Fix some of the unsoundness in Buffer
#1294
Merged
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
Cloning unconditionally increased the refcount in `Buffer::clone`, but only called `napi_reference_unref` on dropping the last Buffer (the one with `strong_count == 1`). This means that the refcount will never drop back to zero after cloning, leaking the Buffer. This commit changes it to also unconditionally unref the buffer.
- `slice::from_raw_parts` may never be created with a null pointer, but `napi_get_buffer_info` was not sufficiently checked → UB when passing an empty Buffer - `&'static mut [u8],` is invalid, as it certainly doesn't live for `'static` Switching to `NonNull<u8>` and a `len` field fixes both of these. - I also don't really understand how the `impl ToNapiValue for &mut Buffer` could have been sound. It creates an entirely new `Arc`, but reuses the same `Vec` allocation, leading to... a double free of the `Vec` on drop? I have replaced it with a simple call to `clone` instead.
As far as I can tell, by just removing the bool and letting the drop code do its thing we clean up correctly in all cases. Because `napi_create_external_buffer` gets an owned `Buffer` attached to it via the Box, we can rely on `from_raw` retrieving it in the `drop_buffer` function.
Brooooooklyn
approved these changes
Sep 5, 2022
h-a-n-a
pushed a commit
to speedy-js/napi-rs
that referenced
this pull request
Sep 9, 2022
* fix leaked napi refcount in `Buffer` when cloning Cloning unconditionally increased the refcount in `Buffer::clone`, but only called `napi_reference_unref` on dropping the last Buffer (the one with `strong_count == 1`). This means that the refcount will never drop back to zero after cloning, leaking the Buffer. This commit changes it to also unconditionally unref the buffer. * fix multiple sources of UB in `Buffer` - `slice::from_raw_parts` may never be created with a null pointer, but `napi_get_buffer_info` was not sufficiently checked → UB when passing an empty Buffer - `&'static mut [u8],` is invalid, as it certainly doesn't live for `'static` Switching to `NonNull<u8>` and a `len` field fixes both of these. - I also don't really understand how the `impl ToNapiValue for &mut Buffer` could have been sound. It creates an entirely new `Arc`, but reuses the same `Vec` allocation, leading to... a double free of the `Vec` on drop? I have replaced it with a simple call to `clone` instead. * remove overcomplicated bool and drop impl As far as I can tell, by just removing the bool and letting the drop code do its thing we clean up correctly in all cases. Because `napi_create_external_buffer` gets an owned `Buffer` attached to it via the Box, we can rely on `from_raw` retrieving it in the `drop_buffer` function.
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.
I'm not 100% sure if all these changes are correct, so please have a close look at the commit messages explaining them, and the associated issue #1295 for more context.
CC @Xaeroxe @jose-acevedoflores