the caller.
case -1:
MakeErrMsg(ErrMsg, "Couldn't fork");
return false;
// Child process: Execute the program.
case 0: {
// Redirect file descriptors...
if (redirects) {
// Redirect stdin
if (RedirectIO(redirects[0], 0, ErrMsg)) { return false; }
// Redirect stdout
if (RedirectIO(redirects[1], 1, ErrMsg)) { return false; }
if (redirects[1] && redirects[2] &&
*(redirects[1]) == *(redirects[2])) {
// If stdout and stderr should go to the same place, redirect stderr
// to the FD already open for stdout.
if (-1 == dup2(1,2)) {
MakeErrMsg(ErrMsg, "Can't redirect stderr to stdout");
return false;
}
} else {
// Just redirect stderr
if (RedirectIO(redirects[2], 2, ErrMsg)) { return false; }
}
}
// Set memory limits
if (memoryLimit!=0) {
SetMemoryLimits(memoryLimit);
}
// Execute!
std::string PathStr = Program;
if (envp != nullptr)
execve(PathStr.c_str(),
const_cast<char **>(args),
const_cast<char **>(envp));
else
execv(PathStr.c_str(),
const_cast<char **>(args));
// If the execve() failed, we should exit. Follow Unix protocol and
// return 127 if the executable was not found, and 126 otherwise.
// Use _exit rather than exit so that atexit functions and static
// object destructors cloned from the parent process aren't
// redundantly run, and so that any data buffered in stdio buffers
// cloned from the parent aren't redundantly written out.
_exit(errno == ENOENT ? 127 : 126);
}
// Parent process: Break out of the switch to do our processing.
default:
break;
}
PI.Pid = child;
return true;
}
namespace llvm {
ProcessInfo sys::Wait(const ProcessInfo &PI, unsigned SecondsToWait,
bool WaitUntilTerminates, std::string *ErrMsg) {
struct sigaction Act, Old;
assert(PI.Pid && "invalid pid to wait on, process not started?");
int WaitPidOptions = 0;
pid_t ChildPid = PI.Pid;
if (WaitUntilTerminates) {
SecondsToWait = 0;
} else if (SecondsToWait) {
// Install a timeout handler. The handler itself does nothing, but the
// simple fact of having a handler at all causes the wait below to return
// with EINTR, unlike if we used SIG_IGN.
memset(&Act, 0, sizeof(Act));
Act.sa_handler = TimeOutHandler;
sigemptyset(&Act.sa_mask);
sigaction(SIGALRM, &Act, &Old);
alarm(SecondsToWait);
} else if (SecondsToWait == 0)
WaitPidOptions = WNOHANG;
// Parent process: Wait for the child process to terminate.
int status;
ProcessInfo WaitResult;
do {
WaitResult.Pid = waitpid(ChildPid, &status, WaitPidOptions);
} while (WaitUntilTerminates && WaitResult.Pid == -1 && errno == EINTR);
if (WaitResult.Pid != PI.Pid) {
if (WaitResult.Pid == 0) {
// Non-blocking wait.
return WaitResult;
} else {
if (SecondsToWait && errno == EINTR) {
// Kill the child.
kill(PI.Pid, SIGKILL);
// Turn off the alarm and restore the signal handler
alarm(0);
sigaction(SIGALRM, &Old, nullptr);
// Wait for child to die
if (wait(&status) != ChildPid)
MakeErrMsg(ErrMsg, "Child timed out but wouldn't die");
else
MakeErrMsg(ErrMsg, "Child timed out", 0);
WaitResult.ReturnCode = -2; // Timeout detected
return WaitResult;
} else if (errno != EINTR) {
MakeErrMsg(ErrMsg, "Error waiting for child process");
WaitResult.ReturnCode = -1;
return WaitResult;
}
}
}
// We exited normally without timeout, so turn off the timer.
if (SecondsToWait && !WaitUntilTerminates) {
alarm(0);
sigaction(SIGALRM, &Old, nullptr);
}
// Return the proper exit status. Detect error conditions
// so we can return -1 for them and set ErrMsg informatively.
int result = 0;
if (WIFEXITED(status)) {
result = WEXITSTATUS(status);
WaitResult.ReturnCode = result;
if (result == 127) {
if (ErrMsg)
*ErrMsg = llvm::sys::StrError(ENOENT);
WaitResult.ReturnCode = -1;
return WaitResult;
}
if (result == 126) {
if (ErrMsg)
*ErrMsg = "Program could not be executed";
WaitResult.ReturnCode = -1;
return WaitResult;
}
} else if (WIFSIGNALED(status)) {
if (ErrMsg) {
*ErrMsg = strsignal(WTERMSIG(status));
#ifdef WCOREDUMP
if (WCOREDUMP(status))
*ErrMsg += " (core dumped)";
#endif
}
// Return a special value to indicate that the process received an unhandled
// signal during execution as opposed to failing to execute.
WaitResult.ReturnCode = -2;
}
return WaitResult;
}
std::error_code sys::ChangeStdinToBinary(){
// Do nothing, as Unix doesn't differentiate between text and binary.
return std::error_code();
}
std::error_code sys::ChangeStdoutToBinary(){
// Do nothing, as Unix doesn't differentiate between text and binary.
return std::error_code();
}
std::error_code
llvm::sys::writeFileWithEncoding(StringRef FileName, StringRef Contents,
WindowsEncodingMethod Encoding /*unused*/) {
std::error_code EC;
llvm::raw_fd_ostream OS(FileName, EC, llvm::sys::fs::OpenFlags::F_Text);
if (EC)
return EC;
OS << Contents;
if (OS.has_error())
return make_error_code(errc::io_error);
return EC;
}
bool llvm::sys::commandLineFitsWithinSystemLimits(StringRef Program, ArrayRef<const char*> Args) {
static long ArgMax = sysconf(_SC_ARG_MAX);
// System says no practical limit.
if (ArgMax == -1)
return true;
// Conservatively account for space required by environment variables.
long HalfArgMax = ArgMax / 2;
size_t ArgLength = Program.size() + 1;
for (ArrayRef<const char*>::iterator I = Args.begin(), E = Args.end();
I != E; ++I) {
ArgLength += strlen(*I) + 1;
if (ArgLength > size_t(HalfArgMax)) {
return false;
}
}
return true;
}
}