lkml.org 
[lkml]   [2019]   [Jul]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v1 1/1] mm/vmalloc.c: Fix percpu free VM area search criteria
On Tue, Jul 30, 2019 at 02:13:25PM -0700, sathyanarayanan kuppuswamy wrote:
>
> On 7/30/19 1:54 PM, Dave Hansen wrote:
> > On 7/30/19 1:46 PM, Uladzislau Rezki wrote:
> > > > + /*
> > > > + * If required width exeeds current VA block, move
> > > > + * base downwards and then recheck.
> > > > + */
> > > > + if (base + end > va->va_end) {
> > > > + base = pvm_determine_end_from_reverse(&va, align) - end;
> > > > + term_area = area;
> > > > + continue;
> > > > + }
> > > > +
> > > > /*
> > > > * If this VA does not fit, move base downwards and recheck.
> > > > */
> > > > - if (base + start < va->va_start || base + end > va->va_end) {
> > > > + if (base + start < va->va_start) {
> > > > va = node_to_va(rb_prev(&va->rb_node));
> > > > base = pvm_determine_end_from_reverse(&va, align) - end;
> > > > term_area = area;
> > > > --
> > > > 2.21.0
> > > >
> > > I guess it is NUMA related issue, i mean when we have several
> > > areas/sizes/offsets. Is that correct?
> > I don't think NUMA has anything to do with it. The vmalloc() area
> > itself doesn't have any NUMA properties I can think of. We don't, for
> > instance, partition it into per-node areas that I know of.
> >
> > I did encounter this issue on a system with ~100 logical CPUs, which is
> > a moderate amount these days.
>
> I agree with Dave. I don't think this issue is related to NUMA. The problem
> here is about the logic we use to find appropriate vm_area that satisfies
> the offset and size requirements of pcpu memory allocator.
>
> In my test case, I can reproduce this issue if we make request with offset
> (ffff000000) and size (600000).
>
> >
> --
> Sathyanarayanan Kuppuswamy
> Linux kernel developer
>

I misspoke earlier. I don't think it's numa related either, but I think
you could trigger this much more easily this way as it could skip more
viable vma space because it'd have to find more holes.

But it seems that pvm_determine_end_from_reverse() will return the free
vma below the address if it is aligned so:

base + end > va->va_end

will always be true and then push down the searching va instead of using
that va first.

Thanks,
Dennis

\
 
 \ /
  Last update: 2019-07-30 23:57    [W:2.553 / U:0.008 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site