Re: [PATCH] lds: align u-boot-nodtb with 8 bytes boundary

[Simon Glass <sjg@chromium.org>]

Hi Eugen,

On Thu, 15 Dec 2022 at 06:37, <Eugen.Hristev@microchip.com> wrote:
>
> On 12/15/22 16:24, Simon Glass wrote:
> > Hi Eugen,
> >
> > On Thu, 15 Dec 2022 at 03:58, Eugen Hristev <eugen.hristev@microchip.com> wrote:
> >>
> >> Newer DTC require that the DTB start address is aligned at 8 bytes.
> >> In the u-boot.bin case, the u-boot-nodtb.bin is concatenated with the
> >> DTB, but there is no alignment/padding to the next 8byte aligned address.
> >> This causes the board to fail booting, because the FDT will claim
> >> that the DTB inside u-boot.bin is not a valid DTB, it will fail with
> >> -FDT_ERR_ALIGNMENT.
> >> To solve this, have the u-boot binary `_end` aligned with 8 bytes.
> >> The objcopy in the Makefile will create the u-boot-nodtb.bin and it has to
> >> be truncated to 8 bytes to correspond to the u-boot.map file which will
> >> have the `_end` aligned to 8 bytes.
> >> The lds files which use devicetrees have been changed to align the `_end`
> >> tag with 8 bytes.
> >>
> >> This patch is also a prerequisite to have the possibility to update the
> >> dtc inside u-boot to newer versions (1.6.1+)
> >>
> >> Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com>
> >> ---
> >> Hi,
> >>
> >> I could not test all affected boards, it's an impossible task.
> >> I tried this on at91 boards which I have, and ran the CI on denx.
> >> I cannot guarantee that no other boards are affected, so this patch is a bit
> >> of an RFC.
> >> If the u-boot-nodtb.bin does not have the size equal with the corresponding
> >> one in u-boot.map, the binary_size_check will fail at build time with
> >> something like this:
> >>
> >> u-boot.map shows a binary size of 502684
> >> but u-boot-nodtb.bin shows 502688
> >>
> >> Thanks,
> >> Eugen
> >>
> >>   Makefile                                    | 2 ++
> >>   arch/arm/cpu/armv8/u-boot.lds               | 4 ++--
> >>   arch/arm/cpu/u-boot-spl.lds                 | 1 +
> >>   arch/arm/cpu/u-boot.lds                     | 1 +
> >>   arch/arm/lib/elf_arm_efi.lds                | 5 +++++
> >>   arch/arm/mach-at91/arm926ejs/u-boot-spl.lds | 2 +-
> >>   arch/arm/mach-at91/armv7/u-boot-spl.lds     | 2 +-
> >>   arch/arm/mach-zynq/u-boot-spl.lds           | 2 +-
> >>   arch/mips/cpu/u-boot.lds                    | 2 +-
> >>   arch/sandbox/cpu/u-boot.lds                 | 6 ++++++
> >>   arch/sh/cpu/u-boot.lds                      | 2 ++
> >>   board/ti/am335x/u-boot.lds                  | 1 +
> >>   tools/binman/test/u_boot_binman_embed.lds   | 2 +-
> >>   13 files changed, 25 insertions(+), 7 deletions(-)
> >>
> >> diff --git a/Makefile b/Makefile
> >> index 9d84f96481..b4d387bcce 100644
> >> --- a/Makefile
> >> +++ b/Makefile
> >> @@ -1317,6 +1317,8 @@ endif
> >>
> >>   u-boot-nodtb.bin: u-boot FORCE
> >>          $(call if_changed,objcopy_uboot)
> >> +# Make sure the size is 8 byte-aligned.
> >> +       @truncate -s %8 $@
> >>          $(BOARD_SIZE_CHECK)
> >
> > I agree this line is needed, since otherwise we will only get 4-byte
> > alignment. But it would be better if we could have the linker scripts
> > fill bytes out to the required alignment. Is that possible?
> >
> > Reviewed-by: Simon Glass <sjg@chromium.org>
> >
> > Regards,
> > Simon
>
>
> Hi Simon,
>
> I tried to check the objcopy option --pad-to , to use it at the time of
> objcopy, but this requires a real number to be passed to it.
> And this number could only be found by inspecting the u-boot.map file,
> since u-boot-nodtb.bin still does not exist.
> And if we pad to the size specified in u-boot.map, then
> binary_size_check does not make much sense anymore, as we will basically
> use the same information to fit the file, and it will always pass with a
> success. (even if we would pad many more bytes than 4 )
> Hence it would lose it's purpose ( binary_size_check ), which I think
> was created to make sure no objects were lost when doing objcopy and
> creating the u-boot-nodtb.bin file.

Yes, I was more thinking of something like:

fill {
    . = ALIGN(8);
    QUAD(0)
};

in the link script, or something that actually writes the padding bytes.

>
> On a side note, do you think I covered all the implied lds files ? I
> would hate to break someone's boards.

If CI passes you should be able to rely on the binary size check.

>
> And also, P.S. : I would require to have the same change when building a
> FIT image with mkimage... all subimages inside a FIT must be aligned to
> 8 bytes. However mkimage only aligns the start address and header of the
> FIT (-B option). Out of your knowledge, is this possible and where could
> I have a look to do this change ?

I lean towards the view that this 8-byte alignment is a bad idea.

Does libfdt itself support this sort of thing? We'll need to be able
to tell it to align the value of a property to 8 bytes, both in the sw
interface and the normal one.

For binman to support it this is pretty easy once the sw interface is
updated - see _add_node() in fit.py.

For mkimage to support it, see fit_import_data(). Perhaps need a
libfdt function like fdt_setprop_align()...?

Regards,
Simon