[U-Boot] [PATCH v2] imx: hab: extend hab_auth_img to calculate ivt_offset

Wed Nov 21 18:42:21 UTC 2018

Hi Parthiban,

Em qua, 21 de nov de 2018 às 15:52, Parthiban Nallathambi <pn at denx.de> escreveu:
>
> Hi Breno,
>
> On 11/21/18 5:45 PM, Breno Matheus Lima wrote:
> > Hi Parthiban,
> >
> > Em qua, 21 de nov de 2018 às 11:50, Parthiban Nallathambi <pn at denx.de> escreveu:
> >>
> >> Current implementation of hab_auth_img command needs ivt_offset to
> >> authenticate the image. But ivt header is placed at the end of image
> >> date after padding.
> >>
> >> This leaves the usage of hab_auth_img command to fixed size or static
> >> offset for ivt header. New function "get_image_ivt_offset" is introduced
> >> to find the ivt offset during runtime. The case conditional check in this
> >> function is same as boot_get_kernel in common/bootm.c
> >>
> >> With this variable length image e.g. FIT image with any random size can
> >> have IVT at the end and ivt_offset option can be left optional
> >>
> >> Can be used as "hab_auth_img $loadaddr $filesize" from u-boot script
> >>
> >> Signed-off-by: Parthiban Nallathambi <pn at denx.de>
> >> ---
> >>
> >> Notes:
> >>      Changelog in v2:
> >>      - Finding IVT offset doesn't need length. Removed the
> >>      length argument from get_image_ivt_offset
> >>
> >>   arch/arm/mach-imx/hab.c | 29 +++++++++++++++++++++++++++--
> >>   1 file changed, 27 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/arch/arm/mach-imx/hab.c b/arch/arm/mach-imx/hab.c
> >> index b88acd13da..dbfd692fa3 100644
> >> --- a/arch/arm/mach-imx/hab.c
> >> +++ b/arch/arm/mach-imx/hab.c
> >> @@ -6,6 +6,8 @@
> >>   #include <common.h>
> >>   #include <config.h>
> >>   #include <fuse.h>
> >> +#include <mapmem.h>
> >> +#include <image.h>
> >>   #include <asm/io.h>
> >>   #include <asm/system.h>
> >>   #include <asm/arch/clock.h>
> >> @@ -302,18 +304,41 @@ static int do_hab_status(cmd_tbl_t *cmdtp, int flag, int argc,
> >>          return 0;
> >>   }
> >>
> >> +static ulong get_image_ivt_offset(ulong img_addr)
> >> +{
> >> +       const void *buf;
> >> +
> >> +       buf = map_sysmem(img_addr, 0);
> >> +       switch (genimg_get_format(buf)) {
> >> +#if defined(CONFIG_IMAGE_FORMAT_LEGACY)
> >> +       case IMAGE_FORMAT_LEGACY:
> >> +               return (image_get_image_size((image_header_t *)img_addr)
> >> +                       + 0x1000 - 1)  & ~(0x1000 - 1);
> >> +#endif
> >> +#if IMAGE_ENABLE_FIT
> >> +       case IMAGE_FORMAT_FIT:
> >> +               return (fit_get_size(buf) + 0x1000 - 1)  & ~(0x1000 - 1);
> >> +#endif
> >> +       default:
> >> +               return 0;
> >> +       }
> >> +}
> >
> >
> > Do you have more details about the image header I should use here?
>
> Is the image signed using CST or similar tool? Is so, the signature data
> (HAB data: CSF, Certs and signature) pads at the end of the kernel
> image.

Yes, my Kernel image contains an IVT and is signed with CST. The image
layout looks like link below:
https://pastebin.com/5qEt7ETa

>
> >
> > I'm trying to get the IVT offset for my Kernel image based on NXP
> > 4.9.11_2.0.0_GA Linux release loaded at 0x80800000:
> >
> > => fatload mmc 0:1 0x80800000 zImage
> > => hab_auth_img 0x80800000 <Len not being used>
>
> Length for hab_auth_img is still mandatory. Length hear means the file
> size or total length of the image, this is required for the HAB API to
> authenticate (HAB_RVT_AUTHENTICATE_IMAGE).

Oh ok, I understood the scenario right now.

>From my first overview I thought we would add IVT_SIZE + CSF_PAD_SIZE
in ivt_offset to calculate the image length, similar approach we have
in spl.c:
http://git.denx.de/?p=u-boot.git;a=blob;f=arch/arm/mach-imx/spl.c;h=a20b30d154d788e4ebd4e22e9a6568a4f24c057e;hb=HEAD#l226

In this case only load addr would be necessary.

>
> >
> > The zImage size in Header looks correct (0x00726690), but
> > get_image_ivt_offset() is returning 0x0
>
> Looks like IVT offset is not found in the image.
>
> >
> > $ hexdump zImage | head
> > 0000000 0000 e1a0 0000 e1a0 0000 e1a0 0000 e1a0
> > *
> > 0000020 0003 ea00 2818 016f 0000 0000 6690 0072
> > 0000030 0201 0403 9000 e10f 04f8 eb00 7001 e1a0
> > 0000040 8002 e1a0 2000 e10f 0003 e312 0001 1a00
> >
> > Seems that genimg_get_format() is returning 0x0.
>
> Image is not signed?

No, my zImage is signed.

>
> >
> > Any ideias if I'm missing something?
> >
> >> +
> >>   static int do_authenticate_image(cmd_tbl_t *cmdtp, int flag, int argc,
> >>                                   char * const argv[])
> >>   {
> >>          ulong   addr, length, ivt_offset;
> >>          int     rcode = 0;
> >>
> >> -       if (argc < 4)
> >> +       if (argc < 3)
> >
> > I think we can also change here to argc < 2, the function
> > get_image_ivt_offset() only requires the img addr now.
>
> No, length is mandatory for authentication. To brief, this patch just
> removes the ivt_offset as argument and make it optional. This is needed
> because, in images like FIT, the location of the ivt header varies
> dynamically depending on the total number of images clubbed.
>
> In such cases, the existing hab_auth_img is hard to use as ivt header
> offset needs to be pre-calculated everytime and fed into.

Yes I understood your scenario right now, as I stated above I thought
your patch would also calculate the image length as we have in spl.c.

I'm only wondering why genimg_get_format() is returning
IMAGE_FORMAT_INVALID, do you happen to know if zImage header should
return IMAGE_FORMAT_LEGACY?

Thanks,
Breno Lima