RR(XS): JDK-8002116 This JdbReadTwiceTest.sh gets an exit 1

Dmitry Samersoff dmitry.samersoff at oracle.com
Wed Jan 22 03:59:07 PST 2014


On 2014-01-22 14:38, David Holmes wrote:
> On 22/01/2014 7:57 PM, Dmitry Samersoff wrote:
>> On 2014-01-22 10:39, David Holmes wrote:
>>> On 22/01/2014 12:00 AM, Dmitry Samersoff wrote:
>>>> Missed CR in subject. Sorry!
>>>>
>>>> http://cr.openjdk.java.net/~dsamersoff/JDK-8002116/webrev.01/
>>>
>>> You already create the file with mkFiles so no need to echo something
>>> into it, a simple cat will suffice to see if it is unreadable.
>>
>> Exit code of cat is not specified. So cat is enough to just warn user,
>> but we have to use grep to branch.
>>
>> Grep should return 2 in case of permission denied, but it's not stable
>> across implementations - grep can return 0 if patter is found and 1 in
>> all other cases.
>>
>> So I just choose the most reliable solution.
> 
> So grep is commonly better than cat but still not guaranteed - okay. But
> can't you simply grep the empty file to get the same result?

If I would grep the empty file I expect the exit code:

1 - pattern not found
2 - file not readable

It usually works, but some of grep implementations return either 0
(pattern found) or 1 (in all other cases)

-Dmitry

> 
> No big deal.
> 
> David
> 
>>> Not sure why this was being run as root though - can't be normal else
>>> we'd have seen this before. There are other tests that fail when run as
>>> root due to still being able to read/write unreadable/unwriteable files.
>>
>> I plan to file infrastructure bug to clarify it. Tests shouldn't run as
>> root in any case.
>>
>> -Dmitry
>>
>>>
>>> BTW the indention in that test case is messed up (in the original).
>>>
>>> David
>>>
>>>>
>>>> On 2014-01-21 17:49, Dmitry Samersoff wrote:
>>>>> Hi Everyone,
>>>>>
>>>>> Please review.
>>>>>
>>>>> http://cr.openjdk.java.net/~dsamersoff/JDK-8002116/webrev.01/
>>>>>
>>>>> I can't reproduce the issue locally, but suspect, the test failed
>>>>> because after chmod a-r the file is sill readable. So check it
>>>>> explicitly before run the test itself.
>>>>>
>>>>
>>>>
>>
>>


-- 
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.


More information about the serviceability-dev mailing list