Re: [PATCH v2] tpm: tpm_ibm_vtpm: Fix unallocated banks

[Jarkko Sakkinen]

On Mon, Jul 08, 2019 at 06:24:04PM -0400, Mimi Zohar wrote:
> > static int tpm_get_pcr_allocation(struct tpm_chip *chip)
> > {
> > 	int rc;
> > 
> > 	rc = (chip->flags & TPM_CHIP_FLAG_TPM2) ?
> >      	     tpm2_get_pcr_allocation(chip) :
> >      	     tpm1_get_pcr_allocation(chip);
> 
> > 
> > 	return rc > 0 ? -ENODEV : rc;
> > }
> > 
> > This addresses the issue that Stefan also pointed out. You have to
> > deal with the TPM error codes.
> 
> Hm, in the past I was told by Christoph not to use the ternary
> operator.  Have things changed?  Other than removing the comment, the
> only other difference is the return.

Lets purge the snippet:

rc = (chip->flags & TPM_CHIP_FLAG_TPM2) ?
     tpm2_get_pcr_allocation(chip) :
     tpm1_get_pcr_allocation(chip);

In this statement ternary operator does make sense because it is the
most readable way to express what is going on. We assign something
selecting one of the two options as the value  of rc based on a
condition.

It is like a natural fit for a ternary-operation and less messy than two
assigment statements.

On the other hand:

return rc > 0 ? -ENODEV : rc;

Here a better form would definitely be:

if (rc > 0)
	return - ENODEV;

return rc;

It is just two hard to grasp the logic when ternary operation is used.

Total ban of any language construct would be just utterly stupid. I
would instead use common sense here.

/Jarkko