qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] [PATCH 2/2] make check: Add qemu-iotests subset


From: Kevin Wolf
Subject: Re: [Qemu-devel] [PATCH 2/2] make check: Add qemu-iotests subset
Date: Thu, 08 Mar 2012 18:01:55 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:10.0.1) Gecko/20120209 Thunderbird/10.0.1

Am 08.03.2012 17:52, schrieb Stefan Hajnoczi:
> On Thu, Mar 8, 2012 at 4:10 PM, Kevin Wolf <address@hidden> wrote:
>> Am 08.03.2012 16:51, schrieb Stefan Hajnoczi:
>>> On Thu, Mar 8, 2012 at 11:32 AM, Kevin Wolf <address@hidden> wrote:
>>>> Run the 'quick' group from qemu-iotests during 'make check'.
>>>>
>>>> Signed-off-by: Kevin Wolf <address@hidden>
>>>> ---
>>>>  tests/Makefile        |    1 +
>>>>  tests/qemu-iotests.sh |   18 ++++++++++++++++++
>>>
>>> I think tests/qemu-iotests-quick.sh would be a clearer name since this
>>> is not a general-purpose qemu-iotests wrapper - it only works when
>>> called from QEMU's root directory and only invokes the 'quick' group.
>>
>> Ok, I'll rename it.
>>
>>>> +./check -T -nocache -raw -g quick || ret=1
>>>> +./check -T -nocache -qcow2 -g quick || ret=1
>>>
>>> Some love for qed?  It adds 10s on my box but we're already up at 26s.
>>
>> I'm used to high expectations, but loving QED is a bit too much... ;-)
>>
>> Not sure what to do about all the formats. Ideally we would test all of
>> them (at least those with better implementations, VMDK, VHD, VDI), but
>> that would definitely take too long. I hope that in the not too distant
>> future, QED will have a similar position to qcow1, but we may consider
>> adding it for now.
>>
>> However, 'make check' is really the quick test that you run when you
>> don't change anything in the image formats. If you do, you should do a
>> full qemu-iotests run. So the important question is whether QED is
>> likely to reveal any breakage outside block/* that the qcow2 tests
>> wouldn't find.
>>
>> With the same reason we can probably drop the raw tests: qcow2 is the
>> most featureful format, so any breakage should be visible there. Most of
>> raw-posix.c should be part of the qcow2 tests already.
> 
> In another thread I think Anthony suggested check-block.  So if we
> just test qcow2 here to make sure the block layer works, then we can
> do a full run in check-block with raw, qed, and friends.

Yes, I think that would make a lot of sense. Do you want me to include a
check-block in v2 (that would be similar to qemu-iotests-quick.sh, just
without -g quick and for more formats) or should we introduce something
more sophisticated later?

Kevin



reply via email to

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