Pkg xen devel - Sep 2014 - Processed: Re: dom0 kernels should suggest irqbalance

(copying debian-kernel for reasons which will hopefully become obvious)

On Mon, 8 Jul 2013 18:10:58 +0200 =?UTF-8?Q?Moritz_M=C3=BChlenhoff?= <jmm at
inutil.org> wrote:
> In current Debian kernel there's no special Xen dom0 kernel image and
depending
> on irqbalance in the kernel package would be overkill.
Would it? I thought irqbalance is actually required even for native with
modern kernels, since the kernel doesn't do any balancing by itself (any
more, it did use to).

Looking on my laptop for instance I see that all interrupts are going to
CPU0 out of the 4 processes. On the other hand my workstation does seem
to have balanced IRQs despite having no irqbalanced running, so I don't
know.

I reckon the kernel probably should recommend irqbalance these days, but
in any case there is no reason for Xen to do something different (since
IRQ balancing should work as on native).

Ian.

On Sun, 2014-08-31 at 03:10 +0100, Ian Campbell wrote:
> (copying debian-kernel for reasons which will hopefully become obvious)
> 
> On Mon, 8 Jul 2013 18:10:58 +0200 =?UTF-8?Q?Moritz_M=C3=BChlenhoff?=
<jmm at inutil.org> wrote:
> > In current Debian kernel there's no special Xen dom0 kernel image
and depending
> > on irqbalance in the kernel package would be overkill.
> 
> Would it? I thought irqbalance is actually required even for native with
> modern kernels, since the kernel doesn't do any balancing by itself
(any
> more, it did use to).
I don't know that it's generally required, but it can be useful if there
is a lot of work done in interrupt or softirq context (and have multiple
processors).
> Looking on my laptop for instance I see that all interrupts are going to
> CPU0 out of the 4 processes. On the other hand my workstation does seem
> to have balanced IRQs despite having no irqbalanced running, so I don't
> know.
> 
> I reckon the kernel probably should recommend irqbalance these days, but
> in any case there is no reason for Xen to do something different (since
> IRQ balancing should work as on native).
At least kernels that support SMP could recommend it.

Ben.

-- 
Ben Hutchings
Experience is directly proportional to the value of equipment destroyed.
                                                         - Carolyn Scheppner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 811 bytes
Desc: This is a digitally signed message part
URL:
<http://lists.alioth.debian.org/pipermail/pkg-xen-devel/attachments/20140904/8b167e46/attachment.sig>

Control: reassign -1 src:linux
COntrol: retitle -1 SMP kernels should suggest irqbalance

On Thu, 2014-09-04 at 00:14 +0100, Ben Hutchings wrote:
> On Sun, 2014-08-31 at 03:10 +0100, Ian Campbell wrote:
> > (copying debian-kernel for reasons which will hopefully become
obvious)
> > 
> > On Mon, 8 Jul 2013 18:10:58 +0200 =?UTF-8?Q?Moritz_M=C3=BChlenhoff?=
<jmm at inutil.org> wrote:
> > > In current Debian kernel there's no special Xen dom0 kernel
image and depending
> > > on irqbalance in the kernel package would be overkill.
> > 
> > Would it? I thought irqbalance is actually required even for native
with
> > modern kernels, since the kernel doesn't do any balancing by
itself (any
> > more, it did use to).
> 
> I don't know that it's generally required, but it can be useful if
there
> is a lot of work done in interrupt or softirq context (and have multiple
> processors).
> 
> > Looking on my laptop for instance I see that all interrupts are going
to
> > CPU0 out of the 4 processes. On the other hand my workstation does
seem
> > to have balanced IRQs despite having no irqbalanced running, so I
don't
> > know.
> > 
> > I reckon the kernel probably should recommend irqbalance these days,
but
> > in any case there is no reason for Xen to do something different
(since
> > IRQ balancing should work as on native).
> 
> At least kernels that support SMP could recommend it.
That sounds like a good idea to me.

Presumably irqbalanced is mostly harmless for UP kernels, but I think we
can avoid that recommends unless the auto-generation of debian/control
gets in the way too much. 

Ian.

Processing control commands:
> reassign -1 src:linux
Bug #577788 [src:xen] dom0 kernels should suggest irqbalance
Bug reassigned from package 'src:xen' to 'src:linux'.
Ignoring request to alter found versions of bug #577788 to the same values
previously set
Ignoring request to alter fixed versions of bug #577788 to the same values
previously set
> retitle -1 SMP kernels should suggest irqbalance
Bug #577788 [src:linux] dom0 kernels should suggest irqbalance
Changed Bug title to 'SMP kernels should suggest irqbalance' from
'dom0 kernels should suggest irqbalance'

-- 
577788: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=577788
Debian Bug Tracking System
Contact owner at bugs.debian.org with problems