Re: [PATCH 6.1 000/194] 6.1.47-rc1 review

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Thu, Aug 24, 2023 at 08:58:46AM -0700, Guenter Roeck wrote:
> On Thu, Aug 24, 2023 at 05:15:30PM +0200, Greg Kroah-Hartman wrote:
> [ ... ]
> > 
> > This might be an issue on AMD chips, but for some reason, in running
> > this kernel on my systems here, I have no boot warnings at all.  I
> > blamed it on them being only AMD chips.  If that's not the issue then I
> > really have no idea, sorry.
> > 
> 
> AFAIK it depends on the compiler used and on the enabled debug options.
> At some point I lost track about what exactly has to be enabled to see
> the problem. We see it a lot with our (ChromeOS) debug kernel testing
> which has various debug options enabled, to the point where we may
> temporarily remove the warning from the kernel to stop the noise.
> 
> Alternatively, as I had suggested earlier, we could revert all srso patches
> because they only seem to cause trouble. So far I have been resisting that
> because I am concerned that it would make things even worse (like one can
> not really revert a heart surgery without doing even more damage).

If you don't want the feature, you can disable it at build time (I
think), and I know you can disable it at runtime.  But there are groups
out that that "need" this, to solve a known security hole in some cpus,
so it is required for them.  To remove the code entirely would not be
good for them, so I will say, if they are the ones that need this, they
should help fix this warning, no one else should be seeing it :)

thanks,

greg k-h