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;
Feedback-ID: i51094778:Fastmail
User-Agent: Cyrus-JMAP/3.7.0-alpha0-1015-gaf7d526680-fm-20220929.001-gaf7d5266
Mime-Version: 1.0
Message-Id: <3398859f-e872-4f1d-8a03-4dcb1e46e010@app.fastmail.com>
In-Reply-To: <20221004112724.31621-2-konrad.dybcio@somainline.org>
References: <20221004112724.31621-1-konrad.dybcio@somainline.org>
 <20221004112724.31621-2-konrad.dybcio@somainline.org>
Date:   Tue, 04 Oct 2022 17:56:38 +0200
From:   "Sven Peter" <sven@svenpeter.dev>
To:     "Konrad Dybcio" <konrad.dybcio@somainline.org>,
        asahi@lists.linux.dev, linux-arm-kernel@lists.infradead.org
Cc:     "Nick Chan" <towinchenmi@gmail.com>,
        "Hector Martin" <marcan@marcan.st>,
        "Alyssa Rosenzweig" <alyssa@rosenzweig.io>,
        "Thomas Gleixner" <tglx@linutronix.de>,
        "Marc Zyngier" <maz@kernel.org>,
        "Rob Herring" <robh+dt@kernel.org>,
        "Krzysztof Kozlowski" <krzysztof.kozlowski+dt@linaro.org>,
        linux-kernel@vger.kernel.org, devicetree@vger.kernel.org
Subject: Re: [PATCH v3 2/2] irqchip/apple-aic: Add support for A7-A11 SoCs
Content-Type: text/plain
Precedence: bulk

Hi,


On Tue, Oct 4, 2022, at 13:27, Konrad Dybcio wrote:
> Add support for A7-A11 SoCs by if-ing out some features only present
> on:
>
> * A11 & newer (implementation-defined IPI & UNCORE registers)
> * A11[1] & newer (fast IPI support).
>
> UNCORE/UNCORE2 and IPI registers conveniently both first appeared on
> A11, so introduce just one check for that.
>
> Knowing whether the SoC supports the latter is necessary, as they are
> written to, even if fast IPI is disabled. 

AFAIK that's only an artifact in this driver: It was added to prevent an FIQ
storm in case there were pending fast ipis (i.e. the bootloader was broken ;))
when this driver didn't support fast ipis yet.

> This in turn causes a crash
> on older platforms, as the implemention-defined registers either do
> something else or are not supposed to be touched - definitely not a
> NOP though.
>
> [1] A11 is supposed to use this feature, but it currently doesn't work
> for reasons unknown and hence remains disabled. It can easily be enabled
> on A11 only, as there is a SoC-specific compatible in the DT with a
> fallback to apple,aic. That said, it is not yet necessary, especially
> with only one core up, and it has worked a-ok so far.

Just to make sure I understand this correctly - we have the following three situations:

- base: no fastipi, no uncore, will work on A11 and M1 though
- A11: fastipi and uncore but fastipi is broken (possibly due to HW errata or some bug in this driver that only happens on A11)
- M1 (or maybe even A12 already, doesn't matter though): fastipi and uncore support

If we figured out _why_ fastipi is broken on A11 we would only need a single
feature flag to enable both uncore and fastipi but for now we need two to
disable fastipi for A11.

I'm also curious: What are the symptoms when you enable fastipi on A11?


Sven