XSUM_hashStream exits on file read errors, breaks multiple input processing #1064

New issue

Closed

opened 2025-08-26 15:56:26 -05:00 by ZoomRmc · 2 comments

ZoomRmc commented

2025-08-26 15:56:26 -05:00

(Migrated from github.com)

github.com/Cyan4973/xxHash@bab7e27f4c/cli/xxhsum.c (L279-L281)

XSUM_hashStream exits on ferror with the Error: a failure occurred reading the input file. message.
This prevents processing multiple input files.

In my case, finding FS/device errors was exactly the point of running xxhsum. When processing files in batches and parsing the output, if the process exited prematurely, the whole batch was marked as failed which resulted in great number of false negatives.

https://github.com/Cyan4973/xxHash/blob/bab7e27f4c6ae4efbb83dd99ae8a554423571635/cli/xxhsum.c#L279-L281 `XSUM_hashStream` exits on `ferror` with the `Error: a failure occurred reading the input file.` message. This prevents processing multiple input files. In my case, finding FS/device errors was exactly the point of running xxhsum. When processing files in batches and parsing the output, if the process exited prematurely, the whole batch was marked as failed which resulted in great number of false negatives.

Cyan4973 commented

2025-08-26 16:02:02 -05:00

(Migrated from github.com)

It requires updating XSUM_hashStream() so that it can return an error. Right now, it can only return successfully, and therefore it exit() when it can't.

It requires updating `XSUM_hashStream()` so that it can return an error. Right now, it can only return successfully, and therefore it `exit()` when it can't.

ZoomRmc commented

2025-08-26 16:11:14 -05:00

(Migrated from github.com)

Sure.
If fixing this is not planned, the docs should mention that outputting any result for every input file is not guaranteed.

Sure. If fixing this is not planned, the docs should mention that outputting any result for every input file is not guaranteed.

No milestone

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

gh-Cyan4973/xxHash#1064

No description provided.

Rows
Columns