WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

Re: [Xen-devel] [RFC PATCH linux-2.6.18-xen] pciback: clean up (MSI-X ve

To: Laszlo Ersek <lersek@xxxxxxxxxx>
Subject: Re: [Xen-devel] [RFC PATCH linux-2.6.18-xen] pciback: clean up (MSI-X vec, entrynr) list when resetting PCI device
From: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
Date: Wed, 1 Jun 2011 10:51:04 -0400
Cc: "xen-devel@xxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxx>
Delivery-date: Wed, 01 Jun 2011 07:52:19 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <4DE60EF8.5060902@xxxxxxxxxx>
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
References: <4DE60EF8.5060902@xxxxxxxxxx>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Mutt/1.5.21 (2010-09-15)
On Wed, Jun 01, 2011 at 12:05:44PM +0200, Laszlo Ersek wrote:
> Hi,
> 
> this is more of an RFC than a patch now for linux-2.6.18-xen. Describing the 
> situation captured in RHBZ#688673, in a nutshell:
> 
> - let's say we have an Intel 82576 card (igb),
> - the card exports virtual functions (igbvf),
> - one virtual function is passed through to a PV guest,
> - the igbvf driver, co-operating with pcifront, pciback, and the hypervisor, 
> sets up three MSI-X vectors for the virtual function PCI device,
> - when the domU is shut down, the MSI-X vectors are "leaked" in dom0, because 
> nobody ever reaches dom0's pci_disable_msix() / msi_remove_pci_irq_vectors() 
> during shutdown,

Could you keep a state in the dom0 of the fact that msi/msi-x were enabled, and 
use those
(instead of dev->msi_enabled) to call pci_disable_msi(). And you could also 
modify the
msi_enabled to be set to 1 before you make the pci_disable_msi() call?

And naturally if the frontend makes a PV call to disable your MSI device, then 
you
set internally your msi_enabled to zero.

Something like this patch (based off stable/2.6.39.x)

diff --git a/drivers/pci/xen-pciback/pciback.h 
b/drivers/pci/xen-pciback/pciback.h
index 788c3ee..69f0961 100644
--- a/drivers/pci/xen-pciback/pciback.h
+++ b/drivers/pci/xen-pciback/pciback.h
@@ -50,11 +50,39 @@ struct xen_pcibk_dev_data {
        unsigned int enable_intx:1;
        unsigned int isr_on:1; /* Whether the IRQ handler is installed. */
        unsigned int ack_intr:1; /* .. and ACK-ing */
+       unsigned int msi_enabled:1;
+       unsigned int msix_enabled:1;
        unsigned long handled;
        unsigned int irq; /* Saved in case device transitions to MSI/MSI-X */
        char irq_name[0]; /* xen-pcibk[000:04:00.0] */
 };
 
+#ifdef CONFIG_MSI
+static inline bool xen_pcibk_force_msi_disable(struct pci_dev *dev)
+{
+       struct xen_pcibk_dev_data *dev_data;
+
+       dev_data = pci_get_drvdata(dev);
+       if (!dev_data)
+               return false;
+
+       if (dev_data->msi_enabled && !dev->msi_enabled)
+               return true;
+       return false;
+}
+static inline bool xen_pcibk_force_msix_disable(struct pci_dev *dev)
+{
+       struct xen_pcibk_dev_data *dev_data;
+
+       dev_data = pci_get_drvdata(dev);
+       if (!dev_data)
+               return false;
+
+       if (dev_data->msix_enabled && !dev->msix_enabled)
+               return true;
+       return false;
+}
+#endif
 /* Used by XenBus and xen_pcibk_ops.c */
 extern wait_queue_head_t xen_pcibk_aer_wait_queue;
 extern struct workqueue_struct *xen_pcibk_wq;
diff --git a/drivers/pci/xen-pciback/pciback_ops.c 
b/drivers/pci/xen-pciback/pciback_ops.c
index 8c95c34..b37935b 100644
--- a/drivers/pci/xen-pciback/pciback_ops.c
+++ b/drivers/pci/xen-pciback/pciback_ops.c
@@ -109,6 +109,11 @@ void xen_pcibk_reset_device(struct pci_dev *dev)
 #ifdef CONFIG_PCI_MSI
                /* The guest could have been abruptly killed without
                 * disabling MSI/MSI-X interrupts.*/
+
+               if (xen_pcibk_force_msi_disable(dev))
+                       dev->msi_enabled = 1;
+               if (xen_pcibk_force_msix_disable(dev))
+                       dev->msix_enabled = 1;
                if (dev->msix_enabled)
                        pci_disable_msix(dev);
                if (dev->msi_enabled)
@@ -160,9 +165,10 @@ int xen_pcibk_enable_msi(struct xen_pcibk_device *pdev,
                        op->value);
 
        dev_data = pci_get_drvdata(dev);
-       if (dev_data)
+       if (dev_data) {
                dev_data->ack_intr = 0;
-
+               dev_data->msix_enabled = 1;
+       }
        return 0;
 }
 
@@ -182,8 +188,10 @@ int xen_pcibk_disable_msi(struct xen_pcibk_device *pdev,
                printk(KERN_DEBUG DRV_NAME ": %s: MSI: %d\n", pci_name(dev),
                        op->value);
        dev_data = pci_get_drvdata(dev);
-       if (dev_data)
+       if (dev_data) {
                dev_data->ack_intr = 1;
+               dev_data->msi_enabled = 0;
+       }
        return 0;
 }
 
@@ -232,9 +240,10 @@ int xen_pcibk_enable_msix(struct xen_pcibk_device *pdev,
 
        op->value = result;
        dev_data = pci_get_drvdata(dev);
-       if (dev_data)
+       if (dev_data) {
                dev_data->ack_intr = 0;
-
+               dev_data->msix_enabled = 1;
+       }
        return result;
 }
 
@@ -257,8 +266,10 @@ int xen_pcibk_disable_msix(struct xen_pcibk_device *pdev,
                printk(KERN_DEBUG DRV_NAME ": %s: MSI-X: %d\n", pci_name(dev),
                        op->value);
        dev_data = pci_get_drvdata(dev);
-       if (dev_data)
+       if (dev_data) {
                dev_data->ack_intr = 1;
+               dev_data->msix_disabled = 0;
+       }
        return 0;
 }
 #endif

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

<Prev in Thread] Current Thread [Next in Thread>