kni: cleanups and improvements
v2 cover kni: cleanups and improvements
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
This is still mostly "putting lipstick on a pig" all users would be better off using virtio_user rather than KNI.
v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (7):
kni: don't need stubs for rx_mode or ioctl
kni: use netdev_alloc_skb
kni: don't keep stats in kni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return -EFAULT if copy_from_user fails
doc: update KNI documentation
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 21 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- 4 files changed, 38 insertions(+), 97 deletions(-)
Signature · 1 line
2.20.1
PATCH v2 1/7 kni: don't need stubs for rx_mode or ioctl
The netdev subsystem already handles case where
network sevice does not support ioctl.
If device has no rx_mode hook it is not called.
--- kernel/linux/kni/kni_net.c | 19 ------------------- 1 file changed, 19 deletions(-)
kernel/linux/kni/kni_net.c +0 -19
@@ -593,23 +593,6 @@ kni_net_tx_timeout(struct net_device *dev)netif_wake_queue (dev); } -/* -* Ioctl commands -*/ -static int -kni_net_ioctl (struct net_device *dev,struct ifreq *rq,int cmd) -{ -pr_debug ("kni_net_ioctl group:%d cmd:%d\n" , - ((struct kni_dev *)netdev_priv (dev))->group_id, cmd); - -return -EOPNOTSUPP ; -} - -static void -kni_net_set_rx_mode (struct net_device *dev) -{ -} -static int kni_net_change_mtu (struct net_device *dev,int new_mtu) {
@@ -758,8 +741,6 @@ static const struct net_device_ops kni_net_netdev_ops = { .ndo_change_rx_flags = kni_net_set_promiscusity, .ndo_start_xmit = kni_net_tx, .ndo_change_mtu = kni_net_change_mtu, - .ndo_do_ioctl = kni_net_ioctl, - .ndo_set_rx_mode = kni_net_set_rx_mode, .ndo_get_stats = kni_net_stats, .ndo_tx_timeout = kni_net_tx_timeout, .ndo_set_mac_address = kni_net_set_mac,
Signature · 1 line
2.20.1
PATCH v2 2/7 kni: use netdev_alloc_skb
In reply to Stephen Hemmingernetdev_alloc_skb is optimized to any alignment or setup of skb->dev that is required. The kernel intentionall does not pad packets on x86, because it is faster.
--- kernel/linux/kni/kni_net.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)
kernel/linux/kni/kni_net.c +3 -14
@@ -340,16 +340,13 @@ kni_net_rx_normal(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (!skb) {/* Update statistics */ kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -368,7 +365,6 @@ kni_net_rx_normal(struct kni_dev *kni) } } - skb->dev = dev; skb->protocol =eth_type_trans (skb, dev); skb->ip_summed =CHECKSUM_UNNECESSARY ;
@@ -512,26 +508,20 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb) { -/* Align IP on 16B boundary */ -skb_reserve (skb,2 );memcpy (skb_put (skb, len), data_kva, len); - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ;dev_kfree_skb (skb); }/* Simulate real usage, allocate/copy skb twice */ - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb ==NULL ) { kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -550,7 +540,6 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) } } - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ; kni->stats.rx_bytes += len;
Signature · 1 line
2.20.1
PATCH v2 3/7 kni: don't keep stats in kni_net
In reply to Stephen HemmingerSince kernel 2.6.28 the network subsystem has provided dev->stats for devices to use statistics handling and is the default if no ndo_get_stats is provided.
This allow allows for 64 bit (rather than just 32 bit) statistics with KNI.
--- kernel/linux/kni/kni_dev.h | 1 - kernel/linux/kni/kni_net.c | 43 +++++++++++++------------------------- 2 files changed, 15 insertions(+), 29 deletions(-)
kernel/linux/kni/kni_dev.h +0 -1
@@ -39,7 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -struct net_device_stats stats;int status;uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */
kernel/linux/kni/kni_net.c +15 -28
118 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v2 4/7 kni: drop unused fields
In reply to Stephen HemmingerThe kni net structure only exists in driver no API/ABI. Several fields were either totally unused or set and never used.
--- kernel/linux/kni/kni_dev.h | 8 -------- kernel/linux/kni/kni_misc.c | 1 - 2 files changed, 9 deletions(-)
kernel/linux/kni/kni_dev.h +0 -8
@@ -39,8 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -int status; -uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */ char name[RTE_KNI_NAMESIZE ];/* Network device name */ struct task_struct *pthread;
@@ -49,9 +47,6 @@ struct kni_dev {wait_queue_head_t wq;struct mutex sync_lock; -/* PCI device id */ -uint16_t device_id; -/* kni device */ struct net_device *net_dev;
@@ -82,9 +77,6 @@ struct kni_dev {/* mbuf size */ uint32_t mbuf_size; -/* synchro for request processing */ -unsigned long synchro; -/* buffers */ void *pa[MBUF_BURST_SZ ];void *va[MBUF_BURST_SZ ];
kernel/linux/kni/kni_misc.c +0 -1
@@ -346,7 +346,6 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num, kni =netdev_priv (net_dev); kni->net_dev = net_dev; - kni->group_id = dev_info.group_id; kni->core_id = dev_info.core_id;strncpy (kni->name, dev_info.name,RTE_KNI_NAMESIZE );
Signature · 1 line
2.20.1
PATCH v2 5/7 kni: use proper type for kni fifo's
In reply to Stephen HemmingerUsing void * instead of proper type is unsafe practice.
--- kernel/linux/kni/kni_dev.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/linux/kni/kni_dev.h +6 -6
@@ -51,22 +51,22 @@ struct kni_dev {struct net_device *net_dev;/* queue for packets to be sent out */ -void *tx_q; +struct rte_kni_fifo *tx_q;/* queue for the packets received */ -void *rx_q; +struct rte_kni_fifo *rx_q;/* queue for the allocated mbufs those can be used to save sk buffs */ -void *alloc_q; +struct rte_kni_fifo *alloc_q;/* free queue for the mbufs to be freed */ -void *free_q; +struct rte_kni_fifo *free_q;/* request queue */ -void *req_q; +struct rte_kni_fifo *req_q;/* response queue */ -void *resp_q; +struct rte_kni_fifo *resp_q;void *sync_kva;void *sync_va;
Signature · 1 line
2.20.1
PATCH v2 6/7 kni: return -EFAULT if copy_from_user fails
In reply to Stephen HemmingerThe correct thing to return if user gives a bad data is to return -EFAULT. Logging is also discouraged because it could be used as a DoS attack.
--- kernel/linux/kni/kni_misc.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
kernel/linux/kni/kni_misc.c +5 -11
@@ -301,11 +301,8 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num,return -EINVAL ;/* Copy kni info from user space */ - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_create" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Check if name is zero-ended */ if (strnlen (dev_info.name,sizeof (dev_info.name)) ==sizeof (dev_info.name)) {
@@ -433,15 +430,12 @@ kni_ioctl_release(struct net *net, uint32_t ioctl_num,if (_IOC_SIZE (ioctl_num) >sizeof (dev_info))return -EINVAL ; - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_release" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Release the network device according to its name */ if (strlen (dev_info.name) ==0 ) -return ret; +return -EINVAL ;down_write (&knet->kni_list_lock);list_for_each_entry_safe (dev, n, &knet->kni_list_head, list) {
Signature · 1 line
2.20.1
PATCH v2 7/7 doc: update KNI documentation
In reply to Stephen HemmingerMake the KNI documentation reflect modern kernel networking. Ifconfig has been superseded by iproute2 for 15 years. Iproute2 is well maintained, supports current feature set.
Ethtool is no longer supported by KNI.
Tshark is a better replacement for tcpdump.
--- .../sample_app_ug/kernel_nic_interface.rst | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
doc/guides/sample_app_ug/kernel_nic_interface.rst +9 -9
@@ -21,14 +21,14 @@ The FIFO queues contain pointers to data packets in the DPDK. This: * Provides a faster mechanism to interface with the kernel net stack and eliminates system calls -* Facilitates the DPDK using standard Linux* userspace net tools (tcpdump, ftp, and so on) +* Facilitates the DPDK using standard Linux* userspace net tools (tshark, rsync, and so on) * Eliminate the copy_to_user and copy_from_user operations on packets. The Kernel NIC Interface sample application is a simple example that demonstrates the use of the DPDK to create a path for packets to go through the Linux* kernel. This is done by creating one or more kernel net devices for each of the DPDK ports. -The application allows the use of standard Linux tools (ethtool, ifconfig, tcpdump) with the DPDK ports and +The application allows the use of standard Linux tools (iproute, tshark) with the DPDK ports and also the exchange of packets between the DPDK application and the Linux* kernel. The Kernel NIC Interface sample application requires that the
@@ -220,13 +220,13 @@ Enable KNI interface and assign an IP address: .. code-block:: console - # ifconfig vEth0_0 192.168.0.1 + # ip addr add dev vEth0_0 192.168.0.1 Show KNI interface configuration and statistics: .. code-block:: console - # ifconfig vEth0_0 + # ip -s -d addr show vEth0_0 The user can also check and reset the packet statistics inside the ``kni`` application by sending the app the USR1 and USR2 signals:
@@ -234,16 +234,16 @@ application by sending the app the USR1 and USR2 signals: .. code-block:: console # Print statistics - # kill -SIGUSR1 `pidof kni` + # pkill -USR1 kni # Zero statistics - # kill -SIGUSR2 `pidof kni` + # pkill -USR2 kni Dump network traffic: .. code-block:: console - # tcpdump -i vEth0_0 + # tshark -n -i vEth0_0 The normal Linux commands can also be used to change the MAC address and MTU size used by the physical NIC which corresponds to the KNI interface.
@@ -254,13 +254,13 @@ Change the MAC address: .. code-block:: console - # ifconfig vEth0_0 hw ether 0C:01:02:03:04:08 + # ip link set dev vEth0_0 lladdr 0C:01:02:03:04:08 Change the MTU size: .. code-block:: console - # ifconfig vEth0_0 mtu 1450 + # ip link set dev vEth0_0 mtu 1450 When the ``kni`` application is closed, all the KNI interfaces are deleted from the Linux kernel.
Signature · 1 line
2.20.1
Quoted 17 lines · On Mon, 10 Jun 2019 10:51:48 -0700 Stephen Hemminger…
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
This is still mostly "putting lipstick on a pig" all users would be better off using
virtio_userrather than KNI.v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (7):
kni: don't need stubs forrx_modeor ioctl
kni: usenetdev_alloc_skb
kni: don't keep stats inkni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return-EFAULTifcopy_from_userfails
doc: update KNI documentation
Quoted 6 lines
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 21 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- 4 files changed, 38 insertions(+), 97 deletions(-)
Don't believe patchwork the patch is fine, it is getting falsely blamed for failures caused by other changes in ice, bnxt which fail FreeBSD build.
Quoted 28 lines · Stephen Hemminger
Quoted 17 lines · On Mon, 10 Jun 2019 10:51:48 -0700 Stephen Hemminger…
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
This is still mostly "putting lipstick on a pig" all users would be better off using
virtio_userrather than KNI.v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (7):
kni: don't need stubs forrx_modeor ioctl
kni: usenetdev_alloc_skb
kni: don't keep stats inkni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return-EFAULTifcopy_from_userfails
doc: update KNI documentationQuoted 6 lines
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 21 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- 4 files changed, 38 insertions(+), 97 deletions(-)
Quoted 3 lines
Don't believe patchwork the patch is fine, it is getting falsely blamed for failures caused by other changes in ice, bnxt which fail FreeBSD build.
Do you mean failures like the ones below? If so, I think think they might be an unintended
consequence of commit a385972c3 "mk: disable warning for packed member pointer".
Lance
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-gcc
CC ice_rxtx.o
CC ice_ethdev.o
cc1: error: unrecognized command line option
'-Wno-address-of-packed-member' [-Werror]
cc1: all warnings being treated as errors
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/e6c3bc062b1045128114bdf0cfdd236b/dpdk/mk/internal/rte.compile-pre.mk:114:
bnxt_ethdev.o] Error 1
gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/e6c3bc062b1045128114bdf0cfdd236b/dpdk/mk/rte.subdir.mk:37:
bnxt] Error 2
gmake[5]: *** Waiting for unfinished jobs....
Quoted 40 lines · On Tue, 11 Jun 2019 17:18:42 -0400 Lance Richardson…
Quoted 28 lines · Stephen Hemminger
Quoted 17 lines · On Mon, 10 Jun 2019 10:51:48 -0700 Stephen Hemminger…
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
This is still mostly "putting lipstick on a pig" all users would be better off using
virtio_userrather than KNI.v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (7):
kni: don't need stubs forrx_modeor ioctl
kni: usenetdev_alloc_skb
kni: don't keep stats inkni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return-EFAULTifcopy_from_userfails
doc: update KNI documentationQuoted 6 lines
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 21 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- 4 files changed, 38 insertions(+), 97 deletions(-)Quoted 3 lines
Don't believe patchwork the patch is fine, it is getting falsely blamed for failures caused by other changes in ice, bnxt which fail FreeBSD build.
Do you mean failures like the ones below? If so, I think think they might be an unintended
consequence of commita385972c3"mk: disable warning for packed member pointer".Lance
Quoted 12 lines
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-gcc
CC ice_rxtx.o
CC ice_ethdev.o
cc1: error: unrecognized command line option
'-Wno-address-of-packed-member' [-Werror]
cc1: all warnings being treated as errors
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/e6c3bc062b1045128114bdf0cfdd236b/dpdk/mk/internal/rte.compile-pre.mk:114:
bnxt_ethdev.o] Error 1
gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/e6c3bc062b1045128114bdf0cfdd236b/dpdk/mk/rte.subdir.mk:37:
bnxt] Error 2
gmake[5]: *** Waiting for unfinished jobs....
More than just that
*Build Failed Build #1:
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-gcc+debug
CC ipn3ke_flow.o
CC lio_23xx_vf.o
cc1: error: unrecognized command line option '-Wno-address-of-packed-member' [-Werror]
cc1: all warnings being treated as errors
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/internal/rte.compile-pre.mk:116: bnxt_ethdev.o] Error 1 gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:37: bnxt] Error 2 gmake[5]: *** Waiting for unfinished jobs....
CC i40e_flow.o
CC rte_eth_null.o
--
LD ixgbe_ethdev.o
AR librte_pmd_ixgbe.a
INSTALL-LIB librte_pmd_ixgbe.a
gmake[4]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:35: net] Error 2 gmake[3]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkbuild.mk:46: drivers] Error 2 gmake[2]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:99: all] Error 2 gmake[1]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkinstall.mk:61: pre_install] Error 2
gmake: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+debug/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:77: install] Error 2
*Build Failed Build #2:
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-gcc+shared
CC rte_eth_null.o.pmd.o
LD rte_eth_null.o
cc1: error: unrecognized command line option '-Wno-address-of-packed-member' [-Werror]
cc1: all warnings being treated as errors
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/internal/rte.compile-pre.mk:116: bnxt_ethdev.o] Error 1 gmake[6]: *** Waiting for unfinished jobs....
CC ice_rxtx.o
== Build drivers/net/ring
--
LD ice_ethdev.o
CC ice_rxtx_vec_sse.o
CC octeontx_pkovf.o
gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:35: bnxt] Error 2 gmake[5]: *** Waiting for unfinished jobs....
CC i40e_tm.o
CC ice_rxtx_vec_avx2.o
--
INSTALL-LIB librte_pmd_ixgbe.so.2.1
LD librte_pmd_qede.so.1.1
INSTALL-LIB librte_pmd_qede.so.1.1
gmake[4]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:35: net] Error 2 gmake[3]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkbuild.mk:46: drivers] Error 2 gmake[2]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:99: all] Error 2 gmake[1]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkinstall.mk:61: pre_install] Error 2
gmake: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc+shared/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:77: install] Error 2
*Build Failed Build #3:
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-gcc
CC ice_dcb.o
LD rte_eth_null.o
cc1: error: unrecognized command line option '-Wno-address-of-packed-member' [-Werror]
cc1: all warnings being treated as errors
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/internal/rte.compile-pre.mk:114: bnxt_ethdev.o] Error 1 gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:37: bnxt] Error 2 gmake[5]: *** Waiting for unfinished jobs....
CC ixgbe_phy.o
CC octeontx_rxtx.o
--
LD qede_ethdev.o
AR librte_pmd_qede.a
INSTALL-LIB librte_pmd_qede.a
gmake[4]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:35: net] Error 2 gmake[3]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkbuild.mk:46: drivers] Error 2 gmake[2]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:99: all] Error 2 gmake[1]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkinstall.mk:61: pre_install] Error 2
gmake: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:77: install] Error 2
*Build Failed Build #4:
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-clang
#define roundup(x, y) ((((x)+((y)-1))/(y))*(y)) /* to any y */
^1 error generated.
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/internal/rte.compile-pre.mk:114: bnxt_ethdev.o] Error 1
CC ice_flex_pipe.o
gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:35: bnxt] Error 2
gmake[5]: *** Waiting for unfinished jobs....
CC lio_mbox.o
AR librte_pmd_null.a
--
LD ixgbe_ethdev.o
AR librte_pmd_ixgbe.a
INSTALL-LIB librte_pmd_ixgbe.a
gmake[4]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.subdir.mk:35: net] Error 2
gmake[3]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkbuild.mk:46: drivers] Error 2
gmake[2]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:99: all] Error 2
gmake[1]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkinstall.mk:61: pre_install] Error 2
gmake: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/rte.sdkroot.mk:77: install] Error 2
*Meson Failed Build #1:
OS: UB1604-32
Target:build-gcc-static
FAILED: examples/c590b3c@@dpdk-rxtx_callbacks at exe/rxtx_callbacks_main.c.o
gcc -Iexamples/c590b3c@@dpdk-rxtx_callbacks at exe -Iexamples -I../examples -Iexamples/rxtx_callbacks -I../examples/rxtx_callbacks -I. -I../ -Iconfig -I../config -Ilib/librte_eal/common/include -I../lib/librte_eal/common/include -I../lib/librte_eal/linux/eal/include -Ilib/librte_eal/common -I../lib/librte_eal/common -Ilib/librte_eal/common/include/arch/x86 -I../lib/librte_eal/common/include/arch/x86 -Ilib/librte_eal -I../lib/librte_eal -Ilib/librte_kvargs -I../lib/librte_kvargs -Ilib/librte_mempool -I../lib/librte_mempool -Ilib/librte_ring -I../lib/librte_ring -Ilib/librte_net -I../lib/librte_net -Ilib/librte_mbuf -I../lib/librte_mbuf -Ilib/librte_ethdev -I../lib/librte_ethdev -Ilib/librte_cmdline -I../lib/librte_cmdline -Ilib/librte_meter -I../lib/librte_meter -fdiagnostics-color=always -pipe -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -Werror -O3 -include rte_config.h -Wunused-parameter -Wsign-compare -Wcast-qual -Wno-pointer-to-int-cast -march=native -D_GNU_SOURCE -DALLOW_EXPERIMENTAL_API -MD -MQ 'examples/c590b3c@@dpdk-rxtx_callbacks at exe/rxtx_callbacks_main.c.o' -MF 'examples/c590b3c@@dpdk-rxtx_callbacks at exe/rxtx_callbacks_main.c.o.d' -o 'examples/c590b3c@@dpdk-rxtx_callbacks at exe/rxtx_callbacks_main.c.o' -c ../examples/rxtx_callbacks/main.c ../examples/rxtx_callbacks/main.c: In function ‘port_init’:../examples/rxtx_callbacks/main.c:172:10: error: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 2 has type ‘uint64_t {aka long long unsigned int}’ [-Werror=format=]
printf("TSC Freq ~= %lu\nHW Freq ~= %lu\nRatio : %f\n",
^
../examples/rxtx_callbacks/main.c:172:10: error: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 3 has type ‘uint64_t {aka long long unsigned int}’ [-Werror=format=]
cc1: all warnings being treated as errors
[1531/1549] Linking target examples/dpdk-vdpa.
[1532/1549] Linking target examples/dpdk-timer.
[1533/1549] Linking target examples/dpdk-tep_termination.
[1534/1549] Compiling C object 'examples/c590b3c@@dpdk-vhost_crypto at exe/vhost_crypto_main.c.o'.
[1535/1549] Compiling C object 'examples/c590b3c@@dpdk-vhost at exe/vhost_main.c.o'.
ninja: build stopped: subcommand failed
Quoted 58 lines · Stephen Hemminger
Quoted 40 lines · On Tue, 11 Jun 2019 17:18:42 -0400 Lance Richardson…
Quoted 28 lines · Stephen Hemminger
Quoted 17 lines · On Mon, 10 Jun 2019 10:51:48 -0700 Stephen Hemminger…
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
This is still mostly "putting lipstick on a pig" all users would be better off using
virtio_userrather than KNI.v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (7):
kni: don't need stubs forrx_modeor ioctl
kni: usenetdev_alloc_skb
kni: don't keep stats inkni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return-EFAULTifcopy_from_userfails
doc: update KNI documentationQuoted 6 lines
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 21 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- 4 files changed, 38 insertions(+), 97 deletions(-)Quoted 3 lines
Don't believe patchwork the patch is fine, it is getting falsely blamed for failures caused by other changes in ice, bnxt which fail FreeBSD build.
Do you mean failures like the ones below? If so, I think think they might be an unintended
consequence of commita385972c3"mk: disable warning for packed member pointer".Lance
Quoted 12 lines
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-gcc
CC ice_rxtx.o
CC ice_ethdev.o
cc1: error: unrecognized command line option
'-Wno-address-of-packed-member' [-Werror]
cc1: all warnings being treated as errors
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/e6c3bc062b1045128114bdf0cfdd236b/dpdk/mk/internal/rte.compile-pre.mk:114:
bnxt_ethdev.o] Error 1
gmake[5]: *** [/tmp/FreeBSD12-64_K19.02_GCC7.3.0/x86_64-native-bsdapp-gcc/e6c3bc062b1045128114bdf0cfdd236b/dpdk/mk/rte.subdir.mk:37:
bnxt] Error 2
gmake[5]: *** Waiting for unfinished jobs....
Quoted 1 line
More than just that
OK, I see now... I'll spin a fix for this one:
Quoted 6 lines
OS: FreeBSD12-64
Target: x86_64-native-bsdapp-clang#define roundup(x, y) ((((x)+((y)-1))/(y))*(y)) /* to any y */ ^1 error generated.
gmake[6]: *** [/tmp/FreeBSD12-64_K19.02_Clang6.0.1/x86_64-native-bsdapp-clang/025b54b548ed4ca894cbf5e754039b68/dpdk/mk/internal/rte.compile-pre.mk:114: bnxt_ethdev.o] Error 1
Thanks,
Lance
v3 cover kni: fixes and cleanups
In reply to Stephen HemmingerWhile testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
v3 - rebase to current master, add style fix patch
v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (8):
kni: don't need stubs for rx_mode or ioctl
kni: use netdev_alloc_skb
kni: don't keep stats in kni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return -EFAULT if copy_from_user fails
doc: update KNI documentation
kni: fix style issues
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 18 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- lib/librte_kni/rte_kni.c | 30 +++---- 5 files changed, 53 insertions(+), 109 deletions(-)
Signature · 1 line
2.20.1
PATCH v3 1/8 kni: don't need stubs for rx_mode or ioctl
The netdev subsystem already handles case where
network sevice does not support ioctl.
If device has no rx_mode hook it is not called.
--- kernel/linux/kni/kni_net.c | 19 ------------------- 1 file changed, 19 deletions(-)
kernel/linux/kni/kni_net.c +0 -19
@@ -593,23 +593,6 @@ kni_net_tx_timeout(struct net_device *dev)netif_wake_queue (dev); } -/* -* Ioctl commands -*/ -static int -kni_net_ioctl (struct net_device *dev,struct ifreq *rq,int cmd) -{ -pr_debug ("kni_net_ioctl group:%d cmd:%d\n" , - ((struct kni_dev *)netdev_priv (dev))->group_id, cmd); - -return -EOPNOTSUPP ; -} - -static void -kni_net_set_rx_mode (struct net_device *dev) -{ -} -static int kni_net_change_mtu (struct net_device *dev,int new_mtu) {
@@ -758,8 +741,6 @@ static const struct net_device_ops kni_net_netdev_ops = { .ndo_change_rx_flags = kni_net_set_promiscusity, .ndo_start_xmit = kni_net_tx, .ndo_change_mtu = kni_net_change_mtu, - .ndo_do_ioctl = kni_net_ioctl, - .ndo_set_rx_mode = kni_net_set_rx_mode, .ndo_get_stats = kni_net_stats, .ndo_tx_timeout = kni_net_tx_timeout, .ndo_set_mac_address = kni_net_set_mac,
Signature · 1 line
2.20.1
PATCH v3 2/8 kni: use netdev_alloc_skb
In reply to Stephen Hemmingernetdev_alloc_skb is optimized to any alignment or setup of skb->dev that is required. The kernel has chosen to not pad packets on x86 (for many years), because it is faster.
--- kernel/linux/kni/kni_net.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)
kernel/linux/kni/kni_net.c +3 -14
@@ -340,16 +340,13 @@ kni_net_rx_normal(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (!skb) {/* Update statistics */ kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -368,7 +365,6 @@ kni_net_rx_normal(struct kni_dev *kni) } } - skb->dev = dev; skb->protocol =eth_type_trans (skb, dev); skb->ip_summed =CHECKSUM_UNNECESSARY ;
@@ -512,26 +508,20 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb) { -/* Align IP on 16B boundary */ -skb_reserve (skb,2 );memcpy (skb_put (skb, len), data_kva, len); - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ;dev_kfree_skb (skb); }/* Simulate real usage, allocate/copy skb twice */ - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb ==NULL ) { kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -550,7 +540,6 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) } } - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ; kni->stats.rx_bytes += len;
Signature · 1 line
2.20.1
PATCH v3 3/8 kni: don't keep stats in kni_net
In reply to Stephen HemmingerSince kernel 2.6.28 the network subsystem has provided dev->stats for devices to use statistics handling and is the default if no ndo_get_stats is provided.
This allow allows for 64 bit (rather than just 32 bit) statistics with KNI.
--- kernel/linux/kni/kni_dev.h | 1 - kernel/linux/kni/kni_net.c | 43 +++++++++++++------------------------- 2 files changed, 15 insertions(+), 29 deletions(-)
kernel/linux/kni/kni_dev.h +0 -1
@@ -39,7 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -struct net_device_stats stats;int status;uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */
kernel/linux/kni/kni_net.c +15 -28
118 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v3 4/8 kni: drop unused fields
In reply to Stephen HemmingerThe kni net structure only exists in driver no API/ABI. Several fields were either totally unused or set and never used.
The fields in dev_info do need to stay in the ABI but kernel can ignore them.
kni: drop unused status element
Yet another ethtool leftover.
--- kernel/linux/kni/kni_dev.h | 5 ----- kernel/linux/kni/kni_misc.c | 1 - 2 files changed, 6 deletions(-)
kernel/linux/kni/kni_dev.h +0 -5
@@ -39,8 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -int status; -uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */ char name[RTE_KNI_NAMESIZE ];/* Network device name */ struct task_struct *pthread;
@@ -79,9 +77,6 @@ struct kni_dev {/* mbuf size */ uint32_t mbuf_size; -/* synchro for request processing */ -unsigned long synchro; -/* buffers */ void *pa[MBUF_BURST_SZ ];void *va[MBUF_BURST_SZ ];
kernel/linux/kni/kni_misc.c +0 -1
@@ -346,7 +346,6 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num, kni =netdev_priv (net_dev); kni->net_dev = net_dev; - kni->group_id = dev_info.group_id; kni->core_id = dev_info.core_id;strncpy (kni->name, dev_info.name,RTE_KNI_NAMESIZE );
Signature · 1 line
2.20.1
PATCH v3 5/8 kni: use proper type for kni fifo's
In reply to Stephen HemmingerUsing void * instead of proper type is unsafe practice.
--- kernel/linux/kni/kni_dev.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/linux/kni/kni_dev.h +6 -6
@@ -51,22 +51,22 @@ struct kni_dev {struct net_device *net_dev;/* queue for packets to be sent out */ -void *tx_q; +struct rte_kni_fifo *tx_q;/* queue for the packets received */ -void *rx_q; +struct rte_kni_fifo *rx_q;/* queue for the allocated mbufs those can be used to save sk buffs */ -void *alloc_q; +struct rte_kni_fifo *alloc_q;/* free queue for the mbufs to be freed */ -void *free_q; +struct rte_kni_fifo *free_q;/* request queue */ -void *req_q; +struct rte_kni_fifo *req_q;/* response queue */ -void *resp_q; +struct rte_kni_fifo *resp_q;void *sync_kva;void *sync_va;
Signature · 1 line
2.20.1
PATCH v3 6/8 kni: return -EFAULT if copy_from_user fails
In reply to Stephen HemmingerThe correct thing to return if user gives a bad data is to return -EFAULT. Logging is also discouraged because it could be used as a DoS attack.
--- kernel/linux/kni/kni_misc.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
kernel/linux/kni/kni_misc.c +5 -11
@@ -301,11 +301,8 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num,return -EINVAL ;/* Copy kni info from user space */ - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_create" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Check if name is zero-ended */ if (strnlen (dev_info.name,sizeof (dev_info.name)) ==sizeof (dev_info.name)) {
@@ -427,15 +424,12 @@ kni_ioctl_release(struct net *net, uint32_t ioctl_num,if (_IOC_SIZE (ioctl_num) >sizeof (dev_info))return -EINVAL ; - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_release" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Release the network device according to its name */ if (strlen (dev_info.name) ==0 ) -return ret; +return -EINVAL ;down_write (&knet->kni_list_lock);list_for_each_entry_safe (dev, n, &knet->kni_list_head, list) {
Signature · 1 line
2.20.1
PATCH v3 7/8 doc: update KNI documentation
In reply to Stephen HemmingerMake the KNI documentation reflect modern kernel networking. Ifconfig has been superseded by iproute2 for 15 years. Iproute2 is well maintained, supports current feature set.
Ethtool is no longer supported by KNI.
Tshark is a better replacement for tcpdump.
--- .../sample_app_ug/kernel_nic_interface.rst | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
doc/guides/sample_app_ug/kernel_nic_interface.rst +9 -9
@@ -21,14 +21,14 @@ The FIFO queues contain pointers to data packets in the DPDK. This: * Provides a faster mechanism to interface with the kernel net stack and eliminates system calls -* Facilitates the DPDK using standard Linux* userspace net tools (tcpdump, ftp, and so on) +* Facilitates the DPDK using standard Linux* userspace net tools (tshark, rsync, and so on) * Eliminate the copy_to_user and copy_from_user operations on packets. The Kernel NIC Interface sample application is a simple example that demonstrates the use of the DPDK to create a path for packets to go through the Linux* kernel. This is done by creating one or more kernel net devices for each of the DPDK ports. -The application allows the use of standard Linux tools (ethtool, ifconfig, tcpdump) with the DPDK ports and +The application allows the use of standard Linux tools (iproute, tshark) with the DPDK ports and also the exchange of packets between the DPDK application and the Linux* kernel. The Kernel NIC Interface sample application requires that the
@@ -220,13 +220,13 @@ Enable KNI interface and assign an IP address: .. code-block:: console - # ifconfig vEth0_0 192.168.0.1 + # ip addr add dev vEth0_0 192.168.0.1 Show KNI interface configuration and statistics: .. code-block:: console - # ifconfig vEth0_0 + # ip -s -d addr show vEth0_0 The user can also check and reset the packet statistics inside the ``kni`` application by sending the app the USR1 and USR2 signals:
@@ -234,16 +234,16 @@ application by sending the app the USR1 and USR2 signals: .. code-block:: console # Print statistics - # kill -SIGUSR1 `pidof kni` + # pkill -USR1 kni # Zero statistics - # kill -SIGUSR2 `pidof kni` + # pkill -USR2 kni Dump network traffic: .. code-block:: console - # tcpdump -i vEth0_0 + # tshark -n -i vEth0_0 The normal Linux commands can also be used to change the MAC address and MTU size used by the physical NIC which corresponds to the KNI interface.
@@ -254,13 +254,13 @@ Change the MAC address: .. code-block:: console - # ifconfig vEth0_0 hw ether 0C:01:02:03:04:08 + # ip link set dev vEth0_0 lladdr 0C:01:02:03:04:08 Change the MTU size: .. code-block:: console - # ifconfig vEth0_0 mtu 1450 + # ip link set dev vEth0_0 mtu 1450 When the ``kni`` application is closed, all the KNI interfaces are deleted from the Linux kernel.
Signature · 1 line
2.20.1
PATCH v3 8/8 kni: fix style issues
In reply to Stephen Hemmingerrte_kni does not follow standard style rules.
Noticed some extra \ line continuation etc.
--- lib/librte_kni/rte_kni.c | 30 +++++++++++++++--------------- 1 file changed, 15 insertions(+), 15 deletions(-)
lib/librte_kni/rte_kni.c +15 -15
@@ -59,7 +59,7 @@ struct rte_kni {uint16_t group_id;/**< Group ID of KNI devices */ uint32_t slot_id;/**< KNI pool slot ID */ struct rte_mempool *pktmbuf_pool;/**< pkt mbuf mempool */ -unsigned mbuf_size;/**< mbuf size */ +unsigned int mbuf_size;/**< mbuf size */ const struct rte_memzone *m_tx_q;/**< TX queue memzone */ const struct rte_memzone *m_rx_q;/**< RX queue memzone */
@@ -78,7 +78,7 @@ struct rte_kni {/* For request & response */ struct rte_kni_fifo *req_q;/**< Request queue */ struct rte_kni_fifo *resp_q;/**< Response queue */ -void * sync_addr;/**< Req/Resp Mem address */ +void *sync_addr;/**< Req/Resp Mem address */ struct rte_kni_ops ops;/**< operations for request */ };
@@ -473,7 +473,7 @@ kni_config_promiscusity(uint16_t port_id, uint8_t to_on)int rte_kni_handle_request (struct rte_kni *kni) { -unsigned ret; +unsigned int ret;struct rte_kni_request *req =NULL ;if (kni ==NULL )
@@ -498,8 +498,8 @@ rte_kni_handle_request(struct rte_kni *kni)break ;case RTE_KNI_REQ_CFG_NETWORK_IF :/* Set network interface up/down */ if (kni->ops.config_network_if) - req->result = kni->ops.config_network_if (\ - kni->ops.port_id, req->if_up); + req->result = kni->ops.config_network_if (kni->ops.port_id, + req->if_up);break ;case RTE_KNI_REQ_CHANGE_MAC_ADDR :/* Change MAC Address */ if (kni->ops.config_mac_address)
@@ -534,7 +534,7 @@ rte_kni_handle_request(struct rte_kni *kni) }unsigned -rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) {void *phy_mbufs[num];unsigned int ret;
@@ -552,9 +552,9 @@ rte_kni_tx_burst(struct rte_kni *kni, struct rte_mbuf **mbufs, unsigned num) }unsigned -rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) { -unsigned ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num); +unsigned int ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num);/* If buffers removed, allocate mbufs and then put them into alloc_q */ if (ret)
@@ -605,7 +605,7 @@ kni_allocate_mbufs(struct rte_kni *kni)return ; } - allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) \ + allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) & (MAX_MBUF_BURST_NUM -1 );for (i =0 ; i < allocq_free; i++) { pkts[i] =rte_pktmbuf_alloc (kni->pktmbuf_pool);
@@ -659,7 +659,7 @@ static enum kni_ops_status kni_check_request_register(struct rte_kni_ops *ops) {/* check if KNI request ops has been registered*/ -if (NULL == ops ) +if (NULL == ops)return KNI_REQ_NO_REGISTER ; if ((ops->change_mtu == NULL)
@@ -672,22 +672,22 @@ kni_check_request_register(struct rte_kni_ops *ops) }int -rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) +rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) {enum kni_ops_status req_status; -if (NULL == ops) { +if (ops ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid KNI request operation.\n" );return -1 ; } -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; } req_status =kni_check_request_register (&kni->ops); -if (KNI_REQ_REGISTERED == req_status) { +if (req_status ==KNI_REQ_REGISTERED ) {RTE_LOG (ERR ,KNI ,"The KNI request operation has already registered.\n" );return -1 ; }
@@ -699,7 +699,7 @@ rte_kni_register_handlers(struct rte_kni *kni,struct rte_kni_ops *ops)int rte_kni_unregister_handlers (struct rte_kni *kni) { -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; }
Signature · 1 line
2.20.1
v3 cover kni: cleanups and fixes
In reply to Stephen HemmingerWhile testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
This is still mostly "putting lipstick on a pig" all users would be better off using virtio_user rather than KNI.
v3 - more style cleanups in last patch
v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (8):
kni: don't need stubs for rx_mode or ioctl
kni: use netdev_alloc_skb
kni: don't keep stats in kni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return -EFAULT if copy_from_user fails
doc: update KNI documentation
kni: fix style issues
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 18 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- lib/librte_kni/rte_kni.c | 38 ++++----- 5 files changed, 57 insertions(+), 113 deletions(-)
Signature · 1 line
2.20.1
PATCH v3 1/8 kni: don't need stubs for rx_mode or ioctl
From: Stephen Hemminger <[email protected]>
The netdev subsystem already handles case where
network sevice does not support ioctl.
If device has no rx_mode hook it is not called.
--- kernel/linux/kni/kni_net.c | 19 ------------------- 1 file changed, 19 deletions(-)
kernel/linux/kni/kni_net.c +0 -19
@@ -593,23 +593,6 @@ kni_net_tx_timeout(struct net_device *dev)netif_wake_queue (dev); } -/* -* Ioctl commands -*/ -static int -kni_net_ioctl (struct net_device *dev,struct ifreq *rq,int cmd) -{ -pr_debug ("kni_net_ioctl group:%d cmd:%d\n" , - ((struct kni_dev *)netdev_priv (dev))->group_id, cmd); - -return -EOPNOTSUPP ; -} - -static void -kni_net_set_rx_mode (struct net_device *dev) -{ -} -static int kni_net_change_mtu (struct net_device *dev,int new_mtu) {
@@ -758,8 +741,6 @@ static const struct net_device_ops kni_net_netdev_ops = { .ndo_change_rx_flags = kni_net_set_promiscusity, .ndo_start_xmit = kni_net_tx, .ndo_change_mtu = kni_net_change_mtu, - .ndo_do_ioctl = kni_net_ioctl, - .ndo_set_rx_mode = kni_net_set_rx_mode, .ndo_get_stats = kni_net_stats, .ndo_tx_timeout = kni_net_tx_timeout, .ndo_set_mac_address = kni_net_set_mac,
Signature · 1 line
2.20.1
PATCH v3 2/8 kni: use netdev_alloc_skb
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
netdev_alloc_skb is optimized to any alignment or setup of skb->dev that is required. The kernel has chosen to not pad packets on x86 (for many years), because it is faster.
--- kernel/linux/kni/kni_net.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)
kernel/linux/kni/kni_net.c +3 -14
@@ -340,16 +340,13 @@ kni_net_rx_normal(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (!skb) {/* Update statistics */ kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -368,7 +365,6 @@ kni_net_rx_normal(struct kni_dev *kni) } } - skb->dev = dev; skb->protocol =eth_type_trans (skb, dev); skb->ip_summed =CHECKSUM_UNNECESSARY ;
@@ -512,26 +508,20 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb) { -/* Align IP on 16B boundary */ -skb_reserve (skb,2 );memcpy (skb_put (skb, len), data_kva, len); - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ;dev_kfree_skb (skb); }/* Simulate real usage, allocate/copy skb twice */ - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb ==NULL ) { kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -550,7 +540,6 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) } } - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ; kni->stats.rx_bytes += len;
Signature · 1 line
2.20.1
PATCH v3 3/8 kni: don't keep stats in kni_net
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
Since kernel 2.6.28 the network subsystem has provided dev->stats for devices to use statistics handling and is the default if no ndo_get_stats is provided.
This allow allows for 64 bit (rather than just 32 bit) statistics with KNI.
--- kernel/linux/kni/kni_dev.h | 1 - kernel/linux/kni/kni_net.c | 43 +++++++++++++------------------------- 2 files changed, 15 insertions(+), 29 deletions(-)
kernel/linux/kni/kni_dev.h +0 -1
@@ -39,7 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -struct net_device_stats stats;int status;uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */
kernel/linux/kni/kni_net.c +15 -28
118 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v3 4/8 kni: drop unused fields
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
The kni net structure only exists in driver no API/ABI. Several fields were either totally unused or set and never used.
The fields in dev_info do need to stay in the ABI but kernel can ignore them.
kni: drop unused status element
Yet another ethtool leftover.
--- kernel/linux/kni/kni_dev.h | 5 ----- kernel/linux/kni/kni_misc.c | 1 - 2 files changed, 6 deletions(-)
kernel/linux/kni/kni_dev.h +0 -5
@@ -39,8 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -int status; -uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */ char name[RTE_KNI_NAMESIZE ];/* Network device name */ struct task_struct *pthread;
@@ -79,9 +77,6 @@ struct kni_dev {/* mbuf size */ uint32_t mbuf_size; -/* synchro for request processing */ -unsigned long synchro; -/* buffers */ void *pa[MBUF_BURST_SZ ];void *va[MBUF_BURST_SZ ];
kernel/linux/kni/kni_misc.c +0 -1
@@ -346,7 +346,6 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num, kni =netdev_priv (net_dev); kni->net_dev = net_dev; - kni->group_id = dev_info.group_id; kni->core_id = dev_info.core_id;strncpy (kni->name, dev_info.name,RTE_KNI_NAMESIZE );
Signature · 1 line
2.20.1
PATCH v3 5/8 kni: use proper type for kni fifo's
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
Using void * instead of proper type is unsafe practice.
--- kernel/linux/kni/kni_dev.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/linux/kni/kni_dev.h +6 -6
@@ -51,22 +51,22 @@ struct kni_dev {struct net_device *net_dev;/* queue for packets to be sent out */ -void *tx_q; +struct rte_kni_fifo *tx_q;/* queue for the packets received */ -void *rx_q; +struct rte_kni_fifo *rx_q;/* queue for the allocated mbufs those can be used to save sk buffs */ -void *alloc_q; +struct rte_kni_fifo *alloc_q;/* free queue for the mbufs to be freed */ -void *free_q; +struct rte_kni_fifo *free_q;/* request queue */ -void *req_q; +struct rte_kni_fifo *req_q;/* response queue */ -void *resp_q; +struct rte_kni_fifo *resp_q;void *sync_kva;void *sync_va;
Signature · 1 line
2.20.1
PATCH v3 6/8 kni: return -EFAULT if copy_from_user fails
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
The correct thing to return if user gives a bad data is to return -EFAULT. Logging is also discouraged because it could be used as a DoS attack.
--- kernel/linux/kni/kni_misc.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
kernel/linux/kni/kni_misc.c +5 -11
@@ -301,11 +301,8 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num,return -EINVAL ;/* Copy kni info from user space */ - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_create" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Check if name is zero-ended */ if (strnlen (dev_info.name,sizeof (dev_info.name)) ==sizeof (dev_info.name)) {
@@ -427,15 +424,12 @@ kni_ioctl_release(struct net *net, uint32_t ioctl_num,if (_IOC_SIZE (ioctl_num) >sizeof (dev_info))return -EINVAL ; - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_release" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Release the network device according to its name */ if (strlen (dev_info.name) ==0 ) -return ret; +return -EINVAL ;down_write (&knet->kni_list_lock);list_for_each_entry_safe (dev, n, &knet->kni_list_head, list) {
Signature · 1 line
2.20.1
PATCH v3 7/8 doc: update KNI documentation
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
Make the KNI documentation reflect modern kernel networking. Ifconfig has been superseded by iproute2 for 15 years. Iproute2 is well maintained, supports current feature set.
Ethtool is no longer supported by KNI.
Tshark is a better replacement for tcpdump.
--- .../sample_app_ug/kernel_nic_interface.rst | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
doc/guides/sample_app_ug/kernel_nic_interface.rst +9 -9
@@ -21,14 +21,14 @@ The FIFO queues contain pointers to data packets in the DPDK. This: * Provides a faster mechanism to interface with the kernel net stack and eliminates system calls -* Facilitates the DPDK using standard Linux* userspace net tools (tcpdump, ftp, and so on) +* Facilitates the DPDK using standard Linux* userspace net tools (tshark, rsync, and so on) * Eliminate the copy_to_user and copy_from_user operations on packets. The Kernel NIC Interface sample application is a simple example that demonstrates the use of the DPDK to create a path for packets to go through the Linux* kernel. This is done by creating one or more kernel net devices for each of the DPDK ports. -The application allows the use of standard Linux tools (ethtool, ifconfig, tcpdump) with the DPDK ports and +The application allows the use of standard Linux tools (iproute, tshark) with the DPDK ports and also the exchange of packets between the DPDK application and the Linux* kernel. The Kernel NIC Interface sample application requires that the
@@ -220,13 +220,13 @@ Enable KNI interface and assign an IP address: .. code-block:: console - # ifconfig vEth0_0 192.168.0.1 + # ip addr add dev vEth0_0 192.168.0.1 Show KNI interface configuration and statistics: .. code-block:: console - # ifconfig vEth0_0 + # ip -s -d addr show vEth0_0 The user can also check and reset the packet statistics inside the ``kni`` application by sending the app the USR1 and USR2 signals:
@@ -234,16 +234,16 @@ application by sending the app the USR1 and USR2 signals: .. code-block:: console # Print statistics - # kill -SIGUSR1 `pidof kni` + # pkill -USR1 kni # Zero statistics - # kill -SIGUSR2 `pidof kni` + # pkill -USR2 kni Dump network traffic: .. code-block:: console - # tcpdump -i vEth0_0 + # tshark -n -i vEth0_0 The normal Linux commands can also be used to change the MAC address and MTU size used by the physical NIC which corresponds to the KNI interface.
@@ -254,13 +254,13 @@ Change the MAC address: .. code-block:: console - # ifconfig vEth0_0 hw ether 0C:01:02:03:04:08 + # ip link set dev vEth0_0 lladdr 0C:01:02:03:04:08 Change the MTU size: .. code-block:: console - # ifconfig vEth0_0 mtu 1450 + # ip link set dev vEth0_0 mtu 1450 When the ``kni`` application is closed, all the KNI interfaces are deleted from the Linux kernel.
Signature · 1 line
2.20.1
PATCH v3 8/8 kni: fix style issues
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
rte_kni does not follow standard style rules.
Noticed some extra \ line continuation etc.
--- lib/librte_kni/rte_kni.c | 38 +++++++++++++++++++------------------- 1 file changed, 19 insertions(+), 19 deletions(-)
lib/librte_kni/rte_kni.c +19 -19
@@ -59,7 +59,7 @@ struct rte_kni {uint16_t group_id;/**< Group ID of KNI devices */ uint32_t slot_id;/**< KNI pool slot ID */ struct rte_mempool *pktmbuf_pool;/**< pkt mbuf mempool */ -unsigned mbuf_size;/**< mbuf size */ +unsigned int mbuf_size;/**< mbuf size */ const struct rte_memzone *m_tx_q;/**< TX queue memzone */ const struct rte_memzone *m_rx_q;/**< RX queue memzone */
@@ -78,7 +78,7 @@ struct rte_kni {/* For request & response */ struct rte_kni_fifo *req_q;/**< Request queue */ struct rte_kni_fifo *resp_q;/**< Response queue */ -void * sync_addr;/**< Req/Resp Mem address */ +void *sync_addr;/**< Req/Resp Mem address */ struct rte_kni_ops ops;/**< operations for request */ };
@@ -473,7 +473,7 @@ kni_config_promiscusity(uint16_t port_id, uint8_t to_on)int rte_kni_handle_request (struct rte_kni *kni) { -unsigned ret; +unsigned int ret;struct rte_kni_request *req =NULL ;if (kni ==NULL )
@@ -498,8 +498,8 @@ rte_kni_handle_request(struct rte_kni *kni)break ;case RTE_KNI_REQ_CFG_NETWORK_IF :/* Set network interface up/down */ if (kni->ops.config_network_if) - req->result = kni->ops.config_network_if (\ - kni->ops.port_id, req->if_up); + req->result = kni->ops.config_network_if (kni->ops.port_id, + req->if_up);break ;case RTE_KNI_REQ_CHANGE_MAC_ADDR :/* Change MAC Address */ if (kni->ops.config_mac_address)
@@ -534,7 +534,7 @@ rte_kni_handle_request(struct rte_kni *kni) }unsigned -rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) {void *phy_mbufs[num];unsigned int ret;
@@ -552,9 +552,9 @@ rte_kni_tx_burst(struct rte_kni *kni, struct rte_mbuf **mbufs, unsigned num) }unsigned -rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) { -unsigned ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num); +unsigned int ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num);/* If buffers removed, allocate mbufs and then put them into alloc_q */ if (ret)
@@ -605,7 +605,7 @@ kni_allocate_mbufs(struct rte_kni *kni)return ; } - allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) \ + allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) & (MAX_MBUF_BURST_NUM -1 );for (i =0 ; i < allocq_free; i++) { pkts[i] =rte_pktmbuf_alloc (kni->pktmbuf_pool);
@@ -659,35 +659,35 @@ static enum kni_ops_status kni_check_request_register(struct rte_kni_ops *ops) {/* check if KNI request ops has been registered*/ -if (NULL == ops ) +if (ops ==NULL )return KNI_REQ_NO_REGISTER ; -if ((ops->change_mtu ==NULL ) - && (ops->config_network_if ==NULL ) - && (ops->config_mac_address ==NULL ) - && (ops->config_promiscusity ==NULL )) +if (ops->change_mtu ==NULL + && ops->config_network_if ==NULL + && ops->config_mac_address ==NULL + && ops->config_promiscusity ==NULL )return KNI_REQ_NO_REGISTER ;return KNI_REQ_REGISTERED ; }int -rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) +rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) {enum kni_ops_status req_status; -if (NULL == ops) { +if (ops ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid KNI request operation.\n" );return -1 ; } -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; } req_status =kni_check_request_register (&kni->ops); -if (KNI_REQ_REGISTERED == req_status) { +if (req_status ==KNI_REQ_REGISTERED ) {RTE_LOG (ERR ,KNI ,"The KNI request operation has already registered.\n" );return -1 ; }
@@ -699,7 +699,7 @@ rte_kni_register_handlers(struct rte_kni *kni,struct rte_kni_ops *ops)int rte_kni_unregister_handlers (struct rte_kni *kni) { -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; }
Signature · 1 line
2.20.1
v4 cover kni: fixes and cleanups
In reply to Stephen HemmingerWhile testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
v4 - add more style fixes
v3 - rebase to current master, add style fix patch
v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (8):
kni: don't need stubs for rx_mode or ioctl
kni: use netdev_alloc_skb
kni: don't keep stats in kni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return -EFAULT if copy_from_user fails
doc: update KNI documentation
kni: fix style issues
.../sample_app_ug/kernel_nic_interface.rst | 18 ++--- kernel/linux/kni/kni_dev.h | 18 ++--- kernel/linux/kni/kni_misc.c | 17 ++-- kernel/linux/kni/kni_net.c | 79 +++++-------------- lib/librte_kni/rte_kni.c | 38 ++++----- 5 files changed, 57 insertions(+), 113 deletions(-)
Signature · 1 line
2.20.1
PATCH v4 1/8 kni: don't need stubs for rx_mode or ioctl
From: Stephen Hemminger <[email protected]>
The netdev subsystem already handles case where
network sevice does not support ioctl.
If device has no rx_mode hook it is not called.
--- kernel/linux/kni/kni_net.c | 19 ------------------- 1 file changed, 19 deletions(-)
kernel/linux/kni/kni_net.c +0 -19
@@ -593,23 +593,6 @@ kni_net_tx_timeout(struct net_device *dev)netif_wake_queue (dev); } -/* -* Ioctl commands -*/ -static int -kni_net_ioctl (struct net_device *dev,struct ifreq *rq,int cmd) -{ -pr_debug ("kni_net_ioctl group:%d cmd:%d\n" , - ((struct kni_dev *)netdev_priv (dev))->group_id, cmd); - -return -EOPNOTSUPP ; -} - -static void -kni_net_set_rx_mode (struct net_device *dev) -{ -} -static int kni_net_change_mtu (struct net_device *dev,int new_mtu) {
@@ -758,8 +741,6 @@ static const struct net_device_ops kni_net_netdev_ops = { .ndo_change_rx_flags = kni_net_set_promiscusity, .ndo_start_xmit = kni_net_tx, .ndo_change_mtu = kni_net_change_mtu, - .ndo_do_ioctl = kni_net_ioctl, - .ndo_set_rx_mode = kni_net_set_rx_mode, .ndo_get_stats = kni_net_stats, .ndo_tx_timeout = kni_net_tx_timeout, .ndo_set_mac_address = kni_net_set_mac,
Signature · 1 line
2.20.1
PATCH v4 2/8 kni: use netdev_alloc_skb
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
netdev_alloc_skb is optimized to any alignment or setup of skb->dev that is required. The kernel has chosen to not pad packets on x86 (for many years), because it is faster.
--- kernel/linux/kni/kni_net.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)
kernel/linux/kni/kni_net.c +3 -14
@@ -340,16 +340,13 @@ kni_net_rx_normal(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (!skb) {/* Update statistics */ kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -368,7 +365,6 @@ kni_net_rx_normal(struct kni_dev *kni) } } - skb->dev = dev; skb->protocol =eth_type_trans (skb, dev); skb->ip_summed =CHECKSUM_UNNECESSARY ;
@@ -512,26 +508,20 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb) { -/* Align IP on 16B boundary */ -skb_reserve (skb,2 );memcpy (skb_put (skb, len), data_kva, len); - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ;dev_kfree_skb (skb); }/* Simulate real usage, allocate/copy skb twice */ - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb ==NULL ) { kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -550,7 +540,6 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) } } - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ; kni->stats.rx_bytes += len;
Signature · 1 line
2.20.1
PATCH v4 3/8 kni: don't keep stats in kni_net
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
Since kernel 2.6.28 the network subsystem has provided dev->stats for devices to use statistics handling and is the default if no ndo_get_stats is provided.
This allow allows for 64 bit (rather than just 32 bit) statistics with KNI.
--- kernel/linux/kni/kni_dev.h | 1 - kernel/linux/kni/kni_net.c | 43 +++++++++++++------------------------- 2 files changed, 15 insertions(+), 29 deletions(-)
kernel/linux/kni/kni_dev.h +0 -1
@@ -39,7 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -struct net_device_stats stats;int status;uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */
kernel/linux/kni/kni_net.c +15 -28
118 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v4 4/8 kni: drop unused fields
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
The kni net structure only exists in driver no API/ABI. Several fields were either totally unused or set and never used.
The fields in dev_info do need to stay in the ABI but kernel can ignore them.
kni: drop unused status element
Yet another ethtool leftover.
--- kernel/linux/kni/kni_dev.h | 5 ----- kernel/linux/kni/kni_misc.c | 1 - 2 files changed, 6 deletions(-)
kernel/linux/kni/kni_dev.h +0 -5
@@ -39,8 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -int status; -uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */ char name[RTE_KNI_NAMESIZE ];/* Network device name */ struct task_struct *pthread;
@@ -79,9 +77,6 @@ struct kni_dev {/* mbuf size */ uint32_t mbuf_size; -/* synchro for request processing */ -unsigned long synchro; -/* buffers */ void *pa[MBUF_BURST_SZ ];void *va[MBUF_BURST_SZ ];
kernel/linux/kni/kni_misc.c +0 -1
@@ -346,7 +346,6 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num, kni =netdev_priv (net_dev); kni->net_dev = net_dev; - kni->group_id = dev_info.group_id; kni->core_id = dev_info.core_id;strncpy (kni->name, dev_info.name,RTE_KNI_NAMESIZE );
Signature · 1 line
2.20.1
PATCH v4 5/8 kni: use proper type for kni fifo's
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
Using void * instead of proper type is unsafe practice.
--- kernel/linux/kni/kni_dev.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/linux/kni/kni_dev.h +6 -6
@@ -51,22 +51,22 @@ struct kni_dev {struct net_device *net_dev;/* queue for packets to be sent out */ -void *tx_q; +struct rte_kni_fifo *tx_q;/* queue for the packets received */ -void *rx_q; +struct rte_kni_fifo *rx_q;/* queue for the allocated mbufs those can be used to save sk buffs */ -void *alloc_q; +struct rte_kni_fifo *alloc_q;/* free queue for the mbufs to be freed */ -void *free_q; +struct rte_kni_fifo *free_q;/* request queue */ -void *req_q; +struct rte_kni_fifo *req_q;/* response queue */ -void *resp_q; +struct rte_kni_fifo *resp_q;void *sync_kva;void *sync_va;
Signature · 1 line
2.20.1
PATCH v4 6/8 kni: return -EFAULT if copy_from_user fails
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
The correct thing to return if user gives a bad data is to return -EFAULT. Logging is also discouraged because it could be used as a DoS attack.
--- kernel/linux/kni/kni_misc.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
kernel/linux/kni/kni_misc.c +5 -11
@@ -301,11 +301,8 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num,return -EINVAL ;/* Copy kni info from user space */ - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_create" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Check if name is zero-ended */ if (strnlen (dev_info.name,sizeof (dev_info.name)) ==sizeof (dev_info.name)) {
@@ -427,15 +424,12 @@ kni_ioctl_release(struct net *net, uint32_t ioctl_num,if (_IOC_SIZE (ioctl_num) >sizeof (dev_info))return -EINVAL ; - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_release" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Release the network device according to its name */ if (strlen (dev_info.name) ==0 ) -return ret; +return -EINVAL ;down_write (&knet->kni_list_lock);list_for_each_entry_safe (dev, n, &knet->kni_list_head, list) {
Signature · 1 line
2.20.1
PATCH v4 7/8 doc: update KNI documentation
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
Make the KNI documentation reflect modern kernel networking. Ifconfig has been superseded by iproute2 for 15 years. Iproute2 is well maintained, supports current feature set.
Ethtool is no longer supported by KNI.
Tshark is a better replacement for tcpdump.
--- .../sample_app_ug/kernel_nic_interface.rst | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
doc/guides/sample_app_ug/kernel_nic_interface.rst +9 -9
@@ -21,14 +21,14 @@ The FIFO queues contain pointers to data packets in the DPDK. This: * Provides a faster mechanism to interface with the kernel net stack and eliminates system calls -* Facilitates the DPDK using standard Linux* userspace net tools (tcpdump, ftp, and so on) +* Facilitates the DPDK using standard Linux* userspace net tools (tshark, rsync, and so on) * Eliminate the copy_to_user and copy_from_user operations on packets. The Kernel NIC Interface sample application is a simple example that demonstrates the use of the DPDK to create a path for packets to go through the Linux* kernel. This is done by creating one or more kernel net devices for each of the DPDK ports. -The application allows the use of standard Linux tools (ethtool, ifconfig, tcpdump) with the DPDK ports and +The application allows the use of standard Linux tools (iproute, tshark) with the DPDK ports and also the exchange of packets between the DPDK application and the Linux* kernel. The Kernel NIC Interface sample application requires that the
@@ -220,13 +220,13 @@ Enable KNI interface and assign an IP address: .. code-block:: console - # ifconfig vEth0_0 192.168.0.1 + # ip addr add dev vEth0_0 192.168.0.1 Show KNI interface configuration and statistics: .. code-block:: console - # ifconfig vEth0_0 + # ip -s -d addr show vEth0_0 The user can also check and reset the packet statistics inside the ``kni`` application by sending the app the USR1 and USR2 signals:
@@ -234,16 +234,16 @@ application by sending the app the USR1 and USR2 signals: .. code-block:: console # Print statistics - # kill -SIGUSR1 `pidof kni` + # pkill -USR1 kni # Zero statistics - # kill -SIGUSR2 `pidof kni` + # pkill -USR2 kni Dump network traffic: .. code-block:: console - # tcpdump -i vEth0_0 + # tshark -n -i vEth0_0 The normal Linux commands can also be used to change the MAC address and MTU size used by the physical NIC which corresponds to the KNI interface.
@@ -254,13 +254,13 @@ Change the MAC address: .. code-block:: console - # ifconfig vEth0_0 hw ether 0C:01:02:03:04:08 + # ip link set dev vEth0_0 lladdr 0C:01:02:03:04:08 Change the MTU size: .. code-block:: console - # ifconfig vEth0_0 mtu 1450 + # ip link set dev vEth0_0 mtu 1450 When the ``kni`` application is closed, all the KNI interfaces are deleted from the Linux kernel.
Signature · 1 line
2.20.1
PATCH v4 8/8 kni: fix style issues
In reply to Stephen HemmingerFrom: Stephen Hemminger <[email protected]>
rte_kni does not follow standard style rules.
Noticed some extra \ line continuation etc.
--- lib/librte_kni/rte_kni.c | 38 +++++++++++++++++++------------------- 1 file changed, 19 insertions(+), 19 deletions(-)
lib/librte_kni/rte_kni.c +19 -19
@@ -59,7 +59,7 @@ struct rte_kni {uint16_t group_id;/**< Group ID of KNI devices */ uint32_t slot_id;/**< KNI pool slot ID */ struct rte_mempool *pktmbuf_pool;/**< pkt mbuf mempool */ -unsigned mbuf_size;/**< mbuf size */ +unsigned int mbuf_size;/**< mbuf size */ const struct rte_memzone *m_tx_q;/**< TX queue memzone */ const struct rte_memzone *m_rx_q;/**< RX queue memzone */
@@ -78,7 +78,7 @@ struct rte_kni {/* For request & response */ struct rte_kni_fifo *req_q;/**< Request queue */ struct rte_kni_fifo *resp_q;/**< Response queue */ -void * sync_addr;/**< Req/Resp Mem address */ +void *sync_addr;/**< Req/Resp Mem address */ struct rte_kni_ops ops;/**< operations for request */ };
@@ -473,7 +473,7 @@ kni_config_promiscusity(uint16_t port_id, uint8_t to_on)int rte_kni_handle_request (struct rte_kni *kni) { -unsigned ret; +unsigned int ret;struct rte_kni_request *req =NULL ;if (kni ==NULL )
@@ -498,8 +498,8 @@ rte_kni_handle_request(struct rte_kni *kni)break ;case RTE_KNI_REQ_CFG_NETWORK_IF :/* Set network interface up/down */ if (kni->ops.config_network_if) - req->result = kni->ops.config_network_if (\ - kni->ops.port_id, req->if_up); + req->result = kni->ops.config_network_if (kni->ops.port_id, + req->if_up);break ;case RTE_KNI_REQ_CHANGE_MAC_ADDR :/* Change MAC Address */ if (kni->ops.config_mac_address)
@@ -534,7 +534,7 @@ rte_kni_handle_request(struct rte_kni *kni) }unsigned -rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) {void *phy_mbufs[num];unsigned int ret;
@@ -552,9 +552,9 @@ rte_kni_tx_burst(struct rte_kni *kni, struct rte_mbuf **mbufs, unsigned num) }unsigned -rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) { -unsigned ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num); +unsigned int ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num);/* If buffers removed, allocate mbufs and then put them into alloc_q */ if (ret)
@@ -605,7 +605,7 @@ kni_allocate_mbufs(struct rte_kni *kni)return ; } - allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) \ + allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) & (MAX_MBUF_BURST_NUM -1 );for (i =0 ; i < allocq_free; i++) { pkts[i] =rte_pktmbuf_alloc (kni->pktmbuf_pool);
@@ -659,35 +659,35 @@ static enum kni_ops_status kni_check_request_register(struct rte_kni_ops *ops) {/* check if KNI request ops has been registered*/ -if (NULL == ops ) +if (ops ==NULL )return KNI_REQ_NO_REGISTER ; -if ((ops->change_mtu ==NULL ) - && (ops->config_network_if ==NULL ) - && (ops->config_mac_address ==NULL ) - && (ops->config_promiscusity ==NULL )) +if (ops->change_mtu ==NULL + && ops->config_network_if ==NULL + && ops->config_mac_address ==NULL + && ops->config_promiscusity ==NULL )return KNI_REQ_NO_REGISTER ;return KNI_REQ_REGISTERED ; }int -rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) +rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) {enum kni_ops_status req_status; -if (NULL == ops) { +if (ops ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid KNI request operation.\n" );return -1 ; } -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; } req_status =kni_check_request_register (&kni->ops); -if (KNI_REQ_REGISTERED == req_status) { +if (req_status ==KNI_REQ_REGISTERED ) {RTE_LOG (ERR ,KNI ,"The KNI request operation has already registered.\n" );return -1 ; }
@@ -699,7 +699,7 @@ rte_kni_register_handlers(struct rte_kni *kni,struct rte_kni_ops *ops)int rte_kni_unregister_handlers (struct rte_kni *kni) { -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; }
Signature · 1 line
2.20.1
v5 cover kni: fixes and cleanups
In reply to Stephen HemmingerWhile testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
v5 - add minimal ethtool, fix checkpath author complaints v4 - add more style fixes
v3 - rebase to current master, add style fix patch
v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (9):
kni: don't need stubs for rx_mode or ioctl
kni: use netdev_alloc_skb
kni: don't keep stats in kni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return -EFAULT if copy_from_user fails
doc: update KNI documentation
kni: fix style issues
kni: add minimal ethtool
.../sample_app_ug/kernel_nic_interface.rst | 18 ++-- kernel/linux/kni/kni_dev.h | 20 ++-- kernel/linux/kni/kni_misc.c | 18 ++-- kernel/linux/kni/kni_net.c | 100 +++++++----------- lib/librte_kni/rte_kni.c | 38 +++---- 5 files changed, 78 insertions(+), 116 deletions(-)
Signature · 1 line
2.20.1
PATCH v5 1/9 kni: don't need stubs for rx_mode or ioctl
The netdev subsystem already handles case where
network sevice does not support ioctl.
If device has no rx_mode hook it is not called.
--- kernel/linux/kni/kni_net.c | 19 ------------------- 1 file changed, 19 deletions(-)
kernel/linux/kni/kni_net.c +0 -19
@@ -593,23 +593,6 @@ kni_net_tx_timeout(struct net_device *dev)netif_wake_queue (dev); } -/* -* Ioctl commands -*/ -static int -kni_net_ioctl (struct net_device *dev,struct ifreq *rq,int cmd) -{ -pr_debug ("kni_net_ioctl group:%d cmd:%d\n" , - ((struct kni_dev *)netdev_priv (dev))->group_id, cmd); - -return -EOPNOTSUPP ; -} - -static void -kni_net_set_rx_mode (struct net_device *dev) -{ -} -static int kni_net_change_mtu (struct net_device *dev,int new_mtu) {
@@ -758,8 +741,6 @@ static const struct net_device_ops kni_net_netdev_ops = { .ndo_change_rx_flags = kni_net_set_promiscusity, .ndo_start_xmit = kni_net_tx, .ndo_change_mtu = kni_net_change_mtu, - .ndo_do_ioctl = kni_net_ioctl, - .ndo_set_rx_mode = kni_net_set_rx_mode, .ndo_get_stats = kni_net_stats, .ndo_tx_timeout = kni_net_tx_timeout, .ndo_set_mac_address = kni_net_set_mac,
Signature · 1 line
2.20.1
PATCH v5 2/9 kni: use netdev_alloc_skb
In reply to Stephen Hemmingernetdev_alloc_skb is optimized to any alignment or setup of skb->dev that is required. The kernel has chosen to not pad packets on x86 (for many years), because it is faster.
--- kernel/linux/kni/kni_net.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)
kernel/linux/kni/kni_net.c +3 -14
@@ -340,16 +340,13 @@ kni_net_rx_normal(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (!skb) {/* Update statistics */ kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -368,7 +365,6 @@ kni_net_rx_normal(struct kni_dev *kni) } } - skb->dev = dev; skb->protocol =eth_type_trans (skb, dev); skb->ip_summed =CHECKSUM_UNNECESSARY ;
@@ -512,26 +508,20 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb) { -/* Align IP on 16B boundary */ -skb_reserve (skb,2 );memcpy (skb_put (skb, len), data_kva, len); - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ;dev_kfree_skb (skb); }/* Simulate real usage, allocate/copy skb twice */ - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb ==NULL ) { kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -550,7 +540,6 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) } } - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ; kni->stats.rx_bytes += len;
Signature · 1 line
2.20.1
PATCH v5 3/9 kni: don't keep stats in kni_net
In reply to Stephen HemmingerSince kernel 2.6.28 the network subsystem has provided dev->stats for devices to use statistics handling and is the default if no ndo_get_stats is provided.
This allow allows for 64 bit (rather than just 32 bit) statistics with KNI.
--- kernel/linux/kni/kni_dev.h | 1 - kernel/linux/kni/kni_net.c | 43 +++++++++++++------------------------- 2 files changed, 15 insertions(+), 29 deletions(-)
kernel/linux/kni/kni_dev.h +0 -1
@@ -39,7 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -struct net_device_stats stats;int status;uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */
kernel/linux/kni/kni_net.c +15 -28
118 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v5 4/9 kni: drop unused fields
In reply to Stephen HemmingerSeveral fields were either totally unused or set and never used.
--- kernel/linux/kni/kni_dev.h | 5 ----- kernel/linux/kni/kni_misc.c | 1 - 2 files changed, 6 deletions(-)
kernel/linux/kni/kni_dev.h +0 -5
@@ -39,8 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -int status; -uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */ char name[RTE_KNI_NAMESIZE ];/* Network device name */ struct task_struct *pthread;
@@ -79,9 +77,6 @@ struct kni_dev {/* mbuf size */ uint32_t mbuf_size; -/* synchro for request processing */ -unsigned long synchro; -/* buffers */ void *pa[MBUF_BURST_SZ ];void *va[MBUF_BURST_SZ ];
kernel/linux/kni/kni_misc.c +0 -1
@@ -346,7 +346,6 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num, kni =netdev_priv (net_dev); kni->net_dev = net_dev; - kni->group_id = dev_info.group_id; kni->core_id = dev_info.core_id;strncpy (kni->name, dev_info.name,RTE_KNI_NAMESIZE );
Signature · 1 line
2.20.1
PATCH v5 5/9 kni: use proper type for kni fifo's
In reply to Stephen HemmingerUsing void * instead of proper type is unsafe practice.
--- kernel/linux/kni/kni_dev.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/linux/kni/kni_dev.h +6 -6
@@ -51,22 +51,22 @@ struct kni_dev {struct net_device *net_dev;/* queue for packets to be sent out */ -void *tx_q; +struct rte_kni_fifo *tx_q;/* queue for the packets received */ -void *rx_q; +struct rte_kni_fifo *rx_q;/* queue for the allocated mbufs those can be used to save sk buffs */ -void *alloc_q; +struct rte_kni_fifo *alloc_q;/* free queue for the mbufs to be freed */ -void *free_q; +struct rte_kni_fifo *free_q;/* request queue */ -void *req_q; +struct rte_kni_fifo *req_q;/* response queue */ -void *resp_q; +struct rte_kni_fifo *resp_q;void *sync_kva;void *sync_va;
Signature · 1 line
2.20.1
PATCH v5 6/9 kni: return -EFAULT if copy_from_user fails
In reply to Stephen HemmingerThe correct thing to return if user gives a bad data is to return -EFAULT. Logging is also discouraged because it could be used as a DoS attack.
--- kernel/linux/kni/kni_misc.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
kernel/linux/kni/kni_misc.c +5 -11
@@ -301,11 +301,8 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num,return -EINVAL ;/* Copy kni info from user space */ - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_create" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Check if name is zero-ended */ if (strnlen (dev_info.name,sizeof (dev_info.name)) ==sizeof (dev_info.name)) {
@@ -427,15 +424,12 @@ kni_ioctl_release(struct net *net, uint32_t ioctl_num,if (_IOC_SIZE (ioctl_num) >sizeof (dev_info))return -EINVAL ; - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_release" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Release the network device according to its name */ if (strlen (dev_info.name) ==0 ) -return ret; +return -EINVAL ;down_write (&knet->kni_list_lock);list_for_each_entry_safe (dev, n, &knet->kni_list_head, list) {
Signature · 1 line
2.20.1
PATCH v5 7/9 doc: update KNI documentation
In reply to Stephen HemmingerMake the KNI documentation reflect modern kernel networking. Ifconfig has been superseded by iproute2 for 15 years. Iproute2 is well maintained, supports current feature set.
Ethtool is no longer supported by KNI.
Tshark is a better replacement for tcpdump.
--- .../sample_app_ug/kernel_nic_interface.rst | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
doc/guides/sample_app_ug/kernel_nic_interface.rst +9 -9
@@ -21,14 +21,14 @@ The FIFO queues contain pointers to data packets in the DPDK. This: * Provides a faster mechanism to interface with the kernel net stack and eliminates system calls -* Facilitates the DPDK using standard Linux* userspace net tools (tcpdump, ftp, and so on) +* Facilitates the DPDK using standard Linux* userspace net tools (tshark, rsync, and so on) * Eliminate the copy_to_user and copy_from_user operations on packets. The Kernel NIC Interface sample application is a simple example that demonstrates the use of the DPDK to create a path for packets to go through the Linux* kernel. This is done by creating one or more kernel net devices for each of the DPDK ports. -The application allows the use of standard Linux tools (ethtool, ifconfig, tcpdump) with the DPDK ports and +The application allows the use of standard Linux tools (iproute, tshark) with the DPDK ports and also the exchange of packets between the DPDK application and the Linux* kernel. The Kernel NIC Interface sample application requires that the
@@ -220,13 +220,13 @@ Enable KNI interface and assign an IP address: .. code-block:: console - # ifconfig vEth0_0 192.168.0.1 + # ip addr add dev vEth0_0 192.168.0.1 Show KNI interface configuration and statistics: .. code-block:: console - # ifconfig vEth0_0 + # ip -s -d addr show vEth0_0 The user can also check and reset the packet statistics inside the ``kni`` application by sending the app the USR1 and USR2 signals:
@@ -234,16 +234,16 @@ application by sending the app the USR1 and USR2 signals: .. code-block:: console # Print statistics - # kill -SIGUSR1 `pidof kni` + # pkill -USR1 kni # Zero statistics - # kill -SIGUSR2 `pidof kni` + # pkill -USR2 kni Dump network traffic: .. code-block:: console - # tcpdump -i vEth0_0 + # tshark -n -i vEth0_0 The normal Linux commands can also be used to change the MAC address and MTU size used by the physical NIC which corresponds to the KNI interface.
@@ -254,13 +254,13 @@ Change the MAC address: .. code-block:: console - # ifconfig vEth0_0 hw ether 0C:01:02:03:04:08 + # ip link set dev vEth0_0 lladdr 0C:01:02:03:04:08 Change the MTU size: .. code-block:: console - # ifconfig vEth0_0 mtu 1450 + # ip link set dev vEth0_0 mtu 1450 When the ``kni`` application is closed, all the KNI interfaces are deleted from the Linux kernel.
Signature · 1 line
2.20.1
PATCH v5 8/9 kni: fix style issues
In reply to Stephen Hemmingerrte_kni does not follow standard style rules.
Noticed some extra \ line continuation etc.
--- kernel/linux/kni/kni_net.c | 7 ++++--- lib/librte_kni/rte_kni.c | 38 +++++++++++++++++++------------------- 2 files changed, 23 insertions(+), 22 deletions(-)
kernel/linux/kni/kni_net.c +4 -3
@@ -401,7 +401,7 @@ kni_net_rx_lo_fifo(struct kni_dev *kni)/* Get the number of entries in rx_q */ num_rq =kni_fifo_count (kni->rx_q); -/* Get the number of free entrie in tx_q */ +/* Get the number of free entries in tx_q */ num_tq =kni_fifo_free_count (kni->tx_q);/* Get the number of entries in alloc_q */
@@ -755,6 +755,7 @@ kni_net_config_lo_mode(char *lo_str) }else if (!strcmp (lo_str,"lo_mode_fifo_skb" )) {pr_debug ("loopback mode=lo_mode_fifo_skb enabled" ); kni_net_rx_func = kni_net_rx_lo_fifo_skb; - }else -pr_debug ("Incognizant parameter, loopback disabled" ); + }else { +pr_debug ("Unknown loopback parameter, disabled" ); + } }
lib/librte_kni/rte_kni.c +19 -19
@@ -59,7 +59,7 @@ struct rte_kni {uint16_t group_id;/**< Group ID of KNI devices */ uint32_t slot_id;/**< KNI pool slot ID */ struct rte_mempool *pktmbuf_pool;/**< pkt mbuf mempool */ -unsigned mbuf_size;/**< mbuf size */ +unsigned int mbuf_size;/**< mbuf size */ const struct rte_memzone *m_tx_q;/**< TX queue memzone */ const struct rte_memzone *m_rx_q;/**< RX queue memzone */
@@ -78,7 +78,7 @@ struct rte_kni {/* For request & response */ struct rte_kni_fifo *req_q;/**< Request queue */ struct rte_kni_fifo *resp_q;/**< Response queue */ -void * sync_addr;/**< Req/Resp Mem address */ +void *sync_addr;/**< Req/Resp Mem address */ struct rte_kni_ops ops;/**< operations for request */ };
@@ -473,7 +473,7 @@ kni_config_promiscusity(uint16_t port_id, uint8_t to_on)int rte_kni_handle_request (struct rte_kni *kni) { -unsigned ret; +unsigned int ret;struct rte_kni_request *req =NULL ;if (kni ==NULL )
@@ -498,8 +498,8 @@ rte_kni_handle_request(struct rte_kni *kni)break ;case RTE_KNI_REQ_CFG_NETWORK_IF :/* Set network interface up/down */ if (kni->ops.config_network_if) - req->result = kni->ops.config_network_if (\ - kni->ops.port_id, req->if_up); + req->result = kni->ops.config_network_if (kni->ops.port_id, + req->if_up);break ;case RTE_KNI_REQ_CHANGE_MAC_ADDR :/* Change MAC Address */ if (kni->ops.config_mac_address)
@@ -534,7 +534,7 @@ rte_kni_handle_request(struct rte_kni *kni) }unsigned -rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) {void *phy_mbufs[num];unsigned int ret;
@@ -552,9 +552,9 @@ rte_kni_tx_burst(struct rte_kni *kni, struct rte_mbuf **mbufs, unsigned num) }unsigned -rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) { -unsigned ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num); +unsigned int ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num);/* If buffers removed, allocate mbufs and then put them into alloc_q */ if (ret)
@@ -605,7 +605,7 @@ kni_allocate_mbufs(struct rte_kni *kni)return ; } - allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) \ + allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) & (MAX_MBUF_BURST_NUM -1 );for (i =0 ; i < allocq_free; i++) { pkts[i] =rte_pktmbuf_alloc (kni->pktmbuf_pool);
@@ -659,35 +659,35 @@ static enum kni_ops_status kni_check_request_register(struct rte_kni_ops *ops) {/* check if KNI request ops has been registered*/ -if (NULL == ops ) +if (ops ==NULL )return KNI_REQ_NO_REGISTER ; -if ((ops->change_mtu ==NULL ) - && (ops->config_network_if ==NULL ) - && (ops->config_mac_address ==NULL ) - && (ops->config_promiscusity ==NULL )) +if (ops->change_mtu ==NULL + && ops->config_network_if ==NULL + && ops->config_mac_address ==NULL + && ops->config_promiscusity ==NULL )return KNI_REQ_NO_REGISTER ;return KNI_REQ_REGISTERED ; }int -rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) +rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) {enum kni_ops_status req_status; -if (NULL == ops) { +if (ops ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid KNI request operation.\n" );return -1 ; } -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; } req_status =kni_check_request_register (&kni->ops); -if (KNI_REQ_REGISTERED == req_status) { +if (req_status ==KNI_REQ_REGISTERED ) {RTE_LOG (ERR ,KNI ,"The KNI request operation has already registered.\n" );return -1 ; }
@@ -699,7 +699,7 @@ rte_kni_register_handlers(struct rte_kni *kni,struct rte_kni_ops *ops)int rte_kni_unregister_handlers (struct rte_kni *kni) { -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; }
Signature · 1 line
2.20.1
PATCH v5 9/9 kni: add minimal ethtool
In reply to Stephen HemmingerSome applications use ethtool so add the minimum ethtool ops.
--- kernel/linux/kni/kni_dev.h | 2 ++ kernel/linux/kni/kni_misc.c | 1 + kernel/linux/kni/kni_net.c | 14 ++++++++++++++ 3 files changed, 17 insertions(+)
kernel/linux/kni/kni_dev.h +2 -0
@@ -11,6 +11,8 @@#endif #define pr_fmt (fmt) KBUILD_MODNAME ": " fmt +#define KNI_VERSION "1.0" +#include "compat.h" #include <linux/if.h>
kernel/linux/kni/kni_misc.c +1 -0
@@ -21,6 +21,7 @@#include "compat.h" #include "kni_dev.h" +MODULE_VERSION (KNI_VERSION );MODULE_LICENSE ("Dual BSD/GPL" );MODULE_AUTHOR ("Intel Corporation" );MODULE_DESCRIPTION ("Kernel Module for managing kni devices" );
kernel/linux/kni/kni_net.c +14 -0
@@ -13,6 +13,7 @@#include <linux/version.h> #include <linux/netdevice.h> #include <linux/etherdevice.h> /* eth_type_trans */ +#include <linux/ethtool.h> #include <linux/skbuff.h> #include <linux/kthread.h> #include <linux/delay.h>
@@ -725,6 +726,18 @@ static const struct net_device_ops kni_net_netdev_ops = {#endif }; +static void kni_get_drvinfo (struct net_device *dev, +struct ethtool_drvinfo *info) +{ +strlcpy (info->version,KNI_VERSION ,sizeof (info->version)); +strlcpy (info->driver,"kni" ,sizeof (info->driver)); +} + +static const struct ethtool_ops kni_net_ethtool_ops = { + .get_drvinfo = kni_get_drvinfo, + .get_link = ethtool_op_get_link, +}; +void kni_net_init (struct net_device *dev) {
@@ -736,6 +749,7 @@ kni_net_init(struct net_device *dev)ether_setup (dev);/* assign some of the fields */ dev->netdev_ops = &kni_net_netdev_ops; dev->header_ops = &kni_net_header_ops; + dev->ethtool_ops = &kni_net_ethtool_ops; dev->watchdog_timeo =WD_TIMEOUT ; }
Signature · 1 line
2.20.1
Re: [dpdk-dev] [PATCH v5 9/9] kni: add minimal ethtool
Hi Stephen,
As ethtool support is restored, I think your patch #7 should be updated to keep ethtool support in docs.
Best regards,
Igor
Quoted 77 lines · Stephen Hemminger
Some applications use ethtool so add the minimum ethtool ops.
Signed-off-by Stephen Hemminger <[email protected]>--- kernel/linux/kni/kni_dev.h | 2 ++ kernel/linux/kni/kni_misc.c | 1 + kernel/linux/kni/kni_net.c | 14 ++++++++++++++ 3 files changed, 17 insertions(+)kernel/linux/kni/kni_dev.h +2
-0@@ -11,6 +11,8 @@#endif #define pr_fmt (fmt) KBUILD_MODNAME ": " fmt +#define KNI_VERSION "1.0" +#include "compat.h" #include <linux/if.h> kernel/linux/kni/kni_misc.c +1
-0@@ -21,6 +21,7 @@#include "compat.h" #include "kni_dev.h" +MODULE_VERSION (KNI_VERSION );MODULE_LICENSE ("Dual BSD/GPL" );MODULE_AUTHOR ("Intel Corporation" );MODULE_DESCRIPTION ("Kernel Module for managing kni devices" );kernel/linux/kni/kni_net.c +14
-1@@ -13,6 +13,7 @@#include <linux/version.h> #include <linux/netdevice.h> #include <linux/etherdevice.h> /* eth_type_trans */ +#include <linux/ethtool.h> #include <linux/skbuff.h> #include <linux/kthread.h> #include <linux/delay.h> @@ -725,6 +726,18 @@ static const struct net_device_ops kni_net_netdev_ops = {#endif }; +static void kni_get_drvinfo (struct net_device *dev, +struct ethtool_drvinfo *info) +{ +strlcpy (info->version,KNI_VERSION ,sizeof (info->version)); +strlcpy (info->driver,"kni" ,sizeof (info->driver)); +} + +static const struct ethtool_ops kni_net_ethtool_ops = { + .get_drvinfo = kni_get_drvinfo, + .get_link = ethtool_op_get_link, +}; +void kni_net_init (struct net_device *dev) {@@ -736,6 +749,7 @@ kni_net_init(struct net_device *dev)ether_setup (dev);/* assign some of the fields */ dev->netdev_ops = &kni_net_netdev_ops; dev->header_ops = &kni_net_header_ops; + dev->ethtool_ops = &kni_net_ethtool_ops; dev->watchdog_timeo =WD_TIMEOUT ; } --2.20.1
Re: [dpdk-dev] [PATCH v5 9/9] kni: add minimal ethtool
Quoted 7 lines · On Fri, 21 Jun 2019 11:26:09 +0300 Igor Ryzhov <[email protected]> wrote:
Hi Stephen,
As ethtool support is restored, I think your patch #7 should be updated to keep ethtool support in docs.
Best regards,
Igor
Good idea, I will add that. This is not like the original ethtool, it is intentionally limited in scope and generic (unlike the original).
v6 cover kni: fixes and cleanups
In reply to Stephen HemmingerWhile testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
v6 - more updates to documentation
v5 - add minimal ethtool, fix checkpath author complaints v4 - add more style fixes
v3 - rebase to current master, add style fix patch
v2 - get rid of unnecessary padding, combine the unused field patches
Stephen Hemminger (9):
kni: don't need stubs for rx_mode or ioctl
kni: use netdev_alloc_skb
kni: don't keep stats in kni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return -EFAULT if copy_from_user fails
kni: fix style issues
kni: add minimal ethtool
doc: update KNI documentation
.../prog_guide/kernel_nic_interface.rst | 9 +- .../sample_app_ug/kernel_nic_interface.rst | 24 +++-- kernel/linux/kni/kni_dev.h | 20 ++-- kernel/linux/kni/kni_misc.c | 18 ++-- kernel/linux/kni/kni_net.c | 100 +++++++----------- lib/librte_kni/rte_kni.c | 38 +++---- 6 files changed, 89 insertions(+), 120 deletions(-)
Signature · 1 line
2.20.1
PATCH v6 1/9 kni: don't need stubs for rx_mode or ioctl
The netdev subsystem already handles case where
network sevice does not support ioctl.
If device has no rx_mode hook it is not called.
--- kernel/linux/kni/kni_net.c | 19 ------------------- 1 file changed, 19 deletions(-)
kernel/linux/kni/kni_net.c +0 -19
@@ -593,23 +593,6 @@ kni_net_tx_timeout(struct net_device *dev)netif_wake_queue (dev); } -/* -* Ioctl commands -*/ -static int -kni_net_ioctl (struct net_device *dev,struct ifreq *rq,int cmd) -{ -pr_debug ("kni_net_ioctl group:%d cmd:%d\n" , - ((struct kni_dev *)netdev_priv (dev))->group_id, cmd); - -return -EOPNOTSUPP ; -} - -static void -kni_net_set_rx_mode (struct net_device *dev) -{ -} -static int kni_net_change_mtu (struct net_device *dev,int new_mtu) {
@@ -758,8 +741,6 @@ static const struct net_device_ops kni_net_netdev_ops = { .ndo_change_rx_flags = kni_net_set_promiscusity, .ndo_start_xmit = kni_net_tx, .ndo_change_mtu = kni_net_change_mtu, - .ndo_do_ioctl = kni_net_ioctl, - .ndo_set_rx_mode = kni_net_set_rx_mode, .ndo_get_stats = kni_net_stats, .ndo_tx_timeout = kni_net_tx_timeout, .ndo_set_mac_address = kni_net_set_mac,
Signature · 1 line
2.20.1
PATCH v6 2/9 kni: use netdev_alloc_skb
In reply to Stephen Hemmingernetdev_alloc_skb is optimized to any alignment or setup of skb->dev that is required. The kernel has chosen to not pad packets on x86 (for many years), because it is faster.
--- kernel/linux/kni/kni_net.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)
kernel/linux/kni/kni_net.c +3 -14
@@ -340,16 +340,13 @@ kni_net_rx_normal(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (!skb) {/* Update statistics */ kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -368,7 +365,6 @@ kni_net_rx_normal(struct kni_dev *kni) } } - skb->dev = dev; skb->protocol =eth_type_trans (skb, dev); skb->ip_summed =CHECKSUM_UNNECESSARY ;
@@ -512,26 +508,20 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) data_kva =kva2data_kva (kva); kni->va[i] =pa2va (kni->pa[i], kva); - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb) { -/* Align IP on 16B boundary */ -skb_reserve (skb,2 );memcpy (skb_put (skb, len), data_kva, len); - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ;dev_kfree_skb (skb); }/* Simulate real usage, allocate/copy skb twice */ - skb =dev_alloc_skb (len +2 ); + skb =netdev_alloc_skb (dev, len);if (skb ==NULL ) { kni->stats.rx_dropped++;continue ; } -/* Align IP on 16B boundary */ -skb_reserve (skb,2 ); -if (kva->nb_segs ==1 ) {memcpy (skb_put (skb, len), data_kva, len); }else {
@@ -550,7 +540,6 @@ kni_net_rx_lo_fifo_skb(struct kni_dev *kni) } } - skb->dev = dev; skb->ip_summed =CHECKSUM_UNNECESSARY ; kni->stats.rx_bytes += len;
Signature · 1 line
2.20.1
PATCH v6 3/9 kni: don't keep stats in kni_net
In reply to Stephen HemmingerSince kernel 2.6.28 the network subsystem has provided dev->stats for devices to use statistics handling and is the default if no ndo_get_stats is provided.
This allow allows for 64 bit (rather than just 32 bit) statistics with KNI.
--- kernel/linux/kni/kni_dev.h | 1 - kernel/linux/kni/kni_net.c | 43 +++++++++++++------------------------- 2 files changed, 15 insertions(+), 29 deletions(-)
kernel/linux/kni/kni_dev.h +0 -1
@@ -39,7 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -struct net_device_stats stats;int status;uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */
kernel/linux/kni/kni_net.c +15 -28
118 lines not on this page · raw message
Signature · 1 line
2.20.1
PATCH v6 4/9 kni: drop unused fields
In reply to Stephen HemmingerSeveral fields were either totally unused or set and never used.
--- kernel/linux/kni/kni_dev.h | 5 ----- kernel/linux/kni/kni_misc.c | 1 - 2 files changed, 6 deletions(-)
kernel/linux/kni/kni_dev.h +0 -5
@@ -39,8 +39,6 @@ struct kni_dev {/* kni list */ struct list_head list; -int status; -uint16_t group_id;/* Group ID of a group of KNI devices */ uint32_t core_id;/* Core ID to bind */ char name[RTE_KNI_NAMESIZE ];/* Network device name */ struct task_struct *pthread;
@@ -79,9 +77,6 @@ struct kni_dev {/* mbuf size */ uint32_t mbuf_size; -/* synchro for request processing */ -unsigned long synchro; -/* buffers */ void *pa[MBUF_BURST_SZ ];void *va[MBUF_BURST_SZ ];
kernel/linux/kni/kni_misc.c +0 -1
@@ -346,7 +346,6 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num, kni =netdev_priv (net_dev); kni->net_dev = net_dev; - kni->group_id = dev_info.group_id; kni->core_id = dev_info.core_id;strncpy (kni->name, dev_info.name,RTE_KNI_NAMESIZE );
Signature · 1 line
2.20.1
PATCH v6 5/9 kni: use proper type for kni fifo's
In reply to Stephen HemmingerUsing void * instead of proper type is unsafe practice.
--- kernel/linux/kni/kni_dev.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/linux/kni/kni_dev.h +6 -6
@@ -51,22 +51,22 @@ struct kni_dev {struct net_device *net_dev;/* queue for packets to be sent out */ -void *tx_q; +struct rte_kni_fifo *tx_q;/* queue for the packets received */ -void *rx_q; +struct rte_kni_fifo *rx_q;/* queue for the allocated mbufs those can be used to save sk buffs */ -void *alloc_q; +struct rte_kni_fifo *alloc_q;/* free queue for the mbufs to be freed */ -void *free_q; +struct rte_kni_fifo *free_q;/* request queue */ -void *req_q; +struct rte_kni_fifo *req_q;/* response queue */ -void *resp_q; +struct rte_kni_fifo *resp_q;void *sync_kva;void *sync_va;
Signature · 1 line
2.20.1
PATCH v6 6/9 kni: return -EFAULT if copy_from_user fails
In reply to Stephen HemmingerThe correct thing to return if user gives a bad data is to return -EFAULT. Logging is also discouraged because it could be used as a DoS attack.
--- kernel/linux/kni/kni_misc.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
kernel/linux/kni/kni_misc.c +5 -11
@@ -301,11 +301,8 @@ kni_ioctl_create(struct net *net, uint32_t ioctl_num,return -EINVAL ;/* Copy kni info from user space */ - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_create" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Check if name is zero-ended */ if (strnlen (dev_info.name,sizeof (dev_info.name)) ==sizeof (dev_info.name)) {
@@ -427,15 +424,12 @@ kni_ioctl_release(struct net *net, uint32_t ioctl_num,if (_IOC_SIZE (ioctl_num) >sizeof (dev_info))return -EINVAL ; - ret =copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info)); -if (ret) { -pr_err ("copy_from_user in kni_ioctl_release" ); -return -EIO ; - } +if (copy_from_user (&dev_info, (void *)ioctl_param,sizeof (dev_info))) +return -EFAULT ;/* Release the network device according to its name */ if (strlen (dev_info.name) ==0 ) -return ret; +return -EINVAL ;down_write (&knet->kni_list_lock);list_for_each_entry_safe (dev, n, &knet->kni_list_head, list) {
Signature · 1 line
2.20.1
PATCH v6 7/9 kni: fix style issues
In reply to Stephen Hemmingerrte_kni does not follow standard style rules.
Noticed some extra \ line continuation etc.
--- kernel/linux/kni/kni_net.c | 7 ++++--- lib/librte_kni/rte_kni.c | 38 +++++++++++++++++++------------------- 2 files changed, 23 insertions(+), 22 deletions(-)
kernel/linux/kni/kni_net.c +4 -3
@@ -401,7 +401,7 @@ kni_net_rx_lo_fifo(struct kni_dev *kni)/* Get the number of entries in rx_q */ num_rq =kni_fifo_count (kni->rx_q); -/* Get the number of free entrie in tx_q */ +/* Get the number of free entries in tx_q */ num_tq =kni_fifo_free_count (kni->tx_q);/* Get the number of entries in alloc_q */
@@ -755,6 +755,7 @@ kni_net_config_lo_mode(char *lo_str) }else if (!strcmp (lo_str,"lo_mode_fifo_skb" )) {pr_debug ("loopback mode=lo_mode_fifo_skb enabled" ); kni_net_rx_func = kni_net_rx_lo_fifo_skb; - }else -pr_debug ("Incognizant parameter, loopback disabled" ); + }else { +pr_debug ("Unknown loopback parameter, disabled" ); + } }
lib/librte_kni/rte_kni.c +19 -19
@@ -59,7 +59,7 @@ struct rte_kni {uint16_t group_id;/**< Group ID of KNI devices */ uint32_t slot_id;/**< KNI pool slot ID */ struct rte_mempool *pktmbuf_pool;/**< pkt mbuf mempool */ -unsigned mbuf_size;/**< mbuf size */ +unsigned int mbuf_size;/**< mbuf size */ const struct rte_memzone *m_tx_q;/**< TX queue memzone */ const struct rte_memzone *m_rx_q;/**< RX queue memzone */
@@ -78,7 +78,7 @@ struct rte_kni {/* For request & response */ struct rte_kni_fifo *req_q;/**< Request queue */ struct rte_kni_fifo *resp_q;/**< Response queue */ -void * sync_addr;/**< Req/Resp Mem address */ +void *sync_addr;/**< Req/Resp Mem address */ struct rte_kni_ops ops;/**< operations for request */ };
@@ -473,7 +473,7 @@ kni_config_promiscusity(uint16_t port_id, uint8_t to_on)int rte_kni_handle_request (struct rte_kni *kni) { -unsigned ret; +unsigned int ret;struct rte_kni_request *req =NULL ;if (kni ==NULL )
@@ -498,8 +498,8 @@ rte_kni_handle_request(struct rte_kni *kni)break ;case RTE_KNI_REQ_CFG_NETWORK_IF :/* Set network interface up/down */ if (kni->ops.config_network_if) - req->result = kni->ops.config_network_if (\ - kni->ops.port_id, req->if_up); + req->result = kni->ops.config_network_if (kni->ops.port_id, + req->if_up);break ;case RTE_KNI_REQ_CHANGE_MAC_ADDR :/* Change MAC Address */ if (kni->ops.config_mac_address)
@@ -534,7 +534,7 @@ rte_kni_handle_request(struct rte_kni *kni) }unsigned -rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_tx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) {void *phy_mbufs[num];unsigned int ret;
@@ -552,9 +552,9 @@ rte_kni_tx_burst(struct rte_kni *kni, struct rte_mbuf **mbufs, unsigned num) }unsigned -rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned num) +rte_kni_rx_burst (struct rte_kni *kni,struct rte_mbuf **mbufs,unsigned int num) { -unsigned ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num); +unsigned int ret =kni_fifo_get (kni->tx_q, (void **)mbufs, num);/* If buffers removed, allocate mbufs and then put them into alloc_q */ if (ret)
@@ -605,7 +605,7 @@ kni_allocate_mbufs(struct rte_kni *kni)return ; } - allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) \ + allocq_free = (kni->alloc_q->read - kni->alloc_q->write -1 ) & (MAX_MBUF_BURST_NUM -1 );for (i =0 ; i < allocq_free; i++) { pkts[i] =rte_pktmbuf_alloc (kni->pktmbuf_pool);
@@ -659,35 +659,35 @@ static enum kni_ops_status kni_check_request_register(struct rte_kni_ops *ops) {/* check if KNI request ops has been registered*/ -if (NULL == ops ) +if (ops ==NULL )return KNI_REQ_NO_REGISTER ; -if ((ops->change_mtu ==NULL ) - && (ops->config_network_if ==NULL ) - && (ops->config_mac_address ==NULL ) - && (ops->config_promiscusity ==NULL )) +if (ops->change_mtu ==NULL + && ops->config_network_if ==NULL + && ops->config_mac_address ==NULL + && ops->config_promiscusity ==NULL )return KNI_REQ_NO_REGISTER ;return KNI_REQ_REGISTERED ; }int -rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) +rte_kni_register_handlers (struct rte_kni *kni,struct rte_kni_ops *ops) {enum kni_ops_status req_status; -if (NULL == ops) { +if (ops ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid KNI request operation.\n" );return -1 ; } -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; } req_status =kni_check_request_register (&kni->ops); -if (KNI_REQ_REGISTERED == req_status) { +if (req_status ==KNI_REQ_REGISTERED ) {RTE_LOG (ERR ,KNI ,"The KNI request operation has already registered.\n" );return -1 ; }
@@ -699,7 +699,7 @@ rte_kni_register_handlers(struct rte_kni *kni,struct rte_kni_ops *ops)int rte_kni_unregister_handlers (struct rte_kni *kni) { -if (NULL == kni) { +if (kni ==NULL ) {RTE_LOG (ERR ,KNI ,"Invalid kni info.\n" );return -1 ; }
Signature · 1 line
2.20.1
PATCH v6 8/9 kni: add minimal ethtool
In reply to Stephen HemmingerSome applications use ethtool so add the minimum ethtool ops.
--- kernel/linux/kni/kni_dev.h | 2 ++ kernel/linux/kni/kni_misc.c | 1 + kernel/linux/kni/kni_net.c | 14 ++++++++++++++ 3 files changed, 17 insertions(+)
kernel/linux/kni/kni_dev.h +2 -0
@@ -11,6 +11,8 @@#endif #define pr_fmt (fmt) KBUILD_MODNAME ": " fmt +#define KNI_VERSION "1.0" +#include "compat.h" #include <linux/if.h>
kernel/linux/kni/kni_misc.c +1 -0
@@ -21,6 +21,7 @@#include "compat.h" #include "kni_dev.h" +MODULE_VERSION (KNI_VERSION );MODULE_LICENSE ("Dual BSD/GPL" );MODULE_AUTHOR ("Intel Corporation" );MODULE_DESCRIPTION ("Kernel Module for managing kni devices" );
kernel/linux/kni/kni_net.c +14 -0
@@ -13,6 +13,7 @@#include <linux/version.h> #include <linux/netdevice.h> #include <linux/etherdevice.h> /* eth_type_trans */ +#include <linux/ethtool.h> #include <linux/skbuff.h> #include <linux/kthread.h> #include <linux/delay.h>
@@ -725,6 +726,18 @@ static const struct net_device_ops kni_net_netdev_ops = {#endif }; +static void kni_get_drvinfo (struct net_device *dev, +struct ethtool_drvinfo *info) +{ +strlcpy (info->version,KNI_VERSION ,sizeof (info->version)); +strlcpy (info->driver,"kni" ,sizeof (info->driver)); +} + +static const struct ethtool_ops kni_net_ethtool_ops = { + .get_drvinfo = kni_get_drvinfo, + .get_link = ethtool_op_get_link, +}; +void kni_net_init (struct net_device *dev) {
@@ -736,6 +749,7 @@ kni_net_init(struct net_device *dev)ether_setup (dev);/* assign some of the fields */ dev->netdev_ops = &kni_net_netdev_ops; dev->header_ops = &kni_net_header_ops; + dev->ethtool_ops = &kni_net_ethtool_ops; dev->watchdog_timeo =WD_TIMEOUT ; }
Signature · 1 line
2.20.1
PATCH v6 9/9 doc: update KNI documentation
In reply to Stephen HemmingerUpdate KNI documentation to reflect current ethtool support.
Replace references to out dated tools (ifconfig) with modern iproute2. Tshark is a better replacement for tcpdump.
--- .../prog_guide/kernel_nic_interface.rst | 9 +++---- .../sample_app_ug/kernel_nic_interface.rst | 24 ++++++++++++------- 2 files changed, 20 insertions(+), 13 deletions(-)
doc/guides/prog_guide/kernel_nic_interface.rst +5 -4
@@ -290,7 +290,8 @@ It then puts the mbuf back in the cache. Ethtool ------- -Ethtool is a Linux-specific tool with corresponding support in the kernel -where each net device must register its own callbacks for the supported operations. -The current implementation uses the igb/ixgbe modified Linux drivers for ethtool support. -Ethtool is not supported in i40e and VMs (VF or EM devices). +Ethtool is a Linux-specific tool with corresponding support in the kernel. +The current version of kni provides minimal ethtool functionality +including querying version and link state. It does not support link +control, statistics, or dumping device registers. +
doc/guides/sample_app_ug/kernel_nic_interface.rst +15 -9
@@ -21,14 +21,14 @@ The FIFO queues contain pointers to data packets in the DPDK. This: * Provides a faster mechanism to interface with the kernel net stack and eliminates system calls -* Facilitates the DPDK using standard Linux* userspace net tools (tcpdump, ftp, and so on) +* Facilitates the DPDK using standard Linux* userspace net tools (tshark, rsync, and so on) * Eliminate the copy_to_user and copy_from_user operations on packets. The Kernel NIC Interface sample application is a simple example that demonstrates the use of the DPDK to create a path for packets to go through the Linux* kernel. This is done by creating one or more kernel net devices for each of the DPDK ports. -The application allows the use of standard Linux tools (ethtool, ifconfig, tcpdump) with the DPDK ports and +The application allows the use of standard Linux tools (ethtool, iproute, tshark) with the DPDK ports and also the exchange of packets between the DPDK application and the Linux* kernel. The Kernel NIC Interface sample application requires that the
@@ -220,13 +220,13 @@ Enable KNI interface and assign an IP address: .. code-block:: console - # ifconfig vEth0_0 192.168.0.1 + # ip addr add dev vEth0_0 192.168.0.1 Show KNI interface configuration and statistics: .. code-block:: console - # ifconfig vEth0_0 + # ip -s -d addr show vEth0_0 The user can also check and reset the packet statistics inside the ``kni`` application by sending the app the USR1 and USR2 signals:
@@ -234,16 +234,16 @@ application by sending the app the USR1 and USR2 signals: .. code-block:: console # Print statistics - # kill -SIGUSR1 `pidof kni` + # pkill -USR1 kni # Zero statistics - # kill -SIGUSR2 `pidof kni` + # pkill -USR2 kni Dump network traffic: .. code-block:: console - # tcpdump -i vEth0_0 + # tshark -n -i vEth0_0 The normal Linux commands can also be used to change the MAC address and MTU size used by the physical NIC which corresponds to the KNI interface.
@@ -254,13 +254,19 @@ Change the MAC address: .. code-block:: console - # ifconfig vEth0_0 hw ether 0C:01:02:03:04:08 + # ip link set dev vEth0_0 lladdr 0C:01:02:03:04:08 Change the MTU size: .. code-block:: console - # ifconfig vEth0_0 mtu 1450 + # ip link set dev vEth0_0 mtu 1450 + +Limited ethtool support: + +.. code-block:: console + + # ethtool -i vEth0_0 When the ``kni`` application is closed, all the KNI interfaces are deleted from the Linux kernel.
Signature · 1 line
2.20.1
Re: [dpdk-dev] [PATCH v6 0/9] kni: fixes and cleanups
In reply to Stephen HemmingerQuoted 9 lines · Stephen Hemminger
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
v6 - more updates to documentation
v5 - add minimal ethtool, fix checkpath author complaints v4 - add more style fixes
v3 - rebase to current master, add style fix patch
v2 - get rid of unnecessary padding, combine the unused field patches
Quoted 10 lines
Stephen Hemminger (9):
kni: don't need stubs forrx_modeor ioctl
kni: usenetdev_alloc_skb
kni: don't keep stats inkni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return-EFAULTifcopy_from_userfails
kni: fix style issues
kni: add minimal ethtool
doc: update KNI documentation
For series,
Thanks for the cleanup.
I am OK to get the set for rc2, but if we will do this please lets merge this early so that we can still have time to test and fix any possible issues before rc2.
Thomas Monjalon
AppliedRe: [dpdk-dev] [PATCH v6 0/9] kni: fixes and cleanups
12/07/2019 19:03, Ferruh Yigit:
Quoted 21 lines · Stephen Hemminger
Quoted 9 lines
While testing KNI with netvsc, saw lots of places more code could be safely removed from KNI kernel driver.
v6 - more updates to documentation
v5 - add minimal ethtool, fix checkpath author complaints v4 - add more style fixes
v3 - rebase to current master, add style fix patch
v2 - get rid of unnecessary padding, combine the unused field patchesQuoted 10 lines
Stephen Hemminger (9):
kni: don't need stubs forrx_modeor ioctl
kni: usenetdev_alloc_skb
kni: don't keep stats inkni_net
kni: drop unused fields
kni: use proper type for kni fifo's
kni: return-EFAULTifcopy_from_userfails
kni: fix style issues
kni: add minimal ethtool
doc: update KNI documentation
Quoted 7 lines
For series,
Acked-by Ferruh Yigit <[email protected]>Thanks for the cleanup.
I am OK to get the set for rc2, but if we will do this please lets merge this early so that we can still have time to test and fix any possible issues before rc2.
Applied, thanks