[Linux-stm32] [PATCH net-next v5 16/19] xsk: allow drivers to retain DMA mappings independently of pools

[Björn Töpel]

James Hilliard <james.hilliard1 at gmail.com> writes:

> An unsuccessful device shutdown does not make its DMA memory safe to
> unmap.

Why does this need an AF_XDP API? Can stmmac fence DMA before releasing
the pool? Please show a case where stop and reset fail, DMA remains
active, and the driver cannot isolate it.


Björn