On Thu, Jan 9, 2020 at 5:30 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote: > > Alexander Potapenko <glider@xxxxxxxxxx> writes: > >> > 2. checkpatch.pl complains a lot about the use of BUG_ON in KMSAN > >> > source. I don't have a strong opinion on this, but KMSAN is a debugging > >> > tool, so any runtime invariant violation in it renders the tool useless. > >> > Therefore it doesn't make much sense to not terminate after a bug in > >> > KMSAN. > >> > >> Such early termination will cause hard time for verification teams > >> execute any major test suite with KMSAN till it will be without any bugs. > >> So yes, I think that the best solution is to avoid BUG_ONs in non-critical code. > >> > > > > To clarify, these BUG_ONs are for bugs in KMSAN itself. KMSAN reports > > don't provoke a BUG(), so many of them can be reported at once. > > Proceeding after a bug in the debugging tool itself might be a bad > > idea - any broken invariant denotes that all following reports may be > > incorrect. > > Correct, unless you can gracefully disable the debugging tool and just > continue which would be the best solution. (sorry for not responding earlier) Not sure gracefully disabling the tool and continuing will add any extra value. In the case KMSAN is used for continuous testing it's important to fail early, so that the developers are aware that the tool is not testing anything. Pretending it still works and proceeding with execution may give us false sense of good testing coverage. -- Alexander Potapenko Software Engineer Google Germany GmbH Erika-Mann-Straße, 33 80636 München Geschäftsführer: Paul Manicle, Halimah DeLaine Prado Registergericht und -nummer: Hamburg, HRB 86891 Sitz der Gesellschaft: Hamburg
