On Fri, Mar 03, 2023 at 10:16:46AM -0700, Aaron Simmons wrote: > I'm trying to bring preempt-rt into a kernel that has had > live-patching for arm64 applied. Specifically, this patch: > https://lkml.org/lkml/2021/6/4/1169 > > There's a conflict in some definitions of arm64's thread_info.h: > ++<<<<<<< HEAD > +#define TIF_PATCH_PENDING 7 /* pending live patching update */ > ++======= > + #define TIF_NEED_RESCHED_LAZY 7 > ++>>>>>>> 3fc990649e4f (PREEMPT_RT 5.15.93-rt58) > > It appears that the TIF_XXX range is getting crowded. :) > > Resolving the conflict seems straightforward -- I should just give > TIF_NEED_RESCHED_LAZY a new value. But I wanted to double-check -- > does the range of these values matter? If the new bit youa re adding is used in _TIF_WORK_MASK amke sure it is contiguous with the other bits in the mask. For reference: https://lore.kernel.org/all/Y8lCspFtQ0wg6uCy@xxxxxxxx/ Luis > I see over in 32-bit arm's thread_info.h a comment about the "range of > 0..15" being important, but I'm not sure if that applies to 64-bit or > not. > > > Thanks, > aaron > ---end quoted text---
