January 2000
- JavaWorld: Build an object database J C Lawrence
- Muq update Cynbe ru Taren
- New link support J C Lawrence
- OS Inspiration Phillip Lenhardt
- OS Inspiration Greg Miller
- Chomsky's recursive theory of grammar J C Lawrence
- concerning tokenization, compilation, performance, and other fun stuff. Nate Cain
- concerning tokenization, compilation, performance, and other fun stuff. cg@ami-cg.GraySage.Edmonton.AB.CA
- concerning tokenization, compilation, performance, and other fun stuff. Joel Dillon
- concerning tokenization, compilation, performance, and other fun stuff. Ola Fosheim Grøstad
- Library submission notification and updates J C Lawrence
- EQ packet sniffer J C Lawrence
- Catalog of Compiler Construction Tools J C Lawrence
- Microthreads for Python J C Lawrence
- For those interested in parsers and compilers J C Lawrence
- EQ packet analyzer is gone? Sellers, Michael
- EQ packet analyzer is gone? J C Lawrence
- EQ packet analyzer is gone? J C Lawrence
- RFC: Worldforge project Scott Clitheroe
- ScryMUD 2.0.9 released (compiles on Windows (cygwin)) Ben Greear
- Clay Shirky's "Playfulness in 3-D Spaces" J C Lawrence
- Clay Shirky's "Playfulness in 3-D Spaces" Raph & Kristen Koster
- Clay Shirky's "Playfulness in 3-D Spaces" Sellers, Michael
- Clay Shirky's "Playfulness in 3-D Spaces" Nick Shaffner
- Clay Shirky's "Playfulness in 3-D Spaces" Sellers, Michael
- Clay Shirky's "Playfulness in 3-D Spaces" Jeremy Music "Sterling"
- Clay Shirky's "Playfulness in 3-D Spaces" msew
- ShowEQ Ashran
- Hello! F. Randall Farmer
- Hello! Cynbe ru Taren
- Y2K archives J C Lawrence
- An introduction... Geoffrey A. MacDougall
- An introduction... Lovecraft
- Laws website moves Raph & Kristen Koster
- Ok, got some brand new core dumps. J C Lawrence
- player politics (was An introduction...) Sellers, Michael
- player politics (was An introduction...) Ola Fosheim Grøstad
- player politics (was An introduction...) Sellers, Michael
- player politics (was An introduction...) Ola Fosheim Grøstad
- player politics (was An introduction...) J C Lawrence
- player politics (was An introduction...) Sellers, Michael
- A quick business question... Geoffrey A. MacDougall
- A quick business question... Matthew Mihaly
- A quick business question... Darrin Hyrup
- Question about multithreaded servers Fabian Lemke
- Question about multithreaded servers AR Schleicher
- Question about multithreaded servers Nick Shaffner
- Question about multithreaded servers Ola Fosheim Grøstad
- Question about multithreaded servers Fabian
- Question about multithreaded servers Jon A. Lambert
- Question about multithreaded servers J C Lawrence
- Question about multithreaded servers Greg Underwood
It's been a while since I posted, so "howdy" to everyone. I've been
lurking since I got back from Europe, but this thread (pardon the pun)
caught my eye.
At 08:34 AM 1/18/00 -0600, "Fabian" <lemkef@execpc.com> wrote:
>
>> -----Original Message-----
>> From: Jon A. Lambert [mailto:jlsysinc@ix.netcom.com]
>>
>> Fabian wrote:
>> >
>> > I suppose just by making a simple requirement that
>> > all objects must be secured at the begining of the
>> > script would take care of that.
No, that's not true. Just because you attempt to lock everything before
you execute the script doesn't mean you avoid deadlocks. Deadlocks occur
when the following happens:
Thread 1 - Get Lock 1 - ok
Thread 2 - Get Lock 2 - ok
Thread 1 - Get Lock 2 - block
Thread 2 - Get Lock 1 - block
Both threads block on their second calls to get the locks that the other
thread has. The reason this occurs is that they both grabbed the locks in
a different order. Thread 1 attempted to get Lock 1 and then 2, while
Thread 2 attempted to get Lock 2 and then 1. No matter when this occurs in
your script execution process, you will get a deadlock. Simply moving the
attempts to get the locks to before the script executes does nothing to
prevent that order of events. The OS can still preempt any thread at any
time, and grabbing the locks in a different order will open your code to
deadlocks.
The only way to gaurentee you will not get a deadlock is to ALWAYS have
EVERY thread attempt to obtain the same locks in the same order. In that
case, this is what happens:
Thread 1 - Get Lock 1 - ok
Thread 2 - Get Lock 1 - block
Thread 1 - Get Lock 2 - ok
Thread 1 - execute required stuff
Thread 1 - Release Lock 2, release Lock 1
Thread 2 - Unblocks, having gotten Lock 1
Thread 2 - Get Lock 2 - ok
Thread 2 - ...
Thread 2 will never deadlock once it comes out of getting Lock 1 because
any other thread that requires both Lock 1 and Lock 2 will be blocked on
the 'Get Lock 1' call, w/o having Lock 2. When the OS decides to preempt
has no effect on which locks you can get.
BTW, I had it realease the Locks in reverse order, simply to avoid minor
thrashing. Assume you release Lock 1 and then 2, but when you release 1,
the OS wakes another thread that is waiting on 1. This thread then
attempts to get Lock 2, and blocks again. The OS then has to switch
control back to the original thread to realease Lock 2, and then switch
control back to the second thread again to get Lock 2. Releasing in
reverse order will prevent those minor, but time consuming context switches.
>>
>> Not a simple requirement though. How about this sort of script construct:
>>
>> for player in connected
>> player.message("How about some spam!?!")
>> endfor
>>
>> Do you lock every player object in the connected list?
>> Would the above method be an atomic transaction requiring
>> completion of all events it would generate?
>> Or would you spawn a new event to each player object,
>> and not worry about completion thus making the method
>> above non-atomic?
Good question, but more one of starvation than deadlock.
>> And what about guaranteeing event sequence?
>>
>> for player in connected
>> player.message("LINE 1")
>> player.message("LINE 2")
>> player.message("LINE 3")
>> endfor
>>
>> Perhaps when sending "LINE 1" the lock cannot be achieved and
>> is rescheduled or terminated. So the Player receives:
>>
>> "LINE 2"
>> "LINE 3"
>> "LINE 1"
>>
>> Just a few things to think about. :)
This is what grabbing all the locks prior to script execution will get you
(as opposed to a gaurentee of no deadlocks). Grabbing them all in advance
will insure that you will get the messages in the right order. However, as
Jon was attempting to point out above, when your script requires that you
grab a zillion locks, you can run into some serious starvation issues.
-Greg - Question about multithreaded servers cg@ami-cg.GraySage.Edmonton.AB.CA
- Question about multithreaded servers J C Lawrence
- Question about multithreaded servers Greg Underwood
- Question about multithreaded servers J C Lawrence
- Question about multithreaded servers Emil Eifrém
- Question about multithreaded servers J C Lawrence
- Community Relations Dundee
- Community Relations Christopher Allen
- Community Relations Matthew Mihaly
- Community Relations Greg Miller
- Community Relations Darrin Hyrup
- Community Relations Dundee
- Community Relations Ola Fosheim Grøstad
- Community Relations Dundee
- Community Relations Ola Fosheim Grøstad
- Community Relations Dundee
- Community Relations Marian Griffith
- Community Relations Koster, Raph
- Community Relations Jon A. Lambert
- Community Relations Matthew Mihaly
- Community Relations Ola Fosheim Grøstad
- Community Relations Matthew Mihaly
- Community Relations Ola Fosheim Grøstad
- Community Relations Matthew Mihaly
- Community Relations Marc Bowden
- Community Relations Eli Stevens {Grey}
- Community Relations Matthew Mihaly
- Community Relations Raph & Kristen Koster
- Community Relations Matthew Mihaly
- Community Relations Koster, Raph
- Community Relations Travis Casey
- Community Relations Ola Fosheim Grøstad
- Community Relations Travis Casey
- Community Relations Ola Fosheim Grøstad
- Community Relations Matthew Mihaly
- Community Relations Rahul Sinha
- Community Relations Matthew Mihaly
- Community Relations Matthew Mihaly
- Community Relations Ola Fosheim Grøstad
- Community Relations Koster, Raph
- Community Relations Matthew Mihaly
- Community Relations Geoffrey A. MacDougall
- Community Relations Jon A. Lambert
- Community Relations Matthew Mihaly
- Community Relations Ola Fosheim Grøstad
- Community Relations Travis S. Casey
- Community Relations Douglas Couch
- Community Relations Dundee
- Community Relations Marian Griffith
- Community Relations Jon A. Lambert
- Community Relations Matthew Mihaly
- Community Relations Jon A. Lambert
- Community Relations Matthew Mihaly
- Community Relations Geoffrey A. MacDougall
- Community Relations Matthew Mihaly
- Community Relations Lovecraft
- Community Relations Geoffrey A. MacDougall
- Community Relations Lovecraft
- Community Relations Marian Griffith
- Community Relations Matthew Mihaly
- Community Relations Sellers, Michael
- Community Relations Lovecraft
- Community Relations J C Lawrence
- Community Relations Lovecraft
- Community Relations J C Lawrence
- Community Relations Koster, Raph
- Community Relations Jon A. Lambert
- Community Relations Ola Fosheim Grøstad
- Community Relations Lovecraft
- Community Relations Jon A. Lambert
- Community Relations Lovecraft
- Community Relations Jon A. Lambert
- Community Relations Jon A. Lambert
- Community Relations J C Lawrence
- Community Relations Wes Connell
- Community Relations Marc Bowden
- Community Relations Dundee
- Community Relations J C Lawrence
- Community Relations J C Lawrence
- Community Relations F. Randall Farmer
- Community Relations Dr. Cat
- Community Relations J C Lawrence
- Community Relations Koster, Raph
- Community Relations Geoffrey A. MacDougall
- Community Relations Matthew Mihaly
- Commercial-use Restrictions on Code Bases (was: help me find 100% fre (fwd) J C Lawrence
- Commercial-use Restrictions on Code Bases (was: help me find 100% fre (fwd) J C Lawrence
- Databases (was Commercial-use Restrictions on Code Bases) Charles Hughes
- Databases (was Commercial-use Restrictions on Code Bases) cg@ami-cg.GraySage.Edmonton.AB.CA
- Databases (was Commercial-use Restrictions on Code Bases) Jon A. Lambert
- Databases (was Commercial-use Restrictions on Code Bases) Jon A. Lambert
- Databases (was Commercial-use Restrictions on Code Bases) J C Lawrence
- Mud-Dev FAQ part I Marian Griffith
- Mud-Dev FAQ part II Marian Griffith
- META: List goals (was OS Inspiration) J C Lawrence
- Valhalla license? CFrancy@aol.com
- Valhalla license? Matthew Mihaly
- Valhalla license? Richard Woolcock
- Planet/Solar System Generation Wes Connell
- Planet/Solar System Generation icecube@ihug.co.nz
- Planet/Solar System Generation Christopher Allen
- Planet/Solar System Generation J C Lawrence
- Planet/Solar System Generation Nolan Darilek
- Planet/Solar System Generation Wes Connell
- Planet/Solar System Generation Richard Woolcock
- How to handle/display partial language skill Joe Kingry
- How to handle/display partial language skill Greg Underwood
- How to handle/display partial language skill Wes Connell
- How to handle/display partial language skill Matt Chatterley
- Signing off... IronWolf
- Simulated Populations phlUID
- Simulated Populations Charles Hughes
- Simulated Populations Vladimir Prelovac
- Simulated Populations phlUID
- Simulated Populations Dundee
- Simulated Populations Wes Connell
- Simulated Populations Nicolai Hansen
- Simulated Populations J C Lawrence
- Simulated Populations J C Lawrence
- Business Licenses CFrancy@aol.com
- Business Licenses Bruce
- Business Licenses J C Lawrence
- [adv-mud] What good is a hero when nobody knows? (fwd) J C Lawrence
- [adv-mud] What good is a hero when nobody knows? (fwd) Matthew Mihaly
- [adv-mud] What good is a hero when nobody knows? (fwd) Wes Connell
- [adv-mud] What good is a hero when nobody knows? (fwd) Matthew Mihaly
- [adv-mud] What good is a hero when nobody knows? (fwd) Erik Jarvi
- Multiply oriented interactive worlds... Justin Rogers
- [adv-mud] What good is a hero when nobody knows? J C Lawrence
- [adv-mud] Topic list repost (fwd) J C Lawrence
- [adv-mud] Spellbound Hierarchy and Keys (fwd) J C Lawrence
- [adv-mud] Spellbound Hierarchy and Keys (fwd) J C Lawrence
- [adv-mud] Better Grammer Detection J C Lawrence
- [adv-mud] MUD-Dev vs. adv-mud J C Lawrence
- MUD-Dev vs. adv-mud phlUID
- [adv-mud] What good is a hero when nobody knows? phlUID
- (fwd) Avatarism and Role-Playing Game Design claw@kanga.nu
- Starlane test Joel Kelso
- hoho Matthew Mihaly
- Current Status of Middle Earth Online Geoffrey A. MacDougall
- ScryMUD 2.0.10 released. Ben Greear
- [adv-mud] What good is a hero when nobody knows? J C Lawrence
- [adv-mud] What good is a hero when nobody knows? Ben Greear
- [adv-mud] What good is a hero when nobody knows? J C Lawrence
- [adv-mud] What good is a hero when nobody knows? Wes Connell
- [adv-mud] What good is a hero when nobody knows? J C Lawrence
- Some new Library references J C Lawrence