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;
Message-ID: <05da00c0-2180-62ee-6276-8553ad9341c4@collabora.com>
Date:   Tue, 2 May 2023 08:56:31 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.10.0
Subject: Re: [PATCH v9 1/6] media: verisilicon: Do not set context src/dst
 formats in reset functions
To:     Thorsten Leemhuis <linux@leemhuis.info>,
        Shreeya Patel <shreeya.patel@collabora.com>
Cc:     linux-media@vger.kernel.org, linux-rockchip@lists.infradead.org,
        linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        kernel@collabora.com, robert.mader@collabora.com,
        nicolas.dufresne@collabora.co.uk, ezequiel@vanguardiasur.com.ar,
        festevam@gmail.com, p.zabel@pengutronix.de, mchehab@kernel.org,
        shawnguo@kernel.org, s.hauer@pengutronix.de, kernel@pengutronix.de,
        hverkuil-cisco@xs4all.nl, linux-imx@nxp.com,
        regressions@lists.linux.dev
References: <20230220104849.398203-1-benjamin.gaignard@collabora.com>
 <20230220104849.398203-2-benjamin.gaignard@collabora.com>
 <26addb7d-bb9d-34e8-d4fe-e323ff488101@collabora.com>
 <dcd317db-3c24-895d-572b-1b139c370ff7@leemhuis.info>
Content-Language: en-US
From:   Benjamin Gaignard <benjamin.gaignard@collabora.com>
In-Reply-To: <dcd317db-3c24-895d-572b-1b139c370ff7@leemhuis.info>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


Le 01/05/2023 à 09:21, Thorsten Leemhuis a écrit :
> On 27.04.23 00:19, Shreeya Patel wrote:
>> On 20/02/23 16:18, Benjamin Gaignard wrote:
>>> Setting context source and destination formats should only be done
>>> in hantro_set_fmt_out() and hantro_set_fmt_cap() after check that
>>> the targeted queue is not busy.
>>> Remove these calls from hantro_reset_encoded_fmt() and
>>> hantro_reset_raw_fmt() to clean the driver.
>>>
>>> Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
>> KernelCI found this patch causes a regression in the
>> baseline.dmesg.alert test [1] on rk3399-rock-pi-4b [2],
>> see the bisection report for more details [3].
>>
>> Let us know if you have any questions.
>>
>>
>> [1]
>> https://github.com/kernelci/kernelci-core/blob/main/config/rootfs/debos/overlays/baseline/opt/kernelci/dmesg.sh
>> [2] https://linux.kernelci.org/test/case/id/6442e825f19134d74c2e865d/
>> [3] https://groups.io/g/kernelci-results/message/40740
> Thx for the report. FWIW, regzbot noticed there is a patch that refers
> to the culprit that might have been landed in next after your test ran
> for the last time (and meanwhile it was mainlined): f100ce3bbd6 ("media:
> verisilicon: Fix crash when probing encoder")

Yes that patch should fix the probing issue.
Marek is working on an additional one to fix pixel format negotiation
but that doesn't impact the boot.

Regards,
Benjamin

>
> I wonder if that is related or might even fix the issue.
>
> Ciao, Thorsten
>
>>> ---
>>>    drivers/media/platform/verisilicon/hantro_v4l2.c | 9 ++-------
>>>    1 file changed, 2 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/drivers/media/platform/verisilicon/hantro_v4l2.c
>>> b/drivers/media/platform/verisilicon/hantro_v4l2.c
>>> index c0d427956210..d8aa42bd4cd4 100644
>>> --- a/drivers/media/platform/verisilicon/hantro_v4l2.c
>>> +++ b/drivers/media/platform/verisilicon/hantro_v4l2.c
>>> @@ -382,13 +382,10 @@ hantro_reset_encoded_fmt(struct hantro_ctx *ctx)
>>>          vpu_fmt = hantro_get_default_fmt(ctx, true);
>>>    -    if (ctx->is_encoder) {
>>> -        ctx->vpu_dst_fmt = vpu_fmt;
>>> +    if (ctx->is_encoder)
>>>            fmt = &ctx->dst_fmt;
>>> -    } else {
>>> -        ctx->vpu_src_fmt = vpu_fmt;
>>> +    else
>>>            fmt = &ctx->src_fmt;
>>> -    }
>>>          hantro_reset_fmt(fmt, vpu_fmt);
>>>        fmt->width = vpu_fmt->frmsize.min_width;
>>> @@ -408,11 +405,9 @@ hantro_reset_raw_fmt(struct hantro_ctx *ctx)
>>>        raw_vpu_fmt = hantro_get_default_fmt(ctx, false);
>>>          if (ctx->is_encoder) {
>>> -        ctx->vpu_src_fmt = raw_vpu_fmt;
>>>            raw_fmt = &ctx->src_fmt;
>>>            encoded_fmt = &ctx->dst_fmt;
>>>        } else {
>>> -        ctx->vpu_dst_fmt = raw_vpu_fmt;
>>>            raw_fmt = &ctx->dst_fmt;
>>>            encoded_fmt = &ctx->src_fmt;
>>>        }
>>