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

Groups > linux.kernel > #1261910

Re: [PATCH v3 14/15] dax: dirty extent notification

Show all headers | View raw

On Wed, Nov 04, 2015 at 07:51:31AM +1100, Dave Chinner wrote:
> On Mon, Nov 02, 2015 at 11:20:49PM -0800, Dan Williams wrote:
> > On Mon, Nov 2, 2015 at 9:40 PM, Dave Chinner <david@fromorbit.com> wrote:
> > > On Mon, Nov 02, 2015 at 08:56:24PM -0800, Dan Williams wrote:
> > >> No, we definitely can't do that.   I think your mental model of the
> > >> cache flushing is similar to the disk model where a small buffer is
> > >> flushed after a large streaming write.  Both Ross' patches and my
> > >> approach suffer from the same horror that the cache flushing is O(N)
> > >> currently, so we don't want to make it responsible for more data
> > >> ranges areas than is strictly necessary.
> > >
> > > I didn't see anything that was O(N) in Ross's patches. What part of
> > > the fsync algorithm that Ross proposed are you refering to here?
> > 
> > We have to issue clflush per touched virtual address rather than a
> > constant number of physical ways, or a flush-all instruction.
> .....
> > > So don't tell me that tracking dirty pages in the radix tree too
> > > slow for DAX and that DAX should not be used for POSIX IO based
> > > applications - it should be as fast as buffered IO, if not faster,
> > > and if it isn't then we've screwed up real bad. And right now, we're
> > > screwing up real bad.
> > 
> > Again, it's not the dirty tracking in the radix I'm worried about it's
> > looping through all the virtual addresses within those pages..
> 
> So, let me summarise what I think you've just said. You are
> 
> 1. fine with looping through the virtual addresses doing cache flushes
>    synchronously when doing IO despite it having significant
>    latency and performance costs.
> 
> 2. Happy to hack a method into DAX to bypass the filesystems by
>    pushing information to the block device for it to track regions that
>    need cache flushes, then add infrastructure to the block device to
>    track those dirty regions and then walk those addresses and issue
>    cache flushes when the filesystem issues a REQ_FLUSH IO regardless
>    of whether the filesystem actually needs those cachelines flushed
>    for that specific IO?
> 
> 3. Not happy to use the generic mm/vfs level infrastructure
>    architectected specifically to provide the exact asynchronous
>    cache flushing/writeback semantics we require because it will
>    cause too many cache flushes, even though the number of cache
>    flushes will be, at worst, the same as in 2).
> 
> 
> 1) will work, but as we can see it is *slow*. 3) is what Ross is
> implementing - it's a tried and tested architecture that all mm/fs
> developers understand, and his explanation of why it will work for
> pmem is pretty solid and completely platform/hardware architecture
> independent.
> 
> Which leaves this question: How does 2) save us anything in terms of
> avoiding iterating virtual addresses and issuing cache flushes
> over 3)? And is it sufficient to justify hacking a bypass into DAX
> and the additional driver level complexity of having to add dirty
> region tracking, flushing and cleaning to REQ_FLUSH operations?

I also don't see a benefit of pushing this into the driver.  The generic
writeback infrastructure that is already in place seems to fit perfectly with
what we are trying to do.  I feel like putting the flushing infrastructure
into the driver, as with my first failed attempt at msync support, ends up
solving one aspect of the problem in a non-generic way that is ultimately
fatally flawed.

The driver inherently doesn't have enough information to solve this problem -
we really do need to involve the filesystem and mm layers.  For example:

1) The driver can't easily mark regions as clean once they have been flushed,
meaning that every time you dirty data you add to an ever increasing list of
things that will be flushed on the next REQ_FLUSH.

2) The driver doesn't know how inodes map to blocks, so when you get a
REQ_FLUSH for an fsync you end up flushing the dirty regions for *the entire
block device*, not just the one inode.

3) The driver doesn't understand how mmap ranges map to block regions, so if
someone msyncs a single page (causing a REQ_FLUSH) on a single mmap you will
once again flush every region that has ever been dirtied on the entire block
device.

Each of these cases is handled by the existing writeback infrastructure.  I'm
strongly in favor of waiting and solving this issue with the radix tree
patches.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

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

Thread

[PATCH v3 00/15] block, dax updates for 4.4 Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
  [PATCH v3 12/15] block: enable dax for raw block devices Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
  [PATCH v3 04/15] libnvdimm,  pmem: move request_queue allocation earlier in probe Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
    Re: [PATCH v3 04/15] libnvdimm, pmem: move request_queue allocation  earlier in probe Ross Zwisler <ross.zwisler@linux.intel.com> - 2015-11-03 20:20 +0100
  [PATCH v3 07/15] kvm: rename pfn_t to kvm_pfn_t Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
  [PATCH v3 13/15] block, dax: make dax mappings opt-in by default Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
    Re: [PATCH v3 13/15] block, dax: make dax mappings opt-in by default Dave Chinner <david@fromorbit.com> - 2015-11-03 01:40 +0100
      Re: [PATCH v3 13/15] block, dax: make dax mappings opt-in by default Dan Williams <dan.j.williams@intel.com> - 2015-11-03 08:40 +0100
        Re: [PATCH v3 13/15] block, dax: make dax mappings opt-in by default Dave Chinner <david@fromorbit.com> - 2015-11-03 21:30 +0100
          Re: [PATCH v3 13/15] block, dax: make dax mappings opt-in by default Dan Williams <dan.j.williams@intel.com> - 2015-11-04 00:10 +0100
            Re: [PATCH v3 13/15] block, dax: make dax mappings opt-in by default Dan Williams <dan.j.williams@intel.com> - 2015-11-04 20:30 +0100
  [PATCH v3 05/15] libnvdimm,  pmem: fix size trim in pmem_direct_access() Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
    Re: [PATCH v3 05/15] libnvdimm, pmem: fix size trim in  pmem_direct_access() Ross Zwisler <ross.zwisler@linux.intel.com> - 2015-11-03 20:40 +0100
      Re: [PATCH v3 05/15] libnvdimm, pmem: fix size trim in pmem_direct_access() Dan Williams <dan.j.williams@intel.com> - 2015-11-03 22:40 +0100
  [PATCH v3 14/15] dax: dirty extent notification Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
    Re: [PATCH v3 14/15] dax: dirty extent notification Dave Chinner <david@fromorbit.com> - 2015-11-03 02:20 +0100
      Re: [PATCH v3 14/15] dax: dirty extent notification Dan Williams <dan.j.williams@intel.com> - 2015-11-03 06:10 +0100
        Re: [PATCH v3 14/15] dax: dirty extent notification Dave Chinner <david@fromorbit.com> - 2015-11-03 06:50 +0100
          Re: [PATCH v3 14/15] dax: dirty extent notification Dan Williams <dan.j.williams@intel.com> - 2015-11-03 08:30 +0100
            Re: [PATCH v3 14/15] dax: dirty extent notification Dave Chinner <david@fromorbit.com> - 2015-11-03 22:00 +0100
              Re: [PATCH v3 14/15] dax: dirty extent notification Dan Williams <dan.j.williams@intel.com> - 2015-11-03 22:20 +0100
              Re: [PATCH v3 14/15] dax: dirty extent notification Ross Zwisler <ross.zwisler@linux.intel.com> - 2015-11-03 22:40 +0100
                Re: [PATCH v3 14/15] dax: dirty extent notification Dan Williams <dan.j.williams@intel.com> - 2015-11-03 22:50 +0100
        Re: [PATCH v3 14/15] dax: dirty extent notification Ross Zwisler <ross.zwisler@linux.intel.com> - 2015-11-03 22:20 +0100
          Re: [PATCH v3 14/15] dax: dirty extent notification Dan Williams <dan.j.williams@intel.com> - 2015-11-03 22:40 +0100
  [PATCH v3 10/15] dax,  pmem: introduce zone_device_revoke() and devm_memunmap_pages() Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100
  [PATCH v3 06/15] um: kill pfn_t Dan Williams <dan.j.williams@intel.com> - 2015-11-02 05:40 +0100

csiph-web