Re: [patch 2.5] tg3.c: pci_{save,restore}_extended_state

[Jeff Garzik]

David S. Miller wrote:
>    From: Ivan Kokshaysky <[e-mail]>
>    Date: Sat, 25 Jan 2003 01:41:02 +0300
>    
>    I don't understand the issue, really. The config register says:
>    "MSIs are enabled". Which means: "My platform is *really* going to
>    use MSI". Why do you want to ignore that?

Let us define "platform."  If you mean ia32 or alpha or sparc64, yes 
this is a quirk.  If you mean tg3, no, this is not a quirk.

> I see, why not code up a generic pci_using_msi(pdev) that
> does this?

Great minds think alike :)  This was going to be my suggestion:
call pci_using_msi(pdev), and require that it be called before 
pci_enable_device().

This fits Jeff's "disabling MSI at config time" AFAICS...

	Jeff