On Thu, 2024-04-18 at 06:52 -0700, Guenter Roeck wrote: > On Thu, Apr 18, 2024 at 01:32:15PM +0000, Winiarska, Iwona wrote: > > On Tue, 2024-04-16 at 23:57 +0000, Luck, Tony wrote: > > > > If the CPU defines and the new macro are to be kept in architecture > > > > code, > > > > maybe include arch/x86/include/asm/cpu_device_id.h from > > > > linux/peci.cpu.h. > > > > That would not be worse than today's include of intel-family.h. > > > > > > Guenter, > > > > > > Looks like I did that to resolve one of the other peci problems. Because I > > > already have: > > > > > > #include "../../arch/x86/include/asm/cpu_device_id.h" > > > #include "../../arch/x86/include/asm/intel-family.h" > > > > > > in <linux/peci_cpu.h> > > > > > > Simply deleting the include from cputemp.c builds OK in the > > > context of all the other changes in my patch series. > > > > Hi Tony, > > > > It won't build on non-x86, as cpu_device_id.h includes <asm/intel-family.h>. > > I think the simplest way to solve the issue is to provide a copy of VFM_* > > macros > > and X86_VENDOR_INTEL in include/linux/peci-cpu.h. > > > > I think the proper fix would really be to move the include files to a > generic directory, such as include/linux/x86/ or include/linux/cpu/x86/. > After all, they _are_ now needed in non-Intel code. Yeah, that was the initial proposal: https://lore.kernel.org/lkml/20210803113134.2262882-2-iwona.winiarska@xxxxxxxxx/ Unfortunately, it ended up being simplified to just include arch/x86 directly. -Iwona > > Guenter
