Re: 6.17 crashes in ipv6 code when booted fips=1 [was: [GIT PULL] Crypto Update for 6.17]

[Vegard Nossum]

On 06/10/2025 18:19, Linus Torvalds wrote:
> I think the other way of writing that is "fips=1 is and will remain 
> irrelevant in the real world as long as it's that black-and-white".

On 06/10/2025 19:11, Linus Torvalds wrote:
> On Mon, 6 Oct 2025 at 09:32, Vegard Nossum <vegard.nossum@xxxxxxxxxx> wrote:
> > Okay, so I get that we don't like fips=1 around here (I'm not a
> > particularly big fan myself), but what's with the snark? fips=1 exists
> > in mainline and obviously has users. I'm just trying to make sure it
> > remains useful and usable.
>
> It literally caused non-bootable machines because of that allegedly
> "remains useful and usable" because it changed something that never
> failed to failing. That's how this thread started.
>
> So that's why the snark. I think you are deluding yourself and others
> if you call that "useful and usable".

Yes, thank you, I've already acknowledged that my patch caused boot
failures and I apologize for that unintentional breakage. Why does this
mean we should throw fips=1 in the bin, though? That's a total non sequitur.

The fact is that fips=1 is not useful if it doesn't actually result
something that complies with the standard; the only purpose of fips=1 is
to allow the kernel to be used and certified as a FIPS module. But as
SHA-1 is still allowed for now, please go ahead and revert my patch,
it's commit 9d50a25eeb05c.


Vegard