Re: [Qemu-devel] [RFC PATCH v4 0/4] qemu-img: add preallocation=full

[Stefan Hajnoczi]

On Fri, Dec 27, 2013 at 11:05:50AM +0800, Hu Tao wrote:
> This series implements full image preallocation to create a non-sparse image
> file at creation time, both for raw and qcow2 format. The purpose is to avoid
> performance deterioration of the guest cause by sparse image.
> 
> v4:  - remove bdrv_preallocate and make preallocation a bdrv_create_file 
> option
>      - prealloc_mode -> PreallocMode and add it to QAPI
>      - fix return value in raw_preallocate2
> 
> Hu Tao (4):
>   qapi: introduce PreallocMode
>   raw,qcow2: don't convert file size to sector size
>   raw-posix: Add full image preallocation option
>   qcow2: Add full image preallocation option
> 
>  block/qcow2.c     | 64 
> +++++++++++++++++++++++++++++++++++++++++++++----------
>  block/raw-posix.c | 50 +++++++++++++++++++++++++++++++++++++++++--
>  qapi-schema.json  | 12 +++++++++++
>  3 files changed, 113 insertions(+), 13 deletions(-)

I've thought more about the lack of progress information during
posix_fallocate(3).  Users creating 10s or 100s of GB images may have to
wait for a long time if posix_fallocate(3) is implemented in userspace
(effectively dd if=/dev/zero of=vm.img).

But I think adding preallocate=full is a good starting point.  Modern
file systems implement it efficiently so the wait time should be
negligible.  If creating images takes too long we could consider a
fancier interface later.

I left comments on specific patches.