Hi Ashkan,

A few thoughts. First, I’d suggest to read the PhD thesis "A Recon gurable 
Mixed-Time-Criticality SDRAM Controller” by Sven Goossens. There is plenty work 
in timing isolation in DRAM controllers already out there. See for example the 
work by Akesson et al. Second, DRAMSim2 is most likely not a good choice. It’s 
slowing down your simulation an order of magnitude compared to the native DRAM 
controller in gem5. Third, you have the masterId in the request (req field in 
the packet), but things are complicated by the caches. Who is responsible for a 
cache eviction is the line was added by core1, updated by core2, and the 
eviction is due to core3?

Andreas

From: gem5-users 
<[email protected]<mailto:[email protected]>> on behalf of 
Ashkan Asgharzade <[email protected]<mailto:[email protected]>>
Reply-To: Ashkan Asgharzade 
<[email protected]<mailto:[email protected]>>, gem5 users mailing 
list <[email protected]<mailto:[email protected]>>
Date: Sunday, 14 February 2016 at 14:40
To: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Subject: [gem5-users] Adding a Core-ID(number) to the packet reaching to Mem. 
Ctrl.

In The Name of God


Hi Dear Users,

I am new to work with gem5. my project is about "avoiding timing channel 
attacks in the Memory Controller of cloud processors".

we are implementing a configuration script file that represents a system with 
these features:

1) two processing cores
2) each has L1(Inst) & L1(Data) , a Unified L2 --> local caches
3) a shared L3 cache
4) Shared Mem. Ctrl. (using DRAMSim2)

the above configuration is got from "Temporal Partitioning approach (HPCA 
2014)",

in memory controller (using DRAMSim2), we want to understand that each packet 
is received, is belonged to which core? by that field, we use 
time-division-multiplexing for memory's transactions-scheduling to deal with 
timing channel attack.


we guess that we should add a new field in "packet.cc" file to represent the 
sender core-id.


our question is that this field will be intact from a core up to reaching to 
shared memory controller or for each step in memory's hierarchy, a new packet 
is generated and we should force that immediate nodes (like L3 cache) doesn't 
change this field(sender id_Number_)?


we appreciate any suggestions to take advance.


Thanks' a lot to read this email.



Ashkan Asgharzadeh Donighi





IMPORTANT NOTICE: The contents of this email and any attachments are 
confidential and may also be privileged. If you are not the intended recipient, 
please notify the sender immediately and do not disclose the contents to any 
other person, use it for any purpose, or store or copy the information in any 
medium. Thank you.
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to