On Wed, Sep 23, 2026 at 03:48:45PM +1000, Herbert Xu wrote: > Please check the Sashiko review, it is not happy with your patches: > https://sashiko.dev/#/patchset/20260917225929.2494111-1-aousherovitch%40rambus.com
Thanks, we went through the whole Sashiko list. A number of the findings are false positives against this tree -- e.g. the dt-bindings "unsatisfiable compatible"/"missing example" items are the eip93 idiom Conor asked us to follow, "kzalloc_obj undefined" is a real slab.h macro, and most of the "unprivileged AF_ALG OOM" DoS class isn't reachable because AF_ALG bind is closed for KERN_DRIVER_ONLY|ASYNC algorithms on this kernel. The substantive ones we did address for the next reroll: the MAC/hash .import() accounting symmetry (incl. a poly1305 counter underflow), several DMA-unmap-of-address-0 guards on PQC/KIC error paths, lost-interrupt ordering in the ring handler, a wedged-mailbox -EIO path, and the MAC finup idempotency fixes. A residual class around async-cancel DMA teardown is called out explicitly rather than papered over. Details will be in the v6 changelog. Thanks, Alex
