[VANILLA-L:523] Re: The server and parallel programming bugs

To: Multiple recipients of list <vanilla-l@anthrax.ecst.csuchico.edu>
Subject: [VANILLA-L:523] Re: The server and parallel programming bugs
From: "NBT Nick B. Trown" <nbt@landacorp.com>
Date: Fri, 18 Apr 1997 19:55:09 -0700 (PDT)
Reply-To: vanilla-l@anthrax.ecst.csuchico.edu
Resent-Date: Fri, 2 May 1997 20:58:11 -0500
Resent-From: tanner@real-time.com
Resent-Message-Id: <199705030158.UAA15573@wizard.real-time.com>
Resent-To: netrek@real-time.com
Sender: vanilla-l@anthrax.ecst.csuchico.edu



>----------
>From: 	cameron@ripper.stl.dec.com[SMTP:cameron@ripper.stl.dec.com]
>Sent: 	Friday, April 18, 1997 4:50 PM
>To: 	Multiple recipients of list
>Subject: 	[VANILLA-L:520] Re: The server and parallel programming bugs
>
>> In particular, the server is a parallel program but is not correct.
>
>Yes, it breaks all the rules for synchronising access to critical data
>structures.  It needs mutexes or semaphores.

Yeah, I've been considering suggesting adding a semaphore or 
two to control the shared memory segment. It will be slower than
the current approach but then the server is suffering on some 
machines. Maybe 1 semaphore for controlling the message 
structure in the shared memory since it is hit so much. 1 more 
semaphore should be used to control the rest of the shared 
memory structure.. If we're going to mess with the shared memory
segment we should do it properly so we don't have to worry this
anymore in the future.

>Nick

Prev by Date: [VANILLA-L:526] Re: Banning whole domains?
Next by Date: [VANILLA-L:522] Re: INL Server Code development
Prev by thread: [VANILLA-L:522] Re: INL Server Code development
Next by thread: [VANILLA-L:526] Re: Banning whole domains?
Index(es):
- Main
- Thread