Re: Xentrace on Xilinx ARM

[Dario Faggioli <dario.faggioli@citrix.com>]

[-- Attachment #1.1: Type: text/plain, Size: 2598 bytes --]

On Tue, 2016-03-08 at 18:04 +0000, Ben Sanda wrote:
> All,
> 
> To update to the current situation. I have been able to get
> xentrace() and
> xenalyze working completely (at least as far as I can tell) on ARM.
> 
Great! :-)

> For xentrace there were changes to the memory allocation routines to
> allow
> mapping of the Xen Heap by dom0, correcting the MFN->PFN
> translations, adding
> the trace buffer initialization to setup.c (init_trace_bufs), and
> correcting the
> get_cycles() call to provide the system TSC. For the get_cycles()
> call I
> gathered that was supposed to return the raw tick count, not a
> translated
> ticks->real time timestamp. I then had to call xenalyze with the core
> frequency
> defined so the timestamps made sence.
> 
> Paul: Was there anything else you did I missed?
> 
Sorry, I can't really comment, as I know few about this part of the
tracing infra, and nothing about ARM.

I do encourage Paul to chime in, but, actually, that would probably be
even easier when we'll have the code posted.

So, you know as they say in Open Source, don't you: release early,
release often! :-)

> > It's not part of any Xen image. It's a command line tool to be
> > used, usually
> > but not necessarily, in dom0, build and installed together with the
> > other
> > tools... At least in my case, for x86 builds and installs. 
> For xenalyze I had to modify the makefile to build xenalyze on the
> ARM platform
> (it was specifically removed from the ARM build). Once that was
> corrected I
> could find and call it from dom0. 
>
Great again!

> It built only locally to Xen though (could
> only run from dom0), I could not use it from the native Linux
> development
> environment (I don't know if you're supposed to be able to? Or since
> I'm running
> ARM it built for ARM not x86 and thus could not be used natively).
> 
Yeah, well, if both dev and test environments are x86, I've done it a
couple of times, but I don't think its critical that we support that.

If architectures don't match, I actually think it is _the_right_thing_
that it does not work (as you say yourself)! :-)

> I plan to push they changes in as a patch to the mainline if that
> seems
> reasonable to everyone.
> 
Looking forward to it. :-)

Thanks and regards,
Dario
-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel