On Mon, 2025-07-21 at 15:02 +0200, Paul Chaignon wrote: > This patch adds two selftests to cover invalid narrower loads on the > context. These used to cause kernel warning before the previous patch. > To trigger the warning, the load had to be aligned, to read an affected > pointer field (ex., skb->sk), and not starting at the beginning of the > pointer field. The new selftests show two such loads of 1B and 4B sizes. > > Signed-off-by: Paul Chaignon <paul.chaignon@xxxxxxxxx> > --- Since we hit a bug here, I think it would be nice to have a test case for each buggy field if that could be done in a terse way. Wrapping `invalid_narrow_skb_sk_load` here in a macro and stamping a few instances seem to be sufficient. [...]
