Joe, thanks for your patch very much.
A small question is, can we check if (!(dev->msi_enabled)) only once in
pci_disable_msi()?
--jyh
Joe Jin wrote:
> Sorry I lost to set @dev->msi_enabled to false in pci_disable_msi,
> here are the patch, please review and comment:
>
> When device driver unload, it may call pci_disable_msi(), if
> msi did not
> enabled but do msi_unmap_pirq(), then later driver reload and without
> msi, then will failed in request_irq() for irq_desc[irq]->chip valie
> is no_irq_chip. So when did not enable msi during driver
> initializing, then
> unloaded driver will not try to disable it.
>
> How to reproduce it:
> At the server with QLogic 25xx, try to reload qla2xxx will hit it.
>
>
> Signed-off-by: Joe Jin <joe.jin@xxxxxxxxxx>
> ---
> msi-xen.c | 13 +++++++++++++
> 1 file changed, 13 insertions(+)
>
>
> diff -r c5c40e80bd7d drivers/pci/msi-xen.c
> --- a/drivers/pci/msi-xen.c Fri Nov 13 22:01:54 2009 +0000
> +++ b/drivers/pci/msi-xen.c Tue Nov 24 09:56:52 2009 +0800 @@ -618,6
> +618,7 @@ return ret;
>
> dev->irq = evtchn_map_pirq(-1, dev->irq);
> + dev->msi_enabled = 1;
> msi_dev_entry->default_irq = temp;
>
> return ret;
> @@ -662,9 +663,15 @@
>
> #ifdef CONFIG_XEN_PCIDEV_FRONTEND
> if (!is_initial_xendomain()) {
> + if (!(dev->msi_enabled)) {
> + printk(KERN_INFO "PCI: %s: Device did
> not enabled MSI.\n",
> + pci_name(dev));
> + return;
> + }
> evtchn_map_pirq(dev->irq, 0);
> pci_frontend_disable_msi(dev);
> dev->irq = msi_dev_entry->default_irq;
> + dev->msi_enabled = 0;
> return;
> }
> #endif
> @@ -673,6 +680,12 @@
> if (!pos)
> return;
>
> + if (!(dev->msi_enabled)) {
> + printk(KERN_INFO "PCI: %s: Device did not
> enabled MSI.\n",
> + pci_name(dev));
> + return;
> + }
> +
> pirq = dev->irq;
> /* Restore dev->irq to its default pin-assertion vector */
> dev->irq = msi_dev_entry->default_irq;
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home