//===-- MachProcess.cpp -----------------------------------------*- C++ -*-===// // // The LLVM Compiler Infrastructure // // This file is distributed under the University of Illinois Open Source // License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// // // Created by Greg Clayton on 6/15/07. // //===----------------------------------------------------------------------===// #include "DNB.h" #include #include #include #include #include #include #include #include #include #include #include "MacOSX/CFUtils.h" #include "SysSignal.h" #include #include #include "DNBDataRef.h" #include "DNBLog.h" #include "DNBThreadResumeActions.h" #include "DNBTimer.h" #include "MachProcess.h" #include "PseudoTerminal.h" #include "CFBundle.h" #include "CFData.h" #include "CFString.h" static CFStringRef CopyBundleIDForPath (const char *app_buncle_path, DNBError &err_str); #ifdef WITH_SPRINGBOARD #include #include #include static bool IsSBProcess (nub_process_t pid) { CFReleaser appIdsForPID (::SBSCopyDisplayIdentifiersForProcessID(pid)); return appIdsForPID.get() != NULL; } #endif #if 0 #define DEBUG_LOG(fmt, ...) printf(fmt, ## __VA_ARGS__) #else #define DEBUG_LOG(fmt, ...) #endif #ifndef MACH_PROCESS_USE_POSIX_SPAWN #define MACH_PROCESS_USE_POSIX_SPAWN 1 #endif #ifndef _POSIX_SPAWN_DISABLE_ASLR #define _POSIX_SPAWN_DISABLE_ASLR 0x0100 #endif MachProcess::MachProcess() : m_pid (0), m_cpu_type (0), m_child_stdin (-1), m_child_stdout (-1), m_child_stderr (-1), m_path (), m_args (), m_task (this), m_flags (eMachProcessFlagsNone), m_stdio_thread (0), m_stdio_mutex (PTHREAD_MUTEX_RECURSIVE), m_stdout_data (), m_thread_actions (), m_profile_enabled (false), m_profile_interval_usec (0), m_profile_thread (0), m_profile_data_mutex(PTHREAD_MUTEX_RECURSIVE), m_profile_data (), m_thread_list (), m_exception_messages (), m_exception_messages_mutex (PTHREAD_MUTEX_RECURSIVE), m_state (eStateUnloaded), m_state_mutex (PTHREAD_MUTEX_RECURSIVE), m_events (0, kAllEventsMask), m_private_events (0, kAllEventsMask), m_breakpoints (), m_watchpoints (), m_name_to_addr_callback(NULL), m_name_to_addr_baton(NULL), m_image_infos_callback(NULL), m_image_infos_baton(NULL), m_did_exec (false) { DNBLogThreadedIf(LOG_PROCESS | LOG_VERBOSE, "%s", __PRETTY_FUNCTION__); } MachProcess::~MachProcess() { DNBLogThreadedIf(LOG_PROCESS | LOG_VERBOSE, "%s", __PRETTY_FUNCTION__); Clear(); } pid_t MachProcess::SetProcessID(pid_t pid) { // Free any previous process specific data or resources Clear(); // Set the current PID appropriately if (pid == 0) m_pid = ::getpid (); else m_pid = pid; return m_pid; // Return actualy PID in case a zero pid was passed in } nub_state_t MachProcess::GetState() { // If any other threads access this we will need a mutex for it PTHREAD_MUTEX_LOCKER(locker, m_state_mutex); return m_state; } const char * MachProcess::ThreadGetName(nub_thread_t tid) { return m_thread_list.GetName(tid); } nub_state_t MachProcess::ThreadGetState(nub_thread_t tid) { return m_thread_list.GetState(tid); } nub_size_t MachProcess::GetNumThreads () const { return m_thread_list.NumThreads(); } nub_thread_t MachProcess::GetThreadAtIndex (nub_size_t thread_idx) const { return m_thread_list.ThreadIDAtIndex(thread_idx); } nub_thread_t MachProcess::GetThreadIDForMachPortNumber (thread_t mach_port_number) const { return m_thread_list.GetThreadIDByMachPortNumber (mach_port_number); } nub_bool_t MachProcess::SyncThreadState (nub_thread_t tid) { MachThreadSP thread_sp(m_thread_list.GetThreadByID(tid)); if (!thread_sp) return false; kern_return_t kret = ::thread_abort_safely(thread_sp->MachPortNumber()); DNBLogThreadedIf (LOG_THREAD, "thread = 0x%8.8" PRIx32 " calling thread_abort_safely (tid) => %u (GetGPRState() for stop_count = %u)", thread_sp->MachPortNumber(), kret, thread_sp->Process()->StopCount()); if (kret == KERN_SUCCESS) return true; else return false; } nub_thread_t MachProcess::GetCurrentThread () { return m_thread_list.CurrentThreadID(); } nub_thread_t MachProcess::GetCurrentThreadMachPort () { return m_thread_list.GetMachPortNumberByThreadID(m_thread_list.CurrentThreadID()); } nub_thread_t MachProcess::SetCurrentThread(nub_thread_t tid) { return m_thread_list.SetCurrentThread(tid); } bool MachProcess::GetThreadStoppedReason(nub_thread_t tid, struct DNBThreadStopInfo *stop_info) { if (m_thread_list.GetThreadStoppedReason(tid, stop_info)) { if (m_did_exec) stop_info->reason = eStopTypeExec; return true; } return false; } void MachProcess::DumpThreadStoppedReason(nub_thread_t tid) const { return m_thread_list.DumpThreadStoppedReason(tid); } const char * MachProcess::GetThreadInfo(nub_thread_t tid) const { return m_thread_list.GetThreadInfo(tid); } uint32_t MachProcess::GetCPUType () { if (m_cpu_type == 0 && m_pid != 0) m_cpu_type = MachProcess::GetCPUTypeForLocalProcess (m_pid); return m_cpu_type; } const DNBRegisterSetInfo * MachProcess::GetRegisterSetInfo (nub_thread_t tid, nub_size_t *num_reg_sets) const { MachThreadSP thread_sp (m_thread_list.GetThreadByID (tid)); if (thread_sp) { DNBArchProtocol *arch = thread_sp->GetArchProtocol(); if (arch) return arch->GetRegisterSetInfo (num_reg_sets); } *num_reg_sets = 0; return NULL; } bool MachProcess::GetRegisterValue ( nub_thread_t tid, uint32_t set, uint32_t reg, DNBRegisterValue *value ) const { return m_thread_list.GetRegisterValue(tid, set, reg, value); } bool MachProcess::SetRegisterValue ( nub_thread_t tid, uint32_t set, uint32_t reg, const DNBRegisterValue *value ) const { return m_thread_list.SetRegisterValue(tid, set, reg, value); } void MachProcess::SetState(nub_state_t new_state) { // If any other threads access this we will need a mutex for it uint32_t event_mask = 0; // Scope for mutex locker { PTHREAD_MUTEX_LOCKER(locker, m_state_mutex); const nub_state_t old_state = m_state; if (old_state == eStateExited) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::SetState(%s) ignoring new state since current state is exited", DNBStateAsString(new_state)); } else if (old_state == new_state) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::SetState(%s) ignoring redundant state change...", DNBStateAsString(new_state)); } else { if (NUB_STATE_IS_STOPPED(new_state)) event_mask = eEventProcessStoppedStateChanged; else event_mask = eEventProcessRunningStateChanged; DNBLogThreadedIf(LOG_PROCESS, "MachProcess::SetState(%s) upating state (previous state was %s), event_mask = 0x%8.8x", DNBStateAsString(new_state), DNBStateAsString(old_state), event_mask); m_state = new_state; if (new_state == eStateStopped) m_stop_count++; } } if (event_mask != 0) { m_events.SetEvents (event_mask); m_private_events.SetEvents (event_mask); if (event_mask == eEventProcessStoppedStateChanged) m_private_events.ResetEvents (eEventProcessRunningStateChanged); else m_private_events.ResetEvents (eEventProcessStoppedStateChanged); // Wait for the event bit to reset if a reset ACK is requested m_events.WaitForResetAck(event_mask); } } void MachProcess::Clear() { // Clear any cached thread list while the pid and task are still valid m_task.Clear(); // Now clear out all member variables m_pid = INVALID_NUB_PROCESS; CloseChildFileDescriptors(); m_path.clear(); m_args.clear(); SetState(eStateUnloaded); m_flags = eMachProcessFlagsNone; m_stop_count = 0; m_thread_list.Clear(); { PTHREAD_MUTEX_LOCKER(locker, m_exception_messages_mutex); m_exception_messages.clear(); } if (m_profile_thread) { pthread_join(m_profile_thread, NULL); m_profile_thread = NULL; } } bool MachProcess::StartSTDIOThread() { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s ( )", __FUNCTION__); // Create the thread that watches for the child STDIO return ::pthread_create (&m_stdio_thread, NULL, MachProcess::STDIOThread, this) == 0; } void MachProcess::SetEnableAsyncProfiling(bool enable, uint64_t interval_usec, DNBProfileDataScanType scan_type) { m_profile_enabled = enable; m_profile_interval_usec = interval_usec; m_profile_scan_type = scan_type; if (m_profile_enabled && (m_profile_thread == NULL)) { StartProfileThread(); } else if (!m_profile_enabled && m_profile_thread) { pthread_join(m_profile_thread, NULL); m_profile_thread = NULL; } } bool MachProcess::StartProfileThread() { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s ( )", __FUNCTION__); // Create the thread that profiles the inferior and reports back if enabled return ::pthread_create (&m_profile_thread, NULL, MachProcess::ProfileThread, this) == 0; } nub_addr_t MachProcess::LookupSymbol(const char *name, const char *shlib) { if (m_name_to_addr_callback != NULL && name && name[0]) return m_name_to_addr_callback(ProcessID(), name, shlib, m_name_to_addr_baton); return INVALID_NUB_ADDRESS; } bool MachProcess::Resume (const DNBThreadResumeActions& thread_actions) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Resume ()"); nub_state_t state = GetState(); if (CanResume(state)) { m_thread_actions = thread_actions; PrivateResume(); return true; } else if (state == eStateRunning) { DNBLog("Resume() - task 0x%x is already running, ignoring...", m_task.TaskPort()); return true; } DNBLog("Resume() - task 0x%x has state %s, can't continue...", m_task.TaskPort(), DNBStateAsString(state)); return false; } bool MachProcess::Kill (const struct timespec *timeout_abstime) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Kill ()"); nub_state_t state = DoSIGSTOP(true, false, NULL); DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Kill() DoSIGSTOP() state = %s", DNBStateAsString(state)); errno = 0; DNBLog ("Sending ptrace PT_KILL to terminate inferior process."); ::ptrace (PT_KILL, m_pid, 0, 0); DNBError err; err.SetErrorToErrno(); DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Kill() DoSIGSTOP() ::ptrace (PT_KILL, pid=%u, 0, 0) => 0x%8.8x (%s)", m_pid, err.Error(), err.AsString()); m_thread_actions = DNBThreadResumeActions (eStateRunning, 0); PrivateResume (); // Try and reap the process without touching our m_events since // we want the code above this to still get the eStateExited event const uint32_t reap_timeout_usec = 1000000; // Wait 1 second and try to reap the process const uint32_t reap_interval_usec = 10000; // uint32_t reap_time_elapsed; for (reap_time_elapsed = 0; reap_time_elapsed < reap_timeout_usec; reap_time_elapsed += reap_interval_usec) { if (GetState() == eStateExited) break; usleep(reap_interval_usec); } DNBLog ("Waited %u ms for process to be reaped (state = %s)", reap_time_elapsed/1000, DNBStateAsString(GetState())); return true; } bool MachProcess::Signal (int signal, const struct timespec *timeout_abstime) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Signal (signal = %d, timeout = %p)", signal, timeout_abstime); nub_state_t state = GetState(); if (::kill (ProcessID(), signal) == 0) { // If we were running and we have a timeout, wait for the signal to stop if (IsRunning(state) && timeout_abstime) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Signal (signal = %d, timeout = %p) waiting for signal to stop process...", signal, timeout_abstime); m_private_events.WaitForSetEvents(eEventProcessStoppedStateChanged, timeout_abstime); state = GetState(); DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Signal (signal = %d, timeout = %p) state = %s", signal, timeout_abstime, DNBStateAsString(state)); return !IsRunning (state); } DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Signal (signal = %d, timeout = %p) not waiting...", signal, timeout_abstime); return true; } DNBError err(errno, DNBError::POSIX); err.LogThreadedIfError("kill (pid = %d, signo = %i)", ProcessID(), signal); return false; } nub_state_t MachProcess::DoSIGSTOP (bool clear_bps_and_wps, bool allow_running, uint32_t *thread_idx_ptr) { nub_state_t state = GetState(); DNBLogThreadedIf(LOG_PROCESS, "MachProcess::DoSIGSTOP() state = %s", DNBStateAsString (state)); if (!IsRunning(state)) { if (clear_bps_and_wps) { DisableAllBreakpoints (true); DisableAllWatchpoints (true); clear_bps_and_wps = false; } // If we already have a thread stopped due to a SIGSTOP, we don't have // to do anything... uint32_t thread_idx = m_thread_list.GetThreadIndexForThreadStoppedWithSignal (SIGSTOP); if (thread_idx_ptr) *thread_idx_ptr = thread_idx; if (thread_idx != UINT32_MAX) return GetState(); // No threads were stopped with a SIGSTOP, we need to run and halt the // process with a signal DNBLogThreadedIf(LOG_PROCESS, "MachProcess::DoSIGSTOP() state = %s -- resuming process", DNBStateAsString (state)); if (allow_running) m_thread_actions = DNBThreadResumeActions (eStateRunning, 0); else m_thread_actions = DNBThreadResumeActions (eStateSuspended, 0); PrivateResume (); // Reset the event that says we were indeed running m_events.ResetEvents(eEventProcessRunningStateChanged); state = GetState(); } // We need to be stopped in order to be able to detach, so we need // to send ourselves a SIGSTOP DNBLogThreadedIf(LOG_PROCESS, "MachProcess::DoSIGSTOP() state = %s -- sending SIGSTOP", DNBStateAsString (state)); struct timespec sigstop_timeout; DNBTimer::OffsetTimeOfDay(&sigstop_timeout, 2, 0); Signal (SIGSTOP, &sigstop_timeout); if (clear_bps_and_wps) { DisableAllBreakpoints (true); DisableAllWatchpoints (true); //clear_bps_and_wps = false; } uint32_t thread_idx = m_thread_list.GetThreadIndexForThreadStoppedWithSignal (SIGSTOP); if (thread_idx_ptr) *thread_idx_ptr = thread_idx; return GetState(); } bool MachProcess::Detach() { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Detach()"); uint32_t thread_idx = UINT32_MAX; nub_state_t state = DoSIGSTOP(true, true, &thread_idx); DNBLogThreadedIf(LOG_PROCESS, "MachProcess::Detach() DoSIGSTOP() returned %s", DNBStateAsString(state)); { m_thread_actions.Clear(); DNBThreadResumeAction thread_action; thread_action.tid = m_thread_list.ThreadIDAtIndex (thread_idx); thread_action.state = eStateRunning; thread_action.signal = -1; thread_action.addr = INVALID_NUB_ADDRESS; m_thread_actions.Append (thread_action); m_thread_actions.SetDefaultThreadActionIfNeeded (eStateRunning, 0); PTHREAD_MUTEX_LOCKER (locker, m_exception_messages_mutex); ReplyToAllExceptions (); } m_task.ShutDownExcecptionThread(); // Detach from our process errno = 0; nub_process_t pid = m_pid; int ret = ::ptrace (PT_DETACH, pid, (caddr_t)1, 0); DNBError err(errno, DNBError::POSIX); if (DNBLogCheckLogBit(LOG_PROCESS) || err.Fail() || (ret != 0)) err.LogThreaded("::ptrace (PT_DETACH, %u, (caddr_t)1, 0)", pid); // Resume our task m_task.Resume(); // NULL our task out as we have already retored all exception ports m_task.Clear(); // Clear out any notion of the process we once were Clear(); SetState(eStateDetached); return true; } //---------------------------------------------------------------------- // ReadMemory from the MachProcess level will always remove any software // breakpoints from the memory buffer before returning. If you wish to // read memory and see those traps, read from the MachTask // (m_task.ReadMemory()) as that version will give you what is actually // in inferior memory. //---------------------------------------------------------------------- nub_size_t MachProcess::ReadMemory (nub_addr_t addr, nub_size_t size, void *buf) { // We need to remove any current software traps (enabled software // breakpoints) that we may have placed in our tasks memory. // First just read the memory as is nub_size_t bytes_read = m_task.ReadMemory(addr, size, buf); // Then place any opcodes that fall into this range back into the buffer // before we return this to callers. if (bytes_read > 0) m_breakpoints.RemoveTrapsFromBuffer (addr, bytes_read, buf); return bytes_read; } //---------------------------------------------------------------------- // WriteMemory from the MachProcess level will always write memory around // any software breakpoints. Any software breakpoints will have their // opcodes modified if they are enabled. Any memory that doesn't overlap // with software breakpoints will be written to. If you wish to write to // inferior memory without this interference, then write to the MachTask // (m_task.WriteMemory()) as that version will always modify inferior // memory. //---------------------------------------------------------------------- nub_size_t MachProcess::WriteMemory (nub_addr_t addr, nub_size_t size, const void *buf) { // We need to write any data that would go where any current software traps // (enabled software breakpoints) any software traps (breakpoints) that we // may have placed in our tasks memory. std::vector bps; const size_t num_bps = m_breakpoints.FindBreakpointsThatOverlapRange(addr, size, bps); if (num_bps == 0) return m_task.WriteMemory(addr, size, buf); nub_size_t bytes_written = 0; nub_addr_t intersect_addr; nub_size_t intersect_size; nub_size_t opcode_offset; const uint8_t *ubuf = (const uint8_t *)buf; for (size_t i=0; iIntersectsRange(addr, size, &intersect_addr, &intersect_size, &opcode_offset); assert(intersects); assert(addr <= intersect_addr && intersect_addr < addr + size); assert(addr < intersect_addr + intersect_size && intersect_addr + intersect_size <= addr + size); assert(opcode_offset + intersect_size <= bp->ByteSize()); // Check for bytes before this breakpoint const nub_addr_t curr_addr = addr + bytes_written; if (intersect_addr > curr_addr) { // There are some bytes before this breakpoint that we need to // just write to memory nub_size_t curr_size = intersect_addr - curr_addr; nub_size_t curr_bytes_written = m_task.WriteMemory(curr_addr, curr_size, ubuf + bytes_written); bytes_written += curr_bytes_written; if (curr_bytes_written != curr_size) { // We weren't able to write all of the requested bytes, we // are done looping and will return the number of bytes that // we have written so far. break; } } // Now write any bytes that would cover up any software breakpoints // directly into the breakpoint opcode buffer ::memcpy(bp->SavedOpcodeBytes() + opcode_offset, ubuf + bytes_written, intersect_size); bytes_written += intersect_size; } // Write any remaining bytes after the last breakpoint if we have any left if (bytes_written < size) bytes_written += m_task.WriteMemory(addr + bytes_written, size - bytes_written, ubuf + bytes_written); return bytes_written; } void MachProcess::ReplyToAllExceptions () { PTHREAD_MUTEX_LOCKER(locker, m_exception_messages_mutex); if (m_exception_messages.empty() == false) { MachException::Message::iterator pos; MachException::Message::iterator begin = m_exception_messages.begin(); MachException::Message::iterator end = m_exception_messages.end(); for (pos = begin; pos != end; ++pos) { DNBLogThreadedIf(LOG_EXCEPTIONS, "Replying to exception %u...", (uint32_t)std::distance(begin, pos)); int thread_reply_signal = 0; nub_thread_t tid = m_thread_list.GetThreadIDByMachPortNumber (pos->state.thread_port); const DNBThreadResumeAction *action = NULL; if (tid != INVALID_NUB_THREAD) { action = m_thread_actions.GetActionForThread (tid, false); } if (action) { thread_reply_signal = action->signal; if (thread_reply_signal) m_thread_actions.SetSignalHandledForThread (tid); } DNBError err (pos->Reply(this, thread_reply_signal)); if (DNBLogCheckLogBit(LOG_EXCEPTIONS)) err.LogThreadedIfError("Error replying to exception"); } // Erase all exception message as we should have used and replied // to them all already. m_exception_messages.clear(); } } void MachProcess::PrivateResume () { PTHREAD_MUTEX_LOCKER (locker, m_exception_messages_mutex); ReplyToAllExceptions (); // bool stepOverBreakInstruction = step; // Let the thread prepare to resume and see if any threads want us to // step over a breakpoint instruction (ProcessWillResume will modify // the value of stepOverBreakInstruction). m_thread_list.ProcessWillResume (this, m_thread_actions); // Set our state accordingly if (m_thread_actions.NumActionsWithState(eStateStepping)) SetState (eStateStepping); else SetState (eStateRunning); // Now resume our task. m_task.Resume(); } DNBBreakpoint * MachProcess::CreateBreakpoint(nub_addr_t addr, nub_size_t length, bool hardware) { DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::CreateBreakpoint ( addr = 0x%8.8llx, length = %llu, hardware = %i)", (uint64_t)addr, (uint64_t)length, hardware); DNBBreakpoint *bp = m_breakpoints.FindByAddress(addr); if (bp) bp->Retain(); else bp = m_breakpoints.Add(addr, length, hardware); if (EnableBreakpoint(addr)) { DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::CreateBreakpoint ( addr = 0x%8.8llx, length = %llu) => %p", (uint64_t)addr, (uint64_t)length, bp); return bp; } else if (bp->Release() == 0) { m_breakpoints.Remove(addr); } // We failed to enable the breakpoint return NULL; } DNBBreakpoint * MachProcess::CreateWatchpoint(nub_addr_t addr, nub_size_t length, uint32_t watch_flags, bool hardware) { DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::CreateWatchpoint ( addr = 0x%8.8llx, length = %llu, flags = 0x%8.8x, hardware = %i)", (uint64_t)addr, (uint64_t)length, watch_flags, hardware); DNBBreakpoint *wp = m_watchpoints.FindByAddress(addr); // since the Z packets only send an address, we can only have one watchpoint at // an address. If there is already one, we must refuse to create another watchpoint if (wp) return NULL; wp = m_watchpoints.Add(addr, length, hardware); wp->SetIsWatchpoint(watch_flags); if (EnableWatchpoint(addr)) { DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::CreateWatchpoint ( addr = 0x%8.8llx, length = %llu) => %p", (uint64_t)addr, (uint64_t)length, wp); return wp; } else { DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::CreateWatchpoint ( addr = 0x%8.8llx, length = %llu) => FAILED", (uint64_t)addr, (uint64_t)length); m_watchpoints.Remove(addr); } // We failed to enable the watchpoint return NULL; } void MachProcess::DisableAllBreakpoints (bool remove) { DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::%s (remove = %d )", __FUNCTION__, remove); m_breakpoints.DisableAllBreakpoints (this); if (remove) m_breakpoints.RemoveDisabled(); } void MachProcess::DisableAllWatchpoints(bool remove) { DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::%s (remove = %d )", __FUNCTION__, remove); m_watchpoints.DisableAllWatchpoints(this); if (remove) m_watchpoints.RemoveDisabled(); } bool MachProcess::DisableBreakpoint(nub_addr_t addr, bool remove) { DNBBreakpoint *bp = m_breakpoints.FindByAddress(addr); if (bp) { // After "exec" we might end up with a bunch of breakpoints that were disabled // manually, just ignore them if (!bp->IsEnabled()) { // Breakpoint might have been disabled by an exec if (remove && bp->Release() == 0) { m_thread_list.NotifyBreakpointChanged(bp); m_breakpoints.Remove(addr); } return true; } // We have multiple references to this breakpoint, decrement the ref count // and if it isn't zero, then return true; if (remove && bp->Release() > 0) return true; DNBLogThreadedIf(LOG_BREAKPOINTS | LOG_VERBOSE, "MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d )", (uint64_t)addr, remove); if (bp->IsHardware()) { bool hw_disable_result = m_thread_list.DisableHardwareBreakpoint (bp); if (hw_disable_result == true) { bp->SetEnabled(false); // Let the thread list know that a breakpoint has been modified if (remove) { m_thread_list.NotifyBreakpointChanged(bp); m_breakpoints.Remove(addr); } DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) (hardware) => success", (uint64_t)addr, remove); return true; } return false; } const nub_size_t break_op_size = bp->ByteSize(); assert (break_op_size > 0); const uint8_t * const break_op = DNBArchProtocol::GetBreakpointOpcode (bp->ByteSize()); if (break_op_size > 0) { // Clear a software breakoint instruction uint8_t curr_break_op[break_op_size]; bool break_op_found = false; // Read the breakpoint opcode if (m_task.ReadMemory(addr, break_op_size, curr_break_op) == break_op_size) { bool verify = false; if (bp->IsEnabled()) { // Make sure we have the a breakpoint opcode exists at this address if (memcmp(curr_break_op, break_op, break_op_size) == 0) { break_op_found = true; // We found a valid breakpoint opcode at this address, now restore // the saved opcode. if (m_task.WriteMemory(addr, break_op_size, bp->SavedOpcodeBytes()) == break_op_size) { verify = true; } else { DNBLogError("MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) memory write failed when restoring original opcode", addr, remove); } } else { DNBLogWarning("MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) expected a breakpoint opcode but didn't find one.", addr, remove); // Set verify to true and so we can check if the original opcode has already been restored verify = true; } } else { DNBLogThreadedIf(LOG_BREAKPOINTS | LOG_VERBOSE, "MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) is not enabled", addr, remove); // Set verify to true and so we can check if the original opcode is there verify = true; } if (verify) { uint8_t verify_opcode[break_op_size]; // Verify that our original opcode made it back to the inferior if (m_task.ReadMemory(addr, break_op_size, verify_opcode) == break_op_size) { // compare the memory we just read with the original opcode if (memcmp(bp->SavedOpcodeBytes(), verify_opcode, break_op_size) == 0) { // SUCCESS bp->SetEnabled(false); // Let the thread list know that a breakpoint has been modified if (remove && bp->Release() == 0) { m_thread_list.NotifyBreakpointChanged(bp); m_breakpoints.Remove(addr); } DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) => success", (uint64_t)addr, remove); return true; } else { if (break_op_found) DNBLogError("MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) : failed to restore original opcode", (uint64_t)addr, remove); else DNBLogError("MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) : opcode changed", (uint64_t)addr, remove); } } else { DNBLogWarning("MachProcess::DisableBreakpoint: unable to disable breakpoint 0x%8.8llx", (uint64_t)addr); } } } else { DNBLogWarning("MachProcess::DisableBreakpoint: unable to read memory at 0x%8.8llx", (uint64_t)addr); } } } else { DNBLogError("MachProcess::DisableBreakpoint ( addr = 0x%8.8llx, remove = %d ) invalid breakpoint address", (uint64_t)addr, remove); } return false; } bool MachProcess::DisableWatchpoint(nub_addr_t addr, bool remove) { DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::%s(addr = 0x%8.8llx, remove = %d)", __FUNCTION__, (uint64_t)addr, remove); DNBBreakpoint *wp = m_watchpoints.FindByAddress(addr); if (wp) { // If we have multiple references to a watchpoint, removing the watchpoint shouldn't clear it if (remove && wp->Release() > 0) return true; nub_addr_t addr = wp->Address(); DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::DisableWatchpoint ( addr = 0x%8.8llx, remove = %d )", (uint64_t)addr, remove); if (wp->IsHardware()) { bool hw_disable_result = m_thread_list.DisableHardwareWatchpoint (wp); if (hw_disable_result == true) { wp->SetEnabled(false); if (remove) m_watchpoints.Remove(addr); DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::Disablewatchpoint ( addr = 0x%8.8llx, remove = %d ) (hardware) => success", (uint64_t)addr, remove); return true; } } // TODO: clear software watchpoints if we implement them } else { DNBLogError("MachProcess::DisableWatchpoint ( addr = 0x%8.8llx, remove = %d ) invalid watchpoint ID", (uint64_t)addr, remove); } return false; } uint32_t MachProcess::GetNumSupportedHardwareWatchpoints () const { return m_thread_list.NumSupportedHardwareWatchpoints(); } bool MachProcess::EnableBreakpoint(nub_addr_t addr) { DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::EnableBreakpoint ( addr = 0x%8.8llx )", (uint64_t)addr); DNBBreakpoint *bp = m_breakpoints.FindByAddress(addr); if (bp) { if (bp->IsEnabled()) { DNBLogWarning("MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ): breakpoint already enabled.", (uint64_t)addr); return true; } else { if (bp->HardwarePreferred()) { bp->SetHardwareIndex(m_thread_list.EnableHardwareBreakpoint(bp)); if (bp->IsHardware()) { bp->SetEnabled(true); return true; } } const nub_size_t break_op_size = bp->ByteSize(); assert (break_op_size != 0); const uint8_t * const break_op = DNBArchProtocol::GetBreakpointOpcode (break_op_size); if (break_op_size > 0) { // Save the original opcode by reading it if (m_task.ReadMemory(addr, break_op_size, bp->SavedOpcodeBytes()) == break_op_size) { // Write a software breakpoint in place of the original opcode if (m_task.WriteMemory(addr, break_op_size, break_op) == break_op_size) { uint8_t verify_break_op[4]; if (m_task.ReadMemory(addr, break_op_size, verify_break_op) == break_op_size) { if (memcmp(break_op, verify_break_op, break_op_size) == 0) { bp->SetEnabled(true); // Let the thread list know that a breakpoint has been modified m_thread_list.NotifyBreakpointChanged(bp); DNBLogThreadedIf(LOG_BREAKPOINTS, "MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ) : SUCCESS.", (uint64_t)addr); return true; } else { DNBLogError("MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ): breakpoint opcode verification failed.", (uint64_t)addr); } } else { DNBLogError("MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ): unable to read memory to verify breakpoint opcode.", (uint64_t)addr); } } else { DNBLogError("MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ): unable to write breakpoint opcode to memory.", (uint64_t)addr); } } else { DNBLogError("MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ): unable to read memory at breakpoint address.", (uint64_t)addr); } } else { DNBLogError("MachProcess::EnableBreakpoint ( addr = 0x%8.8llx ) no software breakpoint opcode for current architecture.", (uint64_t)addr); } } } return false; } bool MachProcess::EnableWatchpoint(nub_addr_t addr) { DNBLogThreadedIf(LOG_WATCHPOINTS, "MachProcess::EnableWatchpoint(addr = 0x%8.8llx)", (uint64_t)addr); DNBBreakpoint *wp = m_watchpoints.FindByAddress(addr); if (wp) { nub_addr_t addr = wp->Address(); if (wp->IsEnabled()) { DNBLogWarning("MachProcess::EnableWatchpoint(addr = 0x%8.8llx): watchpoint already enabled.", (uint64_t)addr); return true; } else { // Currently only try and set hardware watchpoints. wp->SetHardwareIndex(m_thread_list.EnableHardwareWatchpoint(wp)); if (wp->IsHardware()) { wp->SetEnabled(true); return true; } // TODO: Add software watchpoints by doing page protection tricks. } } return false; } // Called by the exception thread when an exception has been received from // our process. The exception message is completely filled and the exception // data has already been copied. void MachProcess::ExceptionMessageReceived (const MachException::Message& exceptionMessage) { PTHREAD_MUTEX_LOCKER (locker, m_exception_messages_mutex); if (m_exception_messages.empty()) m_task.Suspend(); DNBLogThreadedIf(LOG_EXCEPTIONS, "MachProcess::ExceptionMessageReceived ( )"); // Use a locker to automatically unlock our mutex in case of exceptions // Add the exception to our internal exception stack m_exception_messages.push_back(exceptionMessage); } void MachProcess::ExceptionMessageBundleComplete() { // We have a complete bundle of exceptions for our child process. PTHREAD_MUTEX_LOCKER (locker, m_exception_messages_mutex); DNBLogThreadedIf(LOG_EXCEPTIONS, "%s: %llu exception messages.", __PRETTY_FUNCTION__, (uint64_t)m_exception_messages.size()); if (!m_exception_messages.empty()) { m_did_exec = false; // First check for any SIGTRAP and make sure we didn't exec const task_t task = m_task.TaskPort(); size_t i; if (m_pid != 0) { for (i=0; i %s) ...", __FUNCTION__, (uint64_t)len, s); PTHREAD_MUTEX_LOCKER (locker, m_stdio_mutex); m_stdout_data.append(s, len); m_events.SetEvents(eEventStdioAvailable); // Wait for the event bit to reset if a reset ACK is requested m_events.WaitForResetAck(eEventStdioAvailable); } size_t MachProcess::GetAvailableSTDOUT (char *buf, size_t buf_size) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s (&%p[%llu]) ...", __FUNCTION__, buf, (uint64_t)buf_size); PTHREAD_MUTEX_LOCKER (locker, m_stdio_mutex); size_t bytes_available = m_stdout_data.size(); if (bytes_available > 0) { if (bytes_available > buf_size) { memcpy(buf, m_stdout_data.data(), buf_size); m_stdout_data.erase(0, buf_size); bytes_available = buf_size; } else { memcpy(buf, m_stdout_data.data(), bytes_available); m_stdout_data.clear(); } } return bytes_available; } nub_addr_t MachProcess::GetDYLDAllImageInfosAddress () { DNBError err; return m_task.GetDYLDAllImageInfosAddress(err); } size_t MachProcess::GetAvailableSTDERR (char *buf, size_t buf_size) { return 0; } void * MachProcess::STDIOThread(void *arg) { MachProcess *proc = (MachProcess*) arg; DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s ( arg = %p ) thread starting...", __FUNCTION__, arg); // We start use a base and more options so we can control if we // are currently using a timeout on the mach_msg. We do this to get a // bunch of related exceptions on our exception port so we can process // then together. When we have multiple threads, we can get an exception // per thread and they will come in consecutively. The main thread loop // will start by calling mach_msg to without having the MACH_RCV_TIMEOUT // flag set in the options, so we will wait forever for an exception on // our exception port. After we get one exception, we then will use the // MACH_RCV_TIMEOUT option with a zero timeout to grab all other current // exceptions for our process. After we have received the last pending // exception, we will get a timeout which enables us to then notify // our main thread that we have an exception bundle avaiable. We then wait // for the main thread to tell this exception thread to start trying to get // exceptions messages again and we start again with a mach_msg read with // infinite timeout. DNBError err; int stdout_fd = proc->GetStdoutFileDescriptor(); int stderr_fd = proc->GetStderrFileDescriptor(); if (stdout_fd == stderr_fd) stderr_fd = -1; while (stdout_fd >= 0 || stderr_fd >= 0) { ::pthread_testcancel (); fd_set read_fds; FD_ZERO (&read_fds); if (stdout_fd >= 0) FD_SET (stdout_fd, &read_fds); if (stderr_fd >= 0) FD_SET (stderr_fd, &read_fds); int nfds = std::max(stdout_fd, stderr_fd) + 1; int num_set_fds = select (nfds, &read_fds, NULL, NULL, NULL); DNBLogThreadedIf(LOG_PROCESS, "select (nfds, &read_fds, NULL, NULL, NULL) => %d", num_set_fds); if (num_set_fds < 0) { int select_errno = errno; if (DNBLogCheckLogBit(LOG_PROCESS)) { err.SetError (select_errno, DNBError::POSIX); err.LogThreadedIfError("select (nfds, &read_fds, NULL, NULL, NULL) => %d", num_set_fds); } switch (select_errno) { case EAGAIN: // The kernel was (perhaps temporarily) unable to allocate the requested number of file descriptors, or we have non-blocking IO break; case EBADF: // One of the descriptor sets specified an invalid descriptor. return NULL; break; case EINTR: // A signal was delivered before the time limit expired and before any of the selected events occurred. case EINVAL: // The specified time limit is invalid. One of its components is negative or too large. default: // Other unknown error break; } } else if (num_set_fds == 0) { } else { char s[1024]; s[sizeof(s)-1] = '\0'; // Ensure we have NULL termination int bytes_read = 0; if (stdout_fd >= 0 && FD_ISSET (stdout_fd, &read_fds)) { do { bytes_read = ::read (stdout_fd, s, sizeof(s)-1); if (bytes_read < 0) { int read_errno = errno; DNBLogThreadedIf(LOG_PROCESS, "read (stdout_fd, ) => %d errno: %d (%s)", bytes_read, read_errno, strerror(read_errno)); } else if (bytes_read == 0) { // EOF... DNBLogThreadedIf(LOG_PROCESS, "read (stdout_fd, ) => %d (reached EOF for child STDOUT)", bytes_read); stdout_fd = -1; } else if (bytes_read > 0) { proc->AppendSTDOUT(s, bytes_read); } } while (bytes_read > 0); } if (stderr_fd >= 0 && FD_ISSET (stderr_fd, &read_fds)) { do { bytes_read = ::read (stderr_fd, s, sizeof(s)-1); if (bytes_read < 0) { int read_errno = errno; DNBLogThreadedIf(LOG_PROCESS, "read (stderr_fd, ) => %d errno: %d (%s)", bytes_read, read_errno, strerror(read_errno)); } else if (bytes_read == 0) { // EOF... DNBLogThreadedIf(LOG_PROCESS, "read (stderr_fd, ) => %d (reached EOF for child STDERR)", bytes_read); stderr_fd = -1; } else if (bytes_read > 0) { proc->AppendSTDOUT(s, bytes_read); } } while (bytes_read > 0); } } } DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s (%p): thread exiting...", __FUNCTION__, arg); return NULL; } void MachProcess::SignalAsyncProfileData (const char *info) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s (%s) ...", __FUNCTION__, info); PTHREAD_MUTEX_LOCKER (locker, m_profile_data_mutex); m_profile_data.push_back(info); m_events.SetEvents(eEventProfileDataAvailable); // Wait for the event bit to reset if a reset ACK is requested m_events.WaitForResetAck(eEventProfileDataAvailable); } size_t MachProcess::GetAsyncProfileData (char *buf, size_t buf_size) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s (&%p[%llu]) ...", __FUNCTION__, buf, (uint64_t)buf_size); PTHREAD_MUTEX_LOCKER (locker, m_profile_data_mutex); if (m_profile_data.empty()) return 0; size_t bytes_available = m_profile_data.front().size(); if (bytes_available > 0) { if (bytes_available > buf_size) { memcpy(buf, m_profile_data.front().data(), buf_size); m_profile_data.front().erase(0, buf_size); bytes_available = buf_size; } else { memcpy(buf, m_profile_data.front().data(), bytes_available); m_profile_data.erase(m_profile_data.begin()); } } return bytes_available; } void * MachProcess::ProfileThread(void *arg) { MachProcess *proc = (MachProcess*) arg; DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s ( arg = %p ) thread starting...", __FUNCTION__, arg); while (proc->IsProfilingEnabled()) { nub_state_t state = proc->GetState(); if (state == eStateRunning) { std::string data = proc->Task().GetProfileData(proc->GetProfileScanType()); if (!data.empty()) { proc->SignalAsyncProfileData(data.c_str()); } } else if ((state == eStateUnloaded) || (state == eStateDetached) || (state == eStateUnloaded)) { // Done. Get out of this thread. break; } // A simple way to set up the profile interval. We can also use select() or dispatch timer source if necessary. usleep(proc->ProfileInterval()); } return NULL; } pid_t MachProcess::AttachForDebug (pid_t pid, char *err_str, size_t err_len) { // Clear out and clean up from any current state Clear(); if (pid != 0) { DNBError err; // Make sure the process exists... if (::getpgid (pid) < 0) { err.SetErrorToErrno(); const char *err_cstr = err.AsString(); ::snprintf (err_str, err_len, "%s", err_cstr ? err_cstr : "No such process"); return INVALID_NUB_PROCESS; } SetState(eStateAttaching); m_pid = pid; // Let ourselves know we are going to be using SBS if the correct flag bit is set... #ifdef WITH_SPRINGBOARD if (IsSBProcess(pid)) m_flags |= eMachProcessFlagsUsingSBS; #endif if (!m_task.StartExceptionThread(err)) { const char *err_cstr = err.AsString(); ::snprintf (err_str, err_len, "%s", err_cstr ? err_cstr : "unable to start the exception thread"); DNBLogThreadedIf(LOG_PROCESS, "error: failed to attach to pid %d", pid); m_pid = INVALID_NUB_PROCESS; return INVALID_NUB_PROCESS; } errno = 0; if (::ptrace (PT_ATTACHEXC, pid, 0, 0)) err.SetError(errno); else err.Clear(); if (err.Success()) { m_flags |= eMachProcessFlagsAttached; // Sleep a bit to let the exception get received and set our process status // to stopped. ::usleep(250000); DNBLogThreadedIf(LOG_PROCESS, "successfully attached to pid %d", pid); return m_pid; } else { ::snprintf (err_str, err_len, "%s", err.AsString()); DNBLogThreadedIf(LOG_PROCESS, "error: failed to attach to pid %d", pid); } } return INVALID_NUB_PROCESS; } // Do the process specific setup for attach. If this returns NULL, then there's no // platform specific stuff to be done to wait for the attach. If you get non-null, // pass that token to the CheckForProcess method, and then to CleanupAfterAttach. // Call PrepareForAttach before attaching to a process that has not yet launched // This returns a token that can be passed to CheckForProcess, and to CleanupAfterAttach. // You should call CleanupAfterAttach to free the token, and do whatever other // cleanup seems good. const void * MachProcess::PrepareForAttach (const char *path, nub_launch_flavor_t launch_flavor, bool waitfor, DNBError &err_str) { #ifdef WITH_SPRINGBOARD // Tell SpringBoard to halt the next launch of this application on startup. if (!waitfor) return NULL; const char *app_ext = strstr(path, ".app"); const bool is_app = app_ext != NULL && (app_ext[4] == '\0' || app_ext[4] == '/'); if (!is_app) { DNBLogThreadedIf(LOG_PROCESS, "MachProcess::PrepareForAttach(): path '%s' doesn't contain .app, we can't tell springboard to wait for launch...", path); return NULL; } if (launch_flavor != eLaunchFlavorSpringBoard && launch_flavor != eLaunchFlavorDefault) return NULL; std::string app_bundle_path(path, app_ext + strlen(".app")); CFStringRef bundleIDCFStr = CopyBundleIDForPath (app_bundle_path.c_str (), err_str); std::string bundleIDStr; CFString::UTF8(bundleIDCFStr, bundleIDStr); DNBLogThreadedIf(LOG_PROCESS, "CopyBundleIDForPath (%s, err_str) returned @\"%s\"", app_bundle_path.c_str (), bundleIDStr.c_str()); if (bundleIDCFStr == NULL) { return NULL; } SBSApplicationLaunchError sbs_error = 0; const char *stdout_err = "/dev/null"; CFString stdio_path; stdio_path.SetFileSystemRepresentation (stdout_err); DNBLogThreadedIf(LOG_PROCESS, "SBSLaunchApplicationForDebugging ( @\"%s\" , NULL, NULL, NULL, @\"%s\", @\"%s\", SBSApplicationDebugOnNextLaunch | SBSApplicationLaunchWaitForDebugger )", bundleIDStr.c_str(), stdout_err, stdout_err); sbs_error = SBSLaunchApplicationForDebugging (bundleIDCFStr, (CFURLRef)NULL, // openURL NULL, // launch_argv.get(), NULL, // launch_envp.get(), // CFDictionaryRef environment stdio_path.get(), stdio_path.get(), SBSApplicationDebugOnNextLaunch | SBSApplicationLaunchWaitForDebugger); if (sbs_error != SBSApplicationLaunchErrorSuccess) { err_str.SetError(sbs_error, DNBError::SpringBoard); return NULL; } DNBLogThreadedIf(LOG_PROCESS, "Successfully set DebugOnNextLaunch."); return bundleIDCFStr; # else return NULL; #endif } // Pass in the token you got from PrepareForAttach. If there is a process // for that token, then the pid will be returned, otherwise INVALID_NUB_PROCESS // will be returned. nub_process_t MachProcess::CheckForProcess (const void *attach_token) { if (attach_token == NULL) return INVALID_NUB_PROCESS; #ifdef WITH_SPRINGBOARD CFStringRef bundleIDCFStr = (CFStringRef) attach_token; Boolean got_it; nub_process_t attach_pid; got_it = SBSProcessIDForDisplayIdentifier(bundleIDCFStr, &attach_pid); if (got_it) return attach_pid; else return INVALID_NUB_PROCESS; #endif return INVALID_NUB_PROCESS; } // Call this to clean up after you have either attached or given up on the attach. // Pass true for success if you have attached, false if you have not. // The token will also be freed at this point, so you can't use it after calling // this method. void MachProcess::CleanupAfterAttach (const void *attach_token, bool success, DNBError &err_str) { #ifdef WITH_SPRINGBOARD if (attach_token == NULL) return; // Tell SpringBoard to cancel the debug on next launch of this application // if we failed to attach if (!success) { SBSApplicationLaunchError sbs_error = 0; CFStringRef bundleIDCFStr = (CFStringRef) attach_token; sbs_error = SBSLaunchApplicationForDebugging (bundleIDCFStr, (CFURLRef)NULL, NULL, NULL, NULL, NULL, SBSApplicationCancelDebugOnNextLaunch); if (sbs_error != SBSApplicationLaunchErrorSuccess) { err_str.SetError(sbs_error, DNBError::SpringBoard); return; } } CFRelease((CFStringRef) attach_token); #endif } pid_t MachProcess::LaunchForDebug ( const char *path, char const *argv[], char const *envp[], const char *working_directory, // NULL => dont' change, non-NULL => set working directory for inferior to this const char *stdin_path, const char *stdout_path, const char *stderr_path, bool no_stdio, nub_launch_flavor_t launch_flavor, int disable_aslr, DNBError &launch_err ) { // Clear out and clean up from any current state Clear(); DNBLogThreadedIf(LOG_PROCESS, "%s( path = '%s', argv = %p, envp = %p, launch_flavor = %u, disable_aslr = %d )", __FUNCTION__, path, argv, envp, launch_flavor, disable_aslr); // Fork a child process for debugging SetState(eStateLaunching); switch (launch_flavor) { case eLaunchFlavorForkExec: m_pid = MachProcess::ForkChildForPTraceDebugging (path, argv, envp, this, launch_err); break; #ifdef WITH_SPRINGBOARD case eLaunchFlavorSpringBoard: { // .../whatever.app/whatever ? // Or .../com.apple.whatever.app/whatever -- be careful of ".app" in "com.apple.whatever" here const char *app_ext = strstr (path, ".app/"); if (app_ext == NULL) { // .../whatever.app ? int len = strlen (path); if (len > 5) { if (strcmp (path + len - 4, ".app") == 0) { app_ext = path + len - 4; } } } if (app_ext) { std::string app_bundle_path(path, app_ext + strlen(".app")); if (SBLaunchForDebug (app_bundle_path.c_str(), argv, envp, no_stdio, launch_err) != 0) return m_pid; // A successful SBLaunchForDebug() returns and assigns a non-zero m_pid. else break; // We tried a springboard launch, but didn't succeed lets get out } } // In case the executable name has a ".app" fragment which confuses our debugserver, // let's do an intentional fallthrough here... launch_flavor = eLaunchFlavorPosixSpawn; #endif case eLaunchFlavorPosixSpawn: m_pid = MachProcess::PosixSpawnChildForPTraceDebugging (path, DNBArchProtocol::GetArchitecture (), argv, envp, working_directory, stdin_path, stdout_path, stderr_path, no_stdio, this, disable_aslr, launch_err); break; default: // Invalid launch launch_err.SetError(NUB_GENERIC_ERROR, DNBError::Generic); return INVALID_NUB_PROCESS; } if (m_pid == INVALID_NUB_PROCESS) { // If we don't have a valid process ID and no one has set the error, // then return a generic error if (launch_err.Success()) launch_err.SetError(NUB_GENERIC_ERROR, DNBError::Generic); } else { m_path = path; size_t i; char const *arg; for (i=0; (arg = argv[i]) != NULL; i++) m_args.push_back(arg); m_task.StartExceptionThread(launch_err); if (launch_err.Fail()) { if (launch_err.AsString() == NULL) launch_err.SetErrorString("unable to start the exception thread"); DNBLog ("Could not get inferior's Mach exception port, sending ptrace PT_KILL and exiting."); ::ptrace (PT_KILL, m_pid, 0, 0); m_pid = INVALID_NUB_PROCESS; return INVALID_NUB_PROCESS; } StartSTDIOThread(); if (launch_flavor == eLaunchFlavorPosixSpawn) { SetState (eStateAttaching); errno = 0; int err = ::ptrace (PT_ATTACHEXC, m_pid, 0, 0); if (err == 0) { m_flags |= eMachProcessFlagsAttached; DNBLogThreadedIf(LOG_PROCESS, "successfully spawned pid %d", m_pid); launch_err.Clear(); } else { SetState (eStateExited); DNBError ptrace_err(errno, DNBError::POSIX); DNBLogThreadedIf(LOG_PROCESS, "error: failed to attach to spawned pid %d (err = %i, errno = %i (%s))", m_pid, err, ptrace_err.Error(), ptrace_err.AsString()); launch_err.SetError(NUB_GENERIC_ERROR, DNBError::Generic); } } else { launch_err.Clear(); } } return m_pid; } pid_t MachProcess::PosixSpawnChildForPTraceDebugging ( const char *path, cpu_type_t cpu_type, char const *argv[], char const *envp[], const char *working_directory, const char *stdin_path, const char *stdout_path, const char *stderr_path, bool no_stdio, MachProcess* process, int disable_aslr, DNBError& err ) { posix_spawnattr_t attr; short flags; DNBLogThreadedIf(LOG_PROCESS, "%s ( path='%s', argv=%p, envp=%p, working_dir=%s, stdin=%s, stdout=%s stderr=%s, no-stdio=%i)", __FUNCTION__, path, argv, envp, working_directory, stdin_path, stdout_path, stderr_path, no_stdio); err.SetError( ::posix_spawnattr_init (&attr), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit(LOG_PROCESS)) err.LogThreaded("::posix_spawnattr_init ( &attr )"); if (err.Fail()) return INVALID_NUB_PROCESS; flags = POSIX_SPAWN_START_SUSPENDED | POSIX_SPAWN_SETSIGDEF | POSIX_SPAWN_SETSIGMASK; if (disable_aslr) flags |= _POSIX_SPAWN_DISABLE_ASLR; sigset_t no_signals; sigset_t all_signals; sigemptyset (&no_signals); sigfillset (&all_signals); ::posix_spawnattr_setsigmask(&attr, &no_signals); ::posix_spawnattr_setsigdefault(&attr, &all_signals); err.SetError( ::posix_spawnattr_setflags (&attr, flags), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit(LOG_PROCESS)) err.LogThreaded("::posix_spawnattr_setflags ( &attr, POSIX_SPAWN_START_SUSPENDED%s )", flags & _POSIX_SPAWN_DISABLE_ASLR ? " | _POSIX_SPAWN_DISABLE_ASLR" : ""); if (err.Fail()) return INVALID_NUB_PROCESS; // Don't do this on SnowLeopard, _sometimes_ the TASK_BASIC_INFO will fail // and we will fail to continue with our process... // On SnowLeopard we should set "DYLD_NO_PIE" in the inferior environment.... #if !defined(__arm__) // We don't need to do this for ARM, and we really shouldn't now that we // have multiple CPU subtypes and no posix_spawnattr call that allows us // to set which CPU subtype to launch... if (cpu_type != 0) { size_t ocount = 0; err.SetError( ::posix_spawnattr_setbinpref_np (&attr, 1, &cpu_type, &ocount), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit(LOG_PROCESS)) err.LogThreaded("::posix_spawnattr_setbinpref_np ( &attr, 1, cpu_type = 0x%8.8x, count => %llu )", cpu_type, (uint64_t)ocount); if (err.Fail() != 0 || ocount != 1) return INVALID_NUB_PROCESS; } #endif PseudoTerminal pty; posix_spawn_file_actions_t file_actions; err.SetError( ::posix_spawn_file_actions_init (&file_actions), DNBError::POSIX); int file_actions_valid = err.Success(); if (!file_actions_valid || DNBLogCheckLogBit(LOG_PROCESS)) err.LogThreaded("::posix_spawn_file_actions_init ( &file_actions )"); int pty_error = -1; pid_t pid = INVALID_NUB_PROCESS; if (file_actions_valid) { if (stdin_path == NULL && stdout_path == NULL && stderr_path == NULL && !no_stdio) { pty_error = pty.OpenFirstAvailableMaster(O_RDWR|O_NOCTTY); if (pty_error == PseudoTerminal::success) { stdin_path = stdout_path = stderr_path = pty.SlaveName(); } } // if no_stdio or std paths not supplied, then route to "/dev/null". if (no_stdio || stdin_path == NULL || stdin_path[0] == '\0') stdin_path = "/dev/null"; if (no_stdio || stdout_path == NULL || stdout_path[0] == '\0') stdout_path = "/dev/null"; if (no_stdio || stderr_path == NULL || stderr_path[0] == '\0') stderr_path = "/dev/null"; err.SetError( ::posix_spawn_file_actions_addopen (&file_actions, STDIN_FILENO, stdin_path, O_RDONLY | O_NOCTTY, 0), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit (LOG_PROCESS)) err.LogThreaded ("::posix_spawn_file_actions_addopen (&file_actions, filedes=STDIN_FILENO, path='%s')", stdin_path); err.SetError( ::posix_spawn_file_actions_addopen (&file_actions, STDOUT_FILENO, stdout_path, O_WRONLY | O_NOCTTY | O_CREAT, 0640), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit (LOG_PROCESS)) err.LogThreaded ("::posix_spawn_file_actions_addopen (&file_actions, filedes=STDOUT_FILENO, path='%s')", stdout_path); err.SetError( ::posix_spawn_file_actions_addopen (&file_actions, STDERR_FILENO, stderr_path, O_WRONLY | O_NOCTTY | O_CREAT, 0640), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit (LOG_PROCESS)) err.LogThreaded ("::posix_spawn_file_actions_addopen (&file_actions, filedes=STDERR_FILENO, path='%s')", stderr_path); // TODO: Verify if we can set the working directory back immediately // after the posix_spawnp call without creating a race condition??? if (working_directory) ::chdir (working_directory); err.SetError( ::posix_spawnp (&pid, path, &file_actions, &attr, (char * const*)argv, (char * const*)envp), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit(LOG_PROCESS)) err.LogThreaded("::posix_spawnp ( pid => %i, path = '%s', file_actions = %p, attr = %p, argv = %p, envp = %p )", pid, path, &file_actions, &attr, argv, envp); } else { // TODO: Verify if we can set the working directory back immediately // after the posix_spawnp call without creating a race condition??? if (working_directory) ::chdir (working_directory); err.SetError( ::posix_spawnp (&pid, path, NULL, &attr, (char * const*)argv, (char * const*)envp), DNBError::POSIX); if (err.Fail() || DNBLogCheckLogBit(LOG_PROCESS)) err.LogThreaded("::posix_spawnp ( pid => %i, path = '%s', file_actions = %p, attr = %p, argv = %p, envp = %p )", pid, path, NULL, &attr, argv, envp); } // We have seen some cases where posix_spawnp was returning a valid // looking pid even when an error was returned, so clear it out if (err.Fail()) pid = INVALID_NUB_PROCESS; if (pty_error == 0) { if (process != NULL) { int master_fd = pty.ReleaseMasterFD(); process->SetChildFileDescriptors(master_fd, master_fd, master_fd); } } ::posix_spawnattr_destroy (&attr); if (pid != INVALID_NUB_PROCESS) { cpu_type_t pid_cpu_type = MachProcess::GetCPUTypeForLocalProcess (pid); DNBLogThreadedIf(LOG_PROCESS, "MachProcess::%s ( ) pid=%i, cpu_type=0x%8.8x", __FUNCTION__, pid, pid_cpu_type); if (pid_cpu_type) DNBArchProtocol::SetArchitecture (pid_cpu_type); } if (file_actions_valid) { DNBError err2; err2.SetError( ::posix_spawn_file_actions_destroy (&file_actions), DNBError::POSIX); if (err2.Fail() || DNBLogCheckLogBit(LOG_PROCESS)) err2.LogThreaded("::posix_spawn_file_actions_destroy ( &file_actions )"); } return pid; } uint32_t MachProcess::GetCPUTypeForLocalProcess (pid_t pid) { int mib[CTL_MAXNAME]={0,}; size_t len = CTL_MAXNAME; if (::sysctlnametomib("sysctl.proc_cputype", mib, &len)) return 0; mib[len] = pid; len++; cpu_type_t cpu; size_t cpu_len = sizeof(cpu); if (::sysctl (mib, len, &cpu, &cpu_len, 0, 0)) cpu = 0; return cpu; } pid_t MachProcess::ForkChildForPTraceDebugging ( const char *path, char const *argv[], char const *envp[], MachProcess* process, DNBError& launch_err ) { PseudoTerminal::Error pty_error = PseudoTerminal::success; // Use a fork that ties the child process's stdin/out/err to a pseudo // terminal so we can read it in our MachProcess::STDIOThread // as unbuffered io. PseudoTerminal pty; pid_t pid = pty.Fork(pty_error); if (pid < 0) { //-------------------------------------------------------------- // Error during fork. //-------------------------------------------------------------- return pid; } else if (pid == 0) { //-------------------------------------------------------------- // Child process //-------------------------------------------------------------- ::ptrace (PT_TRACE_ME, 0, 0, 0); // Debug this process ::ptrace (PT_SIGEXC, 0, 0, 0); // Get BSD signals as mach exceptions // If our parent is setgid, lets make sure we don't inherit those // extra powers due to nepotism. if (::setgid (getgid ()) == 0) { // Let the child have its own process group. We need to execute // this call in both the child and parent to avoid a race condition // between the two processes. ::setpgid (0, 0); // Set the child process group to match its pid // Sleep a bit to before the exec call ::sleep (1); // Turn this process into ::execv (path, (char * const *)argv); } // Exit with error code. Child process should have taken // over in above exec call and if the exec fails it will // exit the child process below. ::exit (127); } else { //-------------------------------------------------------------- // Parent process //-------------------------------------------------------------- // Let the child have its own process group. We need to execute // this call in both the child and parent to avoid a race condition // between the two processes. ::setpgid (pid, pid); // Set the child process group to match its pid if (process != NULL) { // Release our master pty file descriptor so the pty class doesn't // close it and so we can continue to use it in our STDIO thread int master_fd = pty.ReleaseMasterFD(); process->SetChildFileDescriptors(master_fd, master_fd, master_fd); } } return pid; } #ifdef WITH_SPRINGBOARD pid_t MachProcess::SBLaunchForDebug (const char *path, char const *argv[], char const *envp[], bool no_stdio, DNBError &launch_err) { // Clear out and clean up from any current state Clear(); DNBLogThreadedIf(LOG_PROCESS, "%s( '%s', argv)", __FUNCTION__, path); // Fork a child process for debugging SetState(eStateLaunching); m_pid = MachProcess::SBForkChildForPTraceDebugging(path, argv, envp, no_stdio, this, launch_err); if (m_pid != 0) { m_flags |= eMachProcessFlagsUsingSBS; m_path = path; size_t i; char const *arg; for (i=0; (arg = argv[i]) != NULL; i++) m_args.push_back(arg); m_task.StartExceptionThread(launch_err); if (launch_err.Fail()) { if (launch_err.AsString() == NULL) launch_err.SetErrorString("unable to start the exception thread"); DNBLog ("Could not get inferior's Mach exception port, sending ptrace PT_KILL and exiting."); ::ptrace (PT_KILL, m_pid, 0, 0); m_pid = INVALID_NUB_PROCESS; return INVALID_NUB_PROCESS; } StartSTDIOThread(); SetState (eStateAttaching); int err = ::ptrace (PT_ATTACHEXC, m_pid, 0, 0); if (err == 0) { m_flags |= eMachProcessFlagsAttached; DNBLogThreadedIf(LOG_PROCESS, "successfully attached to pid %d", m_pid); } else { SetState (eStateExited); DNBLogThreadedIf(LOG_PROCESS, "error: failed to attach to pid %d", m_pid); } } return m_pid; } #include // This returns a CFRetained pointer to the Bundle ID for app_bundle_path, // or NULL if there was some problem getting the bundle id. static CFStringRef CopyBundleIDForPath (const char *app_bundle_path, DNBError &err_str) { CFBundle bundle(app_bundle_path); CFStringRef bundleIDCFStr = bundle.GetIdentifier(); std::string bundleID; if (CFString::UTF8(bundleIDCFStr, bundleID) == NULL) { struct stat app_bundle_stat; char err_msg[PATH_MAX]; if (::stat (app_bundle_path, &app_bundle_stat) < 0) { err_str.SetError(errno, DNBError::POSIX); snprintf(err_msg, sizeof(err_msg), "%s: \"%s\"", err_str.AsString(), app_bundle_path); err_str.SetErrorString(err_msg); DNBLogThreadedIf(LOG_PROCESS, "%s() error: %s", __FUNCTION__, err_msg); } else { err_str.SetError(-1, DNBError::Generic); snprintf(err_msg, sizeof(err_msg), "failed to extract CFBundleIdentifier from %s", app_bundle_path); err_str.SetErrorString(err_msg); DNBLogThreadedIf(LOG_PROCESS, "%s() error: failed to extract CFBundleIdentifier from '%s'", __FUNCTION__, app_bundle_path); } return NULL; } DNBLogThreadedIf(LOG_PROCESS, "%s() extracted CFBundleIdentifier: %s", __FUNCTION__, bundleID.c_str()); CFRetain (bundleIDCFStr); return bundleIDCFStr; } pid_t MachProcess::SBForkChildForPTraceDebugging (const char *app_bundle_path, char const *argv[], char const *envp[], bool no_stdio, MachProcess* process, DNBError &launch_err) { DNBLogThreadedIf(LOG_PROCESS, "%s( '%s', argv, %p)", __FUNCTION__, app_bundle_path, process); CFAllocatorRef alloc = kCFAllocatorDefault; if (argv[0] == NULL) return INVALID_NUB_PROCESS; size_t argc = 0; // Count the number of arguments while (argv[argc] != NULL) argc++; // Enumerate the arguments size_t first_launch_arg_idx = 1; CFReleaser launch_argv; if (argv[first_launch_arg_idx]) { size_t launch_argc = argc > 0 ? argc - 1 : 0; launch_argv.reset (::CFArrayCreateMutable (alloc, launch_argc, &kCFTypeArrayCallBacks)); size_t i; char const *arg; CFString launch_arg; for (i=first_launch_arg_idx; (i < argc) && ((arg = argv[i]) != NULL); i++) { launch_arg.reset(::CFStringCreateWithCString (alloc, arg, kCFStringEncodingUTF8)); if (launch_arg.get() != NULL) CFArrayAppendValue(launch_argv.get(), launch_arg.get()); else break; } } // Next fill in the arguments dictionary. Note, the envp array is of the form // Variable=value but SpringBoard wants a CF dictionary. So we have to convert // this here. CFReleaser launch_envp; if (envp[0]) { launch_envp.reset(::CFDictionaryCreateMutable(alloc, 0, &kCFTypeDictionaryKeyCallBacks, &kCFTypeDictionaryValueCallBacks)); const char *value; int name_len; CFString name_string, value_string; for (int i = 0; envp[i] != NULL; i++) { value = strstr (envp[i], "="); // If the name field is empty or there's no =, skip it. Somebody's messing with us. if (value == NULL || value == envp[i]) continue; name_len = value - envp[i]; // Now move value over the "=" value++; name_string.reset(::CFStringCreateWithBytes(alloc, (const UInt8 *) envp[i], name_len, kCFStringEncodingUTF8, false)); value_string.reset(::CFStringCreateWithCString(alloc, value, kCFStringEncodingUTF8)); CFDictionarySetValue (launch_envp.get(), name_string.get(), value_string.get()); } } CFString stdio_path; PseudoTerminal pty; if (!no_stdio) { PseudoTerminal::Error pty_err = pty.OpenFirstAvailableMaster(O_RDWR|O_NOCTTY); if (pty_err == PseudoTerminal::success) { const char* slave_name = pty.SlaveName(); DNBLogThreadedIf(LOG_PROCESS, "%s() successfully opened master pty, slave is %s", __FUNCTION__, slave_name); if (slave_name && slave_name[0]) { ::chmod (slave_name, S_IRWXU | S_IRWXG | S_IRWXO); stdio_path.SetFileSystemRepresentation (slave_name); } } } if (stdio_path.get() == NULL) { stdio_path.SetFileSystemRepresentation ("/dev/null"); } CFStringRef bundleIDCFStr = CopyBundleIDForPath (app_bundle_path, launch_err); if (bundleIDCFStr == NULL) return INVALID_NUB_PROCESS; std::string bundleID; CFString::UTF8(bundleIDCFStr, bundleID); // Find SpringBoard SBSApplicationLaunchError sbs_error = 0; sbs_error = SBSLaunchApplicationForDebugging (bundleIDCFStr, (CFURLRef)NULL, // openURL launch_argv.get(), launch_envp.get(), // CFDictionaryRef environment stdio_path.get(), stdio_path.get(), SBSApplicationLaunchWaitForDebugger | SBSApplicationLaunchUnlockDevice); launch_err.SetError(sbs_error, DNBError::SpringBoard); if (sbs_error == SBSApplicationLaunchErrorSuccess) { static const useconds_t pid_poll_interval = 200000; static const useconds_t pid_poll_timeout = 30000000; useconds_t pid_poll_total = 0; nub_process_t pid = INVALID_NUB_PROCESS; Boolean pid_found = SBSProcessIDForDisplayIdentifier(bundleIDCFStr, &pid); // Poll until the process is running, as long as we are getting valid responses and the timeout hasn't expired // A return PID of 0 means the process is not running, which may be because it hasn't been (asynchronously) started // yet, or that it died very quickly (if you weren't using waitForDebugger). while (!pid_found && pid_poll_total < pid_poll_timeout) { usleep (pid_poll_interval); pid_poll_total += pid_poll_interval; DNBLogThreadedIf(LOG_PROCESS, "%s() polling Springboard for pid for %s...", __FUNCTION__, bundleID.c_str()); pid_found = SBSProcessIDForDisplayIdentifier(bundleIDCFStr, &pid); } CFRelease (bundleIDCFStr); if (pid_found) { if (process != NULL) { // Release our master pty file descriptor so the pty class doesn't // close it and so we can continue to use it in our STDIO thread int master_fd = pty.ReleaseMasterFD(); process->SetChildFileDescriptors(master_fd, master_fd, master_fd); } DNBLogThreadedIf(LOG_PROCESS, "%s() => pid = %4.4x", __FUNCTION__, pid); } else { DNBLogError("failed to lookup the process ID for CFBundleIdentifier %s.", bundleID.c_str()); } return pid; } DNBLogError("unable to launch the application with CFBundleIdentifier '%s' sbs_error = %u", bundleID.c_str(), sbs_error); return INVALID_NUB_PROCESS; } #endif // #ifdef WITH_SPRINGBOARD