On Wed, 11 Dec 2024 at 16:00, Jason Gunthorpe <jgg@ziepe.ca> wrote:
>
> On Wed, Dec 11, 2024 at 09:19:04AM +0100, Stefano Garzarella wrote:
>
> > > After that, there is no meaningful shared code here, and maybe the
> > > TPM_CHIP_FLAG_IRQ hack can be avoided too.
> >
> > IIUC you are proposing the following steps:
> > - extend tpm_class_ops to add a new send_recv() op and use it in
> > tpm_try_transmit()
>
> Yes, that seems to be the majority of your shared code.
>
> > - call the code in tpm_platform_probe() directly in sev
>
> Yes

I tried this, it's not bad, but I have a problem that I'm not sure how 
to solve. Basically, the functions used in tpm_platform_probe() (e.g. 
tpmm_chip_alloc, tpm2_probe, tpm_chip_register) are all defined in 
drivers/char/tpm/tpm.h
And in fact all users are in drivers/char/tpm.

So to use them directly in sev, we would have to move these definitions 
into include/linux/tpm.h or some other file in inlcude/. Is this 
acceptable for TPM maintainers?

Otherwise we need an intermediate module in drivers/char/tpm. Here we 
have 2 options:
1. continue as James did by creating a platform_device.
2. or we could avoid this by just exposing a registration API invoked by 
sev to specify the send_recv() callback to use. I mean something like 
renaming tpm_platform_probe() in tpm_platform_register(), and call it in 
snp_init_platform_device().

WDYT?

Thanks,
Stefano