Re: [PATCH v4 63/66] i915: Use the VMA iterator

From: Liam Howlett
Date: Thu Jan 20 2022 - 12:40:41 EST


* Matthew Wilcox <willy@xxxxxxxxxxxxx> [220120 10:50]:
> On Thu, Jan 20, 2022 at 03:59:11PM +0100, Vlastimil Babka wrote:
> > On 12/1/21 15:30, Liam Howlett wrote:
> > > From: "Matthew Wilcox (Oracle)" <willy@xxxxxxxxxxxxx>
> > >
> > > Replace the O(n.log(n)) loop with an O(n) loop.
> >
> > Not true?
>
> Oh, right, that should have been just the linked-list walk.
> I misread it as calling find_vma() for each iteration instead
> of just the first one. Liam, do you mind updating the changelog
> here?

I will update the changelog for v5.

>
> I wonder whether we want a "for_each_contiguous_vma()" that
> will stop on a hole. It seems like a relatively sensible thing
> to do -- walk across a contiguous range of memory and stop if
> there's no VMA mapping a page. Like gup(), for example.

Hmm, so search for a non-zero gap across a range might be a reasonable
implementation.

>
> > > Signed-off-by: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx>
> > > Signed-off-by: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx>
> >
> > Acked-by: Vlastimil Babka <vbabka@xxxxxxx>
> >
> > > ---
> > > drivers/gpu/drm/i915/gem/i915_gem_userptr.c | 14 +++++---------
> > > 1 file changed, 5 insertions(+), 9 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
> > > index 3173c9f9a040..39960973c130 100644
> > > --- a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
> > > +++ b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
> > > @@ -425,12 +425,11 @@ static const struct drm_i915_gem_object_ops i915_gem_userptr_ops = {
> > > static int
> > > probe_range(struct mm_struct *mm, unsigned long addr, unsigned long len)
> > > {
> > > - const unsigned long end = addr + len;
> > > + VMA_ITERATOR(vmi, mm, addr);
> > > struct vm_area_struct *vma;
> > > - int ret = -EFAULT;
> > >
> > > mmap_read_lock(mm);
> > > - for (vma = find_vma(mm, addr); vma; vma = vma->vm_next) {
> > > + for_each_vma_range(vmi, vma, addr + len) {
> > > /* Check for holes, note that we also update the addr below */
> > > if (vma->vm_start > addr)
> > > break;
> > > @@ -438,16 +437,13 @@ probe_range(struct mm_struct *mm, unsigned long addr, unsigned long len)
> > > if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
> > > break;
> > >
> > > - if (vma->vm_end >= end) {
> > > - ret = 0;
> > > - break;
> > > - }
> > > -
> > > addr = vma->vm_end;
> > > }
> > > mmap_read_unlock(mm);
> > >
> > > - return ret;
> > > + if (vma)
> > > + return -EFAULT;
> > > + return 0;
> > > }
> > >
> > > /*
> >
>