blob: 0c65aa7ab0a7ea828da2afc304857f437cb1209d [file] [log] [blame]
# Copyright 2008 Google Inc, Martin J. Bligh <>,
# Benjamin Poirier, Ryan Stutsman
# Released under the GPL v2
Miscellaneous small functions.
DO NOT import this file directly - it is mixed in by server/,
import that instead
import atexit, os, re, shutil, textwrap, sys, tempfile, types
from autotest_lib.client.common_lib import barrier, utils
from autotest_lib.server import subcommand
# A dictionary of pid and a list of tmpdirs for that pid
__tmp_dirs = {}
def scp_remote_escape(filename):
Escape special characters from a filename so that it can be passed
to scp (within double quotes) as a remote file.
Bis-quoting has to be used with scp for remote files, "bis-quoting"
as in quoting x 2
scp does not support a newline in the filename
filename: the filename string to escape.
The escaped filename string. The required englobing double
quotes are NOT added and so should be added at some point by
the caller.
escape_chars= r' !"$&' "'" r'()*,:;<=>?[\]^`{|}'
new_name= []
for char in filename:
if char in escape_chars:
new_name.append("\\%s" % (char,))
return utils.sh_escape("".join(new_name))
def get(location, local_copy = False):
"""Get a file or directory to a local temporary directory.
location: the source of the material to get. This source may
be one of:
* a local file or directory
* a URL (http or ftp)
* a python file-like object
The location of the file or directory where the requested
content was saved. This will be contained in a temporary
directory on the local host. If the material to get was a
directory, the location will contain a trailing '/'
tmpdir = get_tmp_dir()
# location is a file-like object
if hasattr(location, "read"):
tmpfile = os.path.join(tmpdir, "file")
tmpfileobj = file(tmpfile, 'w')
shutil.copyfileobj(location, tmpfileobj)
return tmpfile
if isinstance(location, types.StringTypes):
# location is a URL
if location.startswith('http') or location.startswith('ftp'):
tmpfile = os.path.join(tmpdir, os.path.basename(location))
utils.urlretrieve(location, tmpfile)
return tmpfile
# location is a local path
elif os.path.exists(os.path.abspath(location)):
if not local_copy:
if os.path.isdir(location):
return location.rstrip('/') + '/'
return location
tmpfile = os.path.join(tmpdir, os.path.basename(location))
if os.path.isdir(location):
tmpfile += '/'
shutil.copytree(location, tmpfile, symlinks=True)
return tmpfile
shutil.copyfile(location, tmpfile)
return tmpfile
# location is just a string, dump it to a file
tmpfd, tmpfile = tempfile.mkstemp(dir=tmpdir)
tmpfileobj = os.fdopen(tmpfd, 'w')
return tmpfile
def get_tmp_dir():
"""Return the pathname of a directory on the host suitable
for temporary file storage.
The directory and its content will be deleted automatically
at the end of the program execution if they are still present.
dir_name = tempfile.mkdtemp(prefix="autoserv-")
pid = os.getpid()
if not pid in __tmp_dirs:
__tmp_dirs[pid] = []
return dir_name
def __clean_tmp_dirs():
"""Erase temporary directories that were created by the get_tmp_dir()
function and that are still present.
pid = os.getpid()
if pid not in __tmp_dirs:
for dir in __tmp_dirs[pid]:
except OSError, e:
if e.errno == 2:
__tmp_dirs[pid] = []
subcommand.subcommand.register_join_hook(lambda _: __clean_tmp_dirs())
def unarchive(host, source_material):
"""Uncompress and untar an archive on a host.
If the "source_material" is compresses (according to the file
extension) it will be uncompressed. Supported compression formats
are gzip and bzip2. Afterwards, if the source_material is a tar
archive, it will be untarred.
host: the host object on which the archive is located
source_material: the path of the archive on the host
The file or directory name of the unarchived source material.
If the material is a tar archive, it will be extracted in the
directory where it is and the path returned will be the first
entry in the archive, assuming it is the topmost directory.
If the material is not an archive, nothing will be done so this
function is "harmless" when it is "useless".
# uncompress
if (source_material.endswith(".gz") or
source_material.endswith(".gzip")):'gunzip "%s"' % (utils.sh_escape(source_material)))
source_material= ".".join(source_material.split(".")[:-1])
elif source_material.endswith("bz2"):'bunzip2 "%s"' % (utils.sh_escape(source_material)))
source_material= ".".join(source_material.split(".")[:-1])
# untar
if source_material.endswith(".tar"):
retval='tar -C "%s" -xvf "%s"' % (
source_material= os.path.join(os.path.dirname(source_material),
return source_material
def get_server_dir():
path = os.path.dirname(sys.modules['autotest_lib.server.utils'].__file__)
return os.path.abspath(path)
def find_pid(command):
for line in utils.system_output('ps -eo pid,cmd').rstrip().split('\n'):
(pid, cmd) = line.split(None, 1)
if, cmd):
return int(pid)
return None
def nohup(command, stdout='/dev/null', stderr='/dev/null', background=True,
env = {}):
cmd = ' '.join(key+'='+val for key, val in env.iteritems())
cmd += ' nohup ' + command
cmd += ' > %s' % stdout
if stdout == stderr:
cmd += ' 2>&1'
cmd += ' 2> %s' % stderr
if background:
cmd += ' &'
def default_mappings(machines):
Returns a simple mapping in which all machines are assigned to the
same key. Provides the default behavior for
form_ntuples_from_machines. """
mappings = {}
failures = []
mach = machines[0]
mappings['ident'] = [mach]
if len(machines) > 1:
machines = machines[1:]
for machine in machines:
return (mappings, failures)
def form_ntuples_from_machines(machines, n=2, mapping_func=default_mappings):
"""Returns a set of ntuples from machines where the machines in an
ntuple are in the same mapping, and a set of failures which are
(machine name, reason) tuples."""
ntuples = []
(mappings, failures) = mapping_func(machines)
# now run through the mappings and create n-tuples.
# throw out the odd guys out
for key in mappings:
key_machines = mappings[key]
total_machines = len(key_machines)
# form n-tuples
while len(key_machines) >= n:
key_machines = key_machines[n:]
for mach in key_machines:
failures.append((mach, "machine can not be tupled"))
return (ntuples, failures)
def parse_machine(machine, user='root', password='', port=22):
Parse the machine string user:pass@host:port and return it separately,
if the machine string is not complete, use the default parameters
when appropriate.
if '@' in machine:
user, machine = machine.split('@', 1)
if ':' in user:
user, password = user.split(':', 1)
# Brackets are required to protect an IPv6 address whenever a
# [xx::xx]:port number (or a file [xx::xx]:/path/) is appended to
# it. Do not attempt to extract a (non-existent) port number from
# an unprotected/bare IPv6 address "xx::xx".
# In the Python >= 3.3 future, 'import ipaddress' will parse
# addresses; and maybe more.
bare_ipv6 = '[' != machine[0] and':.*:', machine)
# Extract trailing :port number if any.
if not bare_ipv6 and':\d*$', machine):
machine, port = machine.rsplit(':', 1)
port = int(port)
# Strip any IPv6 brackets (ssh does not support them).
# We'll add them back later for rsync, scp, etc.
if machine[0] == '[' and machine[-1] == ']':
machine = machine[1:-1]
if not machine or not user:
raise ValueError
return machine, user, password, port
def get_public_key():
Return a valid string ssh public key for the user executing autoserv or
autotest. If there's no DSA or RSA public key, create a DSA keypair with
ssh-keygen and return it.
ssh_conf_path = os.path.expanduser('~/.ssh')
dsa_public_key_path = os.path.join(ssh_conf_path, '')
dsa_private_key_path = os.path.join(ssh_conf_path, 'id_dsa')
rsa_public_key_path = os.path.join(ssh_conf_path, '')
rsa_private_key_path = os.path.join(ssh_conf_path, 'id_rsa')
has_dsa_keypair = os.path.isfile(dsa_public_key_path) and \
has_rsa_keypair = os.path.isfile(rsa_public_key_path) and \
if has_dsa_keypair:
print 'DSA keypair found, using it'
public_key_path = dsa_public_key_path
elif has_rsa_keypair:
print 'RSA keypair found, using it'
public_key_path = rsa_public_key_path
print 'Neither RSA nor DSA keypair found, creating DSA ssh key pair'
utils.system('ssh-keygen -t dsa -q -N "" -f %s' % dsa_private_key_path)
public_key_path = dsa_public_key_path
public_key = open(public_key_path, 'r')
public_key_str =
return public_key_str
def get_sync_control_file(control, host_name, host_num,
instance, num_jobs, port_base=63100):
This function is used when there is a need to run more than one
job simultaneously starting exactly at the same time. It basically returns
a modified control file (containing the synchronization code prepended)
whenever it is ready to run the control file. The synchronization
is done using barriers to make sure that the jobs start at the same time.
Here is how the synchronization is done to make sure that the tests
start at exactly the same time on the client.
sc_bar is a server barrier and s_bar, c_bar are the normal barriers
Job1 Job2 ...... JobN
Server: | sc_bar
Server: | s_bar ...... s_bar
Server: | ......
Client | sc_bar
Client | c_bar c_bar ...... c_bar
Client | <run test> <run test> ...... <run test>
@param control: The control file which to which the above synchronization
code will be prepended.
@param host_name: The host name on which the job is going to run.
@param host_num: (non negative) A number to identify the machine so that
we have different sets of s_bar_ports for each of the machines.
@param instance: The number of the job
@param num_jobs: Total number of jobs that are going to run in parallel
with this job starting at the same time.
@param port_base: Port number that is used to derive the actual barrier
@returns The modified control file.
sc_bar_port = port_base
c_bar_port = port_base
if host_num < 0:
print "Please provide a non negative number for the host"
return None
s_bar_port = port_base + 1 + host_num # The set of s_bar_ports are
# the same for a given machine
sc_bar_timeout = 180
s_bar_timeout = c_bar_timeout = 120
# The barrier code snippet is prepended into the conrol file
# dynamically before is called finally.
control_new = []
# jobid is the unique name used to identify the processes
# trying to reach the barriers
jobid = "%s#%d" % (host_name, instance)
rendv = []
# rendvstr is a temp holder for the rendezvous list of the processes
for n in range(num_jobs):
rendv.append("'%s#%d'" % (host_name, n))
rendvstr = ",".join(rendv)
if instance == 0:
# Do the setup and wait at the server barrier
# Clean up the tmp and the control dirs for the first instance
control_new.append('if os.path.exists(job.tmpdir):')
control_new.append("\t system('umount -f %s > /dev/null"
"2> /dev/null' % job.tmpdir,"
control_new.append("\t system('rm -rf ' + job.tmpdir)")
'b0 = job.barrier("%s", "sc_bar", %d, port=%d)'
% (jobid, sc_bar_timeout, sc_bar_port))
'b0.rendezvous_servers("PARALLEL_MASTER", "%s")'
% jobid)
elif instance == 1:
# Wait at the server barrier to wait for instance=0
# process to complete setup
b0 = barrier.barrier("PARALLEL_MASTER", "sc_bar", sc_bar_timeout,
b0.rendezvous_servers("PARALLEL_MASTER", jobid)
if(num_jobs > 2):
b1 = barrier.barrier(jobid, "s_bar", s_bar_timeout,
# For the rest of the clients
b2 = barrier.barrier(jobid, "s_bar", s_bar_timeout, port=s_bar_port)
# Client side barrier for all the tests to start at the same time
control_new.append('b1 = job.barrier("%s", "c_bar", %d, port=%d)'
% (jobid, c_bar_timeout, c_bar_port))
control_new.append("b1.rendezvous(%s)" % rendvstr)
# Stick in the rest of the control file
return "\n".join(control_new)