Skip to contentSkip to conversation

Toradex Verdin AM62 boot failure on recent master

DISCUSSION Vitor Soares 4 participants 5 messages Jan 23 – Jan 25

lore
Share & tools
lore
#0 lore raw

Hi,

I'm experiencing a boot failure on Toradex Verdin AM62 with the latest master. The last known good commit is a4dc1c3b7d56 and it fails to boot at 64662b99c0eb (current HEAD).

Looking at the 35 commits in this range, I noticed that the first commit after my last known good is 85f586035d75 ("ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB"), which directly modifies the K3/AM62 SPL linker script.

Related commits:
85f586035d75 ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB 8b0ebe054bb3 arm: Update linker scripts to ensure appended device tree is
aligned

Could these changes be related to the boot failure? Any insights would be appreciated.

Thanks,
Vitor Soares

#1 lore raw

Hi Vitor,

Quoted 12 lines · Vitor Soares

Hi,

I'm experiencing a boot failure on Toradex Verdin AM62 with the latest master. The last known good commit is a4dc1c3b7d56 and it fails to boot at 64662b99c0eb (current HEAD).

Looking at the 35 commits in this range, I noticed that the first commit after my last known good is 85f586035d75 ("ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB"), which directly modifies the K3/AM62 SPL linker script.

Quoted 7 lines

Related commits:
85f586035d75 ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB 8b0ebe054bb3 arm: Update linker scripts to ensure appended device tree is
aligned

Could these changes be related to the boot failure? Any insights would be appreciated.

Does reverting 8b0ebe054bb3 ("arm: Update linker scripts to ensure appended device tree is aligned") make the board boot again?

I noticed that this commit broke the boot of a Rockchip board:

https://lore.kernel.org/u-boot/CAOMZO5CLQZ3i+Eit5csv6E+az-_g2eOHfCr-whQTPeg06WtqEg@mail.gmail.com/

In the failing boot case, do you get any SPL message at all?

#2 lore raw
In reply to Vitor Soares
Quoted 11 lines · Vitor Soares

Hi,

I'm experiencing a boot failure on Toradex Verdin AM62 with the latest master. The last known good commit is a4dc1c3b7d56 and it fails to boot at 64662b99c0eb (current HEAD).

Looking at the 35 commits in this range, I noticed that the first commit after my last known good is 85f586035d75 ("ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB"), which directly modifies the K3/AM62 SPL linker script.

Quoted 7 lines

Related commits:
85f586035d75 ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB 8b0ebe054bb3 arm: Update linker scripts to ensure appended device tree is
aligned

Could these changes be related to the boot failure? Any insights would be appreciated.

You likely need
https://patchwork.ozlabs.org/project/uboot/patch/[email protected]/

Signature · 1 line
Tom
#3 lore raw
Quoted 22 lines · Tom Rini
Quoted 20 lines · Vitor Soares

Hi,

I'm experiencing a boot failure on Toradex Verdin AM62 with the latest master.
The last known good commit is a4dc1c3b7d56 and it fails to boot at 64662b99c0eb
(current HEAD).

Looking at the 35 commits in this range, I noticed that the first commit after
my last known good is 85f586035d75 ("ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB"), which directly modifies the K3/AM62 SPL linker script.

Related commits:
   85f586035d75 ARM: OMAP2+: Pad SPL binary to 8-byte alignment before DTB 8b0ebe054bb3 arm: Update linker scripts to ensure appended device tree is aligned

Quoted 6 lines
Quoted 2 lines

Could these changes be related to the boot failure? Any insights would be appreciated.

You likely need
https://patchwork.ozlabs.org/project/uboot/patch/[email protected]/

The v3 patch alone didn't fix the boot issue on Verdin AM62.

However, Marek's suggested approach [1] that only aligns flat_dt type images to 8 bytes does fix it:

tools/fit_image.c +4 -0
@@ -685,6 +685,13 @@ static int fit_extract_data(...)
              const char *data, *type;
              int len;
 
+             /* Ensure flat_dt images are 8-byte aligned per DT spec */
+             if (align_size < 8) {
+                     type = fdt_getprop(fdt, node, FIT_TYPE_PROP, &len);
+                     if (type && len > 0 && !strncmp(type, "flat_dt", len < 8 ?

len : 7))
+ align_size = 8;
+ }
+

              data = fdt_getprop(fdt, node, FIT_DATA_PROP, &len);

+Cc Fabio - this should resolve your boot failure as well.

[1]
https://lore.kernel.org/u-boot/[email protected]/

Best regards,
Vitor Soares

#4 lore raw

Hi Vitor,

Quoted 4 lines · Vitor Soares

+Cc Fabio - this should resolve your boot failure as well.

[1]
https://lore.kernel.org/u-boot/[email protected]/

Thanks for the suggestion, but this does not fix the boot failure on RV1103.

Tom's proposal fixes it:

https://lore.kernel.org/u-boot/CAOMZO5DVFO=A6g1CWOrRbqy4VNgL+VcUw0Q5pgELXm30puGz4w@mail.gmail.com/

Thanks,

Fabio Estevam

Signature · 2 lines
Nabla Software Engineering GmbH	Managing Director: Stefano Babic
Phone: +49 821 45592596 	E-Mail: [email protected]