[LTP] aio_read/8-1.c fails

[LTP] aio_read/8-1.c fails

[Kang Kai]

Hi All,

aio_read/8-1.c "fills in an aiocb with a NULL aio_buf" and then expect 
that the operation is not succesfully queued.
It fails that aio_read() always return 0. And I checked the posix spec 
and didn't find that an aiocb with NULL aio_buf will cause enqueue fails.
So is this case valid?

Thanks,
Kai

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] aio_read/8-1.c fails

[Kang Kai]

On 2012年03月08日 14:00, Kang Kai wrote:
> Hi All,
>
> aio_read/8-1.c "fills in an aiocb with a NULL aio_buf" and then expect
> that the operation is not succesfully queued.
> It fails that aio_read() always return 0. And I checked the posix spec
> and didn't find that an aiocb with NULL aio_buf will cause enqueue fails.
> So is this case valid?
So does the case aio_write/6-1.c.

>
> Thanks,
> Kai
>
> ------------------------------------------------------------------------------
> Virtualization&  Cloud Management Using Capacity Planning
> Cloud computing makes use of virtualization - but cloud computing
> also focuses on allowing computing to be delivered as a service.
> http://www.accelacomm.com/jaw/sfnl/114/51521223/
> _______________________________________________
> Ltp-list mailing list
> Ltp-list@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/ltp-list


------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] aio_read/8-1.c fails

[Cyril Hrubis]

Hi!
> aio_read/8-1.c "fills in an aiocb with a NULL aio_buf" and then expect 
> that the operation is not succesfully queued.
> It fails that aio_read() always return 0. And I checked the posix spec 
> and didn't find that an aiocb with NULL aio_buf will cause enqueue fails.
> So is this case valid?

The testcase passes here. However looking at the sources it tries to
read zero bytes to NULL buffer, which is wrong as the behavior in such
case is undefied by POSIX. I guess this is just another case where the
testcase was coded with "It works so it must be right" attitude.

And out of curiosity, which libc library makes the testcase fail?

-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] aio_read/8-1.c fails

[Cyril Hrubis]

Hi!
> >The testcase passes here. However looking at the sources it tries to
> >read zero bytes to NULL buffer, which is wrong as the behavior in such
> >case is undefied by POSIX. I guess this is just another case where the
> >testcase was coded with "It works so it must be right" attitude.
> >
> >And out of curiosity, which libc library makes the testcase fail?
> Yes, aio_read() return 0, and the case fails.
> 
> In posix spec Line 19887:
> RETURN VALUE
> The aio_read( ) function shall return the value zero if the I/O
> operation is successfully queued;
> otherwise, the function shall return the value ???1 and set errno to
> indicate the error.
> 
> So when the I/O operation enqueued, aio_read() return 0, then that
> the I/O operation will be processing.
> I still doubt the case is invalid.

Well, the POSIX explicitly says that if the pointer to the buffer
becames illegal prior the I/O completion the behavior is undefined. So
strictly speaking when you pass NULL buffer to aio_read() the behavior
is undefined to begin with. At least that is how I understand the
standard.

-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list