On Fri, Nov 18, 2022 at 11:32:36AM +0100, Vlastimil Babka wrote: > On 11/18/22 04:56, Kees Cook wrote: > > With all "silently resizing" callers of ksize() refactored, remove the > > At cursory look seems it's true now in -next (but not mainline?) can you > confirm? Almost, yes. I realized there is 1 case in the BPF verifier that remains. (I thought it was picked up, but only a prereq patch was.) I'm going to resend that one today, but I would expect it to be picked up soon. (But, yes, definitely not for mainline.) > That would probably be safe enough to have slab.git expose this to -next now > and time a PR appropriately in the next merge window? Possibly. I suspect syzkaller might trip KASAN on any larger BPF tests until I get the last one landed. And if you don't want to do the timing of the PR, I can carry this patch in my hardening tree, since I already have to do a two-part early/late-merge-window PR there. -- Kees Cook
