Message ID | 1456153794-2028-1-git-send-email-hau@realtek.com |
---|---|
State | Superseded, archived |
Delegated to: | David Miller |
Headers | show |
Chunhao Lin <hau@realtek.com> : [...] > diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c > index 537974c..a645f8d 100644 > --- a/drivers/net/ethernet/realtek/r8169.c > +++ b/drivers/net/ethernet/realtek/r8169.c > @@ -7730,10 +7730,13 @@ rtl8169_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats) > { > struct rtl8169_private *tp = netdev_priv(dev); > void __iomem *ioaddr = tp->mmio_addr; > + struct pci_dev *pdev = tp->pci_dev; + struct device *d = &tp->pci_dev->dev; (the patch does not use pdev alone) [...] > @@ -7761,7 +7764,8 @@ rtl8169_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats) > * Fetch additonal counter values missing in stats collected by driver > * from tally counters. > */ > - rtl8169_update_counters(dev); > + if (pm_runtime_active(&pdev->dev)) > + rtl8169_update_counters(dev); pm_runtime_active() won't change after pm_runtime_get_noresume(). You may set a boolean active = pm_runtime_active(d) before testing netif_running(). [...] > @@ -7842,6 +7848,12 @@ static int rtl8169_runtime_suspend(struct device *device) > struct pci_dev *pdev = to_pci_dev(device); > struct net_device *dev = pci_get_drvdata(pdev); > struct rtl8169_private *tp = netdev_priv(dev); > + void __iomem *ioaddr = tp->mmio_addr; > + > + /* Update counters before going runtime suspend */ > + if (netif_running(dev)) > + rtl8169_rx_missed(dev, ioaddr); > + rtl8169_update_counters(dev); > > if (!tp->TxDescArray) > return 0; Nits: - the tp->TxDescArray test provides the required synchronization: see rtl8169_{open/close} and their pm_runtime_{get / put}. - ioaddr is not really needed : tp->mmio_addr appears only once and it does not mess the 72..80 cols limit. - even if the device can only be automatically runtime suspended some time after a link down event, you may address davem's point regarding stats reliability and move rtl8169_rx_missed + rtl8169_update_counters after rtl8169_net_suspend.
> Nits: > > - the tp->TxDescArray test provides the required synchronization: see > rtl8169_{open/close} and their pm_runtime_{get / put}. > > - ioaddr is not really needed : tp->mmio_addr appears only once and it does > not mess the 72..80 cols limit. > > - even if the device can only be automatically runtime suspended some time > after a link down event, you may address davem's point regarding stats > reliability and move rtl8169_rx_missed + rtl8169_update_counters after > rtl8169_net_suspend. I will submit the new patch according to your advice. Thanks. Please consider the environment before printing this e-mail.
diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c index 537974c..a645f8d 100644 --- a/drivers/net/ethernet/realtek/r8169.c +++ b/drivers/net/ethernet/realtek/r8169.c @@ -7730,10 +7730,13 @@ rtl8169_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats) { struct rtl8169_private *tp = netdev_priv(dev); void __iomem *ioaddr = tp->mmio_addr; + struct pci_dev *pdev = tp->pci_dev; struct rtl8169_counters *counters = tp->counters; unsigned int start; - if (netif_running(dev)) + pm_runtime_get_noresume(&pdev->dev); + + if (netif_running(dev) && pm_runtime_active(&pdev->dev)) rtl8169_rx_missed(dev, ioaddr); do { @@ -7761,7 +7764,8 @@ rtl8169_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats) * Fetch additonal counter values missing in stats collected by driver * from tally counters. */ - rtl8169_update_counters(dev); + if (pm_runtime_active(&pdev->dev)) + rtl8169_update_counters(dev); /* * Subtract values fetched during initalization. @@ -7774,6 +7778,8 @@ rtl8169_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats) stats->tx_aborted_errors = le16_to_cpu(counters->tx_aborted) - le16_to_cpu(tp->tc_offset.tx_aborted); + pm_runtime_put_noidle(&pdev->dev); + return stats; } @@ -7842,6 +7848,12 @@ static int rtl8169_runtime_suspend(struct device *device) struct pci_dev *pdev = to_pci_dev(device); struct net_device *dev = pci_get_drvdata(pdev); struct rtl8169_private *tp = netdev_priv(dev); + void __iomem *ioaddr = tp->mmio_addr; + + /* Update counters before going runtime suspend */ + if (netif_running(dev)) + rtl8169_rx_missed(dev, ioaddr); + rtl8169_update_counters(dev); if (!tp->TxDescArray) return 0;
There will be a log spam when there is no cable plugged. Please refer to following links. https://bugzilla.kernel.org/show_bug.cgi?id=104351 https://bugzilla.kernel.org/show_bug.cgi?id=107421 This issue is caused by runtime power management. When there is no cable plugged, the driver will be suspend (runtime suspend) by OS and NIC will be put into the D3 state. During this time, if OS call rtl8169_get_stats64() to dump tally counter, because NIC is in D3 state, the register value read by driver will return all 0xff. This will let driver think tally counter flag is not toggled and then sends the warning message "rtl_counters_cond == 1 (loop: 1000, delay: 10)" to kernel log. For fixing this issue, 1.add checking driver's pm runtime status in rtl8169_get_stats64(). 2.dump tally counter before going runtime suspend for counter accuracy in runtime suspend. Signed-off-by: Chunhao Lin <hau@realtek.com> --- drivers/net/ethernet/realtek/r8169.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-)