[PATCH v2 04/34] ARM: put scratch mem area below OP-TEE

[Ahmad Fatoum]

On 17.05.23 15:14, Sascha Hauer wrote:
> On Wed, May 17, 2023 at 02:48:43PM +0200, Ahmad Fatoum wrote:
>> On 17.05.23 11:03, Sascha Hauer wrote:
>>> We want to reserve memory for OP-TEE at the end of available SDRAM,
>>> so move the scratch area below OP-TEE and not above.
>>>
>>> Signed-off-by: Sascha Hauer <s.hauer at pengutronix.de>
>>> ---
>>>  arch/arm/include/asm/barebox-arm.h | 4 ++--
>>>  1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/arch/arm/include/asm/barebox-arm.h b/arch/arm/include/asm/barebox-arm.h
>>> index f446044be6..6e6606d005 100644
>>> --- a/arch/arm/include/asm/barebox-arm.h
>>> +++ b/arch/arm/include/asm/barebox-arm.h
>>> @@ -71,14 +71,14 @@ static inline void arm_fixup_vectors(void)
>>>  
>>>  void *barebox_arm_boot_dtb(void);
>>>  
>>> -#define arm_mem_scratch(endmem) ((endmem) - SZ_32K)
>>> +#define arm_mem_scratch(endmem) ((endmem) - OPTEE_SIZE - SZ_32K)
>>>  
>>>  static inline const void *arm_mem_scratch_get(void)
>>>  {
>>>  	return (const void *)arm_mem_scratch(arm_mem_endmem_get());
>>>  }
>>>  
>>> -#define arm_mem_stack_top(endmem) ((endmem) - SZ_64K - OPTEE_SIZE)
>>> +#define arm_mem_stack_top(endmem) ((endmem) - SZ_64K)
>>
>> I don't understand why you drop OPTEE_SIZE here. Wouldn't the stack
>> now eat into the OP-TEE region?
> 
> I accidently thought that arm_mem_stack_top() is calculated based on the
> region above it, namely arm_mem_scratch(), but really it's calculated
> based on endmem directly.
> 
> Indeed it's wrong like this, it should be:
> 
> #define arm_mem_stack_top(endmem) (arm_mem_scratch(endmem) - SZ_64K)
> 
> I just stumbled upon the SZ_64K here. I followed the value back to 2016
> and found 75c96bd2459e ("ARM: Do not use last 64KiB of address space for
> barebox"). I had a board that time that has SDRAM at the very end of the
> 32bit address space. On that board it happened that we overwrite parts
> of the lowlevel memory with the vector table. It seems that has been
> lost over time as now we put the scratch space and possibly parts of
> OP-TEE into the last 64k.

Scratch space at -SZ_32K leaves 32K for IVT, which should be enough, no?

> 
> Sascha
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |