Builder: avoid unsound buffer reuse #691
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
`toLazyByteString :: Builder -> LazyByteString` had a race condition that could generate wrong results if two threads concurrently evaluated the result. This bug was introduced in haskell#581 (5c4d236) and first present in release 0.11.5.0 (as 0c030bb). Due to the use of `unsafeDupablePerformIO` for performance, it is critical that any IO actions executed when running a `Builder` can be interrupted or executed multiple times. In principle, filling a buffer is safe provided the buffer is used only once and the same bytes are written each time. However, `wrapChunk` in `buildStepToCIOS` would re-use a buffer in the trimming case after copying its contents to produce a new trimmed chunk. This is safe when run in a single thread, but if two threads simultaneously execute the code, one of them may still be copying the contents while the other starts overwriting the buffer. This patch fixes `wrapChunk` to unconditionally allocate a new buffer after trimming, rather than re-using the old buffer. This will presumably come at a slight performance cost for builders inserting many trimmed chunks.
adamgundry
force-pushed
the
wip/690-fix-buffer-reuse
branch
from
f2378bd to
5836a2b
Compare
clyring
approved these changes
Sep 14, 2024
Bodigrim
approved these changes
Sep 14, 2024
Bodigrim
pushed a commit
that referenced
this pull request
Oct 9, 2024
`toLazyByteString :: Builder -> LazyByteString` had a race condition that could generate wrong results if two threads concurrently evaluated the result. This bug was introduced in #581 (5c4d236) and first present in release 0.11.5.0 (as 0c030bb). Due to the use of `unsafeDupablePerformIO` for performance, it is critical that any IO actions executed when running a `Builder` can be interrupted or executed multiple times. In principle, filling a buffer is safe provided the buffer is used only once and the same bytes are written each time. However, `wrapChunk` in `buildStepToCIOS` would re-use a buffer in the trimming case after copying its contents to produce a new trimmed chunk. This is safe when run in a single thread, but if two threads simultaneously execute the code, one of them may still be copying the contents while the other starts overwriting the buffer. This patch fixes `wrapChunk` to unconditionally allocate a new buffer after trimming, rather than re-using the old buffer. This will presumably come at a slight performance cost for builders inserting many trimmed chunks.
Bodigrim
pushed a commit
that referenced
this pull request
Oct 15, 2024
`toLazyByteString :: Builder -> LazyByteString` had a race condition that could generate wrong results if two threads concurrently evaluated the result. This bug was introduced in #581 (5c4d236) and first present in release 0.11.5.0 (as 0c030bb). Due to the use of `unsafeDupablePerformIO` for performance, it is critical that any IO actions executed when running a `Builder` can be interrupted or executed multiple times. In principle, filling a buffer is safe provided the buffer is used only once and the same bytes are written each time. However, `wrapChunk` in `buildStepToCIOS` would re-use a buffer in the trimming case after copying its contents to produce a new trimmed chunk. This is safe when run in a single thread, but if two threads simultaneously execute the code, one of them may still be copying the contents while the other starts overwriting the buffer. This patch fixes `wrapChunk` to unconditionally allocate a new buffer after trimming, rather than re-using the old buffer. This will presumably come at a slight performance cost for builders inserting many trimmed chunks.
|
This is meant to be both 0.11.5.4 and 0.12.2.0, but GitHub allows only one milestone. |
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.
Fixes #690.
toLazyByteString :: Builder -> LazyByteStringhad a race condition that could generate wrong results if two threads concurrently evaluated the result. This bug was introduced in #581 (5c4d236) and first present in release 0.11.5.0 (as 0c030bb).Due to the use of
unsafeDupablePerformIOfor performance, it is critical that any IO actions executed when running aBuildercan be interrupted or executed multiple times. In principle, filling a buffer is safe provided the buffer is used only once and the same bytes are written each time. However,wrapChunkinbuildStepToCIOSwould re-use a buffer in the trimming case after copying its contents to produce a new trimmed chunk. This is safe when run in a single thread, but if two threads simultaneously execute the code, one of them may still be copying the contents while the other starts overwriting the buffer.This patch fixes
wrapChunkto unconditionally allocate a new buffer after trimming, rather than re-using the old buffer. This will presumably come at a slight performance cost for builders inserting many trimmed chunks.