[edk2-devel] [NetworkPkg] Clarification on EFI_TCP6_PROTOCOL Destruction Behavior

[Ayush Singh]

Ok, so I have made some small progress in implementing networking. 
However, I have a few things I don't quite understand since I have never 
really dug so deep into networking before:


1. I am trying to implement a TcpListener[1]. I think TcpListener can be 
implemented by setting the `EFI_TCP6_ACCESS_POINT->ActiveFlag` to false 
during configuration. However, while binding the TcpListenere to an 
address, should the `StationAddress` or the `RemoteAddress` be set to 
the supplied address?


2. Is there any UEFI Shell Application/utility that I can use to listen 
to an address (like `nc` in linux). This would be helpful for me to test 
out whether something is a problem in my implementation or in my qemu 
config.


3. The networking stuff in Rust is synchronous. Currently, I am creating 
a new event during accept with the following parameters:

```

Type = EVT_NOTIFY_SIGNAL

NotifyTpl = TPL_CALLBACK

NotifyFunction = <Empty function that does nothing>

NotifyContext = NULL

```

I pass this event as `ListenToken->CompletionToken->Event`. Then I call 
`EFI_BOOT_SERVICES.WaitForEvent()` on this event. Is there any better 
way to do this?


4. Does the EFI_TCP6_PROTOCOL support IPv4-mapped IPv6 address [2]? I 
was thinking just converting the IPv4 address to IPv6 instead of using 
EFI_TCP4_PROTOCOL.


Ayush Singh


[1]: https://doc.rust-lang.org/std/net/struct.TcpListener.html#

[2]: 
https://doc.rust-lang.org/std/net/struct.Ipv6Addr.html#ipv4-mapped-ipv6-addresses


On 7/14/22 00:00, Ayush Singh wrote:
> Thanks. I'm not sure if I will use it directly, but it gives a very
> good idea about using these protocols in Rust std.
>
> On Wed, Jul 13, 2022 at 11:07 PM Kinney, Michael D
> <michael.d.kinney at intel.com> wrote:
>> It may be simpler to use the following lib class that layers on top of the TCP protocols.
>>
>> https://github.com/tianocore/edk2/blob/master/NetworkPkg/Include/Library/TcpIoLib.h
>>
>> The implementation of this lib shows the correct sequence:
>>
>> https://github.com/tianocore/edk2/blob/master/NetworkPkg/Library/DxeTcpIoLib/DxeTcpIoLib.c
>>
>> It uses NetLib services too
>>
>> https://github.com/tianocore/edk2/blob/master/NetworkPkg/Library/DxeNetLib/DxeNetLib.c
>>
>> Mike
>>
>>
>>> -----Original Message-----
>>> From: Ayush Singh <ayushdevel1325 at gmail.com>
>>> Sent: Wednesday, July 13, 2022 9:45 AM
>>> To: devel at edk2.groups.io
>>> Cc: Kinney, Michael D <michael.d.kinney at intel.com>; mikuback at linux.microsoft.com; Gaibusab, Jabeena B
>>> <jabeena.b.gaibusab at intel.com>; Yao, Jiewen <jiewen.yao at intel.com>
>>> Subject: [NetworkPkg] Clarification on EFI_TCP6_PROTOCOL Destruction Behavior
>>>
>>> Hello everyone. I am trying to implement Network Support in Rust std for
>>> UEFI. While trying to use the TCP6_PROTOCOL, I am not quite sure how to
>>> deal with destructing the protocol. Since TCP6_PROTOCOL is created using
>>> a EFI_SERVICE_BINDING_PROTCOL, it should be destroyed using the
>>> `DestroyChild()` method. However, do I also have to call
>>> `TCP6_PROTOCOL->Close()` before this or will `DestroyChild()` do that
>>> implicitly? If I do have to call this method myself, then do I need to
>>> do `DestroyChild()` in the `CloseToken->Event->NotifyFunction` or is it
>>> fine to `DestroyChild()` after calling `Close()` (event though `Close`
>>> is nonblocking) ?
>>>
>>>
>>> Also, it would be great if someone can point me to some TCP Network
>>> applications. I did find some but most were either drivers, or were very
>>> simple and did not do any Cleanup stuff.
>>>
>>>
>>> Note: Rust does not need the user to close the connection manually. It
>>> is closed once the object is dropped (which is done once the object
>>> owner goes out of scope at compile time).
>>>
>>>
>>> Yours Sincerely
>>>
>>> Ayush Singh


-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#91421): https://edk2.groups.io/g/devel/message/91421
Mute This Topic: https://groups.io/mt/92361755/1813853
Group Owner: devel+owner at edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [edk2-devel-archive at redhat.com]
-=-=-=-=-=-=-=-=-=-=-=-