pchintar opened a new pull request, #9802:
URL: https://github.com/apache/arrow-rs/pull/9802
# Which issue does this PR close?
- Closes #9801 .
# Rationale for this change
The IPC format specifies that compressed buffers are encoded as:
> `[8 bytes uncompressed length] + compressed data`
The current implementation assumes this invariant when reading the prefix
during decompression.
However, in the reader path, buffers are constructed from metadata
(`offset`, `length`) and passed to the decompression logic without validating
that they contain at least the required 8-byte prefix.
In contrast, Parquet defensively validates input before reading fixed-size
prefixes, e.g.:
```rust
while input_len >= PREFIX_LEN { ... }
if input_len < expected_compressed_size { return Err(...) }
```
This ensures malformed or truncated input does not cause panics.
In IPC, however, the prefix is read without a prior length check, so
malformed input (buffer < 8 bytes) can lead to a panic instead of returning an
error.
# What changes are included in this PR?
* Add a bounds check in `read_uncompressed_size` to ensure `buffer.len() >=
8`
* Return `ArrowError::IpcError` when the prefix is missing
* Propagate the error in `decompress_to_buffer`
# Are these changes tested?
Yes.
* Added `test_read_uncompressed_size_rejects_short_prefix`
* Verifies that a buffer shorter than 8 bytes returns an error instead of
panicking
* Existing compression tests remain unchanged and pass
# Are there any user-facing changes?
No.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
