Re: [PATCH RFC v2 06/29] PCI/MSI: Get rid of useless count of msi_desc leftovers

2013-11-22 Thread Alexander Gordeev
On Wed, Nov 20, 2013 at 11:07:16AM -0500, Tejun Heo wrote: > Hmmm... so, before this, the function would have returned the partial > number of irqs that arch set up successfully. Is it okay to lose that > information? If so, can you please elaborate a bit more on why it's > okay in the

Re: [PATCH RFC v2 06/29] PCI/MSI: Get rid of useless count of msi_desc leftovers

2013-11-22 Thread Alexander Gordeev
On Wed, Nov 20, 2013 at 11:07:16AM -0500, Tejun Heo wrote: Hmmm... so, before this, the function would have returned the partial number of irqs that arch set up successfully. Is it okay to lose that information? If so, can you please elaborate a bit more on why it's okay in the description?

Re: [PATCH RFC v2 06/29] PCI/MSI: Get rid of useless count of msi_desc leftovers

2013-11-20 Thread Tejun Heo
Hello, Alexander. On Fri, Oct 18, 2013 at 07:12:06PM +0200, Alexander Gordeev wrote: > @@ -744,23 +744,6 @@ static int msix_capability_init(struct pci_dev *dev, > > return 0; > > -out_avail: > - if (ret < 0) { > - /* > - * If we had some success, report the

Re: [PATCH RFC v2 06/29] PCI/MSI: Get rid of useless count of msi_desc leftovers

2013-11-20 Thread Tejun Heo
Hello, Alexander. On Fri, Oct 18, 2013 at 07:12:06PM +0200, Alexander Gordeev wrote: @@ -744,23 +744,6 @@ static int msix_capability_init(struct pci_dev *dev, return 0; -out_avail: - if (ret 0) { - /* - * If we had some success, report the number of

[PATCH RFC v2 06/29] PCI/MSI: Get rid of useless count of msi_desc leftovers

2013-10-19 Thread Alexander Gordeev
If an architecture failed to allocate requested number of MSI-Xs we should not mislead device drivers and make them retry based on how many MSI-Xs were possibly allocated by the architecture in the process, before the failure. It is rude for an architecture to leave any leftovers anyway.

[PATCH RFC v2 06/29] PCI/MSI: Get rid of useless count of msi_desc leftovers

2013-10-19 Thread Alexander Gordeev
If an architecture failed to allocate requested number of MSI-Xs we should not mislead device drivers and make them retry based on how many MSI-Xs were possibly allocated by the architecture in the process, before the failure. It is rude for an architecture to leave any leftovers anyway.