- User Since
- Apr 14 2017, 1:59 PM (65 w, 6 d)
Tue, Jul 17
Maybe we should keep the idea of one context on hold and focus on the reusable SAT cores for now?
@dblaikie Good point!
LGTM with a nit.
@morehouse unit tests for other sanitizers do though.
I'm a bit confused on what exactly is happening here, and why tests pass on other platforms
@NoQ I think the analyzer would hang if we construct an expression with a complexity of 2^32
@Szelethus false positives are a single biggest problem of the analyzer.
By a *huge* margin, most projects would prefer to err on the side of less, more precise, warnings.
Given that currently in my understanding no actual bugs we are sure about were found by the uninitialized object checker,
I think by default we should err on the "less warnings" side.
Cf. my comments to https://reviews.llvm.org/D49437: is it possible to separate pointer-chasing from the rest of the checker?
Cf. my comments to https://reviews.llvm.org/D49437: while this change looks great, is it possible to separate the pointer chasing from the rest of the checker?
I think a checker for uninitialized values left after a constructor call is very valuable.
Mon, Jul 16
Attempt #2: reduced version of this patch, without ubsan support.
Fri, Jul 13
Thu, Jul 12
Wed, Jul 11
@rnkovacs Do you have evaluation statistics handy for this checker? How many bugs it finds, on which projects? How many of those are real bugs?
That's an awesome idea!
Looks reasonable, but I think @dberris would have to approve.
The imprecision in the built in solver might result in failure to constrain a value to zero while the Z3 might be able to do that.
LGTM with a nit. We'll watch the bots after this is merged.
What issues could it cause since it is guarded by an option?
Tue, Jul 10
@mikhail.ramalho Could you also state how different values affect the effectivity of refutation?
Mon, Jul 9
The overall point is that writing this kind of code is *extremely* error-prone.
We are actually considering going in a different direction and doing a rollback for the previous rearrangement patches due to some issues.
Could you see whether Z3 visitor would meet your needs?
For the first change, let's leave the complexity the same.
Let's discuss alternatives first.
Fri, Jul 6
@rsmith i hope i'm on the right track here.
Thu, Jul 5
Do you think you can instead disable three problematic tests on AARCH64 and revert the revert?
That seems like a more local solution.
@mikhail.ramalho please resubmit with -U999
Tue, Jul 3
Made a copy-paste error. Committed in https://reviews.llvm.org/rL336238.
Oh sorry I've referenced wrong revision in the commit. Re-opening.