llvm-project/lldb/examples/python/process_events.py

255 lines
14 KiB
Python
Raw Normal View History

#!/usr/bin/python
#----------------------------------------------------------------------
# Be sure to add the python path that points to the LLDB shared library.
# On MacOSX csh, tcsh:
# setenv PYTHONPATH /Applications/Xcode.app/Contents/SharedFrameworks/LLDB.framework/Resources/Python
# On MacOSX sh, bash:
# export PYTHONPATH=/Applications/Xcode.app/Contents/SharedFrameworks/LLDB.framework/Resources/Python
#----------------------------------------------------------------------
import commands
import optparse
import os
import platform
import sys
#----------------------------------------------------------------------
# Code that auto imports LLDB
#----------------------------------------------------------------------
try:
# Just try for LLDB in case PYTHONPATH is already correctly setup
import lldb
except ImportError:
lldb_python_dirs = list()
# lldb is not in the PYTHONPATH, try some defaults for the current platform
platform_system = platform.system()
if platform_system == 'Darwin':
# On Darwin, try the currently selected Xcode directory
xcode_dir = commands.getoutput("xcode-select --print-path")
if xcode_dir:
lldb_python_dirs.append(os.path.realpath(xcode_dir + '/../SharedFrameworks/LLDB.framework/Resources/Python'))
lldb_python_dirs.append(xcode_dir + '/Library/PrivateFrameworks/LLDB.framework/Resources/Python')
lldb_python_dirs.append('/System/Library/PrivateFrameworks/LLDB.framework/Resources/Python')
success = False
for lldb_python_dir in lldb_python_dirs:
if os.path.exists(lldb_python_dir):
if not (sys.path.__contains__(lldb_python_dir)):
sys.path.append(lldb_python_dir)
try:
import lldb
except ImportError:
pass
else:
print 'imported lldb from: "%s"' % (lldb_python_dir)
success = True
break
if not success:
print "error: couldn't locate the 'lldb' module, please set PYTHONPATH correctly"
sys.exit(1)
def print_threads(process, options):
if options.show_threads:
for thread in process:
print '%s %s' % (thread, thread.GetFrameAtIndex(0))
def run_commands(command_interpreter, commands):
return_obj = lldb.SBCommandReturnObject()
for command in commands:
command_interpreter.HandleCommand( command, return_obj )
if return_obj.Succeeded():
print return_obj.GetOutput()
else:
print return_obj
if options.stop_on_error:
break
def main(argv):
description='''Debugs a program using the LLDB python API and uses asynchronous broadcast events to watch for process state changes.'''
epilog='''Examples:
#----------------------------------------------------------------------
# Run "/bin/ls" with the arguments "-lAF /tmp/", and set a breakpoint
# at "malloc" and backtrace and read all registers each time we stop
#----------------------------------------------------------------------
% ./process_events.py --breakpoint malloc --stop-command bt --stop-command 'register read' -- /bin/ls -lAF /tmp/
'''
optparse.OptionParser.format_epilog = lambda self, formatter: self.epilog
parser = optparse.OptionParser(description=description, prog='process_events',usage='usage: process_events [options] program [arg1 arg2]', epilog=epilog)
parser.add_option('-v', '--verbose', action='store_true', dest='verbose', help="Enable verbose logging.", default=False)
parser.add_option('-b', '--breakpoint', action='append', type='string', metavar='BPEXPR', dest='breakpoints', help='Breakpoint commands to create after the target has been created, the values will be sent to the "_regexp-break" command which supports breakpoints by name, file:line, and address.')
parser.add_option('-a', '--arch', type='string', dest='arch', help='The architecture to use when creating the debug target.', default=None)
parser.add_option('-l', '--launch-command', action='append', type='string', metavar='CMD', dest='launch_commands', help='LLDB command interpreter commands to run once after the process has launched. This option can be specified more than once.', default=[])
parser.add_option('-s', '--stop-command', action='append', type='string', metavar='CMD', dest='stop_commands', help='LLDB command interpreter commands to run each time the process stops. This option can be specified more than once.', default=[])
parser.add_option('-c', '--crash-command', action='append', type='string', metavar='CMD', dest='crash_commands', help='LLDB command interpreter commands to run in case the process crashes. This option can be specified more than once.', default=[])
parser.add_option('-x', '--exit-command', action='append', type='string', metavar='CMD', dest='exit_commands', help='LLDB command interpreter commands to run once after the process has exited. This option can be specified more than once.', default=[])
parser.add_option('-T', '--no-threads', action='store_false', dest='show_threads', help="Don't show threads when process stops.", default=True)
parser.add_option('--ignore-errors', action='store_false', dest='stop_on_error', help="Don't stop executing LLDB commands if the command returns an error. This applies to all of the LLDB command interpreter commands that get run for launch, stop, crash and exit.", default=True)
parser.add_option('-n', '--run-count', type='int', dest='run_count', metavar='N', help='How many times to run the process in case the process exits.', default=1)
parser.add_option('-t', '--event-timeout', type='int', dest='event_timeout', metavar='SEC', help='Specify the timeout in seconds to wait for process state change events.', default=lldb.UINT32_MAX)
parser.add_option('-e', '--environment', action='append', type='string', metavar='ENV', dest='env_vars', help='Environment variables to set in the inferior process when launching a process.')
parser.add_option('-d', '--working-dir', type='string', metavar='DIR', dest='working_dir', help='The the current working directory when launching a process.', default=None)
parser.add_option('-p', '--attach-pid', type='int', dest='attach_pid', metavar='PID', help='Specify a process to attach to by process ID.', default=-1)
parser.add_option('-P', '--attach-name', type='string', dest='attach_name', metavar='PROCESSNAME', help='Specify a process to attach to by name.', default=None)
parser.add_option('-w', '--attach-wait', action='store_true', dest='attach_wait', help='Wait for the next process to launch when attaching to a process by name.', default=False)
try:
(options, args) = parser.parse_args(argv)
except:
return
attach_info = None
launch_info = None
exe = None
if args:
exe = args.pop(0)
launch_info = lldb.SBLaunchInfo (args)
if options.env_vars:
launch_info.SetEnvironmentEntries(options.env_vars, True)
if options.working_dir:
launch_info.SetWorkingDirectory(options.working_dir)
elif options.attach_pid != -1:
if options.run_count == 1:
attach_info = lldb.SBAttachInfo (options.attach_pid)
else:
print "error: --run-count can't be used with the --attach-pid option"
sys.exit(1)
elif not options.attach_name is None:
if options.run_count == 1:
attach_info = lldb.SBAttachInfo (options.attach_name, options.attach_wait)
else:
print "error: --run-count can't be used with the --attach-name option"
sys.exit(1)
else:
print 'error: a program path for a program to debug and its arguments are required'
sys.exit(1)
# Create a new debugger instance
debugger = lldb.SBDebugger.Create()
command_interpreter = debugger.GetCommandInterpreter()
# Create a target from a file and arch
if exe:
print "Creating a target for '%s'" % exe
target = debugger.CreateTargetWithFileAndArch (exe, options.arch)
if target:
# Set any breakpoints that were specified in the args if we are launching
if launch_info and options.breakpoints:
for bp in options.breakpoints:
debugger.HandleCommand( "_regexp-break %s" % (bp))
run_commands(command_interpreter, ['breakpoint list'])
for run_idx in range(options.run_count):
# Launch the process. Since we specified synchronous mode, we won't return
# from this function until we hit the breakpoint at main
error = lldb.SBError()
if launch_info:
if options.run_count == 1:
print 'Launching "%s"...' % (exe)
else:
print 'Launching "%s"... (launch %u of %u)' % (exe, run_idx + 1, options.run_count)
process = target.Launch (launch_info, error)
else:
if options.attach_pid != -1:
print 'Attaching to process %i...' % (options.attach_pid)
else:
if options.attach_wait:
print 'Waiting for next to process named "%s" to launch...' % (options.attach_name)
else:
print 'Attaching to existing process named "%s"...' % (options.attach_name)
process = target.Attach (attach_info, error)
# Make sure the launch went ok
if process and process.GetProcessID() != lldb.LLDB_INVALID_PROCESS_ID:
pid = process.GetProcessID()
listener = lldb.SBListener("event_listener")
# sign up for process state change events
process.GetBroadcaster().AddListener(listener, lldb.SBProcess.eBroadcastBitStateChanged)
stop_idx = 0
done = False
while not done:
event = lldb.SBEvent()
if listener.WaitForEvent (options.event_timeout, event):
state = lldb.SBProcess.GetStateFromEvent (event)
Initial checkin of a new project: LLDB Performance Testing Infrastructure This is a very basic implementation of a library that easily allows to drive LLDB.framework to write test cases for performance This is separate from the LLDB testsuite in test/ in that: a) this uses C++ instead of Python to avoid measures being affected by SWIG b) this is in very early development and needs lots of tweaking before it can be considered functionally complete c) this is not meant to test correctness but to help catch performance regressions There is a sample application built against the library (in darwin/sketch) that uses the famous sample app Sketch as an inferior to measure certain basic parameters of LLDB's behavior. The resulting output is a PLIST much like the following: <?xml version="1.0" encoding="UTF-8"?> <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd"> <plist version="1.0"> <array> <dict> <key>fetch-frames</key> <real>0.13161715522222225</real> </dict> <dict> <key>file-line-bkpt</key> <real>0.029111678750000002</real> </dict> <dict> <key>fetch-modules</key> <real>0.00026376766666666668</real> </dict> <dict> <key>fetch-vars</key> <real>0.17820429311111111</real> </dict> <dict> <key>run-expr</key> <real>0.029676525769230768</real> </dict> </array> </plist> Areas for improvement: - code cleanups (I will be out of the office for a couple days this coming week, but please keep ideas coming!) - more metrics and test cases - better error checking This toolkit also comprises a simple event-loop-driven controller for LLDB, similar yet much simpler to what the Driver does to implement the lldb command-line tool. llvm-svn: 176715
2013-03-09 04:29:13 +08:00
print "event %s" % (lldb.SBDebugger.StateAsCString(state))
if state == lldb.eStateStopped:
if stop_idx == 0:
if launch_info:
print "process %u launched" % (pid)
else:
print "attached to process %u" % (pid)
for m in target.modules:
print m
if options.breakpoints:
for bp in options.breakpoints:
debugger.HandleCommand( "_regexp-break %s" % (bp))
run_commands(command_interpreter, ['breakpoint list'])
run_commands (command_interpreter, options.launch_commands)
else:
if options.verbose:
print "process %u stopped" % (pid)
run_commands (command_interpreter, options.stop_commands)
stop_idx += 1
print_threads (process, options)
Initial checkin of a new project: LLDB Performance Testing Infrastructure This is a very basic implementation of a library that easily allows to drive LLDB.framework to write test cases for performance This is separate from the LLDB testsuite in test/ in that: a) this uses C++ instead of Python to avoid measures being affected by SWIG b) this is in very early development and needs lots of tweaking before it can be considered functionally complete c) this is not meant to test correctness but to help catch performance regressions There is a sample application built against the library (in darwin/sketch) that uses the famous sample app Sketch as an inferior to measure certain basic parameters of LLDB's behavior. The resulting output is a PLIST much like the following: <?xml version="1.0" encoding="UTF-8"?> <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd"> <plist version="1.0"> <array> <dict> <key>fetch-frames</key> <real>0.13161715522222225</real> </dict> <dict> <key>file-line-bkpt</key> <real>0.029111678750000002</real> </dict> <dict> <key>fetch-modules</key> <real>0.00026376766666666668</real> </dict> <dict> <key>fetch-vars</key> <real>0.17820429311111111</real> </dict> <dict> <key>run-expr</key> <real>0.029676525769230768</real> </dict> </array> </plist> Areas for improvement: - code cleanups (I will be out of the office for a couple days this coming week, but please keep ideas coming!) - more metrics and test cases - better error checking This toolkit also comprises a simple event-loop-driven controller for LLDB, similar yet much simpler to what the Driver does to implement the lldb command-line tool. llvm-svn: 176715
2013-03-09 04:29:13 +08:00
print "continuing process %u" % (pid)
process.Continue()
elif state == lldb.eStateExited:
exit_desc = process.GetExitDescription()
if exit_desc:
print "process %u exited with status %u: %s" % (pid, process.GetExitStatus (), exit_desc)
else:
print "process %u exited with status %u" % (pid, process.GetExitStatus ())
run_commands (command_interpreter, options.exit_commands)
done = True
elif state == lldb.eStateCrashed:
print "process %u crashed" % (pid)
print_threads (process, options)
run_commands (command_interpreter, options.crash_commands)
done = True
elif state == lldb.eStateDetached:
print "process %u detached" % (pid)
done = True
elif state == lldb.eStateRunning:
# process is running, don't say anything, we will always get one of these after resuming
if options.verbose:
print "process %u resumed" % (pid)
elif state == lldb.eStateUnloaded:
print "process %u unloaded, this shouldn't happen" % (pid)
done = True
elif state == lldb.eStateConnected:
print "process connected"
elif state == lldb.eStateAttaching:
print "process attaching"
elif state == lldb.eStateLaunching:
print "process launching"
else:
# timeout waiting for an event
print "no process event for %u seconds, killing the process..." % (options.event_timeout)
done = True
process.Kill() # kill the process
else:
if error:
print error
else:
if launch_info:
print 'error: launch failed'
else:
print 'error: attach failed'
lldb.SBDebugger.Terminate()
if __name__ == '__main__':
main(sys.argv[1:])