Add TCP6 Protocol #34
Conversation
|
Is it correct to represent enums as 32-bit constants? I was not sure about that. Also, I am not sure how to represent the Service Binding Protocol |
Ayush1325
force-pushed
the
network_protocols
branch
from
9e4a25f
to
448c2c2
Compare
Signed-off-by: Ayush Singh <ayushsingh1325@gmail.com>
| hop_limit: u8, | ||
| flow_lable: u32, | ||
| receive_timeout: u32, | ||
| transmit_timeout: u32, |
There was a problem hiding this comment.
All the struct-members lack the pub annotation.
| prefix_length: u8, | ||
| } | ||
|
|
||
| pub type NeighboutState = u32; |
There was a problem hiding this comment.
Neighbout -> Neighbor
Also note that American-English spells it with o instead of ou (as in British English). The spec uses the American style, so I fixed it all over the place.
| pub const EFI_NEIGHBOUR_REACHABLE: NeighboutState = 1; | ||
| pub const EFI_NEIGHBOUR_STATE: NeighboutState = 2; | ||
| pub const EFI_NEIGHBOUR_DELAY: NeighboutState = 3; | ||
| pub const EFI_NEIGHBOUR_PROBE: NeighboutState = 4; |
There was a problem hiding this comment.
The EFI_* prefix can be safely dropped here.
| prefix_table: *mut AddressInfo, | ||
| icmp_type_count: u32, | ||
| icmp_type_list: *mut IcmpType, | ||
| } |
There was a problem hiding this comment.
I added the missing IP6 protocols definitions as well!
| pub flush_queues_on_reset: Boolean, | ||
| pub enable_receive_timestamps: Boolean, | ||
| pub disable_background_polling: Boolean, | ||
| } |
There was a problem hiding this comment.
I completed the protocol with the other bits as well!
|
|
||
| pub type ConnectionState = u32; | ||
|
|
||
| pub const CONNECTION_CLOSED: ConnectionState = 0; |
There was a problem hiding this comment.
I dropped the CONNECTION_* prefix, as it is not used in the other values below. I added the STATE_* prefix though, as that follows how we encode C-enums.
| pub urgent_flag: Boolean, | ||
| pub data_length: u32, | ||
| pub fragment_count: u32, | ||
| pub fragment_table: *mut FragmentData, |
There was a problem hiding this comment.
The last part is not a pointer but inline data. This needs to be [FramentData; 0] to follow our conversion-style.
| pub urgent: Boolean, | ||
| pub data_length: u32, | ||
| pub fragment_count: u32, | ||
| pub fragment_table: *mut FragmentData, |
| #[repr(C)] | ||
| pub struct IoToken { | ||
| pub rx_data: *mut ReceiveData, | ||
| pub tx_data: *mut TransmitData, |
There was a problem hiding this comment.
This is a bit tricky to get right in rust. This needs an extracted union (we cannot nest declarations in rust like the spec does in C; especially anonymous entries are not a stable rust feature).
| 0x9a, | ||
| 0x0d, | ||
| &[0xd2, 0xe4, 0xcc, 0x16, 0xd6, 0x64], | ||
| ); |
There was a problem hiding this comment.
I merged this into tcp6.rs, just like EDK2 does.
I pushed the ServiceBindingProtocol now. Note that https://docs.rs/r-efi/latest/r_efi/ has a bit of information on the transpose-guidelines we follow. If things are unclear, we usually have precedent somewhere in base.rs or system.rs, or in one of the existing protocols. As it might be rather cumbersome to find it, I don't mind fixing up protocols myself. If you have questions regarding my fixups, let me know! I usually just try to keep the different protocols in the same style. I also want to avoid deviations from the spec, so people don't have to guess what the names are in r-efi, but can rely on the spec-wording. Lastly, since things are not always trivial, and since the spec has errors on its own, I generally follow what EDK2 / Tianorcore does. Thanks for the protocols! Very much appreciated! If you want me to push a minor release, let me know! |
|
Thanks, I will keep things in mind from my next PR |
Signed-off-by: Ayush Singh ayushsingh1325@gmail.com