Merge common pieces of code

[rafal-krypa]

Refactoring repeated pieces of code by moving it to internal functions. Merged pieces of the following function pairs:

• apply_rules() and apply_cipso()
• smack_accesses_add() and smack_accesses_add_modify()
• smack_accesses_apply() and smack_accesses_save()

Three patches makes the code shorter by 128 lines total. More importantly modifications to these parts will have to be done in single place, which should improve maintainability.

[rafal-krypa]

This has merged very badly and actually requires a rewrite to work with #53. Please revert this patch for now.

[jarkkojs]

OK, what can you do, this sometimes happens even though I thought I tested it properly :( No worries, I'll revert it and tag v1.0.4 so that we can pull that to tizen. I'll take a revised version of this after I've done v1.0.4.

[jarkkojs]

@rafal-krypa, maybe in situations like this we could use tag like v1.0.3.1. Only in situations where invalid patch slips into release. That kind of documents it. Does that work for you?

[jarkkojs]

@rafal-krypa, commit is now reverted. Should I tag v1.0.3.1 now? Can you do a quick sanity check on v1.0.x branch.

[rafal-krypa]

I have rewritten the commit and added two more in the same manner. These two were inspired by experiments with rule merging. I found that every modification to smack_accesses data structure required similar changes in several places. The issue title and description is now updated to reflect these additions.

[jarkkojs]

I've been thinking that this change should be eventually made (use accesses_apply for both cases) so this does work for me.

Some comments:

• I think internal function name should be kept as accesses_apply. No reason to change that.
• I think now more descriptive parameter name would be use_long or something similar because when saving rules long labels can be used even when smackfs isn't mounted.
• Maybe newline could be instead add_lf (very cosmetic but makes parameter obvious).

[rafal-krypa]

I've introduced function accesses_print() and call it from accesses_apply() and smack_accesses_save(). The latter is now a simple wrapper for accesses_print(), but the former is still handling file opening and fallback from load2 to load. If you prefer some other name than accesses_print, I have no problem with changing that. But accesses_apply is already taken.

[jarkkojs]

You should be able squeeze that patch to a smaller change by putting accesses_print declaration after accesses_apply declaration. Now the patch has unnecessary clutter because place of accesses_apply is swapped.

When writing to a normal file there's no guarantee that write() will write the number of bytes given in the count parameter when writing rules to a file. You should have a loop that would call write has written all the bytes..

[jarkkojs]

Looks great now!