cvs commit: src/lib/libc/rpc svc_vc.c

[Alfred Perlstein]

* Martin Blapp <mb at imp.ch> [030529 16:53] wrote:
> 
> Hi,
> 
> >"for some strange reason"
> 
> Yeah, the comment is not quite correct. I see now EAGAIN errors,
> but the first read on the socket has always 0 lenght for AF_LOCAL:
> 
> recvmsg() = 0
> recvmsg() = 4
> recvmsg() = 92
> recvmsg() = -1, errno = 35
> recvmsg() = 0
> 
> in opposite to
> 
> recvmsg() = 4
> recvmsg() = 40
> recvmsg() = -1, errno = 35
> recvmsg() = 0

If it's a kernel/socket bug we shouldn't really be working around
it in userland.  Is it possible that the read is happening before
the socket is connected or has just been disconnected?  A more detailed
storyboard might get the problem fixed without what appears to be
a hack/workaround in userspace.

-- 
-Alfred Perlstein [alfred at freebsd.org]
'Instead of asking why a piece of software is using "1970s technology,"
 start asking why software is ignoring 30 years of accumulated wisdom.'