v2 ethtool: remove support for ETHTOOL_GRXNTUPLE
authorAlexander Duyck <alexander.h.duyck@intel.com>
Wed, 8 Jun 2011 12:35:08 +0000 (12:35 +0000)
committerDavid S. Miller <davem@davemloft.net>
Wed, 8 Jun 2011 23:45:31 +0000 (16:45 -0700)
This change is meant to remove all support for displaying an ntuple as
strings via ETHTOOL_GRXNTUPLE.  The reason for this change is due to the
fact that multiple issues have been found including:
 - Multiple buffer overruns for strings being displayed.
 - Incorrect filters displayed, cleared filters with ring of -2 are displayed
 - Setting get_rx_ntuple displays no rules if defined.
 - Endianess wrong on displayed values.
 - Hard limit of 1024 filters makes display functionality extremely limited

The only driver that had supported this interface was ixgbe.  Since it no
longer uses the interface and due to the issues mentioned above I am
submitting this patch to remove it.

v2:
Updated based on comments from Ben Hutchings
 - Left ETH_SS_NTUPLE_FILTERS in code but commented on it being deprecated
 - Removed ethtool_rx_ntuple_list and ethtool_rx_ntuple_flow_spec_container
 - Left ETHTOOL_GRXNTUPLE but commented it as deprecated

Also cleaned up set_rx_ntuple since there is no flow spec container to
maintain we can drop all the code for the alloc and free of it and just
return ops->set_rx_ntuple().
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/linux/ethtool.h
include/linux/netdevice.h
net/core/dev.c
net/core/ethtool.c

index c6a850a..dfd3493 100644 (file)
@@ -287,7 +287,7 @@ enum ethtool_stringset {
        ETH_SS_TEST             = 0,
        ETH_SS_STATS,
        ETH_SS_PRIV_FLAGS,
-       ETH_SS_NTUPLE_FILTERS,
+       ETH_SS_NTUPLE_FILTERS,  /* Do not use, GRXNTUPLE is now deprecated */
        ETH_SS_FEATURES,
 };
 
@@ -714,18 +714,6 @@ enum ethtool_sfeatures_retval_bits {
 /* needed by dev_disable_lro() */
 extern int __ethtool_set_flags(struct net_device *dev, u32 flags);
 
-struct ethtool_rx_ntuple_flow_spec_container {
-       struct ethtool_rx_ntuple_flow_spec fs;
-       struct list_head list;
-};
-
-struct ethtool_rx_ntuple_list {
-#define ETHTOOL_MAX_NTUPLE_LIST_ENTRY 1024
-#define ETHTOOL_MAX_NTUPLE_STRING_PER_ENTRY 14
-       struct list_head        list;
-       unsigned int            count;
-};
-
 /**
  * enum ethtool_phys_id_state - indicator state for physical identification
  * @ETHTOOL_ID_INACTIVE: Physical ID indicator should be deactivated
@@ -758,7 +746,6 @@ u32 ethtool_op_get_ufo(struct net_device *dev);
 int ethtool_op_set_ufo(struct net_device *dev, u32 data);
 u32 ethtool_op_get_flags(struct net_device *dev);
 int ethtool_op_set_flags(struct net_device *dev, u32 data, u32 supported);
-void ethtool_ntuple_flush(struct net_device *dev);
 bool ethtool_invalid_flags(struct net_device *dev, u32 data, u32 supported);
 
 /**
@@ -865,7 +852,6 @@ bool ethtool_invalid_flags(struct net_device *dev, u32 data, u32 supported);
  *     error code or zero.
  * @set_rx_ntuple: Set an RX n-tuple rule.  Returns a negative error code
  *     or zero.
- * @get_rx_ntuple: Deprecated.
  * @get_rxfh_indir: Get the contents of the RX flow hash indirection table.
  *     Returns a negative error code or zero.
  * @set_rxfh_indir: Set the contents of the RX flow hash indirection table.
@@ -944,7 +930,6 @@ struct ethtool_ops {
        int     (*reset)(struct net_device *, u32 *);
        int     (*set_rx_ntuple)(struct net_device *,
                                 struct ethtool_rx_ntuple *);
-       int     (*get_rx_ntuple)(struct net_device *, u32 stringset, void *);
        int     (*get_rxfh_indir)(struct net_device *,
                                  struct ethtool_rxfh_indir *);
        int     (*set_rxfh_indir)(struct net_device *,
@@ -1017,7 +1002,7 @@ struct ethtool_ops {
 #define ETHTOOL_FLASHDEV       0x00000033 /* Flash firmware to device */
 #define ETHTOOL_RESET          0x00000034 /* Reset hardware */
 #define ETHTOOL_SRXNTUPLE      0x00000035 /* Add an n-tuple filter to device */
-#define ETHTOOL_GRXNTUPLE      0x00000036 /* Get n-tuple filters from device */
+#define ETHTOOL_GRXNTUPLE      0x00000036 /* deprecated */
 #define ETHTOOL_GSSET_INFO     0x00000037 /* Get string set info */
 #define ETHTOOL_GRXFHINDIR     0x00000038 /* Get RX flow hash indir'n table */
 #define ETHTOOL_SRXFHINDIR     0x00000039 /* Set RX flow hash indir'n table */
index 3362885..6469fa9 100644 (file)
@@ -1348,9 +1348,6 @@ struct net_device {
        /* max exchange id for FCoE LRO by ddp */
        unsigned int            fcoe_ddp_xid;
 #endif
-       /* n-tuple filter list attached to this device */
-       struct ethtool_rx_ntuple_list ethtool_ntuple_list;
-
        /* phy device may attach itself for hardware timestamping */
        struct phy_device *phydev;
 
index 9393078..b3f52d2 100644 (file)
@@ -5867,8 +5867,6 @@ struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
 
        dev->gso_max_size = GSO_MAX_SIZE;
 
-       INIT_LIST_HEAD(&dev->ethtool_ntuple_list.list);
-       dev->ethtool_ntuple_list.count = 0;
        INIT_LIST_HEAD(&dev->napi_list);
        INIT_LIST_HEAD(&dev->unreg_list);
        INIT_LIST_HEAD(&dev->link_watch_list);
@@ -5932,9 +5930,6 @@ void free_netdev(struct net_device *dev)
        /* Flush device addresses */
        dev_addr_flush(dev);
 
-       /* Clear ethtool n-tuple list */
-       ethtool_ntuple_flush(dev);
-
        list_for_each_entry_safe(p, n, &dev->napi_list, dev_list)
                netif_napi_del(p);
 
index fd14116..b7c12a6 100644 (file)
@@ -169,18 +169,6 @@ int ethtool_op_set_flags(struct net_device *dev, u32 data, u32 supported)
 }
 EXPORT_SYMBOL(ethtool_op_set_flags);
 
-void ethtool_ntuple_flush(struct net_device *dev)
-{
-       struct ethtool_rx_ntuple_flow_spec_container *fsc, *f;
-
-       list_for_each_entry_safe(fsc, f, &dev->ethtool_ntuple_list.list, list) {
-               list_del(&fsc->list);
-               kfree(fsc);
-       }
-       dev->ethtool_ntuple_list.count = 0;
-}
-EXPORT_SYMBOL(ethtool_ntuple_flush);
-
 /* Handlers for each ethtool command */
 
 #define ETHTOOL_DEV_FEATURE_WORDS      1
@@ -865,34 +853,6 @@ out:
        return ret;
 }
 
-static void __rx_ntuple_filter_add(struct ethtool_rx_ntuple_list *list,
-                       struct ethtool_rx_ntuple_flow_spec *spec,
-                       struct ethtool_rx_ntuple_flow_spec_container *fsc)
-{
-
-       /* don't add filters forever */
-       if (list->count >= ETHTOOL_MAX_NTUPLE_LIST_ENTRY) {
-               /* free the container */
-               kfree(fsc);
-               return;
-       }
-
-       /* Copy the whole filter over */
-       fsc->fs.flow_type = spec->flow_type;
-       memcpy(&fsc->fs.h_u, &spec->h_u, sizeof(spec->h_u));
-       memcpy(&fsc->fs.m_u, &spec->m_u, sizeof(spec->m_u));
-
-       fsc->fs.vlan_tag = spec->vlan_tag;
-       fsc->fs.vlan_tag_mask = spec->vlan_tag_mask;
-       fsc->fs.data = spec->data;
-       fsc->fs.data_mask = spec->data_mask;
-       fsc->fs.action = spec->action;
-
-       /* add to the list */
-       list_add_tail_rcu(&fsc->list, &list->list);
-       list->count++;
-}
-
 /*
  * ethtool does not (or did not) set masks for flow parameters that are
  * not specified, so if both value and mask are 0 then this must be
@@ -930,8 +890,6 @@ static noinline_for_stack int ethtool_set_rx_ntuple(struct net_device *dev,
 {
        struct ethtool_rx_ntuple cmd;
        const struct ethtool_ops *ops = dev->ethtool_ops;
-       struct ethtool_rx_ntuple_flow_spec_container *fsc = NULL;
-       int ret;
 
        if (!ops->set_rx_ntuple)
                return -EOPNOTSUPP;
@@ -944,269 +902,7 @@ static noinline_for_stack int ethtool_set_rx_ntuple(struct net_device *dev,
 
        rx_ntuple_fix_masks(&cmd.fs);
 
-       /*
-        * Cache filter in dev struct for GET operation only if
-        * the underlying driver doesn't have its own GET operation, and
-        * only if the filter was added successfully.  First make sure we
-        * can allocate the filter, then continue if successful.
-        */
-       if (!ops->get_rx_ntuple) {
-               fsc = kmalloc(sizeof(*fsc), GFP_ATOMIC);
-               if (!fsc)
-                       return -ENOMEM;
-       }
-
-       ret = ops->set_rx_ntuple(dev, &cmd);
-       if (ret) {
-               kfree(fsc);
-               return ret;
-       }
-
-       if (!ops->get_rx_ntuple)
-               __rx_ntuple_filter_add(&dev->ethtool_ntuple_list, &cmd.fs, fsc);
-
-       return ret;
-}
-
-static int ethtool_get_rx_ntuple(struct net_device *dev, void __user *useraddr)
-{
-       struct ethtool_gstrings gstrings;
-       const struct ethtool_ops *ops = dev->ethtool_ops;
-       struct ethtool_rx_ntuple_flow_spec_container *fsc;
-       u8 *data;
-       char *p;
-       int ret, i, num_strings = 0;
-
-       if (!ops->get_sset_count)
-               return -EOPNOTSUPP;
-
-       if (copy_from_user(&gstrings, useraddr, sizeof(gstrings)))
-               return -EFAULT;
-
-       ret = ops->get_sset_count(dev, gstrings.string_set);
-       if (ret < 0)
-               return ret;
-
-       gstrings.len = ret;
-
-       data = kzalloc(gstrings.len * ETH_GSTRING_LEN, GFP_USER);
-       if (!data)
-               return -ENOMEM;
-
-       if (ops->get_rx_ntuple) {
-               /* driver-specific filter grab */
-               ret = ops->get_rx_ntuple(dev, gstrings.string_set, data);
-               goto copy;
-       }
-
-       /* default ethtool filter grab */
-       i = 0;
-       p = (char *)data;
-       list_for_each_entry(fsc, &dev->ethtool_ntuple_list.list, list) {
-               sprintf(p, "Filter %d:\n", i);
-               p += ETH_GSTRING_LEN;
-               num_strings++;
-
-               switch (fsc->fs.flow_type) {
-               case TCP_V4_FLOW:
-                       sprintf(p, "\tFlow Type: TCP\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case UDP_V4_FLOW:
-                       sprintf(p, "\tFlow Type: UDP\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case SCTP_V4_FLOW:
-                       sprintf(p, "\tFlow Type: SCTP\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case AH_ESP_V4_FLOW:
-                       sprintf(p, "\tFlow Type: AH ESP\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case ESP_V4_FLOW:
-                       sprintf(p, "\tFlow Type: ESP\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case IP_USER_FLOW:
-                       sprintf(p, "\tFlow Type: Raw IP\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case IPV4_FLOW:
-                       sprintf(p, "\tFlow Type: IPv4\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               default:
-                       sprintf(p, "\tFlow Type: Unknown\n");
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       goto unknown_filter;
-               }
-
-               /* now the rest of the filters */
-               switch (fsc->fs.flow_type) {
-               case TCP_V4_FLOW:
-               case UDP_V4_FLOW:
-               case SCTP_V4_FLOW:
-                       sprintf(p, "\tSrc IP addr: 0x%x\n",
-                               fsc->fs.h_u.tcp_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tSrc IP mask: 0x%x\n",
-                               fsc->fs.m_u.tcp_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP addr: 0x%x\n",
-                               fsc->fs.h_u.tcp_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP mask: 0x%x\n",
-                               fsc->fs.m_u.tcp_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tSrc Port: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.tcp_ip4_spec.psrc,
-                               fsc->fs.m_u.tcp_ip4_spec.psrc);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest Port: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.tcp_ip4_spec.pdst,
-                               fsc->fs.m_u.tcp_ip4_spec.pdst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tTOS: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.tcp_ip4_spec.tos,
-                               fsc->fs.m_u.tcp_ip4_spec.tos);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case AH_ESP_V4_FLOW:
-               case ESP_V4_FLOW:
-                       sprintf(p, "\tSrc IP addr: 0x%x\n",
-                               fsc->fs.h_u.ah_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tSrc IP mask: 0x%x\n",
-                               fsc->fs.m_u.ah_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP addr: 0x%x\n",
-                               fsc->fs.h_u.ah_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP mask: 0x%x\n",
-                               fsc->fs.m_u.ah_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tSPI: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.ah_ip4_spec.spi,
-                               fsc->fs.m_u.ah_ip4_spec.spi);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tTOS: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.ah_ip4_spec.tos,
-                               fsc->fs.m_u.ah_ip4_spec.tos);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case IP_USER_FLOW:
-                       sprintf(p, "\tSrc IP addr: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tSrc IP mask: 0x%x\n",
-                               fsc->fs.m_u.usr_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP addr: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP mask: 0x%x\n",
-                               fsc->fs.m_u.usr_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               case IPV4_FLOW:
-                       sprintf(p, "\tSrc IP addr: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tSrc IP mask: 0x%x\n",
-                               fsc->fs.m_u.usr_ip4_spec.ip4src);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP addr: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tDest IP mask: 0x%x\n",
-                               fsc->fs.m_u.usr_ip4_spec.ip4dst);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tL4 bytes: 0x%x, mask: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.l4_4_bytes,
-                               fsc->fs.m_u.usr_ip4_spec.l4_4_bytes);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tTOS: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.tos,
-                               fsc->fs.m_u.usr_ip4_spec.tos);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tIP Version: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.ip_ver,
-                               fsc->fs.m_u.usr_ip4_spec.ip_ver);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       sprintf(p, "\tProtocol: %d, mask: 0x%x\n",
-                               fsc->fs.h_u.usr_ip4_spec.proto,
-                               fsc->fs.m_u.usr_ip4_spec.proto);
-                       p += ETH_GSTRING_LEN;
-                       num_strings++;
-                       break;
-               }
-               sprintf(p, "\tVLAN: %d, mask: 0x%x\n",
-                       fsc->fs.vlan_tag, fsc->fs.vlan_tag_mask);
-               p += ETH_GSTRING_LEN;
-               num_strings++;
-               sprintf(p, "\tUser-defined: 0x%Lx\n", fsc->fs.data);
-               p += ETH_GSTRING_LEN;
-               num_strings++;
-               sprintf(p, "\tUser-defined mask: 0x%Lx\n", fsc->fs.data_mask);
-               p += ETH_GSTRING_LEN;
-               num_strings++;
-               if (fsc->fs.action == ETHTOOL_RXNTUPLE_ACTION_DROP)
-                       sprintf(p, "\tAction: Drop\n");
-               else
-                       sprintf(p, "\tAction: Direct to queue %d\n",
-                               fsc->fs.action);
-               p += ETH_GSTRING_LEN;
-               num_strings++;
-unknown_filter:
-               i++;
-       }
-copy:
-       /* indicate to userspace how many strings we actually have */
-       gstrings.len = num_strings;
-       ret = -EFAULT;
-       if (copy_to_user(useraddr, &gstrings, sizeof(gstrings)))
-               goto out;
-       useraddr += sizeof(gstrings);
-       if (copy_to_user(useraddr, data, gstrings.len * ETH_GSTRING_LEN))
-               goto out;
-       ret = 0;
-
-out:
-       kfree(data);
-       return ret;
+       return ops->set_rx_ntuple(dev, &cmd);
 }
 
 static int ethtool_get_regs(struct net_device *dev, char __user *useraddr)
@@ -2101,9 +1797,6 @@ int dev_ethtool(struct net *net, struct ifreq *ifr)
        case ETHTOOL_SRXNTUPLE:
                rc = ethtool_set_rx_ntuple(dev, useraddr);
                break;
-       case ETHTOOL_GRXNTUPLE:
-               rc = ethtool_get_rx_ntuple(dev, useraddr);
-               break;
        case ETHTOOL_GSSET_INFO:
                rc = ethtool_get_sset_info(dev, useraddr);
                break;