W dniu 10 maja 2011 16:09 użytkownik Frank Blaschka <blaschka@xxxxxxxxxxxxxxxxxx> napisał: > On Tue, May 10, 2011 at 03:19:27PM +0200, Michał Mirosław wrote: >> From: Frank Blaschka <frank.blaschka@xxxxxxxxxx> >> > Set rx csum default to hw checksumming again. >> > Remove sysfs interface for rx csum (checksumming) and TSO (large_send). >> > With the new hw_features it does not work to keep the old sysfs >> > interface in parallel. Convert options.checksum_type to new hw_features. >> [...] >> > @@ -1482,32 +1476,28 @@ static int qeth_l3_start_ipa_checksum(st >> [...] >> > - rc = qeth_l3_send_checksum_command(card); >> > - if (!rc) >> > - dev_info(&card->gdev->dev, >> > + card->dev->hw_features &= ~IPA_INBOUND_CHECKSUM; >> > + card->dev->features &= ~IPA_INBOUND_CHECKSUM; >> > + return 0; >> > + } [...] >> Don't modify hw_features. Limit currently available features in >> ndo_fix_features callback instead when checksumming is (temporarily) >> unavailable. >> > We have a recovery operation in our driver. What can we do in case this > recovery detects the checksumming is not available anymore (during runtime > of the device). ndo_fix_features callback does not come into play at this time. > How can we solve this? Just call netdev_update_features() after recovery. >> > static int qeth_l3_set_features(struct net_device *dev, u32 features) >> > { >> > - enum qeth_checksum_types csum_type; >> > struct qeth_card *card = dev->ml_priv; >> > u32 changed = dev->features ^ features; >> > + int on; >> > >> > if (!(changed & NETIF_F_RXCSUM)) >> > return 0; >> > >> > if (features & NETIF_F_RXCSUM) >> > - csum_type = HW_CHECKSUMMING; >> > + on = 1; >> > else >> > - csum_type = SW_CHECKSUMMING; >> > + on = 0; >> > >> > dev->features = features ^ NETIF_F_RXCSUM; >> > - return qeth_l3_set_rx_csum(card, csum_type); >> > + return qeth_l3_set_rx_csum(card, on); >> > } >> >> Since you removed dev->features update from qeth_l3_set_rx_csum(), you >> should also modify this code to match. On exit from ndo_fix_features, >> dev->features should reflect what is currently set, even if part of >> the request failed. > Do you mean qeth_l3_set_rx_csum should set dev->features dependent of > the result of the checksum hardware operation? Yes. If it doesn't, then network core might pass packets needing hardware checksumming for a device with TX checksumming disabled. Best Regards, Michał Mirosław -- To unsubscribe from this list: send the line "unsubscribe linux-s390" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html