From: Michael Goldish <mgoldish@redhat.com>
To: Amos Kong <akong@redhat.com>
Cc: autotest@test.kernel.org, kvm@vger.kernel.org
Subject: Re: [Autotest] [KVM-AUTOTEST PATCH 13/14] KVM test: kvm_monitor.py: add QMP interface
Date: Mon, 14 Jun 2010 09:34:52 +0300 [thread overview]
Message-ID: <4C15CD8C.2060503@redhat.com> (raw)
In-Reply-To: <20100614023900.GA13579@z>
On 06/14/2010 05:39 AM, Amos Kong wrote:
> On Sun, Jun 13, 2010 at 05:33:44PM +0300, Michael Goldish wrote:
>> An initial QMP client implementation.
>> Should be fully functional and supports asynchronous events.
>> However, most tests must be modified to support it, because it returns output
>> in a different format from the human monitor (the human monitor returns strings
>> and the QMP one returns dicts or lists).
>>
>> To enable QMP, set main_monitor to a monitor whose monitor_type is "qmp".
>>
>> For example (a single QMP monitor):
>>
>> monitors = monitor1
>> monitor_type_monitor1 = qmp
>> main_monitor = monitor1
>>
>> Another example (multiple monitors, both human and QMP):
>>
>> monitors = MyMonitor SomeOtherMonitor YetAnotherMonitor # defines 3 monitors
>> monitor_type = human # default for all monitors
>> monitor_type_SomeOtherMonitor = qmp # applies only to SomeOtherMonitor
>> monitor_type_YetAnotherMonitor = qmp # applies only to YetAnotherMonitor
>> main_monitor = SomeOtherMonitor # the main monitor is a QMP one, so
>> # the test will use QMP
>>
>> Note:
>> Monitor methods now raise exceptions such as MonitorLockError and QMPCmdError.
>> If this turns out to be a bad idea, it shouldn't be hard to revert to the old
>> convention of returning a (status, output) tuple.
>>
>> Signed-off-by: Michael Goldish <mgoldish@redhat.com>
>> ---
>> client/tests/kvm/kvm_monitor.py | 275 +++++++++++++++++++++++++++++++++++++++
>> client/tests/kvm/kvm_vm.py | 6 +-
>> 2 files changed, 279 insertions(+), 2 deletions(-)
>>
>> diff --git a/client/tests/kvm/kvm_monitor.py b/client/tests/kvm/kvm_monitor.py
>> index c5cf9c3..76a1a83 100644
>> --- a/client/tests/kvm/kvm_monitor.py
>> +++ b/client/tests/kvm/kvm_monitor.py
>> @@ -6,6 +6,11 @@ Interfaces to the QEMU monitor.
>>
>> import socket, time, threading, logging
>> import kvm_utils
>> +try:
>> + import json
>> +except ImportError:
>> + logging.warning("Could not import json module. "
>> + "QMP monitor functionality disabled.")
>>
>>
>> class MonitorError(Exception):
>> @@ -28,6 +33,10 @@ class MonitorProtocolError(MonitorError):
>> pass
>>
>>
>> +class QMPCmdError(MonitorError):
>> + pass
>> +
>> +
>> class Monitor:
>> """
>> Common code for monitor classes.
>> @@ -114,6 +123,8 @@ class HumanMonitor(Monitor):
>> suppress_exceptions is False
>> @raise MonitorProtocolError: Raised if the initial (qemu) prompt isn't
>> found and suppress_exceptions is False
>> + @note: Other exceptions may be raised. See _get_command_output's
>> + docstring.
>> """
>> try:
>> Monitor.__init__(self, filename)
>> @@ -354,3 +365,267 @@ class HumanMonitor(Monitor):
>> @return: The command's output
>> """
>> return self._get_command_output("mouse_button %d" % state)
>> +
>> +
>> +class QMPMonitor(Monitor):
>> + """
>> + Wraps QMP monitor commands.
>> + """
>> +
>> + def __init__(self, filename, suppress_exceptions=False):
>> + """
>> + Connect to the monitor socket and issue the qmp_capabilities command
>> +
>> + @param filename: Monitor socket filename
>> + @raise MonitorConnectError: Raised if the connection fails and
>> + suppress_exceptions is False
>> + @note: Other exceptions may be raised if the qmp_capabilities command
>> + fails. See _get_command_output's docstring.
>> + """
>> + try:
>> + Monitor.__init__(self, filename)
>> +
>> + self.protocol = "qmp"
>> + self.events = []
>> +
>> + # Issue qmp_capabilities
>> + self._get_command_output("qmp_capabilities")
>> +
>> + except MonitorError, e:
>> + if suppress_exceptions:
>> + logging.warn(e)
>> + else:
>> + raise
>> +
>> +
>> + # Private methods
>> +
>> + def _build_cmd(self, cmd, args=None):
>> + obj = {"execute": cmd}
>> + if args:
>> + obj["arguments"] = args
>> + return obj
>> +
>> +
>> + def _read_objects(self, timeout=5):
>> + """
>> + Read lines from monitor and try to decode them.
>> + Stop when all available lines have been successfully decoded, or when
>> + timeout expires. If any decoded objects are asynchronous events, store
>> + them in self.events. Return all decoded objects.
>> +
>> + @param timeout: Time to wait for all lines to decode successfully
>> + @return: A list of objects
>> + """
>> + s = ""
>> + objs = []
>> + end_time = time.time() + timeout
>> + while time.time() < end_time:
>> + s += self._recvall()
>> + for line in s.splitlines():
>> + if not line:
>> + continue
>> + try:
>> + obj = json.loads(line)
>> + except:
>> + # Found an incomplete or broken line -- keep reading
>> + break
>> + objs += [obj]
>> + else:
>> + # All lines are OK -- stop reading
>> + break
>> + time.sleep(0.1)
>> + # Keep track of asynchronous events
>> + self.events += [obj for obj in objs if "event" in obj]
>> + return objs
>> +
>> +
>> + def _send_command(self, cmd, args=None):
>> + """
>> + Send command without waiting for response.
>> +
>> + @param cmd: Command to send
>> + @param args: A dict containing command arguments, or None
>> + @raise MonitorLockError: Raised if the lock cannot be acquired
>> + @raise MonitorSendError: Raised if the command cannot be sent
>> + """
>> + if not self._acquire_lock(20):
>> + raise MonitorLockError("Could not acquire exclusive lock to send "
>> + "QMP command '%s'" % cmd)
>> +
>> + try:
>> + cmdobj = self._build_cmd(cmd, args)
>> + try:
>> + self.socket.sendall(json.dumps(cmdobj) + "\n")
>> + except socket.error:
>> + raise MonitorSendError("Could not send QMP command '%s'" % cmd)
>> +
>> + finally:
>> + self.lock.release()
>> +
>> +
>> + def _get_command_output(self, cmd, args=None, timeout=20):
>> + """
>> + Send monitor command and wait for response.
>> +
>> + @param cmd: Command to send
>> + @param args: A dict containing command arguments, or None
>> + @param timeout: Time duration to wait for response
>> + @return: The response received
>> + @raise MonitorLockError: Raised if the lock cannot be acquired
>> + @raise MonitorSendError: Raised if the command cannot be sent
>> + @raise MonitorProtocolError: Raised if no response is received
>> + @raise QMPCmdError: Raised if the response is an error message
>> + (the exception's args are (msg, data) where msg is a string and
>> + data is the error data)
>> + """
>> + if not self._acquire_lock(20):
>> + raise MonitorLockError("Could not acquire exclusive lock to send "
>> + "QMP command '%s'" % cmd)
>> +
>> + try:
>> + # Read any data that might be available
>> + self._read_objects()
>> + # Send command
>> + self._send_command(cmd, args)
>> + # Read response
>> + end_time = time.time() + timeout
>> + while time.time() < end_time:
>> + for obj in self._read_objects():
>> + if "return" in obj:
>> + return obj["return"]
>> + elif "error" in obj:
>> + raise QMPCmdError("QMP command '%s' failed" % cmd,
>> + obj["error"])
>> + time.sleep(0.1)
>> + # No response found
>> + raise MonitorProtocolError("Received no response to QMP command "
>> + "'%s'" % cmd)
>> +
>> + finally:
>> + self.lock.release()
>> +
>> +
>> + # Public methods
>> +
>> + def is_responsive(self):
>> + """
>> + Make sure the monitor is responsive by sending a command.
>> +
>> + @return: True if responsive, False otherwise
>> + """
>> + try:
>> + self._get_command_output("query-version")
>> + return True
>> + except MonitorError:
>> + return False
>> +
>> +
>> + def get_events(self):
>> + """
>> + Return a list of the asynchronous events received since the last
>> + clear_events() call.
>> +
>> + @return: A list of events (the objects returned have an "event" key)
>> + @raise MonitorLockError: Raised if the lock cannot be acquired
>> + """
>> + if not self._acquire_lock(20):
>> + raise MonitorLockError("Could not acquire exclusive lock to read "
>> + "events from monitor")
>> + try:
>> + self._read_objects()
>> + return self.events[:]
>> + finally:
>> + self.lock.release()
>> +
>> +
>> + def clear_events(self):
>> + """
>> + Clear the list of asynchronous events.
>> +
>> + @raise MonitorLockError: Raised if the lock cannot be acquired
>> + """
>> + if not self._acquire_lock(20):
>> + raise MonitorLockError("Could not acquire exclusive lock to clear "
>> + "event list")
>> + self.events = []
>> + self.lock.release()
>> +
>> +
>> + # Command wrappers
>> + # Note: all of the following functions raise exceptions in a similar manner
>> + # to cmd() and _get_command_output().
>> +
>> + def cmd(self, command, timeout=20):
>> + """
>> + Send a simple command with no parameters and return its output.
>> + Should only be used for commands that take no parameters and are
>> + implemented under the same name for both the human and QMP monitors.
>> +
>> + @param command: Command to send
>> + @param timeout: Time duration to wait for response
>> + @return: The response to the command
>> + @raise MonitorLockError: Raised if the lock cannot be acquired
>> + @raise MonitorSendError: Raised if the command cannot be sent
>> + @raise MonitorProtocolError: Raised if no response is received
>> + """
>> + return self._get_command_output(command, timeout=timeout)
>> +
>> +
>> + def quit(self):
>> + """
>> + Send "quit" and return the response.
>> + """
>> + return self._get_command_output("quit")
>> +
>> +
>> + def info(self, what):
>> + """
>> + Request info about something and return the response.
>> + """
>> + return self._get_command_output("query-%s" % what)
>> +
>> +
>> + def query(self, what):
>> + """
>> + Alias for info.
>> + """
>> + return self.info(what)
>> +
>> +
>> + def screendump(self, filename):
>> + """
>> + Request a screendump.
>> +
>> + @param filename: Location for the screendump
>> + @return: The response to the command
>> + """
>> + args = {"filename": filename}
>> + return self._get_command_output("screendump", args)
>
> Fail to execute get a screendump. qmp hasn't support this cmd ?
>
> 09:55:14 WARNI| ("QMP command 'screendump' failed", {u'data': {u'name': u'screendump'}, u'class': u'CommandNotFound', u'desc': u'The command screendump has not been found'})
The latest git has screendump support.
BTW, if you're interested in the error data, you can use this syntax:
try:
...
except QMPCmdError, (msg, error):
# here msg is "QMP command ... failed"
# and error is a dict with keys 'data', 'class' and 'desc'
>> +
>> + def migrate(self, uri, full_copy=False, incremental_copy=False, wait=False):
>> + """
>> + Migrate.
>> +
>> + @param uri: destination URI
>> + @param full_copy: If true, migrate with full disk copy
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2010-06-14 6:35 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-06-13 14:33 [KVM-AUTOTEST PATCH 01/14] KVM test: tests_base.cfg.sample: remove inline comments Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 02/14] KVM test: tests_base.cfg.sample: style modifications Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 03/14] KVM test: kvm_utils.py: warn about exceptions raised during unpickling of env Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 04/14] KVM test: kvm_utils.py: remove unnecessary imports Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 05/14] KVM test: kvm_vm.py: make get_pid() return the pid of the VM itself Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 06/14] KVM test: kvm_vm.py: correct get_shared_meminfo() Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 07/14] KVM test: kvm_vm.py: correct add_smp() mistake in make_qemu_command() Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 08/14] KVM test: kvm_vm.py: use shell quoting in make_qemu_command() where appropriate Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 09/14] KVM test: remove reference to _screendump_thread at postprocessing Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 10/14] KVM test: kvm_subprocess.py: use sys.executable instead of hardcoded 'python' Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 11/14] KVM test: add kvm_monitor.py, an interface to QEMU monitors Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 12/14] KVM test: use new monitor interface Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 13/14] KVM test: kvm_monitor.py: add QMP interface Michael Goldish
2010-06-13 14:33 ` [KVM-AUTOTEST PATCH 14/14] KVM test: migration: support QMP Michael Goldish
2010-06-14 2:39 ` [Autotest] [KVM-AUTOTEST PATCH 13/14] KVM test: kvm_monitor.py: add QMP interface Amos Kong
2010-06-14 6:34 ` Michael Goldish [this message]
2010-06-14 1:32 ` [Autotest] [KVM-AUTOTEST PATCH 11/14] KVM test: add kvm_monitor.py, an interface to QEMU monitors Amos Kong
2010-06-14 5:59 ` Michael Goldish
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4C15CD8C.2060503@redhat.com \
--to=mgoldish@redhat.com \
--cc=akong@redhat.com \
--cc=autotest@test.kernel.org \
--cc=kvm@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox