Features scheduled for the 2.6.39 merge window

Features scheduled for the 2.6.39 merge window

[Sander Eikelenboom]

Hi All,

I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.

Of special interest for me:
- blkback (tapdisk2?)
- netback
- pciback

--
Sander

Re: Features scheduled for the 2.6.39 merge window

[Ian Campbell]

On Fri, 2011-03-11 at 10:57 +0000, Sander Eikelenboom wrote:
> Hi All,
> 
> I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.

Konrad posted a state of the world for this last night, see the thread
"2.6.39 merge window (git pulls and what is planned to go in)"

> Of special interest for me:

I'm afraid you are rather out of luck in that respect :-(

> - blkback (tapdisk2?)

Not planned for 2.6.39.

> - netback

Under review on the netdev list, might still make it.

> - pciback

Not planned for 2.6.39.

FWIW blkback, netback and pciback all have relatively clean branches
which could be pulled in for testing purposes though.

Ian.

Re: Features scheduled for the 2.6.39 merge window

[Sander Eikelenboom]

Friday, March 11, 2011, 12:03:39 PM, you wrote:

> On Fri, 2011-03-11 at 10:57 +0000, Sander Eikelenboom wrote:
>> Hi All,
>> 
>> I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.

> Konrad posted a state of the world for this last night, see the thread
> "2.6.39 merge window (git pulls and what is planned to go in)"

Ah missed that, the list is very active lately so haven't had the time to read it all.

>> Of special interest for me:

> I'm afraid you are rather out of luck in that respect :-(

>> - blkback (tapdisk2?)

> Not planned for 2.6.39.

>> - netback

> Under review on the netdev list, might still make it.

>> - pciback

> Not planned for 2.6.39.

> FWIW blkback, netback and pciback all have relatively clean branches
> which could be pulled in for testing purposes though.

OK so the pieces for effectively running 2.6.39 as dom0 for PV-guests will be postponed to 2.6.40 i guess ?
Don't know if Konrad has the time to start a merged branch with all the upcoming stuff, for people who are interested in testing ?


> Ian.





-- 
Best regards,
 Sander                            mailto:linux@eikelenboom.it

Re: Features scheduled for the 2.6.39 merge window

[Sander Eikelenboom]

Hello Ian,

Friday, March 11, 2011, 12:03:39 PM, you wrote:

> On Fri, 2011-03-11 at 10:57 +0000, Sander Eikelenboom wrote:
>> Hi All,
>> 
>> I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.

> Konrad posted a state of the world for this last night, see the thread
> "2.6.39 merge window (git pulls and what is planned to go in)"

>> Of special interest for me:

> I'm afraid you are rather out of luck in that respect :-(

>> - blkback (tapdisk2?)

> Not planned for 2.6.39.

>> - netback

> Under review on the netdev list, might still make it.

>> - pciback

> Not planned for 2.6.39.

> FWIW blkback, netback and pciback all have relatively clean branches
> which could be pulled in for testing purposes though.

Hmmm looking at Konrad's git tree, the master tree seems to be just that in this case :-)
Will try it later this weekend.

> Ian.





-- 
Best regards,
 Sander                            mailto:linux@eikelenboom.it

Re: Re: Features scheduled for the 2.6.39 merge window

[Konrad Rzeszutek Wilk]

On Fri, Mar 11, 2011 at 12:34:03PM +0100, Sander Eikelenboom wrote:
> Hello Ian,
> 
> Friday, March 11, 2011, 12:03:39 PM, you wrote:
> 
> > On Fri, 2011-03-11 at 10:57 +0000, Sander Eikelenboom wrote:
> >> Hi All,
> >> 
> >> I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.
> 
> > Konrad posted a state of the world for this last night, see the thread
> > "2.6.39 merge window (git pulls and what is planned to go in)"
> 
> >> Of special interest for me:
> 
> > I'm afraid you are rather out of luck in that respect :-(
> 
> >> - blkback (tapdisk2?)
> 
> > Not planned for 2.6.39.
> 
> >> - netback
> 
> > Under review on the netdev list, might still make it.
> 
> >> - pciback
> 
> > Not planned for 2.6.39.
> 
> > FWIW blkback, netback and pciback all have relatively clean branches
> > which could be pulled in for testing purposes though.
> 
> Hmmm looking at Konrad's git tree, the master tree seems to be just that in this case :-)
> Will try it later this weekend.

Yeah, except that it does not work right now with legacy IRQ (like USB devices)
passthrough - just found out about that yesterday when I started running it through the paces.

But it works fine for PV, HVM, and Dom 0 case.

Re: Re: Features scheduled for the 2.6.39 merge window

[Sander Eikelenboom]

Hello Konrad,

Friday, March 11, 2011, 3:24:21 PM, you wrote:

> On Fri, Mar 11, 2011 at 12:34:03PM +0100, Sander Eikelenboom wrote:
>> Hello Ian,
>> 
>> Friday, March 11, 2011, 12:03:39 PM, you wrote:
>> 
>> > On Fri, 2011-03-11 at 10:57 +0000, Sander Eikelenboom wrote:
>> >> Hi All,
>> >> 
>> >> I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.
>> 
>> > Konrad posted a state of the world for this last night, see the thread
>> > "2.6.39 merge window (git pulls and what is planned to go in)"
>> 
>> >> Of special interest for me:
>> 
>> > I'm afraid you are rather out of luck in that respect :-(
>> 
>> >> - blkback (tapdisk2?)
>> 
>> > Not planned for 2.6.39.
>> 
>> >> - netback
>> 
>> > Under review on the netdev list, might still make it.
>> 
>> >> - pciback
>> 
>> > Not planned for 2.6.39.
>> 
>> > FWIW blkback, netback and pciback all have relatively clean branches
>> > which could be pulled in for testing purposes though.
>> 
>> Hmmm looking at Konrad's git tree, the master tree seems to be just that in this case :-)
>> Will try it later this weekend.

> Yeah, except that it does not work right now with legacy IRQ (like USB devices)
> passthrough - just found out about that yesterday when I started running it through the paces.

> But it works fine for PV, HVM, and Dom 0 case.

Ok I will give it a spin then and will leave the passthrough VM's out.
Good to know about the legacy IRQ problem, and a bit unfortunate that the msi-irq patches for ohci/ehci controllers didn't make it in, due to objection of Greg KH, since he saw no direct advantages.

Will let you know how it went ..

--
Sander

Re: Re: Features scheduled for the 2.6.39 merge window

[Sander Eikelenboom]

Hello Konrad,

Friday, March 11, 2011, 3:24:21 PM, you wrote:

> On Fri, Mar 11, 2011 at 12:34:03PM +0100, Sander Eikelenboom wrote:
>> Hello Ian,
>> 
>> Friday, March 11, 2011, 12:03:39 PM, you wrote:
>> 
>> > On Fri, 2011-03-11 at 10:57 +0000, Sander Eikelenboom wrote:
>> >> Hi All,
>> >> 
>> >> I have lost a bit of track on what (backend) features are scheduled for the 2.6.39 merge window.
>> 
>> > Konrad posted a state of the world for this last night, see the thread
>> > "2.6.39 merge window (git pulls and what is planned to go in)"
>> 
>> >> Of special interest for me:
>> 
>> > I'm afraid you are rather out of luck in that respect :-(
>> 
>> >> - blkback (tapdisk2?)
>> 
>> > Not planned for 2.6.39.
>> 
>> >> - netback
>> 
>> > Under review on the netdev list, might still make it.
>> 
>> >> - pciback
>> 
>> > Not planned for 2.6.39.
>> 
>> > FWIW blkback, netback and pciback all have relatively clean branches
>> > which could be pulled in for testing purposes though.
>> 
>> Hmmm looking at Konrad's git tree, the master tree seems to be just that in this case :-)
>> Will try it later this weekend.

> Yeah, except that it does not work right now with legacy IRQ (like USB devices)
> passthrough - just found out about that yesterday when I started running it through the paces.

> But it works fine for PV, HVM, and Dom 0 case.

It seems to work for the passthrough of my USB controllers, seems to be working ok.
BLKTAP2 isn't in your tree yet i noticed, so i reverted back to "file:" based blkback.

But with "file:" based blkback it all seems to work OK, including pci-passthrough in my case !

So it's looking good :-)


--
Sander