On 14/08/2018 14:57, Peter Maydell wrote: > On 14 August 2018 at 13:56, Paolo Bonzini <pbonz...@redhat.com> wrote: >> On 14/08/2018 14:06, Peter Maydell wrote: >>> What about the boards that use the serial.c code but do not >>> have PCI, ISA or a superio chip? That is, all the boards/devices >>> that call serial_mm_init() directly to create a memory-mapped >>> 16550. >> >> They just add >> >> CONFIG_SERIAL=y >> >> to the .mak file. > > ...but the patch that has kicked off this thread is *removing* > CONFIG_SERIAL=y from the various .mak files...
Probably because they were including pci.mak. Indeed CONFIG_SERIAL=y should remain in the ARM/HPPA/Microblaze/MIPS/Mozie/NiOS2/OpenRISC/PPC/RiscV/SH/SPARC64/Xtensa .mak files. Another possibility would be to split .mak files per-machine, and include the per-machine file in the toplevel. Thanks, Paolo