On Fri, Jan 6, 2023 at 6:04 PM Luigi Semenzato semenzato@chromium.org wrote:
I worked a fair amount on TPM 1.0 about 10 years ago and I even vaguely remember suspend-related problems. I'd be happy to take a look. The linked thread shows that Peter Huewe was copied. I know Peter well, his opinion can be trusted. Unfortunately I don't immediately see a link to a patch, can you help?
Sorry, I should have included that: https://lore.kernel.org/lkml/20230106030156.3258307-1-Jason@zx2c4.com/ Instead of blocking system suspend when TPM_ORD_SAVESTATE fails, it just lets the system sleep anyway. This means that presumably the system might sleep without having called TPM_ORD_SAVESTATE. Trying to figure out how bad that is.
And yes, Peter Huewe certainly knows about TPMs, especially as he maintains the code in Linux, but the maintainers haven't been so much available, unfortunately. This bug happens to intersect with something mostly related that I work on (the rng), so I'm motivated to at least prevent the worst of the breakage, but I otherwise don't know anything about the Linux TPM driver.
Jason