Re: linux-next: build failure after merge of the amdgpu tree

[Simon Ser]

On Monday, October 11th, 2021 at 09:43, Christoph Hellwig <hch@xxxxxx> wrote:

> On Mon, Oct 11, 2021 at 07:39:52AM +0000, Simon Ser wrote:
> > I don't understand. Can you elaborate why you think this commit is
> > "utter crap"?
>
> A kernel driver has absolutely no business making decissions based
> on current->comm, which can be changed by any userspace process.  This
> is kernel programming 101.
>
> Independent of that a check for a specific program as the callers makes
> no sense whatsoever as a given program and change over time.  This is
> not even something kernel specific but something that ever software
> engineer should do.

Have you heard about the kernel no-regression rule? Here, we can't enable a new
feature because that would regress user-space which mis-uses the kernel uAPI.

This isn't unheard of. Core drm already detects Xorg with current->comm, and
force-disables atomic KMS.

> > I'd also appreciate if you could be a bit less aggressive. There's
> > nothing "obvious" about this from my point of view.
>
> I'm not agressive.  I'm just really disappointed by the amoubt of crap
> that gets shovelled into the kernel and even more disappointed by the
> abslutely lack of knowledge of some of the contributors.

If your reply wasn't aggressive, I don't know what it is.