Fixes #161510 Addresses the Linux parts of #138085 The situation we have to handle here is systems where Yama ptrace_scope set to 1. > 1 - restricted ptrace: a process must have a predefined relationship > with the inferior it wants to call PTRACE_ATTACH on. By default, > this relationship is that of only its descendants when the above > classic criteria is also met. To change the relationship, an > inferior can call prctl(PR_SET_PTRACER, debugger, ...) to declare > an allowed debugger PID to call PTRACE_ATTACH on the inferior. > Using PTRACE_TRACEME is unchanged. (https://www.kernel.org/doc/Documentation/security/Yama.txt) The inferior was addressing this by calling this at the start of main(): prctl(PR_SET_PTRACER, PR_SET_PTRACER_ANY, 0, 0, 0); Which is ok if lldb-server tries to attach after that call has happened, but there was nothing to synchronise this. So if the system was heavily loaded, the inferior may be stalled, delaying the call, causing lldb-server to fail to attach with EPERM (permission denied). We were not using any mechanism to retry the attach or wait for some signal from the inferior. Except we do do this in other tests, even other lldb-server tests. So I have adopted that mechanism to these tests: * The inferior is launched with `syncfile:<path>` as its first argument. * It creates this file at `<path>`, at a point where we know attaching has been allowed. * The test framework launches the inferior then waits for the file to appear. * This check is retried a few times, increasing the delay each time and eventually giving up. * Only once it has seen the file does it start lldb-server and tell it to attach to the inferior. I have tested this by insterting a `sleep()` call before the attach enable call and running the test on a machine with ptrace_scope set to 1. I was able to increase the sleep to 6 seconds before tests failed (when running just these tests, single threaded). With OS scheduling, you could be stalled indefinitely, so we may have to increase this timeout but this is easy to do with wait_for_file_on_target. The alternative is to have the test runner check ptrace_scope and only enable these on systems where it's 0. Would be good to keep them running if we can though.
412 lines
12 KiB
C++
412 lines
12 KiB
C++
#include "attach.h"
|
|
#include <atomic>
|
|
#include <cassert>
|
|
#include <chrono>
|
|
#include <cstdlib>
|
|
#include <cstring>
|
|
#include <errno.h>
|
|
#include <fstream>
|
|
#include <future>
|
|
#include <inttypes.h>
|
|
#include <memory>
|
|
#include <mutex>
|
|
#if !defined(_WIN32)
|
|
#include <pthread.h>
|
|
#include <signal.h>
|
|
#include <unistd.h>
|
|
#endif
|
|
#include "thread.h"
|
|
#include <setjmp.h>
|
|
#include <stdint.h>
|
|
#include <stdio.h>
|
|
#include <string.h>
|
|
#include <string>
|
|
#include <thread>
|
|
#include <time.h>
|
|
#include <vector>
|
|
#if defined(__APPLE__)
|
|
#include <TargetConditionals.h>
|
|
#endif
|
|
|
|
static const char *const PRINT_PID_COMMAND = "print-pid";
|
|
|
|
static bool g_print_thread_ids = false;
|
|
static std::mutex g_print_mutex;
|
|
static bool g_threads_do_segfault = false;
|
|
|
|
static std::mutex g_jump_buffer_mutex;
|
|
static jmp_buf g_jump_buffer;
|
|
static bool g_is_segfaulting = false;
|
|
|
|
static char g_message[256];
|
|
|
|
static volatile char g_c1 = '0';
|
|
static volatile char g_c2 = '1';
|
|
|
|
static void print_pid() {
|
|
#if defined(_WIN32)
|
|
fprintf(stderr, "PID: %d\n", ::GetCurrentProcessId());
|
|
#else
|
|
fprintf(stderr, "PID: %d\n", getpid());
|
|
#endif
|
|
}
|
|
|
|
static void signal_handler(int signo) {
|
|
#if defined(_WIN32)
|
|
// No signal support on Windows.
|
|
#else
|
|
const char *signal_name = nullptr;
|
|
switch (signo) {
|
|
case SIGUSR1:
|
|
signal_name = "SIGUSR1";
|
|
break;
|
|
case SIGSEGV:
|
|
signal_name = "SIGSEGV";
|
|
break;
|
|
default:
|
|
signal_name = nullptr;
|
|
}
|
|
|
|
// Print notice that we received the signal on a given thread.
|
|
char buf[100];
|
|
if (signal_name)
|
|
snprintf(buf, sizeof(buf), "received %s on thread id: %" PRIx64 "\n", signal_name, get_thread_id());
|
|
else
|
|
snprintf(buf, sizeof(buf), "received signo %d (%s) on thread id: %" PRIx64 "\n", signo, strsignal(signo), get_thread_id());
|
|
write(STDOUT_FILENO, buf, strlen(buf));
|
|
|
|
// Reset the signal handler if we're one of the expected signal handlers.
|
|
switch (signo) {
|
|
case SIGSEGV:
|
|
if (g_is_segfaulting) {
|
|
// Fix up the pointer we're writing to. This needs to happen if nothing
|
|
// intercepts the SIGSEGV (i.e. if somebody runs this from the command
|
|
// line).
|
|
longjmp(g_jump_buffer, 1);
|
|
}
|
|
break;
|
|
case SIGUSR1:
|
|
if (g_is_segfaulting) {
|
|
// Fix up the pointer we're writing to. This is used to test gdb remote
|
|
// signal delivery. A SIGSEGV will be raised when the thread is created,
|
|
// switched out for a SIGUSR1, and then this code still needs to fix the
|
|
// seg fault. (i.e. if somebody runs this from the command line).
|
|
longjmp(g_jump_buffer, 1);
|
|
}
|
|
break;
|
|
}
|
|
|
|
// Reset the signal handler.
|
|
sig_t sig_result = signal(signo, signal_handler);
|
|
if (sig_result == SIG_ERR) {
|
|
fprintf(stderr, "failed to set signal handler: errno=%d\n", errno);
|
|
exit(1);
|
|
}
|
|
#endif
|
|
}
|
|
|
|
static void swap_chars() {
|
|
#if defined(__x86_64__) || defined(__i386__)
|
|
asm volatile("movb %1, (%2)\n\t"
|
|
"movb %0, (%3)\n\t"
|
|
"movb %0, (%2)\n\t"
|
|
"movb %1, (%3)\n\t"
|
|
:
|
|
: "i"('0'), "i"('1'), "r"(&g_c1), "r"(&g_c2)
|
|
: "memory");
|
|
#elif defined(__aarch64__)
|
|
asm volatile("strb %w1, [%2]\n\t"
|
|
"strb %w0, [%3]\n\t"
|
|
"strb %w0, [%2]\n\t"
|
|
"strb %w1, [%3]\n\t"
|
|
:
|
|
: "r"('0'), "r"('1'), "r"(&g_c1), "r"(&g_c2)
|
|
: "memory");
|
|
#elif defined(__arm__)
|
|
asm volatile("strb %1, [%2]\n\t"
|
|
"strb %0, [%3]\n\t"
|
|
"strb %0, [%2]\n\t"
|
|
"strb %1, [%3]\n\t"
|
|
:
|
|
: "r"('0'), "r"('1'), "r"(&g_c1), "r"(&g_c2)
|
|
: "memory");
|
|
#elif defined(__riscv)
|
|
asm volatile("sb %1, (%2)\n\t"
|
|
"sb %0, (%3)\n\t"
|
|
"sb %0, (%2)\n\t"
|
|
"sb %1, (%3)\n\t"
|
|
:
|
|
: "r"('0'), "r"('1'), "r"(&g_c1), "r"(&g_c2)
|
|
: "memory");
|
|
|
|
#else
|
|
#warning This may generate unpredictible assembly and cause the single-stepping test to fail.
|
|
#warning Please add appropriate assembly for your target.
|
|
g_c1 = '1';
|
|
g_c2 = '0';
|
|
|
|
g_c1 = '0';
|
|
g_c2 = '1';
|
|
#endif
|
|
}
|
|
|
|
static void trap() {
|
|
#if defined(__x86_64__) || defined(__i386__)
|
|
asm volatile("int3");
|
|
#elif defined(__aarch64__)
|
|
asm volatile("brk #0xf000");
|
|
#elif defined(__arm__)
|
|
asm volatile("udf #254");
|
|
#elif defined(__powerpc__)
|
|
asm volatile("trap");
|
|
#elif __has_builtin(__builtin_debugtrap)
|
|
__builtin_debugtrap();
|
|
#else
|
|
#warning Don't know how to generate a trap. Some tests may fail.
|
|
#endif
|
|
}
|
|
|
|
static void hello() {
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("hello, world\n");
|
|
}
|
|
|
|
static void *thread_func(std::promise<void> ready) {
|
|
ready.set_value();
|
|
static std::atomic<int> s_thread_index(1);
|
|
const int this_thread_index = s_thread_index++;
|
|
if (g_print_thread_ids) {
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("thread %d id: %" PRIx64 "\n", this_thread_index, get_thread_id());
|
|
}
|
|
|
|
if (g_threads_do_segfault) {
|
|
// Sleep for a number of seconds based on the thread index.
|
|
// TODO add ability to send commands to test exe so we can
|
|
// handle timing more precisely. This is clunky. All we're
|
|
// trying to do is add predictability as to the timing of
|
|
// signal generation by created threads.
|
|
int sleep_seconds = 2 * (this_thread_index - 1);
|
|
std::this_thread::sleep_for(std::chrono::seconds(sleep_seconds));
|
|
|
|
// Test creating a SEGV.
|
|
{
|
|
std::lock_guard<std::mutex> lock(g_jump_buffer_mutex);
|
|
g_is_segfaulting = true;
|
|
int *bad_p = nullptr;
|
|
if (setjmp(g_jump_buffer) == 0) {
|
|
// Force a seg fault signal on this thread.
|
|
*bad_p = 0;
|
|
} else {
|
|
// Tell the system we're no longer seg faulting.
|
|
// Used by the SIGUSR1 signal handler that we inject
|
|
// in place of the SIGSEGV so it only tries to
|
|
// recover from the SIGSEGV if this seg fault code
|
|
// was in play.
|
|
g_is_segfaulting = false;
|
|
}
|
|
}
|
|
|
|
{
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("thread %" PRIx64 ": past SIGSEGV\n", get_thread_id());
|
|
}
|
|
}
|
|
|
|
int sleep_seconds_remaining = 60;
|
|
std::this_thread::sleep_for(std::chrono::seconds(sleep_seconds_remaining));
|
|
|
|
return nullptr;
|
|
}
|
|
|
|
static bool consume_front(std::string &str, const std::string &front) {
|
|
if (str.find(front) != 0)
|
|
return false;
|
|
|
|
str = str.substr(front.size());
|
|
return true;
|
|
}
|
|
|
|
int main(int argc, char **argv) {
|
|
lldb_enable_attach();
|
|
|
|
std::vector<std::thread> threads;
|
|
std::unique_ptr<uint8_t[]> heap_array_up;
|
|
int return_value = 0;
|
|
|
|
#if !defined(_WIN32)
|
|
bool is_child = false;
|
|
|
|
// Set the signal handler.
|
|
sig_t sig_result = signal(SIGALRM, signal_handler);
|
|
if (sig_result == SIG_ERR) {
|
|
fprintf(stderr, "failed to set SIGALRM signal handler: errno=%d\n", errno);
|
|
exit(1);
|
|
}
|
|
|
|
sig_result = signal(SIGUSR1, signal_handler);
|
|
if (sig_result == SIG_ERR) {
|
|
fprintf(stderr, "failed to set SIGUSR1 handler: errno=%d\n", errno);
|
|
exit(1);
|
|
}
|
|
|
|
sig_result = signal(SIGSEGV, signal_handler);
|
|
if (sig_result == SIG_ERR) {
|
|
fprintf(stderr, "failed to set SIGSEGV handler: errno=%d\n", errno);
|
|
exit(1);
|
|
}
|
|
|
|
sig_result = signal(SIGCHLD, SIG_IGN);
|
|
if (sig_result == SIG_ERR) {
|
|
fprintf(stderr, "failed to set SIGCHLD handler: errno=%d\n", errno);
|
|
exit(1);
|
|
}
|
|
#endif
|
|
|
|
// Process command line args.
|
|
for (int i = 1; i < argc; ++i) {
|
|
std::string arg = argv[i];
|
|
if (consume_front(arg, "syncfile:")) {
|
|
// Write to this file to tell test framework that attaching is now
|
|
// possible.
|
|
std::ofstream(arg).close();
|
|
} else if (consume_front(arg, "stderr:")) {
|
|
// Treat remainder as text to go to stderr.
|
|
fprintf(stderr, "%s\n", arg.c_str());
|
|
} else if (consume_front(arg, "retval:")) {
|
|
// Treat as the return value for the program.
|
|
return_value = std::atoi(arg.c_str());
|
|
} else if (consume_front(arg, "sleep:")) {
|
|
// Treat as the amount of time to have this process sleep (in seconds).
|
|
int sleep_seconds_remaining = std::atoi(arg.c_str());
|
|
|
|
// Loop around, sleeping until all sleep time is used up. Note that
|
|
// signals will cause sleep to end early with the number of seconds
|
|
// remaining.
|
|
std::this_thread::sleep_for(
|
|
std::chrono::seconds(sleep_seconds_remaining));
|
|
|
|
} else if (consume_front(arg, "set-message:")) {
|
|
// Copy the contents after "set-message:" to the g_message buffer.
|
|
// Used for reading inferior memory and verifying contents match
|
|
// expectations.
|
|
strncpy(g_message, arg.c_str(), sizeof(g_message));
|
|
|
|
// Ensure we're null terminated.
|
|
g_message[sizeof(g_message) - 1] = '\0';
|
|
|
|
} else if (consume_front(arg, "print-message:")) {
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("message: %s\n", g_message);
|
|
} else if (consume_front(arg, "get-data-address-hex:")) {
|
|
volatile void *data_p = nullptr;
|
|
|
|
if (arg == "g_message")
|
|
data_p = &g_message[0];
|
|
else if (arg == "g_c1")
|
|
data_p = &g_c1;
|
|
else if (arg == "g_c2")
|
|
data_p = &g_c2;
|
|
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("data address: %p\n", data_p);
|
|
} else if (consume_front(arg, "get-heap-address-hex:")) {
|
|
// Create a byte array if not already present.
|
|
if (!heap_array_up)
|
|
heap_array_up.reset(new uint8_t[32]);
|
|
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("heap address: %p\n", heap_array_up.get());
|
|
|
|
} else if (consume_front(arg, "get-stack-address-hex:")) {
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("stack address: %p\n", &return_value);
|
|
} else if (consume_front(arg, "get-code-address-hex:")) {
|
|
void (*func_p)() = nullptr;
|
|
|
|
if (arg == "hello")
|
|
func_p = hello;
|
|
else if (arg == "swap_chars")
|
|
func_p = swap_chars;
|
|
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("code address: %p\n", func_p);
|
|
} else if (consume_front(arg, "call-function:")) {
|
|
void (*func_p)() = nullptr;
|
|
|
|
if (arg == "hello")
|
|
func_p = hello;
|
|
else if (arg == "swap_chars")
|
|
func_p = swap_chars;
|
|
func_p();
|
|
#if !defined(_WIN32) && !defined(TARGET_OS_WATCH) && !defined(TARGET_OS_TV)
|
|
} else if (arg == "fork") {
|
|
pid_t fork_pid = fork();
|
|
assert(fork_pid != -1);
|
|
is_child = fork_pid == 0;
|
|
} else if (arg == "vfork") {
|
|
if (vfork() == 0)
|
|
_exit(0);
|
|
} else if (consume_front(arg, "process:sync:")) {
|
|
// this is only valid after fork
|
|
const char *filenames[] = {"parent", "child"};
|
|
std::string my_file = arg + "." + filenames[is_child];
|
|
std::string other_file = arg + "." + filenames[!is_child];
|
|
|
|
// indicate that we're ready
|
|
FILE *f = fopen(my_file.c_str(), "w");
|
|
assert(f);
|
|
fclose(f);
|
|
|
|
// wait for the other process to be ready
|
|
for (int i = 0; i < 5; ++i) {
|
|
f = fopen(other_file.c_str(), "r");
|
|
if (f)
|
|
break;
|
|
std::this_thread::sleep_for(std::chrono::milliseconds(125 * (1<<i)));
|
|
}
|
|
assert(f);
|
|
fclose(f);
|
|
#endif
|
|
} else if (consume_front(arg, "thread:new")) {
|
|
std::promise<void> promise;
|
|
std::future<void> ready = promise.get_future();
|
|
threads.push_back(std::thread(thread_func, std::move(promise)));
|
|
ready.wait();
|
|
} else if (consume_front(arg, "thread:print-ids")) {
|
|
// Turn on thread id announcing.
|
|
g_print_thread_ids = true;
|
|
|
|
// And announce us.
|
|
{
|
|
std::lock_guard<std::mutex> lock(g_print_mutex);
|
|
printf("thread 0 id: %" PRIx64 "\n", get_thread_id());
|
|
}
|
|
} else if (consume_front(arg, "thread:segfault")) {
|
|
g_threads_do_segfault = true;
|
|
} else if (consume_front(arg, "print-pid")) {
|
|
print_pid();
|
|
} else if (consume_front(arg, "print-env:")) {
|
|
// Print the value of specified envvar to stdout.
|
|
const char *value = getenv(arg.c_str());
|
|
printf("%s\n", value ? value : "__unset__");
|
|
} else if (consume_front(arg, "trap")) {
|
|
trap();
|
|
#if !defined(_WIN32)
|
|
} else if (arg == "stop") {
|
|
raise(SIGINT);
|
|
#endif
|
|
} else {
|
|
// Treat the argument as text for stdout.
|
|
printf("%s\n", argv[i]);
|
|
}
|
|
}
|
|
|
|
// If we launched any threads, join them
|
|
for (std::vector<std::thread>::iterator it = threads.begin();
|
|
it != threads.end(); ++it)
|
|
it->join();
|
|
|
|
return return_value;
|
|
}
|