Merge dev/socket into main #1393
Conversation
Implement wasi_addr_resolve function.
Also slightly modify the interface to make it more accessible for the end user:
- replace port argument with the service - so the user can actually get the port for a given service if unknown
- introduce __wasi_addr_info_t and use it as a buffer for addresses, instead of generic buffer
- introduce __wasi_addr_info_hints_t so user can enable filtering on the syscall level (and therefore use smaller buffers for addresses)
- add max_size parameter for the API as an output - in case the number of addresses is bigger than the buffer size, user can repeat the call with bigger buffer
This change is very minimalistic, and it doesn't include the followings:
1. implementation of getaddrinfo in the lib-socket
2. sample application
Which are to be added in the following change #1336
…1327) Fix socket-api byte order issue for systems where host byte order and network byte order are the same: - Document data structures used for storing IP addresses - Fix bug in bind and connect methods by updating code in wasi_socket_ext
Slightly change the __wasi_sock_addr_local interface - since we already have a `__wasi_addr_t` structure which is an union, there's no need for passing the length around - the address buffer will always have the right length (i.e. max of all address families).
Slightly changed the interface sock_addr_remote - since we already have a `__wasi_addr_t` structure which is an union, there's no need for passing length around - the address buffer will always have the right length (i.e. max of all address families).
|
We'll definitely have more PRs for sockets, but none of the changes we have in the branch are breaking WAMR so I think it's safe to merge. Having said that, please don't remove the dev/socket branch as we'll be sending few more PRs (for ipv6, some of the socket properties, add getaddrinfo implementation in socket libc etc.) |
Merge main into dev/socket
|
Tend to keep all modifications on |
|
Some of the changes really impact the current wasm app, for example, the implementation of wasi_sock_ext is changed, which will be linked into wasm app, and the signatures of sock_addr_local/sock_addr_remote/sock_addr_resolve are also changed. These mean that the old wasm app cannot correctly run in the current runtime. |
@wenyongh I understand your concerns and I agree those APIs have changed. However, they did not have the implementation, so I'm not sure if there's a serious risk in releasing those - not only they didn't have an implementation, but they were also not exposed through wasi_sock_ext so not sure if there was any usecase of those functions? Having said that, I don't mind waiting with the merge until all the changes we're waiting for are merged to dev/socket. I guess we can have another WAMR release shortly after we merge dev/socket into main? |
@loganek Yes, thanks for your understanding, my concern is that the current wasm app compiled with recent commits (from WAMR-05-18-2022 to current commit in main branch) cannot correctly run with the runtime compiled with dev/socket, developer may complain about the compatibility, though he can recompile the wasm app to resolve the issue. |
|
agree with that it is a riskless option to release without |
For now this implementation only covers posix platforms, as defined in MVP #1336
The ns-lookup accepts domain names as well as suffixes, e.g.: ``` --allow-resolve=* # allow all domain names --allow-resolve=example.com # only allow example.com name resolution --allow-resolve=example.com --allow-resolve=*.example.com # allow example.com and its subdomains' name resolution ```
So getaddrinfo() can be used when compiling wasm app of C programs.
Added socket send and recv timeout options with implementation for posix platform. This is part of a extending support for sockets in WASI. #1336. Also add sample that sets and reads back the send and receive timeouts using the native function binding.
Add a group of socket options used by cURL and rust stdlib, as well as some UDP multicast options.
#1490) For some implementations (e.g. Mac OS) `bind()` requires the length to be exactly equal to either `sockaddr_in` or `sockaddr_in6` structure. Because we always used `sizeof(struct sockaddr_storage)`, `bind()` was returning errors. In this change we fix the behavior. See StackOverflow [1] for details. [1] https://stackoverflow.com/questions/73707162/socket-bind-failed-with-invalid-argument-error-for-program-running-on-macos
This also fixes debug build
…1497) **What** * Updated `copy_buffer_to_iovec_app` so that it copies as much of the buffer into the iovec as specified * Throw invalid value when allocating an iovec of size 0 **Why** * A bug found from TCP client example which allocates 1024 for the iovec size (where the buf size is also 1024) but received bytes is passed in as the `buf_size` argument to `copy_buffer_to_iovec_app`. This would return early after hitting this check `buf + data->buf_len > buf_begin + buf_size`. However, if the amount to copy is less than the iovec size, we should copy that much of the buf size. Eg TCP client sample receives 27(?) bytes at a time, and this copies 27 bytes into the iovec of size 1024 * The TCP client example attempts to recv bytes of size 0, this attempts to wasm malloc size 0, which outputs a warning. We should early return if recv bytes of size 0
|
We've internally performed the tests and the PR is ready for review and merge. @wenyongh could you remove a |
Implement more socket APIs, refer to bytecodealliance#1336 and below PRs: - Implement wasi_addr_resolve function (bytecodealliance#1319) - Fix socket-api byte order issue when host/network order are the same (bytecodealliance#1327) - Enhance sock_addr_local syscall (bytecodealliance#1320) - Implement sock_addr_remote syscall (bytecodealliance#1360) - Add support for IPv6 in WAMR (bytecodealliance#1411) - Implement ns lookup allowlist (bytecodealliance#1420) - Implement sock_send_to and sock_recv_from system calls (bytecodealliance#1457) - Added http downloader and multicast socket options (bytecodealliance#1467) - Fix `bind()` calls to receive the correct size of `sockaddr` structure (bytecodealliance#1490) - Assert on correct parameters (bytecodealliance#1505) - Copy only received bytes from socket recv buffer into the app buffer (bytecodealliance#1497) Co-authored-by: Marcin Kolny <mkolny@amazon.com> Co-authored-by: Marcin Kolny <marcin.kolny@gmail.com> Co-authored-by: Callum Macmillan <callumimacmillan@gmail.com>
No description provided.