diff options
| author |   Steven <sluong@cisco.com> | 2018-11-02 16:28:52 -0700 |
|---|---|---|
| committer |   Damjan Marion <dmarion@me.com> | 2018-11-07 01:13:01 +0000 |
| commit | 005e4d50416d15368a6687076c7c934f89470f72 (patch) | |
| tree | 0bc050d6c369e6b2298dbe44003605979bfa7cf0 /src/plugins/vmxnet3/input.c | |
| parent | 6ccff02def39f314765f82a39f5dea9a667e47c8 (diff) | |
vmxnet3: reduce calling vmxnet3_reg_write_inline
In output.c, we buffer the descriptors and call vmxnet3_reg_write_inline
once outside the loop. This change improves the performance dramatically.
When refilling the ring, there is no need to inform the device unless
explicitly specified by the device (ctrl.update_prod == 1)
Change-Id: I7031d58bff0d249e913d14236d416c91eb6ab94a
Signed-off-by: Steven <sluong@cisco.com>
Diffstat (limited to 'src/plugins/vmxnet3/input.c')
| -rw-r--r-- | src/plugins/vmxnet3/input.c | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/src/plugins/vmxnet3/input.c b/src/plugins/vmxnet3/input.c index 9392d57747d..dffad6c7d47 100644 --- a/src/plugins/vmxnet3/input.c +++ b/src/plugins/vmxnet3/input.c @@ -100,8 +100,9 @@ vmxnet3_device_input_inline (vlib_main_t * vm, vlib_node_runtime_t * node, next = nexts; rx_comp = &rxq->rx_comp[comp_ring->next]; - while (PREDICT_TRUE (n_rx_packets < VLIB_FRAME_SIZE) && - (comp_ring->gen == (rx_comp->flags & VMXNET3_RXCF_GEN))) + while (PREDICT_TRUE ((n_rx_packets < VLIB_FRAME_SIZE) && + (comp_ring->gen == + (rx_comp->flags & VMXNET3_RXCF_GEN)))) { vlib_buffer_t *b0; u32 bi0; |