Hi, all:
I am trying to measure the scheduling related latency using the xentrace
framework. I add some macros in include/public/trace.h, using mask
0x00024000,
and record the overhead by modifying xen/common/schedule.c.
It compiles and runs.
However, I am not sure how to modify the xenalyze to show the data. I tried
the -a option, seems the new added data is not shown. (but other data works
fine). Could someone kindly give some pointers to me?
I also tried the xentrace_format to dump the raw data, it says:
[xisisu@localhost gedf]$ xentrace_format busy.bin
Traceback (most recent call last):
File "/usr/bin/xentrace_format", line 85, in <module>
defs = read_defs(arg[0])
File "/usr/bin/xentrace_format", line 54, in read_defs
defs[str(eval(m.group(1)))] = m.group(2)
File "<string>", line 1
?
^
SyntaxError: invalid syntax
Does any one know how to fix this?
Thanks very much!
Sisu
--
Sisu Xi, PhD Candidate
http://www.cse.wustl.edu/~xis/
Department of Computer Science and Engineering
Campus Box 1045
Washington University in St. Louis
One Brookings Drive
St. Louis, MO 63130
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
On Fri, Mar 29, 2013 at 9:28 PM, Sisu Xi <xisisu@gmail.com> wrote:> Hi, all: > > I am trying to measure the scheduling related latency using the xentrace > framework. I add some macros in include/public/trace.h, using mask > 0x00024000, > and record the overhead by modifying xen/common/schedule.c. > It compiles and runs. > > However, I am not sure how to modify the xenalyze to show the data. I tried > the -a option, seems the new added data is not shown. (but other data works > fine). Could someone kindly give some pointers to me?Are you saying that you added code to xenalyze to pretty-print the record, but still didn''t see anything? Or that you haven''t tried to add trace records to xenalyze yet? -George
Hi, George: oh, I did not modify xenalyze. I do notice that xenalyze has an option to dump all, but seems it will skip the user added traces. I''ll look into the code to understand it. Thanks. Sisu On Tue, Apr 2, 2013 at 6:17 AM, George Dunlap <dunlapg@umich.edu> wrote:> On Fri, Mar 29, 2013 at 9:28 PM, Sisu Xi <xisisu@gmail.com> wrote: > > Hi, all: > > > > I am trying to measure the scheduling related latency using the xentrace > > framework. I add some macros in include/public/trace.h, using mask > > 0x00024000, > > and record the overhead by modifying xen/common/schedule.c. > > It compiles and runs. > > > > However, I am not sure how to modify the xenalyze to show the data. I > tried > > the -a option, seems the new added data is not shown. (but other data > works > > fine). Could someone kindly give some pointers to me? > > Are you saying that you added code to xenalyze to pretty-print the > record, but still didn''t see anything? Or that you haven''t tried to > add trace records to xenalyze yet? > > -George >-- Sisu Xi, PhD Candidate http://www.cse.wustl.edu/~xis/ Department of Computer Science and Engineering Campus Box 1045 Washington University in St. Louis One Brookings Drive St. Louis, MO 63130 _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
On Wed, Apr 3, 2013 at 4:14 AM, Sisu Xi <xisisu@gmail.com> wrote:> Hi, George: > > oh, I did not modify xenalyze. I do notice that xenalyze has an option to > dump all, but seems it will skip the user added traces. I''ll look into the > code to understand it.No, it does not skip user-added traces. But it has no way of knowing how to "pretty-print" them (i.e., in a human-readable format), so it prints it in raw format, like this: ] 31.846657673 lllllllx d32767v7 1f002(1:f:2) 2 [ c2c2c2c2 c2c2c2c2 ] -George
Hi, George: Yes. You are correct. Seems I used the wrong mask which causes FATAL. I changed the value and can get the value now. Thanks very much! Sisu On Wed, Apr 3, 2013 at 6:17 AM, George Dunlap <George.Dunlap@eu.citrix.com>wrote:> On Wed, Apr 3, 2013 at 4:14 AM, Sisu Xi <xisisu@gmail.com> wrote: > > Hi, George: > > > > oh, I did not modify xenalyze. I do notice that xenalyze has an option to > > dump all, but seems it will skip the user added traces. I''ll look into > the > > code to understand it. > > No, it does not skip user-added traces. But it has no way of knowing > how to "pretty-print" them (i.e., in a human-readable format), so it > prints it in raw format, like this: > > ] 31.846657673 lllllllx d32767v7 1f002(1:f:2) 2 [ c2c2c2c2 c2c2c2c2 ] > > -George >-- Sisu Xi, PhD Candidate http://www.cse.wustl.edu/~xis/ Department of Computer Science and Engineering Campus Box 1045 Washington University in St. Louis One Brookings Drive St. Louis, MO 63130 _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel