On 7/29/25 06:05, Marek Vasut wrote:
> On 6/29/25 9:08 PM, Marek Vasut wrote:
>
> Hi,
>
>>>> diff --git a/arch/arm/mach-stm32mp/Kconfig.13x b/arch/arm/mach-
>>>> stm32mp/Kconfig.13x
>>>> index bc8b3f8cf77..cecf9e3b8c7 100644
>>>> --- a/arch/arm/mach-stm32mp/Kconfig.13x
>>>> +++ b/arch/arm/mach-stm32mp/Kconfig.13x
>>>> @@ -20,7 +20,8 @@ config TARGET_ST_STM32MP13X
>>>> endchoice
>>>> config TEXT_BASE
>>>> - default 0xC0000000
>>>> + default 0xC0000000 if TFABOOT
>>>> + default 0xC0100000 if !TFABOOT
>>>> config PRE_CON_BUF_ADDR
>>>> default 0xC0800000
>>>
>>>
>>> I don't understood why you modify TEXT_BASE here.
>> Loading of data to the very beginning of DRAM may fail in case the data are
>> wrapped in various container image types, because the image loader in SPL
>> may use the area before the data load address to load for example image
>> header(s) in there. This is one example of such header loading, which is
>> triggered on STM32MP13xx in SPL:
>>
>> https://source.denx.de/u-boot/u-boot/-/blob/master/include/spl_load.h#L18
>>
>> "
>> 18 struct legacy_img_hdr *header =
>> 19 spl_get_load_buffer(-sizeof(*header), sizeof(*header));
>> 20 ulong base_offset, image_offset, overhead;
>> 21 int read, ret;
>> 22
>> 23 log_debug("\nloading hdr from %lx to %p\n", (ulong)offset,
>> header);
>> 24 read = info->read(info, offset, ALIGN(sizeof(*header),
>> 25 spl_get_bl_len(info)), header);
>> "
>>
>> The header is loaded at CONFIG_TEXT_BASE - sizeof(*header). If
>> CONFIG_TEXT_BASE is start of DRAM with no usable memory right below it AND
>> sizeof(*header) is non-zero, then this loading will fail.
>>
>> That's why the TEXT_BASE is shifted a bit higher in this patch.
>
> Any news on this series ? It would be good to move this along.
>
Hi Marek
Sorry for the delay, i am back from vacation. I will prepare pull request soon.
Thanks
Patrice
