Ok, pulled the latest M5, and ported over. Unfortunately TM doesn't work again, looks like some the memory model has changed since I last got it working.
But in terms of checkpoints, they still don't work for me. I have no idea why now. They either fail to load (if I dump a checkpoint running in timing mode), or load (atomic cpu dump), but immediately have some form of error when simulating. Could it be all the additions I've made? Extra registers and such for TM? I don't see why that would cause problems with serialization, since the point I checkpoint at it doesn't matter whether the new stuff is saved or not. Geoff -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of nathan binkert Sent: Tuesday, March 25, 2008 11:02 PM To: M5 users mailing list Subject: Re: [m5-users] Full System Checkpoints We'll be happy to have you in the present though. We'd also love for you to keep current and send us diffs! On Tue, Mar 25, 2008 at 7:44 PM, Geoffrey Blake <[EMAIL PROTECTED]> wrote: > I'll try re-cloning the tree and re-applying the TM patches I have to see if > it is just my older version of M5 in a few hours. Not looking forward to > the merge I'll have to do.... > > Geoff > > > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On > > > Behalf Of nathan binkert > Sent: Tuesday, March 25, 2008 10:40 PM > To: M5 users mailing list > Subject: Re: [m5-users] Full System Checkpoints > > I haven't looked at the checkpoint code in the committed M5 scripts in > a while, but there's a chance that things are wrong there. There is > now a requirement to drain the cpus/caches before you checkpoint for > the timing mode stuff. As for the atomic mode, that's odd. Can you > give me enough detail to reproduce this myself? I'd most prefer > something against the head of the tree. > > There might have been problems with the tree. I personally don't recall. > > > Nate > > 2008/3/25 Geoffrey Blake <[EMAIL PROTECTED]>: > > > > > > > > > > Hmmmm, something is broken, I've been trying and nothing seems to work. I > > can't dump a timing_cpu checkpoint, because reloading it causes M5 to > crash > > on initialization, certain serialization elements are missing. When I > dump > > an atomic cpu checkpoint of the same benchmark, I can load it, but the > > kernel immediately panics and crashes. Any ideas? Was checkpointing > broken > > prior to 2.0b5? > > > > > > > > Thanks, > > > > Geoff > > > > > > > > > > > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On > > Behalf Of Geoffrey Blake > > Sent: Tuesday, March 25, 2008 8:31 PM > > > > To: 'M5 users mailing list' > > Subject: RE: [m5-users] Full System Checkpoints > > > > > > > > > > > > I'm pretty sure its loading the correct checkpoints. What is really odd > is > > if I load the atomic CPU checkpoint with Atomic CPU, it still crashed with > a > > page handler error. I'll keep fiddling, perhaps the serialization routines > > are broken on my build. I'm using a version of M5 from the repository that > > is older than 2.0b5, I should probably pull again to get the new changes. > > > > > > > > Geoff > > > > > > > > > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On > > Behalf Of Lisa Hsu > > Sent: Tuesday, March 25, 2008 8:19 PM > > To: M5 users mailing list > > Subject: Re: [m5-users] Full System Checkpoints > > > > > > > > i had this happen to me when i was dumping atomic && timing checkpoints > into > > the same directory, and specifying checkpoint 1 - but it would read the > > wrong one and give me that exact error because it was reading the first > > checkpoint (an atomic checkpoint) rather than the first timing checkpoint. > > if you're doing the same, you'll want to separate them. > > > > lisa > > > > > > 2008/3/25 Geoffrey Blake <[EMAIL PROTECTED]>: > > > > > > > > I'm trying to dump out a checkpoint of multi-core system using Full System > > to allow me to run a profile of the binaries I'm running in M5. I dump the > > checkpoint using the m5 pseudo instruction from inside the application. > The > > checkpoint though seems malformed. If I dump it out running with timing > > simple cpu, the checkpoint will not load, the simulator crashes with a > fatal > > error saying it cannot read _status for system.cpuXX. I've tried making > > checkpoints using the atomic CPU, and they seem load fine, but immediately > > after starting to run once the checkpoint is loaded, the binary crashes > with > > a kernel page handling error, perhaps due to the fact it does not save any > > ITB entries. Is there something I am forgetting? > > > > > > > > Geoff > > > > > > > > > > > > No virus found in this outgoing message. > > Checked by AVG. > > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: > 3/25/2008 > > 10:26 AM > > > > > > _______________________________________________ > > m5-users mailing list > > [email protected] > > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > > > > > > No virus found in this incoming message. > > Checked by AVG. > > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: > 3/25/2008 > > 10:26 AM > > > > > > > > No virus found in this outgoing message. > > Checked by AVG. > > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: > 3/25/2008 > > 10:26 AM > > > > > > > > No virus found in this incoming message. > > Checked by AVG. > > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: > 3/25/2008 > > 10:26 AM > > > > > > > > > > No virus found in this outgoing message. > > Checked by AVG. > > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: > 3/25/2008 > > 10:26 AM > > > > _______________________________________________ > > m5-users mailing list > > [email protected] > > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > > No virus found in this incoming message. > Checked by AVG. > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: 3/25/2008 > 10:26 AM > > > No virus found in this outgoing message. > Checked by AVG. > Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: 3/25/2008 > 10:26 AM > > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users No virus found in this incoming message. Checked by AVG. Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: 3/25/2008 10:26 AM No virus found in this outgoing message. Checked by AVG. Version: 7.5.519 / Virus Database: 269.22.0/1342 - Release Date: 3/25/2008 10:26 AM _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
