Re: [PATCH v1 4/8] xen/riscv: introduce sbi call to putchar to console

[Julien Grall <julien@xen.org>]

Hi Oleksii,

On 09/01/2023 09:04, Oleksii wrote:
> On Fri, 2023-01-06 at 13:40 +0000, Julien Grall wrote:
>> Hi,
>>
>> On 06/01/2023 13:14, Oleksii Kurochko wrote:
>>> The patch introduce sbi_putchar() SBI call which is necessary
>>> to implement initial early_printk
>>>
>>> Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
>>> ---
>>>    xen/arch/riscv/Makefile          |  1 +
>>>    xen/arch/riscv/include/asm/sbi.h | 34 ++++++++++++++++++++++++
>>>    xen/arch/riscv/sbi.c             | 44
>>> ++++++++++++++++++++++++++++++++
>>
>> IMHO, it would be better to implement sbi.c in assembly so you can
>> use
>> print in the console before you jump to C world.
>>
> I thought that we can live with C version as we set up stack from the
> start and then we can call early_printk() from assembly code too.
> Is it bad approach?

It depends on how early you want to call it. For Arm, we chose to use 
assembly because the C code may not be PIE (and even with PIE it may 
need some relocation work).

Andrew suggested that this may not be a problem with RISC-V. I have 
looked a bit more around and notice that the kernel is also calling some 
C function very early (like setup_vm()). But they ensure that the code 
is built with -mcmodel=medany.

It looks like you are already building Xen with this option. So all 
looks good for RISC-V. That said, I would suggest to check that 
__riscv_cmodel_medany is defined in files where you implement C function 
called from early assembly code.

Cheers,

-- 
Julien Grall