On Tue, 11 Apr 2000, Arun Sharma wrote:
> In muc.lists.freebsd.hackers, you wrote:
> >
> > Is there a MT-safe implementation of gethostbyname() in FreeBSD (3.4/4.0)?
> >
> > On Solaris there is gethostbyname_r(). Calling gethostbyname() with in
> > two threads cause both threads to block.
>
> You seem to be talking about two different things:
>
> 1. A reentrant function - two different threads can be in the same function
> simultaneously
>
> 2. A non blocking function.
>
Hi, thanks for the reply. I'm talking about (1). Reentrant functions
are MT-safe, but a non blocking function may not be.
> Reading the code in /usr/src/lib/libc{,_r}, it appears to me that, the
> FreeBSD implementation is not reentrant, and blocking. However,
> gethostbyname calls poll(2), which causes a reschedule to another runnable
> pthread, if any.
>
Sure, and there is no "reimplementation" for gethostbyname() in libc_r as
far as I could tell.
> In your case, both the threads are waiting for a DNS server response,
> so the thread scheduler doesn't have a thread to schedule.
If I only create one thread, then the gethostbyname() returns immediately.
By using truss -p, it seems that gethostbyname() uses a static socket var
for the DNS I/O rather than a local one, and that I think is what caused
the threads to hang. Or, we're talking about the same thing?
Thanks again for the input.
Ming
>
> I think it'd be very useful to have a non-blocking DNS lookup API (one
> which exposes the underlying file descriptor) . Winsock has this. UNIX
> netscape 4.x would freeze half as often if this was done right.
>
> -Arun
>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
