[Backport release-2.17] Fix fragment corruption in the GlobalOrderWriter #4389
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
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
This test case explains part of a bug in the GlobalOrderWriter. When our fragment is full at the beginning of a write to an existing fragment, we would accidentally call `dim_tiles[-1].cell_num()` which would actually manage to return a reasonable value instead of segfaulting. This was due to `num_tiles_to_write` returning 0 coupled with our failure to check for that situation until after a call to set_coords_metadata. Swapping from the `std::vector::operator[]` to `std::vector::at` adds bounds checking and now throws an exception instead of returning a garbage value. This is better since without bounds checking we were corrupting the last tile of every fragment when setting the wrong last tile cell num. (cherry picked from commit f914fd2)
Previously there was an issue with the GlobalOrderWriter when using a maximum fragment size. If we resumed a write where the first tile would cause the partially written fragment to overflow the set maximum size, we failed to check that a single tile could fit. This resulted in a convoluted path where `dim_tiles[-1].cell_num()` return a reasonable result (instead of segfaulting or throwing an exception). This lead us to writing the wrong `last_tile_cell_num` value in the FragmentMetadata. When reading the last tile from one of these corrupted fragments, the logic in `Tile::load_chunk_data` would throw an exception when the expected size was calculated off an invalid `last_tile_cell_num` which caused an exception to be thrown and rendered the last tile of the fragment unreadable. The fix is simply to make sure we check if a single tile can fit and if not, to create a new fragment. For a belt and suspenders approach to safety, we've also changed the code in `WriterBase::set_coords_metadata` to use `std::vector::at` instead of `std::vector::operator[]` so that bounds checking is performed and assertion is thrown rather than writing a bad FragmentMetadata to disk. (cherry picked from commit cc4e7d1)
KiterLuc
approved these changes
Sep 27, 2023
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.
Backport cc4e7d1~2..cc4e7d1 from #4383.