[U-Boot] [PATCH 1/7] arm: K3: Avoid use of MCU_PSRAM0 before SYSFW is loaded

[Andrew F. Davis <afd@ti.com>]

On 2/13/19 9:32 PM, Lokesh Vutla wrote:
> 
> 
> On 14/02/19 12:07 AM, Andrew F. Davis wrote:
>> On HS devices the 512b region of reset isolated memory called
>> MCU_PSRAM0 is firewalled by default. Until SYSFW is loaded we
>> cannot use this memory. It is only used to store a single value
>> left at the end of SRAM by ROM that will be needed later. Save
>> that value to a global variable stored in the .data section.
>> This section is used as .bss will be cleared between saving
>> this value and using it.
>>
>> Signed-off-by: Andrew F. Davis <afd@ti.com>
>> ---
>>  arch/arm/mach-k3/am6_init.c                  | 8 +++-----
>>  arch/arm/mach-k3/include/mach/am6_hardware.h | 3 ---
>>  2 files changed, 3 insertions(+), 8 deletions(-)
>>
>> diff --git a/arch/arm/mach-k3/am6_init.c b/arch/arm/mach-k3/am6_init.c
>> index a5553190b4..462538e45d 100644
>> --- a/arch/arm/mach-k3/am6_init.c
>> +++ b/arch/arm/mach-k3/am6_init.c
>> @@ -49,11 +49,11 @@ static void ctrl_mmr_unlock(void)
>>  	mmr_unlock(CTRL_MMR0_BASE, 7);
>>  }
>>  
>> +u32 bootindex __attribute__((section(".data")));
> 
> After thinking a bit more I realized that backup bootmode might fail. R5 SPL is
> fine but when it jumps to A53 SPL it is trying to read the bootindex again from
> K3_BOOT_PARAM_TABLE_INDEX_VAL(this is already the case and is wrong). It is not
> guaranteed that R5 SPL did not damage this ROM param table. We should somehow
> pass bootindex to A53 SPL using scratchpad space.
> 

I don't know of any super good way to pass data from R5-SPL to A53-SPL,
but I don't think MCU_PSRAM0 will be the way to do it. For all we know
before A53-SPL has started the R5-SPL may have shutdown and something
else has booted, which could have used that MCU local RAM space. I think
we need something that gets passed to A53 anyway, maybe a DT fixup?

>> +
>>  static void store_boot_index_from_rom(void)
>>  {
>> -	u32 *boot_index = (u32 *)K3_BOOT_PARAM_TABLE_INDEX_VAL;
>> -
>> -	*boot_index = *(u32 *)(CONFIG_SYS_K3_BOOT_PARAM_TABLE_INDEX);
>> +	bootindex = *(u32 *)(CONFIG_SYS_K3_BOOT_PARAM_TABLE_INDEX);
>>  }
>>  
>>  void board_init_f(ulong dummy)
>> @@ -94,7 +94,6 @@ u32 spl_boot_mode(const u32 boot_device)
>>  {
>>  #if defined(CONFIG_SUPPORT_EMMC_BOOT)
>>  	u32 devstat = readl(CTRLMMR_MAIN_DEVSTAT);
>> -	u32 bootindex = readl(K3_BOOT_PARAM_TABLE_INDEX_VAL);
>>  
>>  	u32 bootmode = (devstat & CTRLMMR_MAIN_DEVSTAT_BOOTMODE_MASK) >>
>>  			CTRLMMR_MAIN_DEVSTAT_BOOTMODE_SHIFT;
>> @@ -170,7 +169,6 @@ static u32 __get_primary_bootmedia(u32 devstat)
>>  u32 spl_boot_device(void)
>>  {
>>  	u32 devstat = readl(CTRLMMR_MAIN_DEVSTAT);
>> -	u32 bootindex = readl(K3_BOOT_PARAM_TABLE_INDEX_VAL);
>>  
>>  	if (bootindex == K3_PRIMARY_BOOTMODE)
>>  		return __get_primary_bootmedia(devstat);
>> diff --git a/arch/arm/mach-k3/include/mach/am6_hardware.h b/arch/arm/mach-k3/include/mach/am6_hardware.h
>> index b5244609af..3343233aa3 100644
>> --- a/arch/arm/mach-k3/include/mach/am6_hardware.h
>> +++ b/arch/arm/mach-k3/include/mach/am6_hardware.h
>> @@ -44,7 +44,4 @@
>>  #define CTRLMMR_LOCK_KICK1				0x0100c
>>  #define CTRLMMR_LOCK_KICK1_UNLOCK_VAL			0xd172bc5a
>>  
>> -/* MCU SCRATCHPAD usage */
>> -#define K3_BOOT_PARAM_TABLE_INDEX_VAL	CONFIG_SYS_K3_MCU_SCRATCHPAD_BASE
> 
> Can we move scratchpad base to a different location ? This will be used for
> eeprom storing also.
> 

It's okay to use MCU_PSRAM0 after SYSFW has been loaded, does the EEPROM
data need to be moved between SPLs? Or is it local to each?

Andrew

> Thanks and regards,
> Lokesh
> 
>> -
>>  #endif /* __ASM_ARCH_AM6_HARDWARE_H */
>>