[dpdk-dev] [PATCH v8 3/3] i40e: add floating VEB extension support
On Wed, May 25, 2016 at 01:28:06AM +0800, Zhe Tao wrote: > To enable this feature, the user should pass a devargs parameter to the EAL > like "-w 84:00.0,enable_floating=1", and the application will make sure the > PMD > will use the floating VEB feature for all the VFs created by this PF device. > > Also you can specifiy which VF need to connect to this floating veb using > "floating_bitmap", every bit corresponding to one VF (e.g. bitn for VFn). > Like "-w 84:00.0,enable_floating=1,floating_bitmap=1", means only the VF0 > connect > to the floating VEB, VF1 connect to the legacy VEB. Is there only ever one floating VEB inside a PF? That perhaps needs to be called out in the explanation of what the floating VEB is? /Bruce
[dpdk-dev] [PATCH v8 3/3] i40e: add floating VEB extension support
Tested-by: Peng Yuan - Test Commit: f3625976474a595bfbdb44ad1c881b07d722a226 15ef663c2acb7ed576083f97cd2555fc9900c724 d0e57ebaf63fe240391dfafebd28f8e6559da899 - OS/Kernel: Fedora 23/4.2.3 - GCC: gcc version 5.3.1 20151207 (Red Hat 5.3.1-2) (GCC) - CPU: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz - NIC: Fortville 2*40G - Total 6 cases,6 passed, 0 failed. There are three patch about the floating VEB feature, I put all of them, and just send one test-by. Here I listed the test case summary, you can get the detail steps from attachment. I run test cases according to my test plan, all the test cases passed. Test Case1: Floating VEB Inter-VM VF-VF === Summary: DPDK PF, then create 2VFs and 2VMs, assign one VF to one VM, say VF1 in VM1, VF2 in VM2, and make PF link down(the cable can be pluged out). VFs in VMs are running dpdk testpmd, VF1 send traffic, and set the packet's DEST MAC to VF2, check VF2 can receive the packets. Check Inter-VM VF-VF MAC switch when PF is link down as well as up. Test Case2: Floating VEB PF can't get traffic from VF DPDK PF, then create 1VF, assign VF1 to VM1, send traffic from VF1 to PF, then check PF will not see any traffic. Test Case3: Floating VEB VF can't receive traffic from outside world == DPDK PF, then create 1VF, assign VF1 to VM1, send traffic from tester to VF1, in floating mode, check VF1 can't receive traffic from tester. Test Case4: Floating VEB VF can not communicate with legacy VEB VF inter-VM == Summary: DPDK PF, then create 3VFs and 3VMs, assign one VF to one VM, say VF1 in VM1, VF3 in VM3, floating VEB; VF2 in VM2, lagecy VEB. Make PF link down(port stop all), VFs in VMs are running dpdk testpmd. 1. VF1 send traffic, and set the packet's DEST MAC to VF2, check VF2 can not receive the packets. 2. VF1 send traffic, and set the packet's DEST MAC to VF3, check VF3 can receive the packets. 3. VF2 send traffic, and set the packet's DEST MAC to VF1, check VF1 can not receive the packets. Check Inter-VM VF-VF MAC switch when PF is link down as well as up. Test Case5: PF can't get traffic from Floating VEB VF, but PF can get traffic from legacy VEB VF. DPDK PF, then create 2VF, assign VF1 to VM1, VF2 to VM2, set VF1 in floating VEB, VF2 in legacy VEB. 1. Send traffic from VF1 to PF, then check PF will not see any traffic; 2. Send traffic from VF2 to PF, then check PF will receive all the packets. Test Case6: Floating VEB VF can't receive traffic from outside world, while legacy VEB VF can receive traffic from outside world. == DPDK PF, then create 2VF, assign VF1 to VM1, assign VF2 to VM2, set VF1 in floating VEB, set VF2 in legacy VEB 1. send traffic from tester to VF1, check VF1 can't receive traffic from tester. 2. send traffic from tester to VF2, check VF2 can receive all the traffic from tester. Thanks! Yuan. -Original Message- From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Zhe Tao Sent: Wednesday, May 25, 2016 1:28 AM To: dev at dpdk.org Cc: Tao, Zhe ; Wu, Jingjing Subject: [dpdk-dev] [PATCH v8 3/3] i40e: add floating VEB extension support To enable this feature, the user should pass a devargs parameter to the EAL like "-w 84:00.0,enable_floating=1", and the application will make sure the PMD will use the floating VEB feature for all the VFs created by this PF device. Also you can specifiy which VF need to connect to this floating veb using "floating_bitmap", every bit corresponding to one VF (e.g. bitn for VFn). Like "-w 84:00.0,enable_floating=1,floating_bitmap=1", means only the VF0 connect to the floating VEB, VF1 connect to the legacy VEB. Signed-off-by: Zhe Tao --- doc/guides/nics/i40e.rst | 5 +++- drivers/net/i40e/i40e_ethdev.c | 56 -- drivers/net/i40e/i40e_ethdev.h | 1 + drivers/net/i40e/i40e_pf.c | 3 ++- 4 files changed, 61 insertions(+), 4 deletions(-) diff --git a/doc/guides/nics/i40e.rst b/doc/guides/nics/i40e.rst index 49a0598..0919a96 100644 --- a/doc/guides/nics/i40e.rst +++ b/doc/guides/nics/i40e.rst @@ -372,4 +372,7 @@ FVL can support floating VEB feature. To enable this feature, the user should pass a devargs parameter to the EAL like "-w 84:00.0,enable_floating=1", and the application will make sure the PMD will use the floating VEB feature for all the VFs created by this PF device. - +Also you can specify which VF need to connect to this floating veb +using "floating_bitmap", every bit corresponding to one VF (e.g. bitn for VFn). +Like "-w 84:00.0,enable_floating=1,floating_bitm
[dpdk-dev] [PATCH v8 3/3] i40e: add floating VEB extension support
To enable this feature, the user should pass a devargs parameter to the EAL
like "-w 84:00.0,enable_floating=1", and the application will make sure the PMD
will use the floating VEB feature for all the VFs created by this PF device.
Also you can specifiy which VF need to connect to this floating veb using
"floating_bitmap", every bit corresponding to one VF (e.g. bitn for VFn).
Like "-w 84:00.0,enable_floating=1,floating_bitmap=1", means only the VF0
connect
to the floating VEB, VF1 connect to the legacy VEB.
Signed-off-by: Zhe Tao
---
doc/guides/nics/i40e.rst | 5 +++-
drivers/net/i40e/i40e_ethdev.c | 56 --
drivers/net/i40e/i40e_ethdev.h | 1 +
drivers/net/i40e/i40e_pf.c | 3 ++-
4 files changed, 61 insertions(+), 4 deletions(-)
diff --git a/doc/guides/nics/i40e.rst b/doc/guides/nics/i40e.rst
index 49a0598..0919a96 100644
--- a/doc/guides/nics/i40e.rst
+++ b/doc/guides/nics/i40e.rst
@@ -372,4 +372,7 @@ FVL can support floating VEB feature.
To enable this feature, the user should pass a devargs parameter to the EAL
like "-w 84:00.0,enable_floating=1", and the application will make sure the PMD
will use the floating VEB feature for all the VFs created by this PF device.
-
+Also you can specify which VF need to connect to this floating veb using
+"floating_bitmap", every bit corresponding to one VF (e.g. bitn for VFn).
+Like "-w 84:00.0,enable_floating=1,floating_bitmap=1", means only the VF0
connect
+to the floating VEB, VF1 connect to the legacy VEB.
diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 8859936..39da1e0 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -750,6 +750,52 @@ i40e_add_tx_flow_control_drop_filter(struct i40e_pf *pf)
" frames from VSIs.");
}
+static int i40e_check_fbitmap_handler(__rte_unused const char *key,
+ const char *value,
+ void *opaque)
+{
+ errno = 0;
+ *(uint16_t *)opaque = strtoul(value, NULL, 0);
+ if (errno)
+ return -1;
+ return 0;
+}
+
+static uint16_t i40e_check_fbitmap(struct rte_devargs *devargs,
+ uint16_t floating)
+{
+ struct rte_kvargs *kvlist;
+ const char *floating_bitmap = "floating_bitmap";
+ /* default value for vf floating bitmap is -1 */
+ uint16_t vf_fbitmap = (uint16_t)-1;
+ uint16_t new_vf_fbitmap;
+
+ if (floating == false)
+ return 0;
+
+ if (devargs == NULL)
+ return vf_fbitmap;
+
+ kvlist = rte_kvargs_parse(devargs->args, NULL);
+ if (kvlist == NULL)
+ return vf_fbitmap;
+
+ if (!rte_kvargs_count(kvlist, floating_bitmap)) {
+ rte_kvargs_free(kvlist);
+ return vf_fbitmap;
+ }
+ /* Floating is enabled when there's key-value pair: enable_floating=1 */
+ if (rte_kvargs_process(kvlist, floating_bitmap,
+ i40e_check_fbitmap_handler,
+ _vf_fbitmap) < 0) {
+ rte_kvargs_free(kvlist);
+ return vf_fbitmap;
+ }
+ rte_kvargs_free(kvlist);
+
+ return new_vf_fbitmap;
+}
+
static int i40e_check_floating_handler(__rte_unused const char *key,
const char *value,
__rte_unused void *opaque)
@@ -884,8 +930,11 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)
/* Need the special FW version support floating VEB */
if (hw->aq.fw_maj_ver >= FLOATING_FW_MAJ) {
pf->floating = i40e_check_floating(pci_dev->devargs);
+ pf->vf_fbitmap = i40e_check_fbitmap(pci_dev->devargs,
+ pf->floating);
} else {
pf->floating = false;
+ pf->vf_fbitmap = 0;
}
/* Clear PXE mode */
i40e_clear_pxe_mode(hw);
@@ -3855,6 +3904,7 @@ i40e_vsi_release(struct i40e_vsi *vsi)
struct i40e_vsi_list *vsi_list;
int ret;
struct i40e_mac_filter *f;
+ uint16_t user_param = vsi->user_param;
if (!vsi)
return I40E_SUCCESS;
@@ -3886,7 +3936,8 @@ i40e_vsi_release(struct i40e_vsi *vsi)
rte_free(f);
if (vsi->type != I40E_VSI_MAIN &&
- ((vsi->type != I40E_VSI_SRIOV) || !pf->floating)) {
+ ((vsi->type != I40E_VSI_SRIOV) ||
+ !(pf->vf_fbitmap && 1 << user_param))) {
/* Remove vsi from parent's sibling list */
if (vsi->parent_vsi == NULL || vsi->parent_vsi->veb == NULL) {
PMD_DRV_LOG(ERR, "VSI's parent VSI is NULL");
@@ -3901,7 +3952,8 @@ i40e_vsi_release(struct i40e_vsi *vsi)
PMD_DRV_LOG(ERR, "Failed to delete element");
}
- if ((vsi->type ==
