Very Computer

by **Cove Schneid** » Wed, 06 Dec 1995 04:00:00

I was just looking through the source to a ping program when I came
across i = getpid() & 0xffff;. Why would you & pid_t with 0xffff?
All I could think of was it's a strange way to do a cast (Pre-ANSI,
or on some old machine). I'm sure this is historical more than
anything..

Could some one educate me?

Thanks,
any info appreciated.

Cove

--

by **Rene Pijlma** » Wed, 06 Dec 1995 04:00:00

>I was just looking through the source to a ping program when I came
>across i = getpid() & 0xffff;. Why would you & pid_t with 0xffff?
>All I could think of was it's a strange way to do a cast (Pre-ANSI,
>or on some old machine). I'm sure this is historical more than
>anything..

>Could some one educate me?

Cove,

Ping needs to send an ICMP echo request packet with a 16-bit IDENTIFIER
field which uniquely identifies the sending process. The only unique number
readily available is the pid, but unfortunately it's a 32-bit number.
Ping ignores the most signifficant 16 bits, which would be incorrect
if UNIX really used 32-bit pid's. I think in practice typical UNIX'es
cycle the pid on 30,000 (or some other magic number in the 16-bit range)
and the & 0xffff construction is meaningless but also harmless.

HTH,
--
Rene Pijlman

by **Robert Kl** » Thu, 07 Dec 1995 04:00:00

>I was just looking through the source to a ping program when I came
>across i = getpid() & 0xffff;. Why would you & pid_t with 0xffff?

To make sure 'i' isn't higher than 65535.

robert

Very Computer

i = getpid() & 0xffff?

i = getpid() & 0xffff?

i = getpid() & 0xffff?

i = getpid() & 0xffff?