[Bug 294491] netmap vale using offsets when it shouldn't
[email protected] Thu, 09 Jul 2026 14:43:44 +0000
| Newsgroups | gmane.os.freebsd.devel.net |
|---|---|
| Message-ID | <[email protected]/bugzilla/> |
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=294491 Teddy Engel <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #2 from Teddy Engel <[email protected]> --- Thanks for the fixes, Giuseppe. I've backported both commits (ef8fa11e3, e0de5b529) to FreeBSD 16.0-CURRENT and tested them with a custom program that sends packets through VALE with various buffer offsets (0–63), verifying data integrity on vp-to-vp and vp-to-iflib paths. I've opened a PR for review: https://github.com/freebsd/freebsd-src/pull/2317 Regarding NS_INDIRECT — I noticed the patch also stops calling nm_write_offset() for indirect packets, which fixes the bogus offset write the reporter described. The other NS_INDIRECT concerns (copy_len including offset bytes, over-copy) are pre-existing and I left them as-is to match upstream. Happy to look into those separately if useful. -- You are receiving this mail because: You are the assignee for the bug.