Re: [PATCH net-next V3 1/2] igb: add PTP Hardware Clock code

-----Original Message-----
From: Richard Cochran [mailto:richardcochran@gmail.com]
Sent: Tuesday, January 10, 2012 12:45 AM
To: Keller, Jacob E
Cc: netdev@vger.kernel.org; e1000-devel@lists.sourceforge.net; Kirsher,
Jeffrey T; Ronciak, John; John Stultz; Thomas Gleixner
Subject: Re: [PATCH net-next V3 1/2] igb: add PTP Hardware Clock code

On Mon, Jan 09, 2012 at 05:42:20PM +0000, Keller, Jacob E wrote:

quoted

Is there a reason for not using the timecounter structure from the
kernel? It is a layer beneath the timecompare code which is meant to
handle this condition. As far as I can tell this issue is solved in
the timecounter code. If it is not, then that should be a bug in the
timecounter cyclecounter code. I don't know if this issue occurs in
the timecounter structure because it handles the ns conversion
differently.

My only reason is that I am not sure that the timecounter code really
does what we need.  It might well work.  Consider, though, that the
82580 register does not overflow in the usual way. The upper 24 bits
are always zero.

What I wrote does the right thing, I think.  However, duplicated
effort is always bad, so can you show me how to change it?

Sure :) The way to set it up also includes the ability to specify width, so that for example the 82580 can be set to the bit length, and you provide the whole cycle counter value as a 40bit value.

I will send you a copy of the code I will be using for the ixgbe driver.

It *does* handle the conversion in a different way, because it doesn't keep an overflow counter, instead it converts "cycles" to nanoseconds. However, it claims to have support for past timestamps as long as they are within half the cycle max value.

Basically it's a portion of the code used in the timecompare stuff, but without the timecompare averaging scheme which is bad. I don't use the structure the same way that the older igb code did (because it does funky things with the 82580 and doesn't setup the other devices correctly for clock rate adjustment)

Thanks,
Richard

------------------------------------------------------------------------------
Ridiculously easy VDI. With Citrix VDI-in-a-Box, you don't need a complex
infrastructure or vast IT resources to deliver seamless, secure access to
virtual desktops. With this all-in-one solution, easily deploy virtual 
desktops for less than the cost of PCs and save 60% on VDI infrastructure 
costs. Try it free! http://p.sf.net/sfu/Citrix-VDIinabox
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired