> On Sept. 25, 2015, 3:55 p.m., Steve Reinhardt wrote: > > THe fix looks fine to me (though I just skimmed it, obviously). > > > > From a longer-term perspective: we've discussed reference-counting Packet > > and Request objects before, haven't we? I don't recall where that > > discussion ended up though. Did it turn out to cause a performance issue?
Ref counting on packets and requests using shared_ptr costs us around 20% in performance. Not an option in other words. - Andreas ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.gem5.org/r/3140/#review7283 ----------------------------------------------------------- On Sept. 25, 2015, 2:29 p.m., Andreas Hansson wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.gem5.org/r/3140/ > ----------------------------------------------------------- > > (Updated Sept. 25, 2015, 2:29 p.m.) > > > Review request for Default. > > > Repository: gem5 > > > Description > ------- > > Changeset 11139:33043178935c > --------------------------- > mem: Add PacketInfo to be used for packet probe points > > This patch fixes a use-after-delete issue in the packet probe points > by adding a PacketInfo struct to retain the key fields before passing > the packet onwards. We want to probe the packet after it is > successfully sent, but by that time the fields may be modified, and > the packet may even be deleted. > > Amazingly enough the issue has gone undetected for months, and only > recently popped up in our regressions. > > > Diffs > ----- > > src/mem/comm_monitor.cc a611a23c8cc2 > src/mem/probes/base.hh a611a23c8cc2 > src/mem/probes/mem_trace.hh a611a23c8cc2 > src/mem/probes/mem_trace.cc a611a23c8cc2 > src/mem/probes/stack_dist.hh a611a23c8cc2 > src/mem/probes/stack_dist.cc a611a23c8cc2 > src/sim/probe/mem.hh a611a23c8cc2 > > Diff: http://reviews.gem5.org/r/3140/diff/ > > > Testing > ------- > > Affected regressions now pass > > > Thanks, > > Andreas Hansson > > _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
