Re: [PATCH v3 03/13] PCI/TSM: Authenticate devices via platform TSM

[<dan.j.williams@intel.com>]

Alexey Kardashevskiy wrote:
[..]
[trim multiple pages of uncommented context, please trim your replies]

> > +/**
> > + * pci_tsm_pf0_initialize() - common 'struct pci_tsm_pf0' initialization
> > + * @pdev: Physical Function 0 PCI device
> > + * @tsm: context to initialize
> > + */
> > +int pci_tsm_pf0_initialize(struct pci_dev *pdev, struct pci_tsm_pf0 *tsm)
> 
> Here it is: struct pci_tsm_pf0 *tsm  (it is really a "dsm")

All of the context returned by the TSM driver is a "tsm" context, the
only time use "dsm" is in referring to the actual pci_dev that runs the
SPDM session.

> In pci_tsm: struct pci_dev *dsm (alright)
> 
> May be we need some distinction between PF0's pci_dev and pci_tsm_pf0 but still these are DSMs.
> 
> In pci_tsm_pf0 it is: struct pci_tsm tsm, imho "base" is less confusing (I keep catching myself thinking it is a pointer to tsm_dev).

Ok, I can change it to base.

> "tsm" would be what you call "tsm_dev" which is ok but seeing short "tsm" used as "dsm" or "TSM data for this pci_dev" is confusing.
> 
> s/pci_tsm/pci_tsm_ctx/ and s/tsm/tsm_ctx/ ? Thanks,

What is a tsm_ctx? The s/pci_tsm/pci_tsm_ctx/ rename is not adding more
clarity for me. If Aneesh or Yilun also find that rename clarifying I
will switch. For v5 I will stick with 'struct pci_tsm' as the PCI object
that wraps TSM driver produced objects.

The reason I do not think of "pci_tsm" as a "tsm_dev" is because PCI is
always a consumer of an outside of PCI TSM service provided, PCI does
not have a TSM concept internal to it.