[Sheepdog] [PATCH] fix a bug in rx() when read() returns 0 but connection is not marked as closed.

[Huxinwei]

> -----Original Message-----
> From: Liu Yuan [mailto:namei.unix at gmail.com]
> Sent: Wednesday, April 25, 2012 10:15 AM
> To: Huxinwei
> Cc: MORITA Kazutaka; sheepdog at lists.wpkg.org
> Subject: Re: [Sheepdog] [PATCH] fix a bug in rx() when read() returns 0 but
> connection is not marked as closed.
> 
> On 04/25/2012 09:05 AM, Huxinwei wrote:
> 
> > What if reading is interrupted by a signal ?
> > Read() returns 0 while errno is EINTR then ?
> 
> 
> is there any real use case to interrupt read()? If it is killed by
> manual signal, I think it would be too artificial

Er, I was wrong. 
The Linux kernel guarantees that if read is interrupted before reading any data, the return value is -1.
So 0 always means EOF unless the count is 0 too.

It is sufficient to check returning value to determinate the status of the connection. ;)

Thanks.
> Thanks,
> Yuan