Date: | Sat, 26 Apr 2008 00:11:21 +0200 |
---|---|
From: | "Rubén Titos" <rtitos@xxxxxxxxxxx> |
Subject: | Re: [Gems-users] How to abort a transaction at commit time |
Jayaram Thanks for your quick response. I'm not simulating instruction fetches so far. I tried to follow the LL approach and immediately abort the transaction, yet this does not seem to avoid the code following the transaction to be executed (I still got a call to beginTransaction from the aborting processor). Perhaps I didn't do it correctly? I just copy-n-pasted the code from abortLazyTransaction into my notifyCommitNackReceived function, right under "enable_simics_proc" (disabled at the beginning of this "commit request" message exchange) and "beginEscapeAction" (to emulate what's done in trapToHandler so that restartTransaction does not complain). Even if I got it to work in this way, this kind of non-sw-handler abort would be simulated as taking zero cycles to restore the log, wouldn't it? Which is not exactly true... So, for now, I think I'm going to try simulating IFETCH's, and see if that way the trap kicks in before "anything else" is executed. Thanks again for your prompt response, Ruben On Fri, Apr 25, 2008 at 9:36 PM, Jayaram Bobba <bobba@xxxxxxxxxxx> wrote: Ruben, -- Rubén Titos Parallel Computing and Architecture Group Computer Engineering Dept. University of Murcia http://skywalker.inf.um.es/~rtitos/ |
[← Prev in Thread] | Current Thread | [Next in Thread→] |
---|---|---|
|
Previous by Date: | Re: [Gems-users] question about XACT_LOG_BUFFER_SIZE parameter, cge |
---|---|
Next by Date: | Re: [Gems-users] question about XACT_LOG_BUFFER_SIZE parameter, Jayaram Bobba |
Previous by Thread: | Re: [Gems-users] How to abort a transaction at commit time, Jayaram Bobba |
Next by Thread: | [Gems-users] How to increase the network load, Íõ×ô |
Indexes: | [Date] [Thread] |