Re: [PATCH v2 1/2] tpm: Disable a TPM that stops responding

From: Jarkko Sakkinen

Date: Thu Nov 12 2026 - 19:08:02 EST


On Thu, Nov 12, 2026 at 12:40:45PM -0800, Breno Leitao wrote:
> On Thu, Nov 12, 2026 at 04:03:06PM +0200, Jarkko Sakkinen wrote:
> > >
> > > This would be then nr_lost_responses.
> >
> > Oops sorry.
> >
> > I mean nr_missed_responses (and I would prefer "missed" over "lost").
>
> Would you hate if I go with max_missed_responses?

It's a great question. Here I think asymmetry defines things better,
and makes sense also few months from now.

So I think this would be most consistent:

1. kernel command-line: tpm_max_missed_responses
2. tpm_chip: nr_missed_responses.

The first defines the modulo and second counts up to that.

Can't recall (and too lazy to check) how it exactly was in your code
change but I would formalize all this along the lines of:

if (!(++chip->nr_missed_responses % tpm_max_missed_responses)) {
/* chip->flags |= TPM2_CHIP_FLAG_DISABLE etc. and of course bail out. */
}

/* continue doing things */

[those comments are not part of the proposal]

Br, Jarkko