Open huang429 opened 3 years ago
The fourth step in the preparation in the README says that you need to specify the instruction which is considered as the output of the program. For example, if you examine the SDC of your program under test by matching the standard output between the fault-free run and the fault injection runs, you need to specify all the "printf" instructions in the LLVM IR code of your program (you do so by updating the Trident with the LLVM IR index of the printf instruction(s) ). That way Trident knows where to stop when tracking error propagation.
Hope this helps.
Thank you for your explanation about the steps.I will try my programs again.Thanks.
------------------ 原始邮件 ------------------ 发件人: "DependableSystemsLab/Trident" @.>; 发送时间: 2021年6月7日(星期一) 中午12:56 @.>; @.**@.>; 主题: Re: [DependableSystemsLab/Trident] Questions I met when running this code (#7)
The fourth step in the preparation in the README says that you need to specify the instruction which is considered as the output of the program. For example, if you examine the SDC of your program under test by matching the standard output between the fault-free run and the fault injection runs, you need to specify all the "printf" instructions in the LLVM IR code of your program. That way Trident knows where to stop when tracking error propagation.
Hope this helps.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
I'm sorry to say that I don't understand the fourth step in the preparation of Trident.What would I do for choosing the LLVM IR instructions?And when I run the ./test,the log file "fi_breakdown.txt" didnot update the probabilities of SDC,Benign,Crash and breakdown.It just equals to the log file "fi_breakdown.txt" generated in ./inst_count.