Re: [PATCH mm] arm64: kasan: fix MTE symbols exports

From: Andrew Morton <akpm@linux-foundation.org>
To: Catalin Marinas <catalin.marinas@arm.com>
Cc: Andrey Konovalov <andreyknvl@google.com>,
	Vincenzo Frascino <vincenzo.frascino@arm.com>,
	Will Deacon <will.deacon@arm.com>,
	Dmitry Vyukov <dvyukov@google.com>,
	Andrey Ryabinin <aryabinin@virtuozzo.com>,
	Alexander Potapenko <glider@google.com>,
	Marco Elver <elver@google.com>,
	Peter Collingbourne <pcc@google.com>,
	Evgenii Stepanov <eugenis@google.com>,
	Branislav Rankov <Branislav.Rankov@arm.com>,
	Kevin Brodsky <kevin.brodsky@arm.com>,
	Christoph Hellwig <hch@infradead.org>,
	kasan-dev@googlegroups.com, linux-arm-kernel@lists.infradead.org,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH mm] arm64: kasan: fix MTE symbols exports
Date: Tue, 9 Feb 2021 10:45:15 -0800	[thread overview]
Message-ID: <20210209104515.75eaa00dea03175e49e70d6c@linux-foundation.org> (raw)
In-Reply-To: <20210209170255.GG1435@arm.com>

On Tue, 9 Feb 2021 17:02:56 +0000 Catalin Marinas <catalin.marinas@arm.com> wrote:

> On Tue, Feb 09, 2021 at 04:32:30PM +0100, Andrey Konovalov wrote:
> > diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
> > index a66c2806fc4d..788ef0c3a25e 100644
> > --- a/arch/arm64/kernel/mte.c
> > +++ b/arch/arm64/kernel/mte.c
> > @@ -113,13 +113,17 @@ void mte_enable_kernel(void)
> >  	sysreg_clear_set(sctlr_el1, SCTLR_ELx_TCF_MASK, SCTLR_ELx_TCF_SYNC);
> >  	isb();
> >  }
> > +#if IS_ENABLED(CONFIG_KASAN_KUNIT_TEST)
> >  EXPORT_SYMBOL_GPL(mte_enable_kernel);
> > +#endif
> >  
> >  void mte_set_report_once(bool state)
> >  {
> >  	WRITE_ONCE(report_fault_once, state);
> >  }
> > +#if IS_ENABLED(CONFIG_KASAN_KUNIT_TEST)
> >  EXPORT_SYMBOL_GPL(mte_set_report_once);
> > +#endif
> 
> Do we actually care about exporting them when KASAN_KUNIT_TEST=n? It
> looks weird to have these #ifdefs in the arch code. Either the
> arch-kasan API requires these symbols to be exported to modules or not.
> I'm not keen on such kasan internals trickling down into the arch code.
> 
> If you don't want to export them in the KASAN_KUNIT_TEST=n case, add a
> wrapper in the kasan built-in code (e.g. kasan_test_enable_tagging,
> kasan_test_set_report_once) and conditionally compile them based on
> KASAN_KUNIT_TEST.

In other words, the patch's changelog was poor!  It told us what the
patch does (which is often obvious from the code) but it failed to
explain why the patch does what it does.

The same goes for code comments, folks - please explain "why it does
this" rather than "what it does".