On Tue 13 Jun 2017 03:33:26 PM CEST, Stefan Hajnoczi <stefa...@redhat.com>
wrote:
> Use qcow2_calc_prealloc_size() to get the required file size.
>
> Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
> Reviewed-by: Alberto Garcia <be...@igalia.com>
You kept my R-b here but one of the changes was in this patch:
> + info = g_new(BlockMeasureInfo, 1);
> + info->fully_allocated =
> + qcow2_calc_prealloc_size(virtual_size, cluster_size,
> + ctz32(refcount_bits));
> + if (DIV_ROUND_UP(info->fully_allocated, cluster_size) > INT_MAX) {
> + g_free(info);
> + error_setg(&local_err, "The image size is too large "
> + "(try using a larger cluster size)");
> + goto err;
> + }
This has the opposite problem than the previous version: valid image
sizes are now rejected by the 'measure' command.
$ qemu-img create -f qcow2 img.qcow2 1P
Formatting 'img.qcow2', fmt=qcow2 size=1125899906842624 encryption=off
cluster_size=65536 lazy_refcounts=off refcount_bits=16
$ build/qemu-img measure -O qcow2 --size 1P
qemu-img: The image size is too large (try using a larger cluster size)
The actual limit is:
#define QCOW_MAX_L1_SIZE 0x2000000
That's 4194304 entries, each one can address cluster_size^2 / 8 bytes
So using that formula, here is the maximum virtual size depending on the
cluster size:
|--------------+------------------|
| Cluster size | Max virtual size |
|--------------+------------------|
| 512 bytes | 128 GB |
| 1 KB | 512 GB |
| 2 KB | 2 TB |
| 4 KB | 8 TB |
| 8 KB | 32 TB |
| 16 KB | 128 TB |
| 32 KB | 512 TB |
| 64 KB | 2 PB |
| 128 KB | 8 PB |
| 256 KB | 32 PB |
| 512 KB | 128 PB |
| 1 MB | 512 PB |
| 2 MB | 2 EB |
|--------------+------------------|
I just created a 2 EB image and it works fine, Linux can detect it
without problems, I can create a file system, etc.
If you specify a larger size, qcow2_grow_l1_table() fails with -EFIB.
Berto
