On Wed, Jul 21, 2021 at 8:06 AM Julian Wiedmann [off-list ref] wrote:
On 20.07.21 17:46, Arnd Bergmann wrote:
quoted
From: Arnd Bergmann <arnd@arndb.de>
qeth has both standard MII ioctls and custom SIOCDEVPRIVATE ones,
all of which work correctly with compat user space.
Move the private ones over to the new ndo_siocdevprivate callback.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
your get_maintainers scripting seems broken, adding the usual suspects.
Right, I ran the wrong script for sending.
quoted
-int qeth_do_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
+int qeth_siocdevprivate(struct net_device *dev, struct ifreq *rq, void __user *data, int cmd)
{
struct qeth_card *card = dev->ml_priv;
- struct mii_ioctl_data *mii_data;
int rc = 0;
switch (cmd) {
case SIOC_QETH_ADP_SET_SNMP_CONTROL:
- rc = qeth_snmp_command(card, rq->ifr_ifru.ifru_data);
+ rc = qeth_snmp_command(card, data);
break;
case SIOC_QETH_GET_CARD_TYPE:
if ((IS_OSD(card) || IS_OSM(card) || IS_OSX(card)) &&
!IS_VM_NIC(card))
return 1;
return 0;
+ case SIOC_QETH_QUERY_OAT:
+ rc = qeth_query_oat_command(card, data);
+ break;
+ default:
+ if (card->discipline->do_ioctl)
+ rc = card->discipline->do_ioctl(dev, rq, data, cmd);
+ else
+ rc = -EOPNOTSUPP;
+ }
+ if (rc)
+ QETH_CARD_TEXT_(card, 2, "ioce%x", rc);
+ return rc;
+}
+EXPORT_SYMBOL_GPL(qeth_siocdevprivate);
+
Looks like you missed to wire this up in our netdev_ops structs.
Fixed now, thanks! I've gone through the other patches as well
to see if I made the same mistake elsewhere, but it appears this
one was the only time here.
Arnd