[PATCH 3/5] firmware: Add RW section offset in scratch

Jessica Clarke jrtc27 at jrtc27.com
Wed Jan 18 20:43:35 PST 2023


On 19 Jan 2023, at 04:29, Himanshu Chauhan <hchauhan at ventanamicro.com> wrote:
> 
> On Wed, Jan 18, 2023 at 04:23:06PM +0000, Jessica Clarke wrote:
>> On 9 Jan 2023, at 08:40, Himanshu Chauhan <hchauhan at ventanamicro.com> wrote:
>>> 
>>> Add the RW section offset, provided by _fw_rw_offset symbol,
>>> to the scratch structure. This will be used to program
>>> separate pmp entry for RW section.
>>> 
>>> Signed-off-by: Himanshu Chauhan <hchauhan at ventanamicro.com>
>>> ---
>>> firmware/fw_base.S        |  5 +++++
>>> include/sbi/sbi_scratch.h | 24 ++++++++++++++----------
>>> 2 files changed, 19 insertions(+), 10 deletions(-)
>>> 
>>> diff --git a/firmware/fw_base.S b/firmware/fw_base.S
>>> index 3f622b3..ce1f782 100644
>>> --- a/firmware/fw_base.S
>>> +++ b/firmware/fw_base.S
>>> @@ -298,6 +298,11 @@ _scratch_init:
>>> 	sub	a5, t3, a4
>>> 	REG_S	a4, SBI_SCRATCH_FW_START_OFFSET(tp)
>>> 	REG_S	a5, SBI_SCRATCH_FW_SIZE_OFFSET(tp)
>>> +
>>> +	/* Store R/W section's offset in scratch space */
>>> +	lla	a4, _fw_rw_offset
>>> +	REG_S	a4, SBI_SCRATCH_FW_RW_OFFSET(tp)
>>> +
>> 
>> You can’t use LLA for an absolute symbol, especially one whose value
>> isn’t guaranteed to be within 2^31 of the LLA’s address. Use the GOT,
>> use a constant pool (for compatibility with older assemblers) or do the
>> subtraction in the code instead of the linker script.
> 
> This code segment is executed after the relocation. The _fw_rw_offset
> symbol also gets relocated.

No it doesn’t. _fw_rw_offset is absolute. It’s not subject to
relocation. In fact executing this *after* relocation would totally
screw up FW_PIC binaries, because they’d include the run-time
relocation offset in the calculation of _fw_rw_offset, but
_fw_rw_offset is the offset from _fw_base. That is, if your relocation
offset (FreeBSD calls this “relocbase”) is 1 GiB then suddenly
_fw_rw_offset is regarded as 1 GiB + some small value from the start of
OpenSBI’s image, which seems rather nonsensical, when it should be just
that small value.

> Thus the symbol will always be in 2Gig range
> unless the text and RO-data go beyond that size.

That’s not even guaranteed at link time if you have a base address
beyond 2 GiB. It only works with the default base address of 2 GiB
because this lla is in the text segment and thus at a smaller offset
from the start of the file than _fw_rw_offset’s value. For example,
with a base address of 2 GiB, a 4 KiB text segment, and, for argument’s
sake, 32 uncompressed instructions before this lla in the text segment,
_fw_rw_offset will have value 4 KiB, and the lla will be at 2 GiB + 32
* 4 bytes = 2 GiB + 128 bytes, which is less than 2 GiB above 4 KiB.
But if you bump the base address up to 3 GiB then the lla is at 3 GiB +
128 bytes, which is more than 2 GiB above 4 KiB.

Use the GOT or a constant pool.

Jess

> IMHO, I don't see a need
> to go via GOT.
> 
> Regards
> Himanshu
> 
>> Jess
>> 
>>> 	/* Store next arg1 in scratch space */
>>> 	MOV_3R	s0, a0, s1, a1, s2, a2
>>> 	call	fw_next_arg1
>>> diff --git a/include/sbi/sbi_scratch.h b/include/sbi/sbi_scratch.h
>>> index 40a3bc9..2966188 100644
>>> --- a/include/sbi/sbi_scratch.h
>>> +++ b/include/sbi/sbi_scratch.h
>>> @@ -18,26 +18,28 @@
>>> #define SBI_SCRATCH_FW_START_OFFSET		(0 * __SIZEOF_POINTER__)
>>> /** Offset of fw_size member in sbi_scratch */
>>> #define SBI_SCRATCH_FW_SIZE_OFFSET		(1 * __SIZEOF_POINTER__)
>>> +/** Offset (in sbi_scratch) of the R/W Offset */
>>> +#define SBI_SCRATCH_FW_RW_OFFSET		(2 * __SIZEOF_POINTER__)
>>> /** Offset of next_arg1 member in sbi_scratch */
>>> -#define SBI_SCRATCH_NEXT_ARG1_OFFSET		(2 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_NEXT_ARG1_OFFSET		(3 * __SIZEOF_POINTER__)
>>> /** Offset of next_addr member in sbi_scratch */
>>> -#define SBI_SCRATCH_NEXT_ADDR_OFFSET		(3 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_NEXT_ADDR_OFFSET		(4 * __SIZEOF_POINTER__)
>>> /** Offset of next_mode member in sbi_scratch */
>>> -#define SBI_SCRATCH_NEXT_MODE_OFFSET		(4 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_NEXT_MODE_OFFSET		(5 * __SIZEOF_POINTER__)
>>> /** Offset of warmboot_addr member in sbi_scratch */
>>> -#define SBI_SCRATCH_WARMBOOT_ADDR_OFFSET	(5 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_WARMBOOT_ADDR_OFFSET	(6 * __SIZEOF_POINTER__)
>>> /** Offset of platform_addr member in sbi_scratch */
>>> -#define SBI_SCRATCH_PLATFORM_ADDR_OFFSET	(6 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_PLATFORM_ADDR_OFFSET	(7 * __SIZEOF_POINTER__)
>>> /** Offset of hartid_to_scratch member in sbi_scratch */
>>> -#define SBI_SCRATCH_HARTID_TO_SCRATCH_OFFSET	(7 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_HARTID_TO_SCRATCH_OFFSET	(8 * __SIZEOF_POINTER__)
>>> /** Offset of trap_exit member in sbi_scratch */
>>> -#define SBI_SCRATCH_TRAP_EXIT_OFFSET		(8 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_TRAP_EXIT_OFFSET		(9 * __SIZEOF_POINTER__)
>>> /** Offset of tmp0 member in sbi_scratch */
>>> -#define SBI_SCRATCH_TMP0_OFFSET			(9 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_TMP0_OFFSET			(10 * __SIZEOF_POINTER__)
>>> /** Offset of options member in sbi_scratch */
>>> -#define SBI_SCRATCH_OPTIONS_OFFSET		(10 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_OPTIONS_OFFSET		(11 * __SIZEOF_POINTER__)
>>> /** Offset of extra space in sbi_scratch */
>>> -#define SBI_SCRATCH_EXTRA_SPACE_OFFSET		(11 * __SIZEOF_POINTER__)
>>> +#define SBI_SCRATCH_EXTRA_SPACE_OFFSET		(12 * __SIZEOF_POINTER__)
>>> /** Maximum size of sbi_scratch (4KB) */
>>> #define SBI_SCRATCH_SIZE			(0x1000)
>>> 
>>> @@ -53,6 +55,8 @@ struct sbi_scratch {
>>> 	unsigned long fw_start;
>>> 	/** Size (in bytes) of firmware linked to OpenSBI library */
>>> 	unsigned long fw_size;
>>> +	/** Offset (in bytes) of the R/W section */
>>> +	unsigned long fw_rw_offset;
>>> 	/** Arg1 (or 'a1' register) of next booting stage for this HART */
>>> 	unsigned long next_arg1;
>>> 	/** Address of next booting stage for this HART */
>>> -- 
>>> 2.39.0
>>> 
>>> 
>>> -- 
>>> opensbi mailing list
>>> opensbi at lists.infradead.org
>>> http://lists.infradead.org/mailman/listinfo/opensbi




More information about the opensbi mailing list