[PATCH 6/6] drivers:pci:hv: New paravirtual PCI front-end for Hyper-V VMs

Jake Oshins jakeo at microsoft.com
Fri Jun 12 15:11:14 UTC 2015


> -----Original Message-----
> From: Paul Bolle [mailto:pebolle at tiscali.nl]
> Sent: Friday, June 12, 2015 1:44 AM
> To: Jake Oshins
> Cc: gregkh at linuxfoundation.org; KY Srinivasan; linux-
> kernel at vger.kernel.org; devel at linuxdriverproject.org; olaf at aepfle.de;
> apw at canonical.com; vkuznets at redhat.com; linux-pci at vger.kernel.org;
> bhelgaas at google.com; Mike Ebersol; Haiyang Zhang
> Subject: Re: [PATCH 6/6] drivers:pci:hv: New paravirtual PCI front-end for
> Hyper-V VMs
> 
> Greg has already asked you to resend. So here follow a few remarks to
> take into account for that resend.
> 

Thank you.  I'll fix everything you've mentioned before resending.  I do have one more question, below.

> 
> > +EXPORT_SYMBOL(hv_read_config_block);
> 
> > +EXPORT_SYMBOL(hv_write_config_block);
> 
> > +EXPORT_SYMBOL(hv_register_block_invalidate);
> 
> I couldn't spot any users of these exports. Actually, I couldn't even
> spot any users of these three functions. Why were they added?
> 

This driver is intended to support both full PCI Express device pass through and also be the basis for SR-IOV networking on top of Hyper-V.  These functions would allow somebody trying to make their NIC driver work on top of Hyper-V to exchange messages with their back-end Windows driver.

My question is this.  How does somebody delivering a platform usually work with the Linux community to deliver enablement code like this?  I'm trying to work in the open, and go upstream early (or at least I think that understand what these things mean.)  If the community doesn't want functions that have no callers (and I understand that, too) then how should I provide them to the NIC vendors?

Thanks,
Jake Oshins



More information about the devel mailing list