<div dir="ltr">That's true but you're losing the scan-build like, colored html output (use debatable) with explanations under what circumstance the issue happens (branches taken, etc.) (very helpful). </div><div class="gmail_extra">
<br><br><div class="gmail_quote">On Thu, May 2, 2013 at 4:49 AM, PaX Team <span dir="ltr"><<a href="mailto:pageexec@freemail.hu" target="_blank">pageexec@freemail.hu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="HOEnZb"><div class="h5">On 1 May 2013 at 19:47, Behan Webster wrote:<br>
<br>
> >>> I was able to x64 just fine the other day. (I also tried it<br>
> >>> with allyesconfig but that (obviously) failed).<br>
> >> We don't support all configurations yet. Getting to the point<br>
> >> where we are already has taken a lot of work already.<br>
> >><br>
> >><br>
> >> I understand. It was also on of the ideas on the GSoC page to<br>
> >> have allyesconfig work. I was just curious how long until the<br>
> >> error (longer than I expected; it was a VLAS).<br>
> > I would expect allyesconfig would fail very quickly. :)<br>
> ><br>
> ><br>
> > Once I actually went to JOBS=1 that was indeed the case :-D.<br>
> LOL.<br>
><br>
> >>> As far as the scope for implementing checker for the<br>
> >>> kernel, many of the frequent contributors just have not<br>
> >>> had the time to try it. From those who have tried it, we<br>
> >>> know it doesn't just work out of the box.<br>
> >>><br>
> >>><br>
> >>> I see. When ever I used the analyzer it was using the<br>
> >>> scan-build tool. Need to check whether it works with the<br>
> >>> custom build system<br>
> >> We don't need to run it on our makefiles. We need it run on<br>
> >> the kernel makefiles (Kbuild). The analyzer would be driven<br>
> >> by our build system.<br>
> >><br>
> >><br>
> >> The way I understand the system, it basically replaces env<br>
> >> variables but allows customization, does error-handling, sets<br>
> >> defaults, etc. I was just wondering if it would propagate through<br>
> >> down to where the `make` that actually builds the kernel.<br>
> > That would be the gig: Figuring this out, then ultimately tuning<br>
> > it for the kernel if time permitted (adding specific<br>
> > analysis/checks which are Linux kernel specific).<br>
> ><br>
> ><br>
> > Alright, with some experimentation I figured: it does. Well, to some<br>
> > extent at least. After disabling the hardcoded CC variable in<br>
> > make-kernel.sh. I was able to run the analyzer using the scan-build<br>
> > wrapper (ccc-analyzer).<br>
> Hmm. Maybe it isn't as much work as we thought? :)<br>
<br>
</div></div>sorry to chime in this late, but if you just want to build the kernel<br>
with the analyzer, it's very simple as i described it some time ago<br>
(<a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2010-October/011742.html" target="_blank">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2010-October/011742.html</a>):<br>
<br>
the easiest way to run the analyzer is to issue<br>
make CC=.../clang C=2 CHECK="clang --analyze"<br>
<br>
</blockquote></div><br></div>