2014-10-29 03:21, Wu, Jingjing: > > > -----Original Message----- > > From: Thomas Monjalon [mailto:thomas.monjalon at 6wind.com] > > Sent: Tuesday, October 28, 2014 10:15 PM > > To: Wu, Jingjing > > Cc: dev at dpdk.org > > Subject: Re: [dpdk-dev] [PATCH v4 16/21] ethdev: define structures for > > configuring flexible payload > > > > 2014-10-22 09:01, Jingjing Wu: > > > +/** > > > + * A structure defined a field vector to specify each field. > > > + */ > > > +struct rte_eth_field_vector { > > > + uint8_t offset; /**< Source word offset */ > > > + uint8_t size; /**< Field Size defined in word units */ > > > +}; > > > > I'm sorry but I don't understand this patch at all. > > I think the reason is that I need more information about flex filter. > > > > Yes, it is a little difficult to understand the behavior. Let me explain it > as I can. > <1> flex words: > We can select where the 8 flex words are. It can be constituted of 1-3 > partitions. > Each partition is defined as rte_eth_field_vector. > For example: > if we set > rte_eth_flex_payload_cfg = { > .type = RTE_ETH_L4_PAYLOAD; > .nb_field = 1; > .field = { > {0, 8}, /*offset = 0, size =8*/ > }; > }; > This means the 8 words from the beginning of l4 payload will be consider as > flex words. > If we set > rte_eth_flex_payload_cfg = { > .type = RTE_ETH_L4_PAYLOAD; > .nb_field = 1; > .field = { > {0, 2}, /*offset = 0, size =2*/ > {4, 6}, /*offset = 4, size =6*/ > }; > }; > This means the 2 words from the beginning of l4 payload plus 6 words form the > offset 4 > will be consider as flex words. > > <2> flex mask: > As the <1> configuration, we already select 8 words as flex words. Then the > flex mask > Will help us to define whether these word will participate in comparing. > The flex mask can be constituted of 2 partitions. One is word mask, another > is bit mask. > Word mask is defined in rte_eth_fdir_flex_masks.words_mask, it means bit i > enables word i of flex words. > rte_eth_fdir_flex_masks.field specify each word's bit mask, Fortville support > 2 word bitmask.
OK thanks. The most difficult task will be to explain it in the doxygen comment :) Please try to be precise and concise in doxygen comments. It must be a reference when using the API. If more examples are required, the developer can look into the programmer's guide. -- Thomas
