Merge with upstream #266
Merged
Merge with upstream #266
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* Update QEMU used in CI I'm getting segfaults on s390x in bytecodealliance#9702 and I can't reproduce them locally in QEMU. Try updating QEMU to see if it's a transient bug. Historical updates haven't always gone well, so let's see what CI says. prtest:full * Update risc-v qemu arguments
In upcoming work we're going to centralize memory management inside `Mmap`. In order to do that, we need memory that logically borrows from the `Mmap` to have access to it. That turns out to be possible in some situations but very difficult in others (see bytecodealliance#9681 for an attempt that was a real hassle to get working). To prepare for this work, start moving these uses to be behind `Arc`. These aren't cloned at the moment, but they will be in the future.
…ance#9712) * Burn down the `allow_attributes_without_reason` backlog Just a bit, not everything fixed. * Fix wasi-nn annotations * Tweak `#[cfg]`
* Rely on `core::error::Error` With Wasmtime's new MSRV at 1.81 this means that `core::error::Error` is available which means that in `no_std` mode the `Error` trait can be used. This has been integrated into `anyhow::Error` already upstream and means that we can remove our own local hacks such as the `Err2Anyhow` trait. This commit removes the `Err2Anyhow` trait and all usage, going back to idiomatic Rust error propagation and conversion even in the `no_std` world. This should make code more portable by default and remove some weird idioms we had for supporting this. prtest:full * Add some trusted vets * Audit object crate update * Disable backtraces on CI
* pulley: Implement `get_frame_pointer` This commit implements the CLIF `get_frame_pointer` instruction which will be needed by trampolines in Wasmtime. This is implemented by generalizing the preexisting `GetSp` `MInst` into a "get special". This the additionally removes the extended `get_sp` opcode in Pulley as it's not necessary as `xmov` can operate on this register. * Review comments
Fill out `get_return_address` lowering in Cranelift.
…e#9724) Similar to bytecodealliance/wasm-tools#1934 but for `wasmtime::FieldType`.
This fixes the output of the pulley cranelift backend with respect to virtual stack offsets and adjustments for calls with stack parameters and results. Previously the virtual stack pointer was updated but the rest of the backend assumed the real stack pointer was updated. This updates the pulley backend to match the other backends in this respects and fixes a test on my "more complete pulley integration" branch.
…yout (bytecodealliance#9697) * cranelift-entity: Make PackedOption and EntityList have guaranteed layout * cranelift-entity: Add docs for PackedOption and EntityList layout
For the larger test/build jobs use some variables in the `*.js` files calculating the matrix to pin the images to fixed versions of GitHub actions images. For the `main.yml` file instead of changing everything over to a pin only do the ones that are currently failing on the `ubuntu-24.04` update. The hope is that there's only a few locations to update pinned versions in the future, and we'll need to keep an eye on CI warnings and such to know when to update these pins in the future.
Unsure what will happen, but it probably can't break things too much... right? cc bytecodealliance#9698
On s390x, the unwound SP is always at current CFA - 160. Therefore, the default rule used on most other platforms (which sets the unwound SP to the current CFA) is incorrect, so we need to provide an explicit DWARF CFI rule to unwind SP. With the platform ABI, the caller's SP is always stored in the register save area like other call-saved GPRs, so we can simply use a normal DW_CFA_offset rule. However, with the new tail-call ABI, the value saved in that slot is incorrect - it is not corrected for the incoming tail-call stack arguments that will have been removed as the tail call returns. To fix this without introducing unnecessary run-time overhead, we can simply use a DW_CFA_val_offset rule that will set the unwound SP to CFA - 160, which is always correct. However, the current UnwindInst abstraction does not allow any way to generate this DWARF CFI instruction. Therefore, we introduce a new UnwindInst::RegStackOffset rule for this purpose. Fixes: bytecodealliance#9719
Manually updating some images after the pinning in bytecodealliance#9717. There's a few more images after this but this should be the lion's share of the update. prtest:full
…nce#9709) This commit is a follow-up to bytecodealliance#9472. It ensures that the fuel checking code makes use of `Context::without` to retrieve the current fuel register. Even though no issues have been found, it is preferable to use `Context::without` in scenarios in which we have to introduce special out-of-band control flow and function calls to ensure that all the register allocation invariants are respected i.e., avoid holding onto to registers that: are not in the stack and that might be needed for function calls.
Update from Ubuntu 22.04 to Ubuntu 24.04, and update the version of clang used accordingly.
…liance#9718) * pulley: Fill out more bits and pieces of stack allocation * Use `push_frame` and `pop_frame` in the Cranelift backend. * Use dedicated instructions for modifying the stack pointer * decrementing the stack pointer is always checked * incrementing the stack pointer is checked in debug mode * Disregard stack probes and stack_limit checks as they're both unnecessary for Pulley (all stack allocations are checked). * Fix pulley fuzz * Fix some tests from a rebase * Fix more tests
…#9721) * Add basic pulley disassembly support to `clif-util` Helpful when running `clif-util compile -D --target pulley64` over the input. * Fix non-pulley build
This commit addresses issues identified while working on issue bytecodealliance#8091. It improves the frame handling in Winch to prevent subtle bugs and enhance the robustness of the code generation process. Previously, there was no clear mechanism to verify when the frame was fully set up and safe to access the local slots allocated for register arguments, including the special slots used for the `VMContext`. As a result, it was possible to inadvertently read from uninitialized memory if calls were made before the frame was properly set up and sealed. This commit introduces two main changes with the objective to help reduce the risk of introducing bugs related to the above: * A `CodeGenPhase` trait, used via the type state pattern to clearly gate the operations allowed during each phase of the code generation process. * Improve the semantics of locals, by clearly separating the notion of Wasm locals and special locals used by the compiler. This specialization allows a more accurate representation of the semantics of Wasm locals and their index space.
… Wasmtime. (bytecodealliance#9689) This PR allows a `no_std` Wasmtime build to be configured with the `async` feature. (Previously, a minimal `no_std` configuration could only run with sync entry points, without suspending of stacks.) The main hurdle to this support was the `wasmtime-fiber` crate. Fortunately, the "unix" variant of fibers was almost entirely portable to a `no_std` environment, owing to the fact that it implements stack-switching manually in assembly itself. I moved the per-ISA implementations to a shared submodule and built the nostd platform backend for `wasmtime-fiber` with a stripped-down version of the unix backend. The nostd backend does not support mmap'd stacks, does not support custom stack allocators, and does not propagate panics. prtest:full
Co-authored-by: Wasmtime Publish <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.