diff mbox

[KVM-AUTOTEST] timedrift support

Message ID 4A0D98B8.2070705@redhat.com (mailing list archive)
State New, archived
Headers show

Commit Message

Bear Yang May 15, 2009, 4:30 p.m. UTC
Good morning you all. :)

Lucas:
Very appreciated your help on try...except block implementation. I am 
finally understand your means when I read the mail.  :(

I modified the timedrift case in several place again.
1. add more comments in script.
2. functional the snippnet of script which get clock resource from guest 
and host.
3. using vm.get_command_status replace the sendline API when complie the 
stress program in guest area.

If there is something need to be done by me for the script. feel free 
just kick me.

Happy weekend.

Comments

Yaniv Kaul May 16, 2009, 8:36 p.m. UTC | #1
On 5/15/2009 7:30 PM, bear wrote:
> Good morning you all. :)
>
> Lucas:
> Very appreciated your help on try...except block implementation. I am 
> finally understand your means when I read the mail.  :(
>
> I modified the timedrift case in several place again.
> 1. add more comments in script.
> 2. functional the snippnet of script which get clock resource from 
> guest and host.
> 3. using vm.get_command_status replace the sendline API when complie 
> the stress program in guest area.
>
> If there is something need to be done by me for the script. feel free 
> just kick me.
>
> Happy weekend.
Hi Bear,

Please drop the NTP stuff. There's no need for it.
Llet me again suggest my algorithm (pseudo-code, of course):

MeasureHostParams() // don't forget to measure the host's CPU, memory, 
I/O and network data - always useful.
while (!enough)
{
     RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
     ForEachVM // this runs in parallel on the VMs
     {
         before = TimeOnHost()
         RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
         after = TimeOnHost()
         drift = after - before
         If(drift >= TooMuch) // How much is too much? Could be in 
percentage or absolute values.
             enough = true
     }
}

Simple algorithm, works for us, does not require NTP, and loads the host 
in the most natural way it will be loaded in real life - by more VMs.
Slight variations are possible. For example, use the first VM as a 
'canary' VM - and measure only on it - the others can do random load; 
pin down VM to CPUs; vary the load, the VMs used, and the time, etc.

HTH,
Y.
--
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
Bear Yang May 18, 2009, 1:54 a.m. UTC | #2
Hello Yaniv Kaul:
I am understand your means for timedrift testcase.
* give up using ntp to query and sync the clock from guest to host. my 
question is below:
1. why we can not using the NTP staff in timedrift test case? Is there 
some positive bugs were point on NTP or we found that something of NTP 
definite blocked us to using it? Or could you give me what exact we need 
give up NTP in timedrift test case?

2. for your  algorithm:
MeasureHostParams() // don't forget to measure the host's CPU, memory, 
I/O and network data - always useful.
while (!enough)
{
    RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
    ForEachVM // this runs in parallel on the VMs
    {
        before = TimeOnHost()
        RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
        after = TimeOnHost()
        drift = after - before
        If(drift >= TooMuch) // How much is too much? Could be in 
percentage or absolute values.
            enough = true
    }
}
Why get time from host(TimeOnHost()), why?I think it should be 
TimeOnGuest().  Could you kindly tell me the details about how to get 
the value of before/after and drift?

3. you suggest that the loaded is runing on host is prefer than running 
on guest and running several guest on host to make load on host. right? 
why? nowadays, I using the stress program which provided by Lucas just 
running in guest.

4. I also want to know the result that the script works on your 
side.Maybe it is more easy to help me understand your algorithm than 
read the script.

thanks for your explanation


Yaniv Kaul wrote:
> On 5/15/2009 7:30 PM, bear wrote:
>> Good morning you all. :)
>>
>> Lucas:
>> Very appreciated your help on try...except block implementation. I am 
>> finally understand your means when I read the mail.  :(
>>
>> I modified the timedrift case in several place again.
>> 1. add more comments in script.
>> 2. functional the snippnet of script which get clock resource from 
>> guest and host.
>> 3. using vm.get_command_status replace the sendline API when complie 
>> the stress program in guest area.
>>
>> If there is something need to be done by me for the script. feel free 
>> just kick me.
>>
>> Happy weekend.
> Hi Bear,
>
> Please drop the NTP stuff. There's no need for it.
> Llet me again suggest my algorithm (pseudo-code, of course):
>
> MeasureHostParams() // don't forget to measure the host's CPU, memory, 
> I/O and network data - always useful.
> while (!enough)
> {
>     RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
>     ForEachVM // this runs in parallel on the VMs
>     {
>         before = TimeOnHost()
>         RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
> synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
>         after = TimeOnHost()
>         drift = after - before
>         If(drift >= TooMuch) // How much is too much? Could be in 
> percentage or absolute values.
>             enough = true
>     }
> }
>
> Simple algorithm, works for us, does not require NTP, and loads the 
> host in the most natural way it will be loaded in real life - by more 
> VMs.
> Slight variations are possible. For example, use the first VM as a 
> 'canary' VM - and measure only on it - the others can do random load; 
> pin down VM to CPUs; vary the load, the VMs used, and the time, etc.
>
> HTH,
> Y.
> -- 
> 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

--
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
Yaniv Kaul May 18, 2009, 4:20 a.m. UTC | #3
On 5/18/2009 4:54 AM, Bear Yang wrote:
> Hello Yaniv Kaul:
> I am understand your means for timedrift testcase.
> * give up using ntp to query and sync the clock from guest to host. my 
> question is below:
> 1. why we can not using the NTP staff in timedrift test case? Is there 
> some positive bugs were point on NTP or we found that something of NTP 
> definite blocked us to using it? Or could you give me what exact we 
> need give up NTP in timedrift test case?
Just another requirement from the host we do not really need. And a bit 
more hassle to configure on Windows. And who knows how to configure on 
operating system this-or-that.
>
> 2. for your  algorithm:
> MeasureHostParams() // don't forget to measure the host's CPU, memory, 
> I/O and network data - always useful.
> while (!enough)
> {
>    RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
>    ForEachVM // this runs in parallel on the VMs
>    {
>        before = TimeOnHost()
>        RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
> synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
>        after = TimeOnHost()
>        drift = after - before
>        If(drift >= TooMuch) // How much is too much? Could be in 
> percentage or absolute values.
>            enough = true
>    }
> }
> Why get time from host(TimeOnHost()), why?I think it should be 
> TimeOnGuest().  Could you kindly tell me the details about how to get 
> the value of before/after and drift?
I don't need to get the time on the guest. I know I've asked it to run a 
load for X minutes. The guest has run that for what it think is X 
minutes - but may be X+drift minutes - because its clock drifted. If I 
want to know by how much, I need to compare to a reliable source - the 
host (or in our case, the test server actually).
>
> 3. you suggest that the loaded is runing on host is prefer than 
> running on guest and running several guest on host to make load on 
> host. right? why? nowadays, I using the stress program which provided 
> by Lucas just running in guest.
Good, load the guest. But if the host can satisfy the guest's 
requirements, it won't be drifting (I hope!). We need to overload a bit 
here. The most natural way is with more VMs.
>
> 4. I also want to know the result that the script works on your 
> side.Maybe it is more easy to help me understand your algorithm than 
> read the script.
We'll need to run it. Haven't run it for a while actually.

Y.
>
> thanks for your explanation
>
>
> Yaniv Kaul wrote:
>> On 5/15/2009 7:30 PM, bear wrote:
>>> Good morning you all. :)
>>>
>>> Lucas:
>>> Very appreciated your help on try...except block implementation. I 
>>> am finally understand your means when I read the mail.  :(
>>>
>>> I modified the timedrift case in several place again.
>>> 1. add more comments in script.
>>> 2. functional the snippnet of script which get clock resource from 
>>> guest and host.
>>> 3. using vm.get_command_status replace the sendline API when complie 
>>> the stress program in guest area.
>>>
>>> If there is something need to be done by me for the script. feel 
>>> free just kick me.
>>>
>>> Happy weekend.
>> Hi Bear,
>>
>> Please drop the NTP stuff. There's no need for it.
>> Llet me again suggest my algorithm (pseudo-code, of course):
>>
>> MeasureHostParams() // don't forget to measure the host's CPU, 
>> memory, I/O and network data - always useful.
>> while (!enough)
>> {
>>     RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
>>     ForEachVM // this runs in parallel on the VMs
>>     {
>>         before = TimeOnHost()
>>         RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
>> synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
>>         after = TimeOnHost()
>>         drift = after - before
>>         If(drift >= TooMuch) // How much is too much? Could be in 
>> percentage or absolute values.
>>             enough = true
>>     }
>> }
>>
>> Simple algorithm, works for us, does not require NTP, and loads the 
>> host in the most natural way it will be loaded in real life - by more 
>> VMs.
>> Slight variations are possible. For example, use the first VM as a 
>> 'canary' VM - and measure only on it - the others can do random load; 
>> pin down VM to CPUs; vary the load, the VMs used, and the time, etc.
>>
>> HTH,
>> Y.
>> -- 
>> 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
>

--
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
Bear Yang May 18, 2009, 5:18 a.m. UTC | #4
Hello Yaniv Kaul:
my comments is below.

Yaniv Kaul wrote:
> On 5/18/2009 4:54 AM, Bear Yang wrote:
>> Hello Yaniv Kaul:
>> I am understand your means for timedrift testcase.
>> * give up using ntp to query and sync the clock from guest to host. 
>> my question is below:
>> 1. why we can not using the NTP staff in timedrift test case? Is 
>> there some positive bugs were point on NTP or we found that something 
>> of NTP definite blocked us to using it? Or could you give me what 
>> exact we need give up NTP in timedrift test case?
> Just another requirement from the host we do not really need. And a 
> bit more hassle to configure on Windows. And who knows how to 
> configure on operating system this-or-that.
Actually, I still have not a clear realize about why not give  up NTP in 
timedrift test case. just for could not run this mode on windows guest? 
Except this NTP have not problem internally to block us to use it? right?

I have working on TimeDrift problem base on KVM for several months, now 
you pointed out that we do not need NTP.  you really need give me a 
reasonable reason or explanation for this at first. because If I give up 
NTP, that mean I need rewrite the TimeDrift case. the bug related my 
case will be abolished and my work on this will be abolished.
>>
>> 2. for your  algorithm:
>> MeasureHostParams() // don't forget to measure the host's CPU, 
>> memory, I/O and network data - always useful.
>> while (!enough)
>> {
>>    RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
>>    ForEachVM // this runs in parallel on the VMs
>>    {
>>        before = TimeOnHost()
>>        RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
>> synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
>>        after = TimeOnHost()
>>        drift = after - before
>>        If(drift >= TooMuch) // How much is too much? Could be in 
>> percentage or absolute values.
>>            enough = true
>>    }
>> }
>> Why get time from host(TimeOnHost()), why?I think it should be 
>> TimeOnGuest().  Could you kindly tell me the details about how to get 
>> the value of before/after and drift?
> I don't need to get the time on the guest. I know I've asked it to run 
> a load for X minutes. The guest has run that for what it think is X 
> minutes - but may be X+drift minutes - because its clock drifted. If I 
> want to know by how much, I need to compare to a reliable source - the 
> host (or in our case, the test server actually).
Yes, we need get the compare between host clock and guest clock whatever 
we useing NTP or not, please tell me how to get this comparison in your 
algorithm clearly. and show me how exactly you implement it for the 
value(before/after and drift)

We purpose is found out if *Guest* has time drift problem. KVM is full 
virtualizaiton technology, guest does not know it running as a guest. so 
Why we need add the loaded to Host?

BTW, my script now have the ability to configre option from control file 
to running stress testing just for X minutes or X hours?
>>
>> 3. you suggest that the loaded is runing on host is prefer than 
>> running on guest and running several guest on host to make load on 
>> host. right? why? nowadays, I using the stress program which provided 
>> by Lucas just running in guest.
> Good, load the guest. But if the host can satisfy the guest's 
> requirements, it won't be drifting (I hope!). We need to overload a 
> bit here. The most natural way is with more VMs.
I don't catch your means for your sentence "load the guest. But if the 
host can satisfy the guest's requirements, it won't be drifting (I 
hope!)." Could you please explain it for me?
>
>>
>> 4. I also want to know the result that the script works on your 
>> side.Maybe it is more easy to help me understand your algorithm than 
>> read the script.
> We'll need to run it. Haven't run it for a while actually.
>
> Y.
>>
>> thanks for your explanation
>>
>>
>> Yaniv Kaul wrote:
>>> On 5/15/2009 7:30 PM, bear wrote:
>>>> Good morning you all. :)
>>>>
>>>> Lucas:
>>>> Very appreciated your help on try...except block implementation. I 
>>>> am finally understand your means when I read the mail.  :(
>>>>
>>>> I modified the timedrift case in several place again.
>>>> 1. add more comments in script.
>>>> 2. functional the snippnet of script which get clock resource from 
>>>> guest and host.
>>>> 3. using vm.get_command_status replace the sendline API when 
>>>> complie the stress program in guest area.
>>>>
>>>> If there is something need to be done by me for the script. feel 
>>>> free just kick me.
>>>>
>>>> Happy weekend.
>>> Hi Bear,
>>>
>>> Please drop the NTP stuff. There's no need for it.
>>> Llet me again suggest my algorithm (pseudo-code, of course):
>>>
>>> MeasureHostParams() // don't forget to measure the host's CPU, 
>>> memory, I/O and network data - always useful.
>>> while (!enough)
>>> {
>>>     RunAnotherVM() // don't forget to call SleepTillVmIsFullyRunning()
>>>     ForEachVM // this runs in parallel on the VMs
>>>     {
>>>         before = TimeOnHost()
>>>         RunXMinutesLoadOnGuest(load, minutes) // load can vary - 
>>> synthetic CPU, I/O or 'real life' load. Can also vary from VM to VM.
>>>         after = TimeOnHost()
>>>         drift = after - before
>>>         If(drift >= TooMuch) // How much is too much? Could be in 
>>> percentage or absolute values.
>>>             enough = true
>>>     }
>>> }
>>>
>>> Simple algorithm, works for us, does not require NTP, and loads the 
>>> host in the most natural way it will be loaded in real life - by 
>>> more VMs.
>>> Slight variations are possible. For example, use the first VM as a 
>>> 'canary' VM - and measure only on it - the others can do random 
>>> load; pin down VM to CPUs; vary the load, the VMs used, and the 
>>> time, etc.
>>>
>>> HTH,
>>> Y.
>>> -- 
>>> 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
>>
>

--
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
diff mbox

Patch

diff -urN kvm_runtest_2.bak/genload/Makefile kvm_runtest_2/genload/Makefile
--- kvm_runtest_2.bak/genload/Makefile	1969-12-31 19:00:00.000000000 -0500
+++ kvm_runtest_2/genload/Makefile	2008-09-29 12:22:14.000000000 -0400
@@ -0,0 +1,14 @@ 
+CFLAGS+= -DPACKAGE=\"stress\" -DVERSION=\"0.17pre11\" 
+
+LDLIBS+= -lm
+
+SRCS=$(wildcard *.c)
+TARGETS=$(patsubst %.c,%,$(SRCS))
+
+all: $(TARGETS)
+
+install:
+	@ln -f $(TARGETS) ../../testcases/bin/genload
+
+clean:
+	rm -fr $(TARGETS)
diff -urN kvm_runtest_2.bak/genload/README kvm_runtest_2/genload/README
--- kvm_runtest_2.bak/genload/README	1969-12-31 19:00:00.000000000 -0500
+++ kvm_runtest_2/genload/README	2002-12-13 16:34:13.000000000 -0500
@@ -0,0 +1,72 @@ 
+USAGE
+
+See the program's usage statement by invoking with --help.
+
+NOTES
+
+This program works really well for me, but it might not have some of the
+features that you want.  If you would like, please extend the code and send
+me the patch[1].  Enjoy the program :-)
+
+Please use the context diff format.  That is: save the original program
+as stress.c.orig, then make and test your desired changes to stress.c, then
+run 'diff -u stress.c.orig stress.c' to produce a context patch.  Thanks.
+
+Amos Waterland <apw@rossby.metr.ou.edu>
+Norman, Oklahoma
+27 Nov 2001
+
+EXAMPLES
+[examples]
+
+The simple case is that you just want to bring the system load average up to
+an arbitrary value.  The following forks 13 processes, each of which spins
+in a tight loop calculating the sqrt() of a random number acquired with
+rand().
+
+  % stress -c 13
+
+Long options are supported, as well as is making the output less verbose.
+The following forks 1024 processes, and only reports error messages if any.
+
+  % stress --quiet --hogcpu 1k
+
+To see how your system performs when it is I/O bound, use the -i switch.
+The following forks 4 processes, each of which spins in a tight loop calling
+sync(), which is a system call that flushes memory buffers to disk.
+
+  % stress -i 4
+
+Multiple hogs may be combined on the same command line.  The following does
+everything the preceding examples did in one command, but also turns up the
+verbosity level as well as showing how to cause the command to
+self-terminate after 1 minute.
+
+  % stress -c 13 -i 4 --verbose --timeout 1m
+
+An value of 0 normally denotes infinity.  The following is how to do a fork
+bomb (be careful with this).
+
+  % stress -c 0
+
+For the -m and -d options, a value of 0 means to redo their operation an
+infinite number of times.  To allocate and free 128MB in a redo loop use the
+following command.  This can be useful for "bouncing" against the system RAM
+ceiling.
+
+  % stress -m 0 --hogvm-bytes 128M
+
+For the -m and -d options, a negative value of n means to redo the operation
+abs(n) times.  Here is now to allocate and free 5MB three times in a row.
+
+  % stress -m -3 --hogvm-bytes 5m
+
+You can write a file of arbitrary length to disk.  The file is created with
+mkstemp() in the current directory, the default is to unlink it, but
+unlinking can be overridden with the --hoghdd-noclean flag.
+
+  % stress -d 1 --hoghdd-noclean --hoghdd-bytes 13
+
+Large file support is enabled.
+
+  % stress -d 1 --hoghdd-noclean --hoghdd-bytes 3G
diff -urN kvm_runtest_2.bak/genload/stress.c kvm_runtest_2/genload/stress.c
--- kvm_runtest_2.bak/genload/stress.c	1969-12-31 19:00:00.000000000 -0500
+++ kvm_runtest_2/genload/stress.c	2007-07-26 08:40:17.000000000 -0400
@@ -0,0 +1,983 @@ 
+/* A program to put stress on a POSIX system (stress).
+ *
+ * Copyright (C) 2001, 2002 Amos Waterland <awaterl@yahoo.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation; either version 2 of the License, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc., 59
+ * Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include <ctype.h>
+#include <errno.h>
+#include <libgen.h>
+#include <math.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <signal.h>
+#include <time.h>
+#include <unistd.h>
+#include <sys/wait.h>
+
+/* By default, print all messages of severity info and above.  */
+static int global_debug = 2;
+
+/* By default, just print warning for non-critical errors.  */
+static int global_ignore = 1;
+
+/* By default, retry on non-critical errors every 50ms.  */
+static int global_retry = 50000;
+
+/* By default, use this as backoff coefficient for good fork throughput.  */
+static int global_backoff = 3000;
+
+/* By default, do not timeout.  */
+static int global_timeout = 0;
+
+/* Name of this program */
+static char *global_progname = PACKAGE;
+
+/* By default, do not hang after allocating memory.  */
+static int global_vmhang = 0;
+
+/* Implemention of runtime-selectable severity message printing.  */
+#define dbg if (global_debug >= 3) \
+            fprintf (stdout, "%s: debug: (%d) ", global_progname, __LINE__), \
+            fprintf
+#define out if (global_debug >= 2) \
+            fprintf (stdout, "%s: info: ", global_progname), \
+            fprintf
+#define wrn if (global_debug >= 1) \
+            fprintf (stderr, "%s: warn: (%d) ", global_progname, __LINE__), \
+            fprintf
+#define err if (global_debug >= 0) \
+            fprintf (stderr, "%s: error: (%d) ", global_progname, __LINE__), \
+            fprintf
+
+/* Implementation of check for option argument correctness.  */
+#define assert_arg(A) \
+          if (++i == argc || ((arg = argv[i])[0] == '-' && \
+              !isdigit ((int)arg[1]) )) \
+            { \
+              err (stderr, "missing argument to option '%s'\n", A); \
+              exit (1); \
+            }
+
+/* Prototypes for utility functions.  */
+int usage (int status);
+int version (int status);
+long long atoll_s (const char *nptr);
+long long atoll_b (const char *nptr);
+
+/* Prototypes for the worker functions.  */
+int hogcpu (long long forks);
+int hogio (long long forks);
+int hogvm (long long forks, long long chunks, long long bytes);
+int hoghdd (long long forks, int clean, long long files, long long bytes);
+
+int
+main (int argc, char **argv)
+{
+  int i, pid, children = 0, retval = 0;
+  long starttime, stoptime, runtime;
+
+  /* Variables that indicate which options have been selected.  */
+  int do_dryrun = 0;
+  int do_timeout = 0;
+  int do_cpu = 0;               /* Default to 1 fork. */
+  long long do_cpu_forks = 1;
+  int do_io = 0;                /* Default to 1 fork. */
+  long long do_io_forks = 1;
+  int do_vm = 0;                /* Default to 1 fork, 1 chunk of 256MB.  */
+  long long do_vm_forks = 1;
+  long long do_vm_chunks = 1;
+  long long do_vm_bytes = 256 * 1024 * 1024;
+  int do_hdd = 0;               /* Default to 1 fork, clean, 1 file of 1GB.  */
+  long long do_hdd_forks = 1;
+  int do_hdd_clean = 0;
+  long long do_hdd_files = 1;
+  long long do_hdd_bytes = 1024 * 1024 * 1024;
+
+  /* Record our start time.  */
+  if ((starttime = time (NULL)) == -1)
+    {
+      err (stderr, "failed to acquire current time\n");
+      exit (1);
+    }
+
+  /* SuSv3 does not define any error conditions for this function.  */
+  global_progname = basename (argv[0]);
+
+  /* For portability, parse command line options without getopt_long.  */
+  for (i = 1; i < argc; i++)
+    {
+      char *arg = argv[i];
+
+      if (strcmp (arg, "--help") == 0 || strcmp (arg, "-?") == 0)
+        {
+          usage (0);
+        }
+      else if (strcmp (arg, "--version") == 0)
+        {
+          version (0);
+        }
+      else if (strcmp (arg, "--verbose") == 0 || strcmp (arg, "-v") == 0)
+        {
+          global_debug = 3;
+        }
+      else if (strcmp (arg, "--quiet") == 0 || strcmp (arg, "-q") == 0)
+        {
+          global_debug = 0;
+        }
+      else if (strcmp (arg, "--dry-run") == 0 || strcmp (arg, "-n") == 0)
+        {
+          do_dryrun = 1;
+        }
+      else if (strcmp (arg, "--no-retry") == 0)
+        {
+          global_ignore = 0;
+          dbg (stdout, "turning off ignore of non-critical errors");
+        }
+      else if (strcmp (arg, "--retry-delay") == 0)
+        {
+          assert_arg ("--retry-delay");
+          global_retry = atoll (arg);
+          dbg (stdout, "setting retry delay to %dus\n", global_retry);
+        }
+      else if (strcmp (arg, "--backoff") == 0)
+        {
+          assert_arg ("--backoff");
+          global_backoff = atoll (arg);
+          if (global_backoff < 0)
+            {
+              err (stderr, "invalid backoff factor: %i\n", global_backoff);
+              exit (1);
+            }
+          dbg (stdout, "setting backoff coeffient to %dus\n", global_backoff);
+        }
+      else if (strcmp (arg, "--timeout") == 0 || strcmp (arg, "-t") == 0)
+        {
+          do_timeout = 1;
+          assert_arg ("--timeout");
+          global_timeout = atoll_s (arg);
+          dbg (stdout, "setting timeout to %ds\n", global_timeout);
+        }
+      else if (strcmp (arg, "--cpu") == 0 || strcmp (arg, "-c") == 0)
+        {
+          do_cpu = 1;
+          assert_arg ("--cpu");
+          do_cpu_forks = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--io") == 0 || strcmp (arg, "-i") == 0)
+        {
+          do_io = 1;
+          assert_arg ("--io");
+          do_io_forks = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--vm") == 0 || strcmp (arg, "-m") == 0)
+        {
+          do_vm = 1;
+          assert_arg ("--vm");
+          do_vm_forks = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--vm-chunks") == 0)
+        {
+          assert_arg ("--vm-chunks");
+          do_vm_chunks = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--vm-bytes") == 0)
+        {
+          assert_arg ("--vm-bytes");
+          do_vm_bytes = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--vm-hang") == 0)
+        {
+          global_vmhang = 1;
+        }
+      else if (strcmp (arg, "--hdd") == 0 || strcmp (arg, "-d") == 0)
+        {
+          do_hdd = 1;
+          assert_arg ("--hdd");
+          do_hdd_forks = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--hdd-noclean") == 0)
+        {
+          do_hdd_clean = 2;
+        }
+      else if (strcmp (arg, "--hdd-files") == 0)
+        {
+          assert_arg ("--hdd-files");
+          do_hdd_files = atoll_b (arg);
+        }
+      else if (strcmp (arg, "--hdd-bytes") == 0)
+        {
+          assert_arg ("--hdd-bytes");
+          do_hdd_bytes = atoll_b (arg);
+        }
+      else
+        {
+          err (stderr, "unrecognized option: %s\n", arg);
+          exit (1);
+        }
+    }
+
+  /* Hog CPU option.  */
+  if (do_cpu)
+    {
+      out (stdout, "dispatching %lli hogcpu forks\n", do_cpu_forks);
+
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          if (do_dryrun)
+            exit (0);
+          exit (hogcpu (do_cpu_forks));
+        case -1:               /* error */
+          err (stderr, "hogcpu dispatcher fork failed\n");
+          exit (1);
+        default:               /* parent */
+          children++;
+          dbg (stdout, "--> hogcpu dispatcher forked (%i)\n", pid);
+        }
+    }
+
+  /* Hog I/O option.  */
+  if (do_io)
+    {
+      out (stdout, "dispatching %lli hogio forks\n", do_io_forks);
+
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          if (do_dryrun)
+            exit (0);
+          exit (hogio (do_io_forks));
+        case -1:               /* error */
+          err (stderr, "hogio dispatcher fork failed\n");
+          exit (1);
+        default:               /* parent */
+          children++;
+          dbg (stdout, "--> hogio dispatcher forked (%i)\n", pid);
+        }
+    }
+
+  /* Hog VM option.  */
+  if (do_vm)
+    {
+      out (stdout,
+           "dispatching %lli hogvm forks, each %lli chunks of %lli bytes\n",
+           do_vm_forks, do_vm_chunks, do_vm_bytes);
+
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          if (do_dryrun)
+            exit (0);
+          exit (hogvm (do_vm_forks, do_vm_chunks, do_vm_bytes));
+        case -1:               /* error */
+          err (stderr, "hogvm dispatcher fork failed\n");
+          exit (1);
+        default:               /* parent */
+          children++;
+          dbg (stdout, "--> hogvm dispatcher forked (%i)\n", pid);
+        }
+    }
+
+  /* Hog HDD option.  */
+  if (do_hdd)
+    {
+      out (stdout, "dispatching %lli hoghdd forks, each %lli files of "
+           "%lli bytes\n", do_hdd_forks, do_hdd_files, do_hdd_bytes);
+
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          if (do_dryrun)
+            exit (0);
+          exit (hoghdd
+                (do_hdd_forks, do_hdd_clean, do_hdd_files, do_hdd_bytes));
+        case -1:               /* error */
+          err (stderr, "hoghdd dispatcher fork failed\n");
+          exit (1);
+        default:               /* parent */
+          children++;
+          dbg (stdout, "--> hoghdd dispatcher forked (%i)\n", pid);
+        }
+    }
+
+  /* We have no work to do, so bail out.  */
+  if (children == 0)
+    usage (0);
+
+  /* Wait for our children to exit.  */
+  while (children)
+    {
+      int status, ret;
+
+      if ((pid = wait (&status)) > 0)
+        {
+          if ((WIFEXITED (status)) != 0)
+            {
+              if ((ret = WEXITSTATUS (status)) != 0)
+                {
+                  err (stderr, "dispatcher %i returned error %i\n", pid, ret);
+                  retval += ret;
+                }
+              else
+                {
+                  dbg (stdout, "<-- dispatcher return (%i)\n", pid);
+                }
+            }
+          else
+            {
+              err (stderr, "dispatcher did not exit normally\n");
+              ++retval;
+            }
+
+          --children;
+        }
+      else
+        {
+          dbg (stdout, "wait() returned error: %s\n", strerror (errno));
+          err (stderr, "detected missing dispatcher children\n");
+          ++retval;
+          break;
+        }
+    }
+
+  /* Record our stop time.  */
+  if ((stoptime = time (NULL)) == -1)
+    {
+      err (stderr, "failed to acquire current time\n");
+      exit (1);
+    }
+
+  /* Calculate our runtime.  */
+  runtime = stoptime - starttime;
+
+  /* Print final status message.  */
+  if (retval)
+    {
+      err (stderr, "failed run completed in %lis\n", runtime);
+    }
+  else
+    {
+      out (stdout, "successful run completed in %lis\n", runtime);
+    }
+
+  exit (retval);
+}
+
+int
+usage (int status)
+{
+  char *mesg =
+    "`%s' imposes certain types of compute stress on your system\n\n"
+    "Usage: %s [OPTION [ARG]] ...\n\n"
+    " -?, --help            show this help statement\n"
+    "     --version         show version statement\n"
+    " -v, --verbose         be verbose\n"
+    " -q, --quiet           be quiet\n"
+    " -n, --dry-run         show what would have been done\n"
+    "     --no-retry        exit rather than retry non-critical errors\n"
+    "     --retry-delay n   wait n us before continuing past error\n"
+    " -t, --timeout n       timeout after n seconds\n"
+    "     --backoff n       wait for factor of n us before starting work\n"
+    " -c, --cpu n           spawn n procs spinning on sqrt()\n"
+    " -i, --io n            spawn n procs spinning on sync()\n"
+    " -m, --vm n            spawn n procs spinning on malloc()\n"
+    "     --vm-chunks c     malloc c chunks (default is 1)\n"
+    "     --vm-bytes b      malloc chunks of b bytes (default is 256MB)\n"
+    "     --vm-hang         hang in a sleep loop after memory allocated\n"
+    " -d, --hdd n           spawn n procs spinning on write()\n"
+    "     --hdd-noclean     do not unlink file to which random data written\n"
+    "     --hdd-files f     write to f files (default is 1)\n"
+    "     --hdd-bytes b     write b bytes (default is 1GB)\n\n"
+    "Infinity is denoted with 0.  For -m, -d: n=0 means infinite redo,\n"
+    "n<0 means redo abs(n) times. Valid suffixes are m,h,d,y for time;\n"
+    "k,m,g for size.\n\n";
+
+  fprintf (stdout, mesg, global_progname, global_progname);
+
+  if (status <= 0)
+    exit (-1 * status);
+
+  return 0;
+}
+
+int
+version (int status)
+{
+  char *mesg = "%s %s\n";
+
+  fprintf (stdout, mesg, global_progname, VERSION);
+
+  if (status <= 0)
+    exit (-1 * status);
+
+  return 0;
+}
+
+/* Convert a string representation of a number with an optional size suffix
+ * to a long long.
+ */
+long long
+atoll_b (const char *nptr)
+{
+  int pos;
+  char suffix;
+  long long factor = 1;
+
+  if ((pos = strlen (nptr) - 1) < 0)
+    {
+      err (stderr, "invalid string\n");
+      exit (1);
+    }
+
+  switch (suffix = nptr[pos])
+    {
+    case 'k':
+    case 'K':
+      factor = 1024;
+      break;
+    case 'm':
+    case 'M':
+      factor = 1024 * 1024;
+      break;
+    case 'g':
+    case 'G':
+      factor = 1024 * 1024 * 1024;
+      break;
+    default:
+      if (suffix < '0' || suffix > '9')
+        {
+          err (stderr, "unrecognized suffix: %c\n", suffix);
+          exit (1);
+        }
+    }
+
+  factor = atoll (nptr) * factor;
+
+  return factor;
+}
+
+/* Convert a string representation of a number with an optional time suffix
+ * to a long long.
+ */
+long long
+atoll_s (const char *nptr)
+{
+  int pos;
+  char suffix;
+  long long factor = 1;
+
+  if ((pos = strlen (nptr) - 1) < 0)
+    {
+      err (stderr, "invalid string\n");
+      exit (1);
+    }
+
+  switch (suffix = nptr[pos])
+    {
+    case 's':
+    case 'S':
+      factor = 1;
+      break;
+    case 'm':
+    case 'M':
+      factor = 60;
+      break;
+    case 'h':
+    case 'H':
+      factor = 60 * 60;
+      break;
+    case 'd':
+    case 'D':
+      factor = 60 * 60 * 24;
+      break;
+    case 'y':
+    case 'Y':
+      factor = 60 * 60 * 24 * 360;
+      break;
+    default:
+      if (suffix < '0' || suffix > '9')
+        {
+          err (stderr, "unrecognized suffix: %c\n", suffix);
+          exit (1);
+        }
+    }
+
+  factor = atoll (nptr) * factor;
+
+  return factor;
+}
+
+int
+hogcpu (long long forks)
+{
+  long long i;
+  double d;
+  int pid, retval = 0;
+
+  /* Make local copies of global variables.  */
+  int ignore = global_ignore;
+  int retry = global_retry;
+  int timeout = global_timeout;
+  long backoff = global_backoff * forks;
+
+  dbg (stdout, "using backoff sleep of %lius for hogcpu\n", backoff);
+
+  for (i = 0; forks == 0 || i < forks; i++)
+    {
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          alarm (timeout);
+
+          /* Use a backoff sleep to ensure we get good fork throughput.  */
+          usleep (backoff);
+
+          while (1)
+            d = sqrt (rand ());
+
+          /* This case never falls through; alarm signal can cause exit.  */
+        case -1:               /* error */
+          if (ignore)
+            {
+              ++retval;
+              wrn (stderr, "hogcpu worker fork failed, continuing\n");
+              usleep (retry);
+              continue;
+            }
+
+          err (stderr, "hogcpu worker fork failed\n");
+          return 1;
+        default:               /* parent */
+          dbg (stdout, "--> hogcpu worker forked (%i)\n", pid);
+        }
+    }
+
+  /* Wait for our children to exit.  */
+  while (i)
+    {
+      int status, ret;
+
+      if ((pid = wait (&status)) > 0)
+        {
+          if ((WIFEXITED (status)) != 0)
+            {
+              if ((ret = WEXITSTATUS (status)) != 0)
+                {
+                  err (stderr, "hogcpu worker %i exited %i\n", pid, ret);
+                  retval += ret;
+                }
+              else
+                {
+                  dbg (stdout, "<-- hogcpu worker exited (%i)\n", pid);
+                }
+            }
+          else
+            {
+              dbg (stdout, "<-- hogcpu worker signalled (%i)\n", pid);
+            }
+
+          --i;
+        }
+      else
+        {
+          dbg (stdout, "wait() returned error: %s\n", strerror (errno));
+          err (stderr, "detected missing hogcpu worker children\n");
+          ++retval;
+          break;
+        }
+    }
+
+  return retval;
+}
+
+int
+hogio (long long forks)
+{
+  long long i;
+  int pid, retval = 0;
+
+  /* Make local copies of global variables.  */
+  int ignore = global_ignore;
+  int retry = global_retry;
+  int timeout = global_timeout;
+  long backoff = global_backoff * forks;
+
+  dbg (stdout, "using backoff sleep of %lius for hogio\n", backoff);
+
+  for (i = 0; forks == 0 || i < forks; i++)
+    {
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          alarm (timeout);
+
+          /* Use a backoff sleep to ensure we get good fork throughput.  */
+          usleep (backoff);
+
+          while (1)
+            sync ();
+
+          /* This case never falls through; alarm signal can cause exit.  */
+        case -1:               /* error */
+          if (ignore)
+            {
+              ++retval;
+              wrn (stderr, "hogio worker fork failed, continuing\n");
+              usleep (retry);
+              continue;
+            }
+
+          err (stderr, "hogio worker fork failed\n");
+          return 1;
+        default:               /* parent */
+          dbg (stdout, "--> hogio worker forked (%i)\n", pid);
+        }
+    }
+
+  /* Wait for our children to exit.  */
+  while (i)
+    {
+      int status, ret;
+
+      if ((pid = wait (&status)) > 0)
+        {
+          if ((WIFEXITED (status)) != 0)
+            {
+              if ((ret = WEXITSTATUS (status)) != 0)
+                {
+                  err (stderr, "hogio worker %i exited %i\n", pid, ret);
+                  retval += ret;
+                }
+              else
+                {
+                  dbg (stdout, "<-- hogio worker exited (%i)\n", pid);
+                }
+            }
+          else
+            {
+              dbg (stdout, "<-- hogio worker signalled (%i)\n", pid);
+            }
+
+          --i;
+        }
+      else
+        {
+          dbg (stdout, "wait() returned error: %s\n", strerror (errno));
+          err (stderr, "detected missing hogio worker children\n");
+          ++retval;
+          break;
+        }
+    }
+
+  return retval;
+}
+
+int
+hogvm (long long forks, long long chunks, long long bytes)
+{
+  long long i, j, k;
+  int pid, retval = 0;
+  char **ptr;
+
+  /* Make local copies of global variables.  */
+  int ignore = global_ignore;
+  int retry = global_retry;
+  int timeout = global_timeout;
+  long backoff = global_backoff * forks;
+
+  dbg (stdout, "using backoff sleep of %lius for hogvm\n", backoff);
+
+  if (bytes == 0)
+    {
+      /* 512MB is guess at the largest value can than be malloced at once.  */
+      bytes = 512 * 1024 * 1024;
+    }
+
+  for (i = 0; forks == 0 || i < forks; i++)
+    {
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          alarm (timeout);
+
+          /* Use a backoff sleep to ensure we get good fork throughput.  */
+          usleep (backoff);
+
+          while (1)
+            {
+              ptr = (char **) malloc ( chunks * 2);
+              for (j = 0; chunks == 0 || j < chunks; j++)
+                {
+                  if ((ptr[j] = (char *) malloc (bytes * sizeof (char))))
+                    {
+                      for (k = 0; k < bytes; k++)
+                        ptr[j][k] = 'Z';   /* Ensure that COW happens.  */
+                      dbg (stdout, "hogvm worker malloced %lli bytes\n", k);
+                    }
+                  else if (ignore)
+                    {
+                      ++retval;
+                      wrn (stderr, "hogvm malloc failed, continuing\n");
+                      usleep (retry);
+                      continue;
+                    }
+                  else
+                    {
+                      ++retval;
+                      err (stderr, "hogvm malloc failed\n");
+                      break;
+                    }
+                }
+              if (global_vmhang && retval == 0)
+                {
+                  dbg (stdout, "sleeping forever with allocated memory\n");
+                  while (1)
+                    sleep (1024);
+                }
+              if (retval == 0)
+                {
+                  dbg (stdout,
+                       "hogvm worker freeing memory and starting over\n");
+                  for (j = 0; chunks == 0 || j < chunks; j++) {
+                      free (ptr[j]);
+                  }
+                  free(ptr);
+                  continue;
+                }
+
+              exit (retval);
+            }
+
+          /* This case never falls through; alarm signal can cause exit.  */
+        case -1:               /* error */
+          if (ignore)
+            {
+              ++retval;
+              wrn (stderr, "hogvm worker fork failed, continuing\n");
+              usleep (retry);
+              continue;
+            }
+
+          err (stderr, "hogvm worker fork failed\n");
+          return 1;
+        default:               /* parent */
+          dbg (stdout, "--> hogvm worker forked (%i)\n", pid);
+        }
+    }
+
+  /* Wait for our children to exit.  */
+  while (i)
+    {
+      int status, ret;
+
+      if ((pid = wait (&status)) > 0)
+        {
+          if ((WIFEXITED (status)) != 0)
+            {
+              if ((ret = WEXITSTATUS (status)) != 0)
+                {
+                  err (stderr, "hogvm worker %i exited %i\n", pid, ret);
+                  retval += ret;
+                }
+              else
+                {
+                  dbg (stdout, "<-- hogvm worker exited (%i)\n", pid);
+                }
+            }
+          else
+            {
+              dbg (stdout, "<-- hogvm worker signalled (%i)\n", pid);
+            }
+
+          --i;
+        }
+      else
+        {
+          dbg (stdout, "wait() returned error: %s\n", strerror (errno));
+          err (stderr, "detected missing hogvm worker children\n");
+          ++retval;
+          break;
+        }
+    }
+
+  return retval;
+}
+
+int
+hoghdd (long long forks, int clean, long long files, long long bytes)
+{
+  long long i, j;
+  int fd, pid, retval = 0;
+  int chunk = (1024 * 1024) - 1;        /* Minimize slow writing.  */
+  char buff[chunk];
+
+  /* Make local copies of global variables.  */
+  int ignore = global_ignore;
+  int retry = global_retry;
+  int timeout = global_timeout;
+  long backoff = global_backoff * forks;
+
+  /* Initialize buffer with some random ASCII data.  */
+  dbg (stdout, "seeding buffer with random data\n");
+  for (i = 0; i < chunk - 1; i++)
+    {
+      j = rand ();
+      j = (j < 0) ? -j : j;
+      j %= 95;
+      j += 32;
+      buff[i] = j;
+    }
+  buff[i] = '\n';
+
+  dbg (stdout, "using backoff sleep of %lius for hoghdd\n", backoff);
+
+  for (i = 0; forks == 0 || i < forks; i++)
+    {
+      switch (pid = fork ())
+        {
+        case 0:                /* child */
+          alarm (timeout);
+
+          /* Use a backoff sleep to ensure we get good fork throughput.  */
+          usleep (backoff);
+
+          while (1)
+            {
+              for (i = 0; i < files; i++)
+                {
+                  char name[] = "./stress.XXXXXX";
+
+                  if ((fd = mkstemp (name)) < 0)
+                    {
+                      perror ("mkstemp");
+                      err (stderr, "mkstemp failed\n");
+                      exit (1);
+                    }
+
+                  if (clean == 0)
+                    {
+                      dbg (stdout, "unlinking %s\n", name);
+                      if (unlink (name))
+                        {
+                          err (stderr, "unlink failed\n");
+                          exit (1);
+                        }
+                    }
+
+                  dbg (stdout, "fast writing to %s\n", name);
+                  for (j = 0; bytes == 0 || j + chunk < bytes; j += chunk)
+                    {
+                      if (write (fd, buff, chunk) != chunk)
+                        {
+                          err (stderr, "write failed\n");
+                          exit (1);
+                        }
+                    }
+
+                  dbg (stdout, "slow writing to %s\n", name);
+                  for (; bytes == 0 || j < bytes - 1; j++)
+                    {
+                      if (write (fd, "Z", 1) != 1)
+                        {
+                          err (stderr, "write failed\n");
+                          exit (1);
+                        }
+                    }
+                  if (write (fd, "\n", 1) != 1)
+                    {
+                      err (stderr, "write failed\n");
+                      exit (1);
+                    }
+                  ++j;
+
+                  dbg (stdout, "closing %s after writing %lli bytes\n", name,
+                       j);
+                  close (fd);
+
+                  if (clean == 1)
+                    {
+                      if (unlink (name))
+                        {
+                          err (stderr, "unlink failed\n");
+                          exit (1);
+                        }
+                    }
+                }
+              if (retval == 0)
+                {
+                  dbg (stdout, "hoghdd worker starting over\n");
+                  continue;
+                }
+
+              exit (retval);
+            }
+
+          /* This case never falls through; alarm signal can cause exit.  */
+        case -1:               /* error */
+          if (ignore)
+            {
+              ++retval;
+              wrn (stderr, "hoghdd worker fork failed, continuing\n");
+              usleep (retry);
+              continue;
+            }
+
+          err (stderr, "hoghdd worker fork failed\n");
+          return 1;
+        default:               /* parent */
+          dbg (stdout, "--> hoghdd worker forked (%i)\n", pid);
+        }
+    }
+
+  /* Wait for our children to exit.  */
+  while (i)
+    {
+      int status, ret;
+
+      if ((pid = wait (&status)) > 0)
+        {
+          if ((WIFEXITED (status)) != 0)
+            {
+              if ((ret = WEXITSTATUS (status)) != 0)
+                {
+                  err (stderr, "hoghdd worker %i exited %i\n", pid, ret);
+                  retval += ret;
+                }
+              else
+                {
+                  dbg (stdout, "<-- hoghdd worker exited (%i)\n", pid);
+                }
+            }
+          else
+            {
+              dbg (stdout, "<-- hoghdd worker signalled (%i)\n", pid);
+            }
+
+          --i;
+        }
+      else
+        {
+          dbg (stdout, "wait() returned error: %s\n", strerror (errno));
+          err (stderr, "detected missing hoghdd worker children\n");
+          ++retval;
+          break;
+        }
+    }
+
+  return retval;
+}
diff -urN kvm_runtest_2.bak/kvm_runtest_2.py kvm_runtest_2/kvm_runtest_2.py
--- kvm_runtest_2.bak/kvm_runtest_2.py	2009-04-29 06:17:29.000000000 -0400
+++ kvm_runtest_2/kvm_runtest_2.py	2009-04-29 08:06:32.000000000 -0400
@@ -36,6 +36,8 @@ 
                 "autotest":     test_routine("kvm_tests",           "run_autotest"),
                 "kvm_install":  test_routine("kvm_install",         "run_kvm_install"),
                 "linux_s3":     test_routine("kvm_tests",           "run_linux_s3"),
+                "ntp_server_setup": test_routine("kvm_tests",       "run_ntp_server_setup"),
+                "timedrift":    test_routine("kvm_tests",           "run_timedrift"),
                 }
 
         # Make it possible to import modules from the test's bindir
diff -urN kvm_runtest_2.bak/kvm_tests.cfg.sample kvm_runtest_2/kvm_tests.cfg.sample
--- kvm_runtest_2.bak/kvm_tests.cfg.sample	2009-04-29 06:17:29.000000000 -0400
+++ kvm_runtest_2/kvm_tests.cfg.sample	2009-05-14 02:23:31.000000000 -0400
@@ -81,6 +81,11 @@ 
     - linux_s3:      install setup
         type = linux_s3
 
+    - ntp_server_setup:
+        type = ntp_server_setup
+    - timedrift:      ntp_server_setup
+        type = timedrift
+        stress_test_specify = './stress -c 10 --timeout 240m'
 # NICs
 variants:
     - @rtl8139:
diff -urN kvm_runtest_2.bak/kvm_tests.py kvm_runtest_2/kvm_tests.py
--- kvm_runtest_2.bak/kvm_tests.py	2009-04-29 06:17:29.000000000 -0400
+++ kvm_runtest_2/kvm_tests.py	2009-05-15 05:04:35.000000000 -0400
@@ -394,3 +394,279 @@ 
     kvm_log.info("VM resumed after S3")
 
     session.close()
+
+def run_ntp_server_setup(test, params, env):
+    
+    """NTP server configuration and related network file modification
+    """
+
+    kvm_log.info("stop the iptables service if it is running for timedrift testing")
+
+    if utils.system("/etc/init.d/iptables status", ignore_status=True) == 0:
+        utils.system("/etc/init.d/iptables stop", ignore_status=True)
+
+
+    # prevent dhcp client modify the ntp.conf
+    kvm_log.info("prevent dhcp client modify the ntp.conf")
+
+    config_file = "/etc/sysconfig/network"
+    network_file = open("/etc/sysconfig/network", "a")
+    string = "PEERNTP=no"
+
+    if os.system("grep %s %s" % (string, config_file)):
+        network_file.writelines(str(string)+'\n')
+    
+    network_file.close()
+  
+    # stop the ntp service if it is running
+    kvm_log.info("stop ntp service if it is running")
+
+    if utils.system("/etc/init.d/ntpd status", ignore_status=True) == 0:
+        utils.system("/etc/init.d/ntpd stop", ignore_status=True)
+
+    kvm_log.info("start ntp server on host with the custom config file.")
+
+    ntp_cmd = '''
+        echo "restrict default kod nomodify notrap nopeer noquery" > /etc/timedrift.ntp.conf;\
+        echo "restrict 127.0.0.1" >> /etc/timedrift.ntp.conf;\
+        echo "driftfile /var/lib/ntp/drift" >> /etc/timedrift.ntp.conf;\
+        echo "keys /etc/ntp/keys" >> /etc/timedrift.ntp.conf;\
+        echo "server 127.127.1.0" >> /etc/timedrift.ntp.conf;\
+        echo "fudge 127.127.1.0 stratum 1" >> /etc/timedrift.ntp.conf;\
+        ntpd -c /etc/timedrift.ntp.conf;
+        '''
+
+    try:
+        utils.system(ntp_cmd)
+    except:
+        raise error.TestFail, "NTP server has not starting correctly..."
+
+def run_timedrift(test, params, env):
+    """judge whether the guest clock will encounter timedrift prblem or not. including three stages:
+       1: try to sync the clock with host, if the offset value of guest clock is large than 1 sec.
+       2: running the stress testing program which is hack from LTP on guest.
+       3: then run analyze loop totally 20 times to determine if the clock on guest has time drift.
+    """
+    # variables using in timedrift testcase
+    flag = True
+
+    cpu_stress_program = 'genload'
+    remote_dir = '/'
+
+    clock_resource_used = 'cat /sys/devices/system/clocksource/clocksource0/current_clocksource'
+    clock_resource_available = 'cat /sys/devices/system/clocksource/clocksource0/available_clocksource'
+
+    pwd = os.path.join(os.environ['AUTODIR'],'tests/kvm_runtest_2')
+    stress_test_dir = os.path.join(pwd, cpu_stress_program)
+
+    stress_test_specify = params.get('stress_test_specify')
+    kvm_log.info("stress testing will running :%s" % stress_test_specify)
+    stress_cmdline = 'cd %s%s&&make&&%s &' % (remote_dir, os.path.basename(stress_test_dir), stress_test_specify)
+
+    stress_search_cmdline = "ps -ef|grep 'stress'|grep -v grep"
+
+    hostname = socket.gethostname()
+    if "localhost.localdomain" == hostname:
+        hostname = os.popen('hostname').read().split('\n')[0]
+        kvm_log.info("since get wrong hostname from python evnironment, then use the hostname get from system call(hostname).")
+
+    kvm_log.info("get host name :%s" % hostname)
+
+    # ntpdate info command and ntpdate sync command
+    ntpdate_info_cmd = "ntpdate -q %s" % hostname
+    ntpdate_sync_cmd = "ntpdate %s" % hostname
+
+    # get vm handle
+    vm = kvm_utils.env_get_vm(env,params.get("main_vm"))
+    if not vm:
+        raise error.TestError, "VM object not found in environment"
+    if not vm.is_alive():
+        raise error.TestError, "VM seems to be dead; Test requires a living VM"
+
+    kvm_log.info("Waiting for guest to be up...")
+
+    pxssh = kvm_utils.wait_for(vm.ssh_login, 240, 0, 2)
+    if not pxssh:
+        raise error.TestFail, "Could not log into guest"
+
+    kvm_log.info("Logged into guest IN run_timedrift function.")
+
+    # clock resource get from host and guest
+    def clock_source_info(pxssh,clock_resource_used,flag):
+        kvm_log.info("****************** clock resource info ******************")
+        host_clock_resource = os.popen(clock_resource_used).read().split('\n')[0]
+        kvm_log.info("the clock resource on HOST is :%s" % host_clock_resource)
+    
+        pxssh.sendline(clock_resource_used)
+        s, o = pxssh.read_up_to_prompt()
+        guest_clock_resource = o.splitlines()[-2]
+        kvm_log.info("the clock resource using on GUEST :%s" % guest_clock_resource)
+
+        pxssh.sendline(clock_resource_available)
+        s, o = pxssh.read_up_to_prompt()
+        guest_clock_resource_available = o.splitlines()[-2]
+        kvm_log.info("clock resource available on GUEST :%s" % guest_clock_resource_available)
+        kvm_log.info("*********************************************************")
+
+        if host_clock_resource != guest_clock_resource:
+            #raise error.TestFail, "Host and Guest using different clock resource"
+            kvm_log.info("Host and Guest using different clock resource,Let's move on.")
+        else:
+            kvm_log.info("Host and Guest using same clock resource,Let's move on.")
+    
+        if flag:
+            flag = False
+            return guest_clock_resource, guest_clock_resource_available, flag
+    
+    # a entire process to get ntpdate command line result from guest.
+    def ntpdate_op(command):
+        output = []
+        try:
+            pxssh = kvm_utils.wait_for(vm.ssh_login, 240, 0, 2)
+            if not pxssh:
+                raise error.TestFail, "Could not log into guest"
+
+            kvm_log.info("Logged in:(ntpdate_op)")
+
+            while True:
+                pxssh.sendline(command)
+                s, output = pxssh.read_up_to_prompt()
+                if "time server" in output:
+                    # output is a string contain the (ntpdate -q) infor on guest
+                    pxssh.close()
+                    return True, output
+                else:
+                    continue
+        except:
+            pxssh.close()
+            return False, output
+        return False, output
+
+    # get the numeric handing by regular expression and make timedrift calulation.
+    def time_drift_or_not(output):
+        date_string = re.findall(r'offset [+-]?(.*) sec', output, re.M)
+        num = float(date_string[0])
+        if num >= 1:
+            kvm_log.info("guest clock has drifted in this scenario :%s %s" % (date_string, num))
+            return False
+        else:
+            kvm_log.info("guest clock running veracious in now stage :%s %s" % (date_string, num))
+            return True
+
+    # get clock source infor from function
+    (guest_clock_resource, guest_clock_resource_available, flag) = clock_source_info(pxssh, clock_resource_used, flag)
+
+    # send the command and get the ouput from guest
+    # this loop will pick out several conditions need to be process
+    # Actually, we want to get the info match "time server", then script can analyzing it to
+    # determine if guest's clock need sync with host or not.
+    while True:
+        pxssh.sendline(ntpdate_info_cmd)
+        s, output = pxssh.read_up_to_prompt()
+        kvm_log.info("the ntpdate query info get from guest is below: \n%s" %output)
+        if ("no server suitable" not in output) and ("time server" not in output):
+            kvm_log.info("very creazying output got. let's try again")
+            continue
+        elif "no server suitable" in output:
+            kvm_log.info("seems NTP server is not ready for servicing")
+            time.sleep(30)
+            continue
+        elif "time server" in output:
+            # get the ntpdate info from guest
+            # kvm_log.info("Got the correct output for analyze. The output is below: \n%s" %output) 
+            break
+
+    kvm_log.info("get the ntpdate infomation from guest successfully :%s" % os.popen('date').read())
+
+    # judge the clock need to sync with host or not
+    while True:
+        date_string = re.findall(r'offset [+-]?(.*) sec', output, re.M)
+        num = float(date_string[0])
+        if num >= 1:
+            kvm_log.info("guest need sync with the server: %s" % hostname)
+            s, output = ntpdate_op(ntpdate_sync_cmd)
+            #print "$$$$$$$$$$$$$$$$$$$"
+            #print output
+            #print "$$$$$$$$$$$$$$$$$$$"
+            if s:
+                continue
+        else:
+            #pxssh.sendline("hwclock --systohc")
+            #kvm_log.info("guest clock sync prcdure is finished. then sync the guest clock to guest bios.")
+            #pxssh.sendline("hwclock --show")
+            #s, o = pxssh.read_up_to_prompt()
+            #kvm_log.info("the date infomation get from guest bios is :\n%s" % o)
+
+            pxssh.sendline(ntpdate_info_cmd)
+            s, o = pxssh.read_up_to_prompt()
+            kvm_log.info("guest clock after sync with host is :\n%s" % o)
+
+            break
+
+    kvm_log.info("Timedrift Preparation *Finished* at last :%s" % os.popen('date').read())
+
+    if not vm.scp_to_remote(stress_test_dir, remote_dir):
+        raise error.TestError, "Could not copy program to guest."
+
+    # running the ntp query command before the stress process
+    pxssh.sendline(ntpdate_info_cmd)
+    s, o = pxssh.read_up_to_prompt()
+    kvm_log.info("the ntpdate query from host *BEFORE* running the cpu stress program.\n%s" % o)
+
+    # start running stress process
+    if pxssh.get_command_status(stress_cmdline) == 0:
+        s, o = pxssh.read_up_to_prompt()
+        kvm_log.info("running command line on guest and sleeping for 1200 secs.\n%s" % o)
+    else:
+        s, o = pxssh.read_up_to_prompt()
+        raise error.TestError, "Could not found GCC on the guest.\n%s" % o
+
+    time.sleep(1200)
+
+    # query for the stress process every 60 secs to indetify whether the stress process is finished or not
+    while True:
+        if pxssh.get_command_status(stress_search_cmdline) == 0:
+            print "stress testing on CPU has not finished yet.waiting for next detect after sleep 60 secs."
+            time.sleep(60)
+            continue
+        else:
+            #(s, o) = pxssh.get_command_status_output(stress_search_cmdline)
+            #print "s is :%s" % s
+            #print "o is :%s" % o
+            #print "--------------------------------------------"
+            #aaa = pxssh.get_command_status(stress_search_cmdline)
+            #print "aaa is :%s" % aaa
+            #print "--------------------------------------------"
+
+            print "stress testing process has been completed and quit."
+            break
+
+    # runing the ntp query commnd after the stress process
+    pxssh.sendline(ntpdate_info_cmd)
+    s, o = pxssh.read_up_to_prompt()
+    kvm_log.info("the ntpdate query from host *AFTER* running the cpu stress program.\n%s" % o)
+
+    # close the vm handle finally
+    pxssh.close()
+
+    # Sleep for analyze...
+    kvm_log.info("sleeping(180 secs) Starting... :%s" % os.popen('date').read())
+    time.sleep(180)
+    kvm_log.info("wakeup to get the analyzing... :%s" % os.popen('date').read())
+    count = 0
+    for i in range(1, 21):
+        kvm_log.info("this is %s time to get clock info from guest." % i)
+        s, o = ntpdate_op(ntpdate_info_cmd)
+        
+        if not s:
+            raise error.TestFail, "Guest seems hang or ssh service based on guest has been crash down"
+        
+        if not time_drift_or_not(o):
+            count += 1
+
+        if count == 5:
+            raise error.TestFail, "TimeDrift testing Abort because guest's clock has drift too much"
+
+        kvm_log.info("*********************** Sleep 60 seconds for next loop *************************")
+        time.sleep(60)
+