Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210913081424.48613-1-vincenzo.frascino@arm.com>
 <20210913081424.48613-6-vincenzo.frascino@arm.com> <CANpmjNN5atO1u6+Y71EiEvr9V8+WhdOGzC_8gvviac+BDkP+sA@mail.gmail.com>
 <f789ede2-3fa2-8a50-3d82-8b2dc2f12386@arm.com>
In-Reply-To: <f789ede2-3fa2-8a50-3d82-8b2dc2f12386@arm.com>
From:   Andrey Konovalov <andreyknvl@gmail.com>
Date:   Sun, 3 Oct 2021 19:16:04 +0200
Message-ID: <CA+fCnZe-gogW1yMuiHhXmKXTsmfkb+-iWp1Vf9K6ZY9madtxfw@mail.gmail.com>
Subject: Re: [PATCH 5/5] kasan: Extend KASAN mode kernel parameter
To:     Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc:     Marco Elver <elver@google.com>,
        Linux ARM <linux-arm-kernel@lists.infradead.org>,
        LKML <linux-kernel@vger.kernel.org>,
        kasan-dev <kasan-dev@googlegroups.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        Dmitry Vyukov <dvyukov@google.com>,
        Andrey Ryabinin <aryabinin@virtuozzo.com>,
        Alexander Potapenko <glider@google.com>,
        Evgenii Stepanov <eugenis@google.com>,
        Branislav Rankov <Branislav.Rankov@arm.com>,
        Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Sep 20, 2021 at 9:46 AM Vincenzo Frascino
<vincenzo.frascino@arm.com> wrote:
>
> On 9/16/21 12:43 PM, Marco Elver wrote:
> >
> > Shouldn't kasan.h also define kasan_asymm_mode_enabled() similar to
> > kasan_async_mode_enabled()?
> >
> > And based on that, also use it where kasan_async_mode_enabled() is
> > used in tests to ensure the tests do not fail. Otherwise, there is no
> > purpose for kasan_flag_asymm.
> >
>
> I was not planning to have the tests shipped as part of this series, they will
> come in a future one.
>
> For what concerns kasan_flag_asymm, I agree with you it is meaningful only if
> the tests are implemented hence I will remove it in v2.

Hi Vincenzo,

Up till now, the code assumes that not having the async mode enabled
means that the sync mode is enabled. There are two callers to
kasan_async_mode_enabled(): lib/test_kasan.c and mm/kasan/report.c.
Assuming tests support will be added later, at least the second one
should be adjusted.

Maybe we should rename kasan_async_mode_enabled() to
kasan_async_fault_possible(), make it return true for both async and
asymm modes, and use that in mm/kasan/report.c. And also add
kasan_sync_fault_possible() returning true for sync and asymm, and use
that in lib/test_kasan.c. (However, it seems that the tests don't work
with async faults right now.)

Thanks!