On 09/10/2018 15:27, Thomas Huth wrote: > On 2018-10-09 13:45, Philippe Mathieu-Daudé wrote: >> Hi Thomas, >> >> On 08/10/2018 14:39, Thomas Huth wrote: >>> The spapr-rng device is suboptimal when compared to virtio-rng, so >>> users might want to disable it in their builds. Thus let's introduce >>> a proper CONFIG switch to allow us to compile QEMU without this device. >>> The function spapr_rng_populate_dt is required for linking, so move it >>> to a different location. >> >> Without CONFIG_SPAPR_RNG this function is not reachable, can you use a >> stub instead? > > I'd like to avoid stubs if possible. stubs are fine if there is no easy > other solution, but they are a little bit confusing ("which of these > functions is now linked into the executable?"), so in case there is no > urgent need, it's IMHO way nicer to get along without them. In this > case, there is no real urgent need - the function is just small, and > does not have any other dependencies into the disabled code, so why > should we complicate things here and introduce a stub?
In this case this is fair enough because this function uses the same API than the surrounding functions (no extra headers/libs required). [I just read v1 comments on this series] Reviewed-by: Philippe Mathieu-Daudé <phi...@redhat.com>