On Sat, Aug 01, 2015 at 09:23:17PM +0800, Fengguang Wu wrote:
> On Sat, Aug 01, 2015 at 04:12:20PM +0300, Dan Carpenter wrote:
> > On Sat, Aug 01, 2015 at 08:22:41PM +0800, Fengguang Wu wrote:
> > > Hi Dan,
> > >
> > > On Fri, Jul 31, 2015 at 05:03:20PM +0300, Dan Carpenter wrote:
> > > > We're getting a lot of false positives today because the database is
> > > > out of sync with the kernel we are testing.
> > >
> > > Sorry we just fixed a bug that prevented smatch checks for long time.
> > > So here comes a bunch of smatch reports.
> > >
> > > Since the database is expensive to build, we typically reuse the same
> > > database across a number of kernels, which may lead to the false
> > > positives you noticed.
> > >
> > > I'll try improving the test process.
> > >
> >
> > I understand. I'm not complaining, I'm just explaining why.
> >
> > Most of the false positives were things like:
> >
> > memset(foo, 0, sizeof(*foo));
> >
> > That should have been fixable on my end. It seemed like a simple thing
> > and I tried to do that some months back but it just made Smatch segfault
> > and I was never able to figure it out. I will look again next week.
>
> OK, thank you!
>
> I just did a patch to force rebuild smatch db on every branch, which
> would add cost by about 1+ hour. It's reasonably acceptable for me.
> However it will no longer be a big problem, I can still revert that
> change.
That 1+ hour is mainly the time spend in this function.
# borrowed from Dan's smatch_scripts/build_kernel_data.sh
update_smatchdb()
{
# Smatch db won't be complete for a broken build. We may either
# 1) reuse any old smatch db
# 2) don't do smatch check at all
# Prefer (2) for now.
[[ "$build_error" ]] && ! grep -q vmlinux.o $kbuild_log && return
local SCRIPT_DIR=/c/smatch/smatch_scripts
local DATA_DIR=/c/smatch/smatch_data
local PROJECT=kernel
local info_file=/c/smatch/smatch-warns.txt
find $SMATCH_OUT_ROOT -type f | xargs cat > $info_file
du -s $info_file
for i in $SCRIPT_DIR/gen_* ; do
$i $info_file &
done
wait
mv ${PROJECT}.* $DATA_DIR
mv no_return_funcs $DATA_DIR/${PROJECT}.no_return_funcs
$DATA_DIR/db/create_db.sh -p=kernel $info_file
rm $info_file
}
The log file is
Fri Jul 31 12:32:26 CST 2015
9658260 /c/smatch/smatch-warns.txt
Done. List saved as 'kernel.expects_err_ptr'
Done. List saved as 'kernel.returns_err_ptr'
Done. List saved as 'kernel.returns_held_funcs'
Done. List saved as 'no_return_funcs'
Copy it to smatch_data/<project>.no_return_funcs
Done. List saved as 'kernel.frees_argument'
Done. List saved as 'kernel.allocation_funcs'
Done. List saved as 'kernel.unwind_functions'
Done. List saved as 'kernel.sizeof_param'
Done. List saved as 'kernel.bit_shifters'
Done. List saved as 'kernel.puts_argument'
Done. List saved as 'kernel.dma_funcs'
Done. List saved as 'kernel.rosenberg_funcs'
Done. List saved as 'kernel.gfp_flags'
2147483646
off
off
Fri Jul 31 14:06:36 CST 2015
Thanks,
Fengguang
--
To unsubscribe from this list: send the line "unsubscribe smatch" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
