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;
Date:   Mon, 17 Jul 2023 09:05:06 +0200
From:   Boris Brezillon <boris.brezillon@collabora.com>
To:     Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc:     Rob Herring <robh@kernel.org>, Steven Price <steven.price@arm.com>,
        dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org,
        kernel@collabora.com
Subject: Re: [PATCH v1] drm/panfrost: Sync IRQ by job's timeout handler
Message-ID: <20230717090506.2ded4594@collabora.com>
In-Reply-To: <20230717065254.1061033-1-dmitry.osipenko@collabora.com>
References: <20230717065254.1061033-1-dmitry.osipenko@collabora.com>
Organization: Collabora
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hi Dmitry,

On Mon, 17 Jul 2023 09:52:54 +0300
Dmitry Osipenko <dmitry.osipenko@collabora.com> wrote:

> Panfrost IRQ handler may stuck for a long time, for example this happens
> when there is a bad HDMI connection and HDMI handler takes a long time to
> finish processing, holding Panfrost. Make Panfrost's job timeout handler
> to sync IRQ before checking fence signal status in order to prevent
> spurious job timeouts due to a slow IRQ processing.

Feels like the problem should be fixed in the HDMI encoder driver
instead, so it doesn't stall the whole system when processing its
IRQs (use threaded irqs, maybe). I honestly don't think blocking in the
job timeout path to flush IRQs is a good strategy.

Regards,

Boris

> 
> Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
> ---
>  drivers/gpu/drm/panfrost/panfrost_job.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
> index dbc597ab46fb..a356163da22d 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_job.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_job.c
> @@ -713,6 +713,8 @@ static enum drm_gpu_sched_stat panfrost_job_timedout(struct drm_sched_job
>  	struct panfrost_device *pfdev = job->pfdev;
>  	int js = panfrost_job_get_slot(job);
>  
> +	synchronize_irq(pfdev->js->irq);
> +
>  	/*
>  	 * If the GPU managed to complete this jobs fence, the timeout is
>  	 * spurious. Bail out.