Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220722171534.3576419-1-dlatypov@google.com> <20220722171534.3576419-3-dlatypov@google.com>
In-Reply-To: <20220722171534.3576419-3-dlatypov@google.com>
From:   Brendan Higgins <brendanhiggins@google.com>
Date:   Wed, 5 Oct 2022 16:36:15 -0400
Message-ID: <CAFd5g450Jgxu=HYVpj8YmrtF9Hd3K9fuTHBp-Q5ZK+w=YVxO+w@mail.gmail.com>
Subject: Re: [PATCH v2 3/5] kunit: make kunit_kfree() only work on pointers
 from kunit_malloc() and friends
To:     Daniel Latypov <dlatypov@google.com>
Cc:     davidgow@google.com, linux-kernel@vger.kernel.org,
        kunit-dev@googlegroups.com, linux-kselftest@vger.kernel.org,
        skhan@linuxfoundation.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Jul 22, 2022 at 1:15 PM Daniel Latypov <dlatypov@google.com> wrote:
>
> kunit_kfree() exists to clean up allocations from kunit_kmalloc() and
> friends early instead of waiting for this to happen automatically at the
> end of the test.
>
> But it can be used on *anything* registered with the kunit resource API.
>
> E.g. the last 2 statements are equivalent:
>   struct kunit_resource *res = something();
>   kfree(res->data);
>   kunit_put_resource(res);
>
> The problem is that there could be multiple resources that point to the
> same `data`.
>
> E.g. you can have a named resource acting as a pseudo-global variable in
> a test. If you point it to data allocated with kunit_kmalloc(), then
> calling `kunit_kfree(ptr)` has the chance to delete either the named
> resource or to kfree `ptr`.
> Which one it does depends on the order the resources are registered as
> kunit_kfree() will delete resources in LIFO order.
>
> So this patch restricts kunit_kfree() to only working on resources
> created by kunit_kmalloc(). Calling it is therefore guaranteed to free
> the memory, not do anything else.
>
> Note: kunit_resource_instance_match() wasn't used outside of KUnit, so
> it should be safe to remove from the public interface. It's also
> generally dangerous, as shown above, and shouldn't be used.
>
> Signed-off-by: Daniel Latypov <dlatypov@google.com>
> Reviewed-by: David Gow <davidgow@google.com>

Reviewed-by: Brendan Higgins <brendanhiggins@google.com>