Hi Stefano,
On Fri, Jan 3, 2020 at 1:23 PM Stefano Babic sbabic@denx.de wrote:
Hi Igor,
On 27/12/19 10:23, Igor Opaniuk wrote:
Hi Breno,
On Mon, Dec 23, 2019 at 7:07 PM Breno Matheus Lima brenomatheus@gmail.com wrote:
Hi Igor,
Em qui., 19 de dez. de 2019 às 07:55, Igor Opaniuk igor.opaniuk@gmail.com escreveu:
From: Igor Opaniuk igor.opaniuk@toradex.com
Currently Colibri iMX7 NAND version doesn't boot at all with HABv4 support enabled. If CSF section is included in the final imx binary, BootROM every time switches to usb recovery mode. However eMMC version of the same SoM works without any issues.
Disable HAB and CAAM support for now until the problem is properly investigated and fixed.
This issue is also happening with i.MX6ULL, seems that padding the U-Boot binary to the size defined in boot data is addressing this issue.
Please follow example below.
- Dump boot data:
$ hexdump u-boot-dtb.imx | head 0000000 00d1 4020 0000 8780 0000 0000 f42c 877f 0000010 f420 877f f400 877f 6000 878d 0000 0000 0000020 f000 877f b000 000d 0000 0000 01d2 40e8 0000030 01cc 04e4 0c02 6840 ffff ffff 0c02 6c40
IVT self = 0x877ff400 Boot data addr = 0x877ff000 Boot data size = 0x000db000
- Calculate image size:
Image offset = IVT self(0x877ff400) - Boot data addr(0x877ff000) = 0x400 Total image size = Boot data size(0x000db000) - Image offset(0x400) = 0xdac00
- Pad U-Boot image:
$ objcopy -I binary -O binary --pad-to 0xdac00 --gap-fill=0x00 u-boot-dtb.imx u-boot-dtb.imx.pad
Could you please try similar in your i.MX7D board?
Thanks, Breno Lima
Finally, it works!
Thanks a lot for your help! This should be documented (I'll create a patch for this) or padding should implicitly be done in makefiles when building the final imx image, including initial padding for 0x400, like here [1].
I left [1] in my queue and I have always asked myself if we really need it. Frankly speaking, I prefer that the case should be clearly documented instead of adding an initial padding. Padding could forbid (or makes more difficult) to pack u-boot inside another container.
I would say that [1] is a dirty workaround. I'm currently trying to understand if it's possible to address both issues in tools/imximage.c, so please just ignore [1].
Regards, Stefano
Regards, Igor
--
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-53 Fax: +49-8142-66989-80 Email: sbabic@denx.de =====================================================================
Thanks