On 4/29/19 7:29 AM, Richard W.M. Jones wrote:
On Sat, Apr 27, 2019 at 04:26:45PM -0500, Eric Blake wrote:
> Test the just-fixed bug in the truncate filter. Although it is in
> general unsafe to shrink an image being served by the file plugin, the
> test is careful to only issue operations on connection A after the
> file has been resized back to its original length, in order to prove
> whether the smaller size observed during the short-lived connection B
> bleeds over into the pread results seen by connection A. If this test
> is applied without the previous patch, the second read from A sees all
> zeroes instead of the desired pattern 2.
>
> Signed-off-by: Eric Blake <eblake(a)redhat.com>
I guess this test will break in future if we implement and return
‘can_resize = true’ in the file plugin? Or will the (also updated)
truncate filter be able to cope?
If nothing else, this test will remind us of things to test as we
implement resize support :)
Anyway since we can adjust or remove the test if necessary later,
ACK
Thanks for the review; you were right that <pthread.h> is not needed in
patch 2, so with that fix, I'm pushing 1-3. I'll decide if I want to try
a different wording for patch 4, since you were hesitant about the
initial wording attempt.
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3226
Virtualization:
qemu.org |
libvirt.org