Re: [PATCH v3 3/6] ext4: cache extent hole in extent status tree for ext4_da_map_blocks()

[Theodore Ts'o <tytso@mit.edu>]

On Thu, Sep 04, 2014 at 09:04:48PM +0800, Zheng Liu wrote:
> > I thought the reason why we have the EXT4_GET_BLOCKS_NO_PUT_HOLE flag
> > is because in ext4_da_map_blocks(), if there is a hole, we will be
> > immediately following it up with a call to ext4_es_insert_extent() to
> > fill in the hole with the EXTENT_STATUS_DELAYED flag.  The only time
> > we don't is if we run into an ENOSPC error.
> > 
> > Am I missing something?
> 
> Yes, you are right.  The purpose is used to do the work like you said
> above.  As the commit log described this flag brings a huge number of
> cache misses when an empty file is written.  

Right, and I was trying to figure out why that was happening, because
it looked like in the normal case we would immediately fill in the
hole afterwards.  I was goign to try doing some tracing to understand
why this was resulting a huge number of cache misses, but I haven't
had time to do that investigation yet.  Can you sketch out the
scenario where this was leading to so many cache misses?

Thanks,

					- Ted