Re: real POSIX.1b semaphores

David S. Miller (
Tue, 19 Nov 1996 07:39:21 -0500

Date: Tue, 19 Nov 1996 12:32:57 +0000
From: ". Tethys "SYSTEM ADMIN" X" <>

Except surely this breaks any future move to a longer pid_t. You
could argue that if we have a 64 bit pid_t, then the lower 32 bits
should be more than enough for the PID, with the upper 32 bits for
thread_id and any other information. But then again, that's
probably what they thought about 16 bit PIDs all those years ago...

4 billion processes, times sizeof(struct task_struct), that alone is
around 2 terabytes. In reality 4 billion processes would take up at a
minimum around 6 or 7 terabytes. I give it 5 or 6 years for this to
matter at all. ;-)

Even with a million users on a single machine (even a NUMA one!) each
of them would need to spawn 4,000 processes to reach this level of

Yow! 10.49 MB/s remote host TCP bandwidth ////
over 100Mb/s ethernet. Beat that! ////
-----------------------------------------////__________ o
David S. Miller, /_____________/ / // /_/ ><