See https://github.com/google/rowhammer-test - this is an adaption of the ''double_sided_rowhammer'' program from their repository.
Also see our paper ''Rowhammer.js: A Remote Software-Induced Fault Attack in JavaScript'' on Arxiv: http://arxiv.org/abs/1507.06955
cd native
make
./double_sided_rowhammer_ivy -d 1 # -d number of dimms
# or
./double_sided_rowhammer_haswell -d 1 # -d number of dimms
The test should work on x86-64 Linux.
If you have found a reproducible bitflip, look for the ''Print this for the watch_firefox tool'' comment.
Edit tools/watch_firefox.cc to contain the addresses from your native eviction-based rowhammer test.
Start firefox with rowhammer.html
cd tools
make
./watch_firefox
The program outputs physical address mappings and the time since the last allocation. Based on this and the virtual address printed you can determine where the array starts.
Allocate memory in a large array in JavaScript. If using ''rowhammer.html'': Click the ''Allocate'' button.
If you have much noise before you press the button, just restart ''watch_firefox'' and try again.
As soon as it has found the indices it asks you to enter the virtual address of the array start. This is not yet automated.
The program prints the array indices to use in JavaScript. In case of ''rowhammer.html'' just copy them into the editbox and click ''Parse''. Then you can start Hammering.
In the ''javascript'' folder you find the Rowhammer.js version for Haswell CPUs with a 16-way L3 cache and no L4 cache ''rowhammer.html''. It will probably not work on other CPUs without modifications.
Open ''rowhammer.html'' in a browser, paste the hammering array indices in the editbox (you can use the ''watch_firefox'' program for this).
You can modify ''rowhammer.js'' while the page is still loaded and click the ''Refresh'' button to only reload the ''rowhammer.js'' file. This way you keep the array and the array indices and you can experiment with different settings while not having to search for the array indices anew.
Also, in the ''javascript'' folder you will find the ''rowhammer_scan.html''. It is the pure JavaScript proof-of-concept for Haswell CPUs with a 16-way L3 cache and no L4 cache, memory in single-channel mode (this is the case if you have only one DIMM). Furthermore, it assumes that JavaScript memory is physically contiguous in blocks of 2 megabytes - this is the case if your OS allocates 2M anonymous pages (all our Linux systems do) or if it allocates physically contiguous 4K pages. It will probably not work on other systems without modifications.
Open ''rowhammer_scan.html'' in a browser. Click the ''Allocate'' button. Wait a second to let Firefox allocate memory and click ''Hammer'' to start the hammering.
For reference we have added a screenshot:
You can modify ''rowhammer_scan.js'' while the page is still loaded and click the ''Refresh'' button to only reload the ''rowhammer.js'' file. This way you keep the array and the array indices and you can experiment with different settings while not having to allocate the array anew.
This version is not adaptive to all CPUs. As we said in the paper, the eviction strategy finding algorithm is very slow. We still try different optimizations and we will evaluate it's performance. However, there is not much use in doing this search in JavaScript as it takes hours and there is not much benefit once you know the strategy. A more realistic adaptive approach would be to try different strategies that are already known to work on some CPUs. This way you can find the right strategy adaptively without having to execute the generic eviction strategy finding algorithm. However, if you like to try, you will find the ''cached'' function already implemented in the ''rowhammer_scan.js''.
Same warnings as in the original https://github.com/google/rowhammer-test repository:
Warning #1: We are providing this code as-is. You are responsible for protecting yourself, your property and data, and others from any risks caused by this code. This code may cause unexpected and undesirable behavior to occur on your machine. This code may not detect the vulnerability on your machine.
Be careful not to run this test on machines that contain important data. On machines that are susceptible to the rowhammer problem, this test could cause bit flips that crash the machine, or worse, cause bit flips in data that gets written back to disc.
Warning #2: If you find that a computer is susceptible to the rowhammer problem, you may want to avoid using it as a multi-user system. Bit flips caused by row hammering breach the CPU's memory protection. On a machine that is susceptible to the rowhammer problem, one process can corrupt pages used by other processes or by the kernel.
Additionally, if your computer is susceptible to the rowhammer bug, disable JavaScript in your browser! Attackers could exploit this bug through JavaScript and take control over your machine.