Messages in this thread | | | From | Radu Greab <> | Date | Sat, 31 Mar 2001 19:26:00 +0300 (EEST) | Subject | Re: bug report: select on unconnected sockets |
| |
On Sun, 1 Apr 2001 02:15 +1000, john slee wrote: > On Sat, Mar 31, 2001 at 06:44:20PM +0300, Radu Greab wrote: > > Sorry if this is already known: on a RH 7.0 system with kernel 2.4.2 > > or 2.4.3, a select on an unconnected socket incorrectly says that the > > socket is ready for input and output. Of course, reading from the socket > > file descriptor returns -1 and errno is set to ENOTCONN as shown in > > the strace output: > > > > socket(PF_INET, SOCK_STREAM, IPPROTO_IP) = 3 > > select(4, [3], [3], [3], {0, 0}) = 2 (in [3], out [3], left {0, 0}) > > read(3, 0xbffff668, 1024) = -1 ENOTCONN (Transport endpoint is not connected) > > > > I attached a small example program to reproduce the bug. > > bleah. which one is supposed to be right?
I think that the Solaris one is right.
> > linux 2.4 > --------- > $ uname -a > Linux XXXXX 2.4.2-ac20 #8 Wed Mar 14 01:53:05 EST 2001 i686 unknown > $ ./t > select result=2 > read: Transport endpoint is not connected
Select says that the socket is ready for both input and output. A read results in ENOTCONN, a write results in EPIPE.
> linux 2.2 > --------- > $ uname -a > Linux XXXXX 2.2.18 #1 Thu Dec 21 21:13:10 EST 2000 i586 unknown > $ ./t > select result=1
Select says that the socket is ready for output, but a write results in EPIPE.
> > solaris > ------- > $ uname -a > SunOS XXXXX 5.7 Generic_106541-07 sun4m sparc sun4m > $ ./t > select result=0
And Digital Unix is right too:
$ uname -a OSF1 XXX V4.0 1229 alpha $ ./a.out select result=0
Thanks, Radu Greab - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |