tisonkun commented on PR #224: URL: https://github.com/apache/datasketches-rust/pull/224#issuecomment-5412742095
One more issue with the accepted boundary: `(lg_max_map_size, lg_cur_map_size) = (30, 30)` still passes the new validation and immediately attempts a very large allocation. A minimal 8-byte empty header is: ```rust let bytes = [1, 1, 10, 30, 30, 5, 0, 0]; FrequentItemsSketch::<i64>::deserialize(&bytes); ``` This reaches `ReversePurgeItemHashMap::new(1usize << 30)`. On a 64-bit target with `i64` items, the three backing arrays require roughly 26 GiB in total (`Option<i64>` + `u64` + `u16` per slot), so the process may abort or be killed rather than return `Error::deserial`. I did not execute this exact case without a memory limit because it is intentionally OOM-inducing. Also, the test described as exercising the “exact upper bound” currently uses `lg_max = 10`, so it does not cover this path. Could you review whether accepting `lg_cur = 30` is really the intended resource limit? -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
