On Fri, Jul 10, 2020 at 09:20:59AM +0800, Zhenyu Ye wrote: > On 2020/7/10 0:48, Catalin Marinas wrote: > > On Thu, Jun 25, 2020 at 04:03:11PM +0800, Zhenyu Ye wrote: > >> @@ -189,8 +195,9 @@ static inline void flush_tlb_page_nosync(struct vm_area_struct *vma, > >> unsigned long addr = __TLBI_VADDR(uaddr, ASID(vma->vm_mm)); > >> > >> dsb(ishst); > >> - __tlbi(vale1is, addr); > >> - __tlbi_user(vale1is, addr); > >> + /* This function is only called on a small page */ > >> + __tlbi_level(vale1is, addr, 3); > >> + __tlbi_user_level(vale1is, addr, 3); > >> } > > > > Actually, that's incorrect. It was ok in v2 of your patches when I > > suggested to drop level 0, just leave the function unchanged but I > > missed that you updated it to pass level 3. > > > > pmdp_set_access_flags -> ptep_set_access_flags -> > > flush_tlb_fix_spurious_fault -> flush_tlb_page -> flush_tlb_page_nosync. > > How do you want to fix this error? I notice that this series have been applied > to arm64 (for-next/tlbi). Should I send a new series based on arm64 (for-next/tlbi)? Just a patch on top with a Fixes: tag. Thanks. -- Catalin
