Re: [PATCH] sysfs: add per pci device msi[x] irq listing (v3)

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Thu, Sep 22, 2011 at 06:57:06AM -0400, Neil Horman wrote:
> On Thu, Sep 22, 2011 at 06:49:02AM -0400, Konrad Rzeszutek Wilk wrote:
> > On Mon, Sep 19, 2011 at 11:47:15AM -0400, Neil Horman wrote:
> > > So a while back, I wanted to provide a way for irqbalance (and other apps) to
> > > definitively map irqs to devices, which, for msi[x] irqs is currently not really
> > > possible in user space.  My first attempt wen't not so well:
> > > https://lkml.org/lkml/2011/4/21/308
> > > 
> > > It was plauged by the same issues that prior attempts were, namely that it
> > > violated the one-file-one-value sysfs rule.  I wandered off but have recently
> > > come back to this.  I've got a new implementation here that exports a new
> > > subdirectory for every pci device,  called msi_irqs.  This subdirectory contanis
> > > a variable number of numbered subdirectories, in which the number represents an
> > > msi irq.  Each numbered subdirectory contains attributes for that irq, which
> > > currently is only the mode it is operating in (msi vs. msix).  I think fits
> > > within the constraints sysfs requires, and will allow irqbalance to properly map
> > > msi irqs to devices without having to rely on rickety, best guess methods like
> > > interface name matching.
> > 
> > Are there irqbalance patches that correspond to this? Where would they be available?
> > 
> I've got them here locally, shemminger and I are testing them out, when I'm comfortable with
> them, I'll be comitting them to the public repository at code.google.com

Right.. asking b/c it might be a good idea to include that in the git
description of the patch.

BTW, I also had some question on the patch itself - not sure if you had a chance
to read them.
> 
> Note, the changes to support this kernel update in irqbalance is also comming
> with a major gutting that I'm doing of the daemon.  Things like making workload
> bias adjustments based on packets received per interface, since theres no
> guaranteed 1:1 correlation between network interfaces and irqs.  The point being
> that while irq identification will definately be fixed, you might notice other
> behavioral differences that we may or may not want to address.  Bug reports
> would be greatly appreciated at the project site.
> 
> Thanks!
> Neil