On 02/28/2014 11:30 AM, Andreas Färber wrote:
Am 26.02.2014 18:18, schrieb Jason J. Herne:
From: "Jason J. Herne" <jjhe...@us.ibm.com>
Modify s390_cpu_addr2state to allow fetching state information for cpu addresses
above smp_cpus. Hotplug requires this capability.
Also add s390_cpu_set_state function to allow modification of ipi_state entries
during hotplug.
Signed-off-by: Jason J. Herne <jjhe...@us.ibm.com>
This patch is *still* present despite your previous response that you
incorporated my review comments. You are adding a custom s390x API here
rather than reusing the generic QOM API as requested. I don't even read
further in such cases - like I pointed out, a pure resend is not helping
address review feedback!
Andreas
Andreas,
My apologies, I am not trying to ruffle any feathers and I certainly
have not intended to make you feel like I've ignored your review
comments. I believe, if you look at patches 3 and 4 you will find that
I've address the majority of your comments. I've even completed the work
that adds the link property as you have suggested.
The reason the ipi_array still exists in my patch set is because I was
under the impression that we needed it to still exist. This belief
stemmed from a misunderstanding I had related to an earlier e-mail
discussion. I apologize for getting that detail wrong. I had assumed we
would make use of the link property when possible but still keep the
array around too.
At this point, I only have one concern about removing the array.
ipi_states is accessed from some high frequency code paths dealing with
interrupt handling. I have not yet looked into it myself, but I trust I
would find that retrieving the link property value would not
significantly affect performance?
Thank you for taking the time to read this patch set. Please let me know
if you have any more concerns. I'm happy to work with you to address them.
P.S. Sorry for taking so long to respond, I was away on vacation for a
few days and I was not checking mail.
--
-- Jason J. Herne (jjhe...@linux.vnet.ibm.com)
