Add support to LAVA-M dataset

[Yhcrown]

fix #102.
I add more intercepted functions so that the lava-m programs can properly get input tainted.
If the code is not quite standardized, I hope to get your test and modification.
Thanks & Best regards.

[aurelf]

Thanks for the PR, I added a few comments.
Also it would be great to add a test case, like extending test/file_input.c those are run by the CI and allow to detect regressions.
Thanks !

[Yhcrown]

Some functions are difficult for me to write test cases for (like getutent). I’m still learing it. 😢

[aurelf]

Thanks for the changes! Would be great if the test is possible to add for that function too... Do you think you can add it ?

[Yhcrown]

Thanks for the changes! Would be great if the test is possible to add for that function too... Do you think you can add it ?

It may be a bit difficult for me now 😭 , I haven't figured out how symcc testing works yet. It would be great to have your help!.

[sebastianpoeplau]

I haven't figured out how symcc testing works yet.

We have some documentation here: https://github.com/eurecom-s3/symcc/blob/master/docs/Testing.txt. Let us know if you have questions!

[zhou1615]

Sorry for trouble but this PR does not seem to work in the current version of SymCC.
It seems that g_config.fullyConcrete has been removed from Config.cpp.
I am not quite familiar with SymCC, thus I am not sure how to make it work.

Maybe a possible change is use maybeSetInputFile like this commit d5891b0

[sebastianpoeplau]

Sorry for trouble but this PR does not seem to work in the current version of SymCC. It seems that g_config.fullyConcrete has been removed from Config.cpp. I am not quite familiar with SymCC, thus I am not sure how to make it work.

I can take it over, but I've got a question about the symbolic version of getutent. If I understand correctly, it treats the data returned from this libc function as symbolic input. Is this intended? In other words, do you expect SymCC to generate alternate contents of /var/run/utmp? Users can't write this file anyway, so wouldn't it make more sense to treat any data read from there as concrete?

I don't have a LAVA-M setup to check ☹️

[aurelf]

I'm trying to look at this PR, @Yhcrown in case you can give me push access to your branch that would save me quite some time :) It's quite manual now... I seem to be only able modify your pr from the online editor.

[AndrewQuijano]

Just curious, is this still being worked on? LAVA is currently in the process of being upgraded, and perhaps I can help with this. For now, LAVA is being upgraded to work with LLVM-11 and at least there is potentially a plan to have LAVA leverage symcc to be able to plant bugs in more places in code.

@Yhcrown

[Yhcrown]

Just curious, is this still being worked on? LAVA is currently in the process of being upgraded, and perhaps I can help with this. For now, LAVA is being upgraded to work with LLVM-11 and at least there is potentially a plan to have LAVA leverage symcc to be able to plant bugs in more places in code.

@Yhcrown

Sorry for the late reply. This sounds interesting, but I am not currently pursuing work in this direction. You can discuss advancing this direction with the members here.

[Yhcrown]

I'm trying to look at this PR, @Yhcrown in case you can give me push access to your branch that would save me quite some time :) It's quite manual now... I seem to be only able modify your pr from the online editor.

I'm very sorry for such a late reply, I missed this notification before. I have added you as a collaborator, although it is quite late. I hope this may still be of help.