Re: [RFT PATCH v3 21/27] tty: serial: samsung_tty: IRQ rework

From: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
To: Arnd Bergmann <arnd@kernel.org>
Cc: Hector Martin <marcan@marcan.st>,
	Andy Shevchenko <andy.shevchenko@gmail.com>,
	linux-arm Mailing List <linux-arm-kernel@lists.infradead.org>,
	Marc Zyngier <maz@kernel.org>, Rob Herring <robh@kernel.org>,
	Olof Johansson <olof@lixom.net>,
	Mark Kettenis <mark.kettenis@xs4all.nl>,
	Tony Lindgren <tony@atomide.com>,
	Mohamed Mediouni <mohamed.mediouni@caramail.com>,
	Stan Skowronek <stan@corellium.com>,
	Alexander Graf <graf@amazon.com>, Will Deacon <will@kernel.org>,
	Linus Walleij <linus.walleij@linaro.org>,
	Mark Rutland <mark.rutland@arm.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Jonathan Corbet <corbet@lwn.net>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Christoph Hellwig <hch@infradead.org>,
	"David S. Miller" <davem@davemloft.net>,
	devicetree <devicetree@vger.kernel.org>,
	"open list:SERIAL DRIVERS" <linux-serial@vger.kernel.org>,
	Linux Documentation List <linux-doc@vger.kernel.org>,
	Linux Samsung SOC <linux-samsung-soc@vger.kernel.org>,
	Linux-Arch <linux-arch@vger.kernel.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [RFT PATCH v3 21/27] tty: serial: samsung_tty: IRQ rework
Date: Sun, 7 Mar 2021 20:51:03 +0100	[thread overview]
Message-ID: <20a8ce3e-bffa-4fc0-1f1d-185c04be8d9f@canonical.com> (raw)
In-Reply-To: <CAK8P3a0o4NHjXZ+ePj_Xpcw6ZmonoiR1dfkcsv=3i1JBEF4arA@mail.gmail.com>

On 07/03/2021 17:01, Arnd Bergmann wrote:
> On Sun, Mar 7, 2021 at 12:34 PM Krzysztof Kozlowski
> <krzysztof.kozlowski@canonical.com> wrote:
>> On 05/03/2021 17:29, Hector Martin wrote:
>>> On 06/03/2021 01.20, Andy Shevchenko wrote:
>>>>> I am just splitting an
>>>>> existing function into two, where one takes the lock and the other does
>>>>> the work. Do you mean using a different locking function? I'm not
>>>>> entirely sure what you're suggesting.
>>>>
>>>> Yes, as a prerequisite
>>>>
>>>> spin_lock_irqsave -> spin_lock().
>>>
>>> Krzysztof, is this something you want in this series? I was trying to
>>> avoid logic changes to the non-Apple paths.
>>
>> I don't quite get the need for such change (the code will be still
>> called in interrupt handler, right?), but assuming the "why?" is
>> properly documented, it can be a separate patch here.
> 
> This is only for readability: the common rule is to not disable
> interrupts when they are already disabled, so a reader might wonder
> if this instance of the handler is special in some case that it might
> be called with interrupts enabled.
> 
> There is also a small overhead in accessing the global irq mask
> register on some architectures, but for a uart that does not make
> any difference of course.
> 
> While I'm generally in favor of that kind of cleanup, I'd also
> prefer to leave it out of this series -- once you get into details
> like this the series gets harder to review.

So it's only about the spinlock in the IRQ handler (which does not need
to disable the IRQs). Makes sense but not related at all to the topic of
bringing up Apple M1, therefore should not stop the review/merging.

Best regards,
Krzysztof