Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1373791

Re: [PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD mmap

From Toshi Kani <toshi.kani@hpe.com>
Newsgroups linux.kernel
Subject Re: [PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD mmap
Date 2016-04-07 23:30 +0200
Message-ID <rlpF0-1zn-13@gated-at.bofh.it> (permalink)
References <rkWjE-52m-15@gated-at.bofh.it> <rkYY9-6QA-1@gated-at.bofh.it> <rkZUd-7Dl-5@gated-at.bofh.it> <rlme6-7wf-11@gated-at.bofh.it>
Organization linux.* mail to news gateway

Show all headers | View raw


On Thu, 2016-04-07 at 13:41 -0400, Matthew Wilcox wrote:
> On Wed, Apr 06, 2016 at 11:44:32AM -0600, Toshi Kani wrote:
> > > 
> > > The NVML chooses appropriate addresses and gets a properly aligned
> > > address without any kernel code.
> >
> > An application like NVML can continue to specify a specific address to
> > mmap().  Most existing applications, however, do not specify an address
> > to mmap().  With this patch, specifying an address will remain
> > optional.
>
> The point is that this *can* be done in userspace.  You need to sell us
> on the advantages of doing it in the kernel.

Sure.  As I said, the point is that we do not need to modify existing
applications for using DAX PMD mappings.

For instance, fio with "ioengine=mmap" performs I/Os with mmap().
https://github.com/caius/fio/blob/master/engines/mmap.c

With this change, unmodified fio can be used for testing with DAX PMD
mappings.  There are many examples like this, and I do not think we want to
modify all applications that we want to evaluate/test with.

> > > I think this is the wrong place for it, if we decide that this is the
> > > right thing to do.  The filesystem has a get_unmapped_area() which
> > > should be used instead.
> >
> > Yes, I considered adding a filesystem entry point, but decided going
> > this way because:
> >  - arch_get_unmapped_area() and arch_get_unmapped_area_topdown() are
> > arch-specific code.  Therefore, this filesystem entry point will need
> > arch-specific implementation. 
> >  - There is nothing filesystem specific about requesting PMD alignment.
>
> See http://article.gmane.org/gmane.linux.kernel.mm/149227 for Hugh's
> approach for shmem.  I strongly believe that if we're going to do this
> i the kernel, we should build on this approach, and not hack something
> into each architecture's generic get_unmapped_area.

Thanks for the pointer.  Yes, we can call current->mm->get_unmapped_area()
with size + PMD_SIZE, and adjust with the alignment in a filesystem entry
point.  I will update the patch with this approach.

-Toshi

Back to linux.kernel | Previous | Next — Previous in thread | Find similar | Unroll thread


Thread

[PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD mmap Toshi Kani <toshi.kani@hpe.com> - 2016-04-06 16:10 +0200
  Re: [PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD mmap Matthew Wilcox <willy@linux.intel.com> - 2016-04-06 19:00 +0200
    Re: [PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD  mmap Toshi Kani <toshi.kani@hpe.com> - 2016-04-06 20:00 +0200
      Re: [PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD mmap Matthew Wilcox <willy@linux.intel.com> - 2016-04-07 19:50 +0200
        Re: [PATCH] x86 get_unmapped_area: Add PMD alignment for DAX PMD  mmap Toshi Kani <toshi.kani@hpe.com> - 2016-04-07 23:30 +0200

csiph-web