Hi David,
On Mon, 22 Jul 2019 at 01:25, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Sat, Jul 20, 2019 at 05:31:08PM -0600, Simon Glass wrote:
> > This file provides read-only access to the device tree and includes many
> > sanity checks. Allow these checks to be disabled to reduce code size.
> >
> > Signed-off-by: Simon Glass <sjg@xxxxxxxxxxxx>
> > ---
> >
> > libfdt/fdt_ro.c | 73 +++++++++++++++++++++++++++++++------------------
> > 1 file changed, 47 insertions(+), 26 deletions(-)
> >
> > diff --git a/libfdt/fdt_ro.c b/libfdt/fdt_ro.c
> > index 6fd9ec1..bc1b4e0 100644
> > --- a/libfdt/fdt_ro.c
> > +++ b/libfdt/fdt_ro.c
> > @@ -14,9 +14,13 @@ static int fdt_nodename_eq_(const void *fdt, int offset,
> > const char *s, int len)
> > {
> > int olen;
> > - const char *p = fdt_get_name(fdt, offset, &olen);
> > + const char *p;
> >
> > - if (!p || olen < len)
> > + if (_check2())
> > + p = fdt_get_name(fdt, offset, &olen);
> > + else
> > + p = fdt_offset_ptr(fdt, offset + FDT_TAGSIZE, len+1);
>
> Here we're not just excising code based on check level but taking a
> different path. This makes me nervous about fragility - I'd be a lot
> more comfortable if we could somehow extend the testsuite to run at
> all the different check levels (except for tests that clearly rely on
> the extra checking, obviously).
Yes, and I think I might as well revert this as there is not much
code-size improvement. I just noticed that the old code was a single
line so thought I'd use that.
Re tests, yes I think we need to check all check levels. Let's see
what you think of the mask vs. level thing in v2, and once we have
that figured out, i'll take a look.
[..]
Regards,
Simon
