On Fri, Oct 25, 2024 at 09:22:24PM -0700, Andrew Morton wrote: > On Sat, 26 Oct 2024 03:34:36 +0000 Wei Yang <richard.weiyang@xxxxxxxxx> wrote: > > > >------------------------------------------------------ > > >From: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > > >Subject: mm-vma-the-pgoff-is-correct-if-can_merge_right-fix > > >Date: Fri Oct 25 07:39:41 PM PDT 2024 > > > > > >remove now-unused local > > > > > >mm/vma.c: In function 'vma_merge_new_range': > > >mm/vma.c:965:23: error: unused variable 'start' [-Werror=unused-variable] > > > 965 | unsigned long start = vmg->start; > > > | ^~~~~ > > > > Andrew > > > > The removal of start usage maybe come from another patch. > > > > If my understanding is correct, 'start' would be used to restore the state on > > expansion failure. > > I don't think so. This patch did Yeah I removed the logic that would do this in a recent series. We can drop start. > > > --- a/mm/vma.c > > +++ b/mm/vma.c > > @@ -915,7 +915,6 @@ struct vm_area_struct *vma_merge_new_range(struct vma_merge_struct *vmg) > > unsigned long start = vmg->start; > > unsigned long end = vmg->end; > > pgoff_t pgoff = vmg->pgoff; > > - pgoff_t pglen = PHYS_PFN(end - start); > > bool can_merge_left, can_merge_right; > > > > mmap_assert_write_locked(vmg->mm); > > which was the only usage of `start' in vma_merge_new_range(). Yes this is true now. We're good to drop it, thanks!
