Re: [PATCH] Enable MSI/MSI-X caps and disable MSI interrupts at PCI probe time - code move

[Bjorn Helgaas <helgaas@kernel.org>]

On Tue, Dec 01, 2015 at 10:41:42AM -0200, Guilherme G. Piccoli wrote:
> On 11/30/2015 11:35 PM, Bjorn Helgaas wrote:
> >Thanks for your patience, Guilherme.
> 
> Hello Bjorn, thank you for the very deep review of my patch. Much
> appreciated!
> 
> 
> >I applied this to pci/msi for v4.5.  I reworked the changelog and made
> >pci_msi_setup_pci_dev() static, since I think it's now only called
> >from drivers/pci/probe.c.
> 
> Thanks. Regarding pci_msi_setup_pci_dev(), it really should be
> static. I changed it because PowerPC arch was calling this since my
> last patch, but this patch remove the call from PowerPC arch, so
> it's great to see this function static again.
> 
> 
> >I didn't quite follow the last paragraph about reverting 1851617cd2da.
> >1851617cd2da moved the MSI init (dev->msi_cap init and MSI disable)
> >from pci/msi.c (where it was only done when CONFIG_PCI_MSI=y) to
> >pci/probe.c (where we did it for all non-OF arches, regardless of
> >CONFIG_PCI_MSI).  So I would characterize this patch as doing the MSI
> >init for *all* arches, regardless of CONFIG_PCI_MSI.
> 
> Well, my English skills aren't so good, so I possibly I wrote
> confusing text. What I meant is that, from practical viewpoint, what
> we did here has the same effect of reverting those patches (in the
> sense of initializing MSI caps). But you can ignore my statement,
> and I appreciate that you removed it from the commit message if it
> was confusing heh

No problem, just making sure I wasn't missing something.  It's not a
literal revert, but it certainly does touch the same code.

Bjorn