Re: [RFC PATCH V3 05/16] netback: add module get/put operations along with vif connect/disconnect.
From: Ian Campbell <hidden>
Date: 2012-01-31 10:24:38
Also in:
xen-devel
On Mon, 2012-01-30 at 14:45 +0000, Wei Liu wrote:
If there is vif running and user unloads netback, it will certainly cause problems -- guest's network interface just mysteriously stops working.
This seems like a bug fix for 02/16 "netback: add module unload function". Please could you fold back such fixes where appropriate? I think there's a handful of these sorts of patches in the series.
quoted hunk ↗ jump to hunk
v2: fix module_put path disconnect function may get called by the generic framework even before vif connects. Tested-by: Konrad Rzeszutek Wilk <redacted> Signed-off-by: Wei Liu <redacted> --- drivers/net/xen-netback/interface.c | 11 ++++++++++- 1 files changed, 10 insertions(+), 1 deletions(-)diff --git a/drivers/net/xen-netback/interface.c b/drivers/net/xen-netback/interface.c index dfc04f8..7914f60 100644 --- a/drivers/net/xen-netback/interface.c +++ b/drivers/net/xen-netback/interface.c@@ -323,6 +323,8 @@ int xenvif_connect(struct xenvif *vif, unsigned long tx_ring_ref, if (vif->irq) return 0; + __module_get(THIS_MODULE); + err = xen_netbk_map_frontend_rings(vif, tx_ring_ref, rx_ring_ref); if (err < 0) goto err;@@ -372,12 +374,14 @@ err_unbind: err_unmap: xen_netbk_unmap_frontend_rings(vif); err: + module_put(THIS_MODULE); return err; } void xenvif_disconnect(struct xenvif *vif) { struct net_device *dev = vif->dev; + int need_module_put = 0; if (netif_carrier_ok(dev)) { rtnl_lock();@@ -397,12 +401,17 @@ void xenvif_disconnect(struct xenvif *vif) del_timer_sync(&vif->credit_timeout); - if (vif->irq) + if (vif->irq) { unbind_from_irqhandler(vif->irq, vif); + need_module_put = 1;
This seems like a slightly odd condition. Why is the put not unconditional?
+ } unregister_netdev(vif->dev); xen_netbk_unmap_frontend_rings(vif); free_netdev(vif->dev); + + if (need_module_put) + module_put(THIS_MODULE); }