Re: [Qemu-devel] [PATCH] block: Formats don't need CONSISTENT_READ with NO_IO

[Eric Blake <eblake@redhat.com>]

On 12/01/2017 06:41 AM, Kevin Wolf wrote:
>>
>> I guess it is the block of writes/resizes that prevents metadata from
>> getting inconsistent; CONSISTENT_READ does indeed make more sense if
>> interpreted solely in light of will the guest read consistent data
>> (and not will the format layer see consistent contents from the
>> protocol layer).
> 
> Yes, that's what the write/resize locks are meant for.
> 
> Consistent read is more about "this image may not contain the useful
> data you're expecting".
> 

> 
>> I'm not opposed to your patch, but am trying to make sure that I'm not
>> overlooking any problem before giving R-b.  Maybe it's just that the
>> comment needs updating in v2.
> 
> Do you have a suggestion for the comment?

Perhaps:

Commit 1f4ad7d fixed 'qemu-img info' for raw images that are currently
in use as a mirror target. It is not enough for image formats, though,
as these still unconditionally request BLK_PERM_CONSISTENT_READ.

As this permission is geared towards whether the guest-visible data is 
consistent, and has no impact on whether the metadata is sane, and 
'qemu-img info' does not read guest-visible data (except for the raw 
format), it makes sense to not require BLK_PERM_CONSISTENT_READ if there 
is not going to be any guest I/O performed, regardless of image format.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org