Re: Tread/Task IDs

[Marcus Brinkmann]

On Thu, May 08, 2003 at 01:42:31PM +0200, Joshua LeVasseur wrote:
> - the semantics of the TID space has a direct effect on the speed at 
> which the kernel can map a TID to its kernel thread-control-block.

This only affects the thread number though, which is 17 bit on ia32, giving
the advertised 131,072 KTCBs.  It was very interesting to hear the reason
behind the 17 bit value, though.  Thanks for the explanation!

The discussion was in context of the need for a larger version component if
you want to use it to grant (reliably) unicity in time.  However, giving that
the thread id data types are carefully designed for maximum performance, I see
the reason behind the limitation.

That's why I suggested to use the version ID to hold more information (the
PID) about the thread.  The PID also has the unique over time property
(although even less reliably), and eliminates IPC to the task server to find
the PID for a thread (to identify the caller).

Thanks,
Marcus

-- 
`Rhubarb is no Egyptian god.' GNU      http://www.gnu.org    address@hidden
Marcus Brinkmann              The Hurd http://www.gnu.org/software/hurd/
address@hidden
http://www.marcus-brinkmann.de/