[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Xen 4.3 development update CODE FREEZE

>>> On 23.04.13 at 11:09, George Dunlap <George.Dunlap@xxxxxxxxxxxxx> wrote:
> On Tue, Apr 23, 2013 at 8:38 AM, Jan Beulich <JBeulich@xxxxxxxx> wrote:
>>>>> On 22.04.13 at 17:01, George Dunlap <George.Dunlap@xxxxxxxxxxxxx> wrote:
>>> On Mon, Apr 22, 2013 at 4:00 PM, George Dunlap
>>> <George.Dunlap@xxxxxxxxxxxxx> wrote:
>>>> * Multi-vector PCI MSI (support at least for Dom0)
>>>>   owner: jan@suse
>>>>   status: ?
>>> Jan, what's the status of this one?  Are you still pushing for this to
>>> make it into 4.3, or will you wait until 4.4?
>> As also said on irc earlier this morning (I was gone already when
>> you asked yesterday), I'm still hoping to get this in, but it's not
>> really under my control - I'm depending on AMD to help out
>> debugging on their hardware.
>> Suravee actually pointed out that the problem with the earlier
>> posted prerequisite patches was apparently spurious, but he
>> found a problem with the final patch of the full series, which we'll
>> have to work out.
> Right -- then from a release perspective, should we take it off the table?
> Remember the criteria:
> 1. Will we find all the bugs before the release?
> 2. What is the risk of finding bugs that slip the release?
> 3. Is it valuable enough that it's worth the risk of slipping the release?
> Will this be enabled on typical hardware, or will it only be present
> if the user specifically does something to enable it?  If the second,
> #1 is probably less important; otherwise, it seems pretty risky.

On suitable hardware it would get used by default, provided the
kernel also has been enabled accordingly.

> Regarding 2 and 3, I admit I don't have a good handle on how valuable
> the feature is, but absent a pretty high value, it seems like it might
> be better to wait.

It's primarily a load spreading feature, allowing to distribute
interrupts more evenly across (v)CPUs.

And yes, since we're making so slow progress with this, I'm
inclined to agree, however much I regret that.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.