I can't work on the code right now since I'm at work, but by inspection it looks like I forgot to serialize bmiAddr and bmiSize. It would be a good idea for us to have a regression test for this sort of thing.
Gabe Quoting Rick Strong <[EMAIL PROTECTED]>: > Hi, > > After resuming a checkpoint, I keep getting the same panic for any > bechmark I try: > panic: IDE controller access to invalid address: 0x801fc008016 > [dispatchAccess:/home/rstrong/build/m5powerfs/build/ALPHA_FS/dev/ide_ctrl.cc, > line 467] > > I know that Gabe has been cleaning this up lately, so any ideas? I get > the same error for a simple > getpid benchmark and subset of parsec that worked fine a week ago. > > Best, > -Rick > > Detailed error reports per application. Not very informative, but just > corroborating the same > invalid address happens in several applications. > > Simple getpid microbenchmark that calls getpid 100 times: > panic: IDE controller access to invalid address: 0x801fc008016 > > > run:parsec_bodytrack:1:10G:dual:ss:mesh:vanilla > > 836 panic: IDE controller access to invalid address: 0x801fc008016 > 837 @ cycle 3004894384513 > 838 > [dispatchAccess:/home/rstrong/build/m5powerfs/build/ALPHA_FS/dev/ide_ctrl.cc, > line 467] > > run:parsec_ferret:1:10G:dual:ss:mesh:vanilla > 524 panic: IDE controller access to invalid address: 0x801fc008016 > @ cycle 2968677743013 > > [dispatchAccess:/home/rstrong/build/m5powerfs/build/ALPHA_FS/dev/ide_ctrl.cc, > line 467] > > run:parsec_freqmine:1:10G:dual:ss:mesh:vanilla > panic: IDE controller access to invalid address: 0x801fc008016 > @ cycle 2976949207513 > > [dispatchAccess:/home/rstrong/build/m5powerfs/build/ALPHA_FS/dev/ide_ctrl.cc, > line 467] > _______________________________________________ > m5-users mailing list > m5-users@m5sim.org > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > _______________________________________________ m5-users mailing list m5-users@m5sim.org http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
