From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1756243AbcCOO3t (ORCPT <rfc822;w@1wt.eu>);
	Tue, 15 Mar 2016 10:29:49 -0400
Received: from mx1.redhat.com ([209.132.183.28]:37266 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1753472AbcCOO3r (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Tue, 15 Mar 2016 10:29:47 -0400
Date: Tue, 15 Mar 2016 08:29:41 -0600
From: Alex Williamson <alex.williamson@redhat.com>
To: Julia Lawall <julia.lawall@lip6.fr>
Cc: kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
        Alexander Gordeev <agordeev@redhat.com>,
        Bjorn Helgaas <bhelgaas@google.com>
Subject: Re: question about pci_enable_msix_range
Message-ID: <20160315082941.11216f19@t450s.home>
In-Reply-To: <alpine.DEB.2.10.1603151123340.2531@hadrien>
References: <alpine.DEB.2.10.1603151123340.2531@hadrien>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Tue, 15 Mar 2016 11:26:50 +0100 (CET)
Julia Lawall <julia.lawall@lip6.fr> wrote:

> Hello,
> 
> I was looking at the following code in the file
> drivers/vfio/pci/vfio_pci_intrs.c:
> 
>                 ret = pci_enable_msix_range(pdev, vdev->msix, 1, nvec);
>                 if (ret < nvec) {
>                         if (ret > 0)
>                                 pci_disable_msix(pdev);
>                         kfree(vdev->msix);
>                         kfree(vdev->ctx);
>                         return ret;
>                 }
> 
> I was wondering what is the point of using a range of 1 .. nvec if there
> is going to be a failure if the number of allocated irqs is less than
> nvec?

Hi Julia,

The intention is that on failure we can indicate to the user a value
that might work.  If we were to call with {nvec, nvec} we'd only get
back -ENOSPC and the user could only arbitrarily decrease the request
by some amount and try again.  By using {1, nvec} we can hopefully
provide a useful next step.  On the other hand, we haven't enabled the
number of vectors the user requested, so it doesn't seem to make sense
to leave any enabled.  Thanks,

Alex