On Tue, Oct 20, 2026 at 12:19 PM Jakub Kicinski <[email protected]> wrote: > > On Thu, 15 Oct 2026 16:39:14 +0000 Mina Almasry wrote: > > Set a 64-bit DMA mask on nsim_bus_dev, enable NETMEM_TX_DMA on the > > netdevsim net_device, and configure page_pool instances with > > PP_FLAG_ALLOW_UNREADABLE_NETMEM and queue_idx, plus PP_FLAG_DMA_MAP | > > PP_FLAG_DMA_SYNC_DEV when CONFIG_HAS_DMA is set, so devmem bind-rx and > > bind-tx netlink operations can be tested against netdevsim in software > > selftests. > > Slightly worried this will bring a bunch of maintenance burden and > "bugs" in netdevsim. IOW that this will be more pain than gain.
I felt the same way when I noticed many reported bugs against this series because I didn't add proper support. But on the other hand the test coverage is nice, and maybe one day if we add enough devmem support in netdevsim we'll be able to fuzz the data path or something. Is there a way to get the best of both worlds? Like put a comment on top of netdevsim that this is test simulator device and any missing functionality and/or bugs in it should be medium criticality max or something? Or is it too much trouble and I should drop the netdevsim support and tests? -- Thanks, Mina
