| 5764da5f | 04-Sep-2025 |
Joel Dice <[email protected]> |
Revamp component model stream/future host API (again) (#11515)
* Revamp component model stream/future host API (again)
This changes the host APIs for dealing with futures and streams from a "rendez
Revamp component model stream/future host API (again) (#11515)
* Revamp component model stream/future host API (again)
This changes the host APIs for dealing with futures and streams from a "rendezvous"-style API to a callback-oriented one.
Previously you would create e.g. a `StreamReader`/`StreamWriter` pair and call their `read` and `write` methods, respectively, and those methods would return `Future`s that resolved when the operation was matched with a corresponding `write` or `read` operation on the other end.
With the new API, you instead provide a `StreamProducer` trait implementation whe creating the stream, whose `produce` method will be called as soon as a read happens, giving the implementation a chance to respond immediately without making the reader wait for a rendezvous. Likewise, you can match the read end of a stream to a `StreamConsumer` to respond immediately to writes. This model should reduce scheduling overhead and make it easier to e.g. pipe items to/from `AsyncWrite`/`AsyncRead` or `Sink`/`Stream` implementations without needing to explicitly spawn background tasks. In addition, the new API provides direct access to guest read and write buffers for `stream<u8>` operations, enabling zero-copy operations.
Other changes:
- I've removed the `HostTaskOutput`; we were using it to run extra code with access to the store after a host task completes, but we can do that more elegantly inside the future using `tls::get`. This also allowed me to simplify `Instance::poll_until` a bit.
- I've removed the `watch_{reader,writer}` functionality; it's not needed now given that the runtime will automatically dispose of the producer or consumer when the other end of the stream or future is closed -- no need for embedder code to manage that.
- In order to make `UntypedWriteBuffer` `Send`, I had to wrap its raw pointer `buf` field in a `SendSyncPtr`.
- I've removed `{Future,Stream}Writer` entirely and moved `Instance::{future,stream}` to `{Future,Stream}Reader::new`, respectively.
- I've added a bounds check to the beginnings of `Instance::guest_read` and `Instance::guest_write` so that we need not do it later in `Guest{Source,Destination}::remaining`, meaning those functions can be infallible.
Note that I haven't updated `wasmtime-wasi` yet to match; that will happen in one or more follow-up commits.
Signed-off-by: Joel Dice <[email protected]>
* Add `Accessor::getter`, rename `with_data` to `with_getter`
* fixup bindgen invocation
Signed-off-by: Roman Volosatovs <[email protected]>
* add support for zero-length writes/reads to/from host
I've added a test to cover this; it also tests direct buffer access for `stream<u8>`, which I realized I forgot to cover earlier. And of course there was a bug :facepalm:.
Signed-off-by: Joel Dice <[email protected]>
* add `{Destination,Source}::remaining` methods
This can help `Stream{Producer,Consumer}` implementations determine how many items to write or read, respectively.
Signed-off-by: Joel Dice <[email protected]>
* wasi: migrate sockets to new API
Signed-off-by: Roman Volosatovs <[email protected]>
* tests: read the socket stream until EOF
Signed-off-by: Roman Volosatovs <[email protected]>
* p3-sockets: account for cancellation
Signed-off-by: Roman Volosatovs <[email protected]>
* p3-sockets: mostly ensure byte buffer cancellation-safety
Signed-off-by: Roman Volosatovs <[email protected]>
* p3-filesystem: switch to new API
Signed-off-by: Roman Volosatovs <[email protected]>
* fixup! p3-sockets: mostly ensure byte buffer cancellation-safety
* p3-cli: switch to new API
Signed-off-by: Roman Volosatovs <[email protected]>
* p3: limit maximum buffer size
Signed-off-by: Roman Volosatovs <[email protected]>
* p3-sockets: remove reuseaddr test loop workaround
Signed-off-by: Roman Volosatovs <[email protected]>
* p3: drive I/O in `when_ready`
Signed-off-by: Roman Volosatovs <[email protected]>
* fixup! p3: drive I/O in `when_ready`
* Refine `Stream{Producer,Consumer}` APIs
Per conversations last week with Roman, Alex, and Lann, I've updated these traits to present a lower-level API based on `poll_{consume,produce}` functions and have documented the implementation requirements for various scenarios which have come up in `wasmtime-wasi`, particularly around graceful cancellation. See the doc comments for those functions for details.
Signed-off-by: Joel Dice <[email protected]>
* being integration of new API
Signed-off-by: Roman Volosatovs <[email protected]>
* update wasi/src/p3/filesystem to use new stream API
This is totally untested so far; I'll run the tests once we have everything else compiling.
Signed-off-by: Joel Dice <[email protected]>
* update wasi/src/p3/cli to use new stream API
This is totally untested and doesn't even compile yet due to a lifetime issue I don't have time to address yet. I'll follow up later with a fix.
Signed-off-by: Joel Dice <[email protected]>
* fix: remove `'a` bound on `&self`
Signed-off-by: Roman Volosatovs <[email protected]>
* finish `wasi:sockets` adaptation
Signed-off-by: Roman Volosatovs <[email protected]>
* finish `wasi:cli` adaptation
Note, that this removes the read optimization - let's get the implementation complete first and optimize later
Signed-off-by: Roman Volosatovs <[email protected]>
* remove redundant loop in sockets
Signed-off-by: Roman Volosatovs <[email protected]>
* wasi: buffer on 0-length reads
Signed-off-by: Roman Volosatovs <[email protected]>
* finish `wasi:filesystem` adaptation
Signed-off-by: Roman Volosatovs <[email protected]>
* remove `MAX_BUFFER_CAPACITY`
Signed-off-by: Roman Volosatovs <[email protected]>
* refactor `Cursor` usage
Signed-off-by: Roman Volosatovs <[email protected]>
* impl Default for VecBuffer
Signed-off-by: Roman Volosatovs <[email protected]>
* refactor: use consistent import styling
Signed-off-by: Roman Volosatovs <[email protected]>
* feature-gate fs Arc accessors
Signed-off-by: Roman Volosatovs <[email protected]>
* Update test expectations
---------
Signed-off-by: Joel Dice <[email protected]> Signed-off-by: Roman Volosatovs <[email protected]> Co-authored-by: Alex Crichton <[email protected]> Co-authored-by: Roman Volosatovs <[email protected]>
show more ...
|