[PATCH] spl: Jump to image at end of board_init_r

[Heinrich Schuchardt]

On 29.09.23 13:56, Tom Rini wrote:
> On Fri, Sep 29, 2023 at 02:57:42AM +0200, Heinrich Schuchardt wrote:
>> On 9/27/23 23:44, Jonas Karlman wrote:
>>> spl_board_prepare_for_boot() is not called before jumping/invoking atf,
>>> optee, opensbi or linux images.
>>>
>>> Jump to image at the end of board_init_r() to fix this.
>>>
>>> Signed-off-by: Jonas Karlman <jonas at kwiboo.se>
>>> ---
>>> This patch have dependencies on the following patches:
>>>
>>> spl: add __noreturn attribute to spl_invoke_opensbi function
>>> https://patchwork.ozlabs.org/patch/1827057/
>>>
>>> spl: add __noreturn attribute to spl_invoke_atf function
>>> https://patchwork.ozlabs.org/patch/1831366/
>>>
>>> spl: Drop the switch() statement for OS selection
>>> from the "spl: Preparation for Universal Payload" series
>>> https://patchwork.ozlabs.org/patch/1839731/
>>> ---
>>>    common/spl/spl.c | 12 +++++++-----
>>>    1 file changed, 7 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/common/spl/spl.c b/common/spl/spl.c
>>> index f7608f14e365..79c39820262a 100644
>>> --- a/common/spl/spl.c
>>> +++ b/common/spl/spl.c
>>> @@ -647,6 +647,8 @@ void board_init_r(gd_t *dummy1, ulong dummy2)
>>>    		BOOT_DEVICE_NONE,
>>>    		BOOT_DEVICE_NONE,
>>>    	};
>>> +	typedef void __noreturn (*jump_to_image_t)(struct spl_image_info *);
>>> +	jump_to_image_t jump_to_image = &jump_to_image_no_args;
>>>    	struct spl_image_info spl_image;
>>>    	int ret, os;
>>>
>>> @@ -735,20 +737,20 @@ void board_init_r(gd_t *dummy1, ulong dummy2)
>>>    	} else if (CONFIG_IS_ENABLED(ATF) && os == IH_OS_ARM_TRUSTED_FIRMWARE) {
>>>    		debug("Jumping to U-Boot via ARM Trusted Firmware\n");
>>>    		spl_fixup_fdt(spl_image_fdt_addr(&spl_image));
>>> -		spl_invoke_atf(&spl_image);
>>> +		jump_to_image = &spl_invoke_atf;
>>>    	} else if (CONFIG_IS_ENABLED(OPTEE_IMAGE) && os == IH_OS_TEE) {
>>>    		debug("Jumping to U-Boot via OP-TEE\n");
>>>    		spl_board_prepare_for_optee(spl_image_fdt_addr(&spl_image));
>>> -		jump_to_image_optee(&spl_image);
>>> +		jump_to_image = &jump_to_image_optee;
>>>    	} else if (CONFIG_IS_ENABLED(OPENSBI) && os == IH_OS_OPENSBI) {
>>>    		debug("Jumping to U-Boot via RISC-V OpenSBI\n");
>>> -		spl_invoke_opensbi(&spl_image);
>>> +		jump_to_image = &spl_invoke_opensbi;
>>>    	} else if (CONFIG_IS_ENABLED(OS_BOOT) && os == IH_OS_LINUX) {
>>>    		debug("Jumping to Linux\n");
>>>    		if (IS_ENABLED(CONFIG_SPL_OS_BOOT))
>>>    			spl_fixup_fdt((void *)SPL_PAYLOAD_ARGS_ADDR);
>>>    		spl_board_prepare_for_linux();
>>> -		jump_to_image_linux(&spl_image);
>>> +		jump_to_image = &jump_to_image_linux;
>>>    	} else {
>>>    		debug("Unsupported OS image.. Jumping nevertheless..\n");
>>>    	}
>>> @@ -788,7 +790,7 @@ void board_init_r(gd_t *dummy1, ulong dummy2)
>>>    	}
>>>
>>>    	spl_board_prepare_for_boot();
>>> -	jump_to_image_no_args(&spl_image);
>>> +	jump_to_image(&spl_image);
>>>    }
>>>
>>>    /*
>>
>> In SPL we are fighting for every byte of binary size.
>>
>> What is the impact of this change on the code size?
>>
>> I would expect that your increasing it; especially if only one of the
>> CONFIG_OPTIONS is enabled.
>>
>> If so, NAK to this patch despite all elegance.
>
> We aren't _that_ strict, no.  And a very quick peek shows that this
> seems fine overall.  Since you raised the question I'll do a quick
> world build but socfpga_agilex_vab (as a config I had size change
> results for in front of me for something else) shrank by 4 bytes with
> just the prerequsites and this patch applied.
>

Shrinking the size sounds great. Thanks for measuring.

Best regards

Heinrich