So far, the pipeline supports the following operations on IPv6 addresses: -using an IPv6 address as a table match field of exact/ternary/LPM match type -assignment of IPv6 address to another IPv6 address -conversion between an IPv4 address and an IPv6 address
The reason for this limited support is the fact that CPUs have 64-bit registers, so supporting operations on 128-bit variables requires additional logic. We are now adding support for the following operations involving IPv6 addresses: -logic operations: AND, OR, XOR -shift left/right -equality/inequality tests The way we do this is by introducing a new instruction (called "movh") to read and write the upper half of a 128-bit operand and optimizing the existing instruction (called "mov"). This way, we can split an IPv6 address into its upper half and lower half and use the existing 64-bit instructions to handle the required operations. We are still not supporting the following operations on IPv6 addresses, as they seem of very little practical use: addition, subtraction, multiplication and division. Cristian Dumitrescu (3): pipeline: add new instruction for upper half of IPv6 address pipeline: optimize conversion between IPv4 and IPv6 addresses examples/pipeline: add example for IPv6 address swap examples/pipeline/examples/ipv6_addr_swap.cli | 35 ++++ .../pipeline/examples/ipv6_addr_swap.spec | 83 ++++++++++ lib/pipeline/rte_swx_pipeline.c | 155 ++++++++++++++++++ lib/pipeline/rte_swx_pipeline_internal.h | 104 +++++++++++- 4 files changed, 373 insertions(+), 4 deletions(-) create mode 100644 examples/pipeline/examples/ipv6_addr_swap.cli create mode 100644 examples/pipeline/examples/ipv6_addr_swap.spec -- 2.34.1
