The shlwapi:istream tests and Microsoft Security Essentials

[Reece Dunn]

On 16 March 2010 16:36, Paul Vriens <paul.vriens.wine at gmail.com> wrote:
> On 03/16/2010 03:56 PM, Paul Vriens wrote:
>>
>> On 03/16/2010 03:20 PM, Reece Dunn wrote:
>>>
>>> On 16 March 2010 13:27, Paul Vriens<paul.vriens.wine at gmail.com> wrote:
>>>>
>>>> Hi,
>>>>
>>>> I was wondering why I (out of all those boxes) had several failures for
>>>> these tests on some of my XP and Vista boxes.
>>>
>>> I was wondering the same.
>>>
>>>> Turns out that you get these failures when MSE is active. I just ran the
>>>> tests on a clean XP-SP3 box (up-to-date) and no failures. Installing MSE
>>>> (and downloading the latest signatures) is enough to show a multitude of
>>>> test failures. Turning 'real time monitoring' off or excluding the test
>>>> executable makes the tests succeed again.
>>>>
>>>> Most of the failures are of the E_ACCESSDENIED nature btw.
>>>
>>> This is interesting. I wonder if it is because MSE is locking the file
>>> so it can scan it for viruses, so when the next test gets run, the
>>> istream test cannot access that file and so gets an E_ACCESSDENIED
>>> back.
>>
>> I never had this issue with either "Microsoft Defender" or the Avast!
>> Home Edition virus scanner.
>>
>>>
>>>> Does this make sense?
>>>
>>> Yes.
>>>
>>>> If the test failures are valid, should I change the tests to cope
>>>> with this?
>>>
>>> The test failures look valid in this case (MSE may be installed on a
>>> test system), so a change similar to
>>>
>>> http://source.winehq.org/git/wine.git/?a=commitdiff;h=1571337c0366b27e819851d7cdec4a7bc1ce7690
>>>
>>> should be done.
>>
>> The thing is that depending on the mode/flag the failures start at a
>> different place. I'll see if I can create something generic or whether
>> the flags indicate something (don't think so as several runs produces
>> different output).
>>
>
> The attached patch fixes the issue on my boxes. Looks like it's just a
> timing issue (MSE being sllloooowww I guess).

Looks good to me.

> I should maybe raise the timeout to something higher, it's only affecting a
> handful of boxes anyway?

If it works on your machine over 2-3 runs, it should be ok. We can
always increase the Sleep if needed. Raising it too high means that
you are more likely to get the wine test program failing the test
because it is taking too long.

- Reece