Re: [PATCH v3 05/10] iov_iter: Create a function to prepare userspace VM for UBUF/IOVEC tests

From: Linus Torvalds <torvalds@linux-foundation.org>
To: David Howells <dhowells@redhat.com>
Cc: Christian Brauner <christian@brauner.io>,
	Jens Axboe <axboe@kernel.dk>, Al Viro <viro@zeniv.linux.org.uk>,
	Christoph Hellwig <hch@lst.de>,
	David Laight <David.Laight@aculab.com>,
	Matthew Wilcox <willy@infradead.org>,
	Brendan Higgins <brendanhiggins@google.com>,
	David Gow <davidgow@google.com>,
	linux-fsdevel@vger.kernel.org, linux-block@vger.kernel.org,
	linux-mm@kvack.org, netdev@vger.kernel.org,
	linux-kselftest@vger.kernel.org, kunit-dev@googlegroups.com,
	linux-kernel@vger.kernel.org,
	Andrew Morton <akpm@linux-foundation.org>,
	Christian Brauner <brauner@kernel.org>,
	David Hildenbrand <david@redhat.com>,
	John Hubbard <jhubbard@nvidia.com>,
	Huacai Chen <chenhuacai@kernel.org>,
	WANG Xuerui <kernel@xen0n.name>,
	Heiko Carstens <hca@linux.ibm.com>,
	Vasily Gorbik <gor@linux.ibm.com>,
	Alexander Gordeev <agordeev@linux.ibm.com>,
	Christian Borntraeger <borntraeger@linux.ibm.com>,
	Sven Schnelle <svens@linux.ibm.com>,
	loongarch@lists.linux.dev, linux-s390@vger.kernel.org
Subject: Re: [PATCH v3 05/10] iov_iter: Create a function to prepare userspace VM for UBUF/IOVEC tests
Date: Wed, 15 Nov 2023 11:58:48 -0500	[thread overview]
Message-ID: <CAHk-=whEj_+oP0mwNr7eArnOzWf_380-+-6LD9RtQXVs29fYJQ@mail.gmail.com> (raw)
In-Reply-To: <3936726.1700066370@warthog.procyon.org.uk>

On Wed, 15 Nov 2023 at 11:39, David Howells <dhowells@redhat.com> wrote:
>
> I was trying to make it possible to do these tests before starting userspace
> as there's a good chance that if the UBUF/IOVEC iterators don't work right
> then your system can't be booted.

Oh, I don't think that any unit test should bother to check for that
kind of catastrophic case.

If something is so broken that the kernel doesn't boot properly even
into some basic test infrastructure, then bisection will trivially
find where that breakage was introduced.

And if it's something as core as the iov iterators, it won't even get
past the initial developer unless it's some odd build system
interaction.

So extreme cases aren't even worth checking for. What's worth testing
is "the system boots and works, but I want to check the edge cases".

IOW, when it comes to things like user copies, it's things like
alignment, and the page fault edge cases with EFAULT in particular.
You can easily get the return value wrong for a user copy that ends up
with an unaligned fault at the end of the last mapped page. Everything
normal will still work fine, because nobody does something that odd.

But those are best handled as user mode tests.

           Linus