Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <CGME20210811074838eucas1p2a0e8625af27c10209d9bcfc732254ae7@eucas1p2.samsung.com>
 <20210810103336.114077-1-robert.foss@linaro.org> <0b694e24-5cc8-4944-d3a2-115306ae7b89@samsung.com>
 <CAG3jFys+ch86Y7338-DH1+8Q4w5eK83revVsNwoVCugwXeqjmQ@mail.gmail.com>
 <CAG3jFytOQQBnnJQU9qDdQedrrcPz=SQPeXHX1HJQ8c5U94feCg@mail.gmail.com> <de1df486-255f-6405-5ea9-4dddd5507006@samsung.com>
In-Reply-To: <de1df486-255f-6405-5ea9-4dddd5507006@samsung.com>
From:   Robert Foss <robert.foss@linaro.org>
Date:   Thu, 12 Aug 2021 11:22:01 +0200
Message-ID: <CAG3jFyt7KB0JxTzimKA3Oi2dLU1Uj5uJ4aPFkQONbST8eFk6gg@mail.gmail.com>
Subject: Re: [PATCH v1] media: camss: vfe: Don't use vfe->base before it's assigned
To:     Marek Szyprowski <m.szyprowski@samsung.com>,
        Naresh Kamboju <naresh.kamboju@linaro.org>
Cc:     Todor Tomov <todor.too@gmail.com>, Andy Gross <agross@kernel.org>,
        Bjorn Andersson <bjorn.andersson@linaro.org>,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        Hans Verkuil <hverkuil-cisco@xs4all.nl>,
        linux-media <linux-media@vger.kernel.org>,
        MSM <linux-arm-msm@vger.kernel.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Hans Verkuil <hverkuil@xs4all.nl>,
        Linux Kernel Functional Testing <lkft@linaro.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

> >>>
> >>> [   18.480608] qcom-venus 1d00000.video-codec: Adding to iommu group 1
> >>> [   18.536167] qcom-camss 1b0ac00.camss: Adding to iommu group 2
> >>> [   18.600373] Internal error: synchronous external abort: 96000010 [#1]
> >>> PREEMPT SMP
> > After testing this patch + linux-next[1] I'm not able to replicate the
> > 'Internal error' above on the db410c. And I don't think it is related
> > to this patch.
> >
> > Are you seeing the same error on [1]? And are you seeing it before the
> > b10b5334528a9 ("media: camss: vfe: Don't read hardware version
> > needlessly") patch?
>
> I've checked once again on your branch. Yes, it is fully reproducible on
> my DragonBoard410c. On your branch I get the above synchronous external
> abort, while without your last patch I get Null ptr dereference.
>
> Are you sure you have deployed the kernel modules for doing the test?
> This issue happens when udev triggers loading kernel modules for the
> detected hardware.
>
> Here is the kernel configuration used for my tests on ARM64 based board:
>
> # make ARCH=arm64 defconfig && ./scripts/config --set-val
> CMA_SIZE_MBYTES 96 -e PROVE_LOCKING -e DEBUG_ATOMIC_SLEEP -e PM_DEBUG -e
> PM_ADVANCED_DEBUG -d ARCH_SUNXI -d ARCH_ALPINE -d DRM_NOUVEAU -d
> ARCH_BCM_IPROC -d ARCH_BERLIN -d ARCH_BRCMSTB -d ARCH_LAYERSCAPE -d
> ARCH_LG1K -d ARCH_HISI -d ARCH_MEDIATEK -d ARCH_MVEBU -d ARCH_ROCKCHIP
> -d ARCH_SEATTLE -d ARCH_SYNQUACER -d ARCH_RENESAS -d ARCH_STRATIX10 -d
> ARCH_TEGRA -d ARCH_SPRD -d ARCH_THUNDER -d ARCH_THUNDER2 -d
> ARCH_UNIPHIER -d ARCH_XGENE -d ARCH_ZX -d ARCH_ZYNQMP -d HIBERNATION -d
> CLK_SUNXI -e BLK_DEV_RAM --set-val BLK_DEV_RAM_COUNT 4 --set-val
> BLK_DEV_RAM_SIZE 65536 -d CONFIG_EFI -d CONFIG_TEE
>
> Comparing to the arm64's defconfig, I've enabled some debugging options
> and disabled some unused boards.
>

I made a mistake when testing on the db410c earlier, but with it fixed
I can replicate the issue. It seems to stem from the hardware not
actually being powered up when the hw_version() call is being made.

Unfortunately there is no simple way to achieve the original intention
of the series that reduced the frequency of the hw_version() being
called, while avoiding the above issue. So let's revert to the
previous behaviour for the time being.

Thank you for your help Marek! I'll submit a v2 shortly.


Rob