[U-Boot] [U-Boot,v2] spl: implement CRC check on U-Boot uImage

[Simon Goldschmidt]

On Sun, Feb 10, 2019 at 1:43 AM Tom Rini <trini at konsulko.com> wrote:
>
> On Sat, Feb 09, 2019 at 10:56:40PM +0100, Simon Goldschmidt wrote:
> > On Fri, Feb 8, 2019 at 10:20 PM Tom Rini <trini at konsulko.com> wrote:
> > >
> > > On Fri, Feb 08, 2019 at 10:05:41PM +0100, Simon Goldschmidt wrote:
> > > > On Fri, Feb 8, 2019 at 8:46 PM Tom Rini <trini at konsulko.com> wrote:
> > > > >
> > > > > On Wed, Nov 28, 2018 at 09:52:45PM +0100, Simon Goldschmidt wrote:
> > > > >
> > > > > > SPL currently does not check uImage CRCs when loading U-Boot.
> > > > > >
> > > > > > This patch adds checking the uImage CRC when SPL loads U-Boot. It does
> > > > > > this by reusing the existing config option SPL_CRC32_SUPPORT to allow
> > > > > > leaving out the CRC check on boards where the additional code size or
> > > > > > boot time is a problem (adding the CRC check currently adds ~1.4 kByte
> > > > > > to flash).
> > > > > >
> > > > > > The SPL_CRC32_SUPPORT config option now gets enabled by default if SPL
> > > > > > support for legacy images is enabled to check the CRC on all boards
> > > > > > that don't actively take countermeasures.
> > > > > >
> > > > > > Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt at gmail.com>
> > > > > > Reviewed-by: Simon Glass <sjg at chromium.org>
> > > > >
> > > > > Really sorry for the delay on this, especially as I've found one or two
> > > > > problems.  The first problem is that with this vyasa-rk3288 and a few
> > > > > others fail to build due to a number of errors such as:
> > > > > ../common/spl/spl.c:269:12: error: 'struct spl_image_info' has no member
> > > > > named 'dcrc_data'
> > > >
> > > > Hmm, let me check what's wrong there.
> >
> > OK, so the vyasa-rk3288 uses TPL, that's what I got wrong.
> >
> > > > >
> > > > > Second, I believe this is causing a number of platforms with very tight
> > > > > SPL constraints, namely all of 64bit sunxi, to fail to link as they
> > > > > overflow SRAM now.  This can be fixed at least by making the new
> > > > > behavior opt-in, but I would fix the vyasa-rk3288 problem first.  Thanks
> > > > > in advance!
> > > >
> > > > Well, I thought it would be better to have it default to 'y' since I think it
> > > > is what is expected. Shouldn't we explicitly work on those platforms?
> > > > I must say I was pretty shocked to see that SPL did not detect an invalid
> > > > CRC...
> > > >
> > > > Can you point me to a config that fails?
> > >
> > > In general, pine64_plus_defconfig is one that might (as I re-run my
> > > build without this patch, I still see some failures but they've scrolled
> > > off screen, but 64bit sunxi _is_ very sensitive to SPL growth).
> >
> > And I also did compile that one but I don't see any error message. How
> > am I supposed to detect the SRAM overflowing?
> >
> > I assume it's .text or .rodata which makes it grow too big? If so, we could
> > still check the CRC by using a smaller algorithm that does not rely on a
> > precalculated table...
>
> That could have been the other patch then.  Does all of sunxi build for
> you?  If so, we're good then.

Well, I really can't tell: especially with other patches in the queue that might
get puhed before mine. text and rodata do grow by ~1.6 KiB summed up, so
that might of course trigger some platforms.

On the other hand, I think this is a critical feature for reliable boot unless
using FIT. If you don't think so (and in fact, noone seems to have noticed this
for some years), then I'm happy to re-post this without default 'y'.

Regards,
Simon