qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: qcow2: Zero-initialization of external data files


From: Eric Blake
Subject: Re: qcow2: Zero-initialization of external data files
Date: Mon, 6 Apr 2020 17:22:46 -0500
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.6.0

On 2/17/20 10:56 AM, Max Reitz wrote:
Hi,

AFAIU, external data files with data_file_raw=on are supposed to return
the same data as the qcow2 file when read.  But we still use the qcow2
metadata structures (which are by default initialized to “everything
unallocated”), even though we never ensure that the external data file
is zero, too, so this can happen:

$ dd if=/dev/urandom of=foo.raw 64M
[...]

$ sudo losetup -f --show foo.raw
/dev/loop0

$ sudo ./qemu-img create -f qcow2 -o \
     data_file=/dev/loop0,data_file_raw=on foo.qcow2 64M
[...]

$ sudo ./qemu-io -c 'read -P 0 0 64M' foo.qcow2
read 67108864/67108864 bytes at offset 0
64 MiB, 1 ops; 00.00 sec (25.036 GiB/sec and 400.5751 ops/sec)

This looks like a bug (and we should fix it for 5.0 if possible) - read of a data_file_raw=on should not treat unallocated clusters as reading 0, but rather as reading whatever the raw data contains.


$ sudo ./qemu-io -c 'read -P 0 0 64M' -f raw foo.raw
Pattern verification failed at offset 0, 67108864 bytes
read 67108864/67108864 bytes at offset 0
64 MiB, 1 ops; 00.01 sec (5.547 GiB/sec and 88.7484 ops/sec)

I suppose this behavior is fine for blockdev-create because I guess it’s
the user’s responsibility to ensure that the external data file is zero.
  But maybe it isn’t, so that’s my first question: Is it really the
user’s responsibility or should we always ensure it’s zero?

I'd argue that requiring the user to pre-zero the raw data file is undesirable; and that we should instead fix our code to not report the image as reading all zeroes when creating with data_file_raw=on.


My second question is: If we decide that this is fine for
blockdev-create, should at least qcow2_co_create_opts() ensure the data
file it just created is zero?

Having an option to make qemu force-zero the raw image during qcow2_co_create_opts seems reasonable, but for performance reasons, I don't think the flag should be on by default.

--
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org




reply via email to

[Prev in Thread] Current Thread [Next in Thread]