Fix testcase set_filename

[SpaceWhite]

fix #1084

add rename file logic in set_filename when new file name is set

[domenukk]

If we store, do we want to remove the old version (if that exists)?

[SpaceWhite]

To remove old file versions, we could implement a delete function for the Input trait.
But I'm wondering why this issue is even a problem on the first place?
Isn't set_filename is only called when testcase.filename does not exist? Hence it is impossible to have self.filename with some string value.

@VTCAKAVSMoACE

[addisoncrump]

I think the solution here is to check if the file .<filename>.lafl_lock exists. If it does not, then it is safe to remove. Otherwise, return an illegal state error.

[addisoncrump]

I believe in this case, we also must not simply store_input but actually to move the input file, the .metadata file, and the .lafl_lock (if each exist). Otherwise, this will simply create a copy.

[domenukk]

wouldn't we maybe have created the lock file, though?

[addisoncrump]

Yes, we would have made the lockfile. But even so, if we are renaming to something that we currently use, it would be unsafe to remove/replace the destination file.

[domenukk]

So what to do?

[SpaceWhite]

I've added rename file logic. If either old lock file or new lock file exist, then set_filename will return error.

[domenukk]

Will that ever work, then? When is the lock file removed?

[SpaceWhite]

I have added some more code to the functions.
So the idea is that old/new lock files are created and deleted in set_filename function.

If lock file are already created for old/new testcase then set_filename called, then it throws an error.
By doing this testcases could be safely created and removed.

[SpaceWhite]

Is it okay for testcase.rs to depend on std?

[domenukk]

It needs to work without (for inmemory), but for storing to disk it needs std yes

[SpaceWhite]

added set_Filename function for no_std.

Since there is no file creation/deletion, it just assigns value to struct.

[domenukk]

clippy complains

error: the borrowed expression implements the required traits
   --> libafl/src/corpus/testcase.rs:176:24
    |
176 |             fs::rename(&old_filename, &new_filename)?;
    |                        ^^^^^^^^^^^^^ help: change this to: `old_filename`
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#needless_borrow
note: the lint level is defined here
   --> libafl/src/lib.rs:16:9
    |
16  | #![deny(clippy::all)]
    |         ^^^^^^^^^^^
    = note: `#[deny(clippy::needless_borrow)]` implied by `#[deny(clippy::all)]`

error: the borrowed expression implements the required traits
   --> libafl/src/corpus/testcase.rs:176:39
    |
176 |             fs::rename(&old_filename, &new_filename)?;
    |                                       ^^^^^^^^^^^^^ help: change this to: `new_filename`
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#needless_borrow

error: the borrowed expression implements the required traits
   --> libafl/src/corpus/testcase.rs:180:24
    |
180 |             fs::rename(&old_metadata_filename, &new_metadata_filename)?;
    |                        ^^^^^^^^^^^^^^^^^^^^^^ help: change this to: `old_metadata_filename`
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#needless_borrow

error: the borrowed expression implements the required traits
   --> libafl/src/corpus/testcase.rs:180:48
    |
180 |             fs::rename(&old_metadata_filename, &new_metadata_filename)?;
    |                                                ^^^^^^^^^^^^^^^^^^^^^^ help: change this to: `new_metadata_filename`
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#needless_borrow

[SpaceWhite]

@domenukk

fixed

[domenukk]

@addisoncrump do you think this is correct now? thx :)

[domenukk]

Sorry, missed the update. Looks good to me now, thanks! 👍

[tokatoka]

fuzzbench_qemu broken after this

[tokatoka]

@SpaceWhite can you check this again?

[SpaceWhite]

@tokatoka Sorry for the long wait. I just sent the PR to fix this.