On Thu, Jul 16, 2020 at 12:53:23PM +0530, Naresh Kamboju wrote: > On Thu, 16 Jul 2020 at 12:07, Naresh Kamboju <naresh.kamboju@xxxxxxxxxx> wrote: > > > > On Thu, 16 Jul 2020 at 04:49, Linus Torvalds > > <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote: > > > > > > On Wed, Jul 15, 2020 at 4:04 PM Linus Torvalds > > > <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote: > > > > > > > > It *might* be as simple as this incremental thing on top > > > > > > No, it needs to be > > > > > > + if (*old_addr + *len < old->vm_end) > > > + return; > > > > > > in try_to_align_end(), of course. > > > > > > Now I'm going for a lie-down, because this cross-eyed thing isn't working. > > > > > > Just want to double check. > > Here is the diff after those two patches applied. Please correct me if > > it is wrong. > > This patch applied on top of Linus mainline master branch. > > I am starting my test cycles. > > Sorry this patch (the below pasted ) did not solve the reported problem. As Linus said, it does not trigger on the stack movement anymore and it is not going to fix the issue, but may help to increase coverage of the optimization. -- Kirill A. Shutemov
