linux-kernel - Re: [PATCH] ARM: exynos_defconfig: Disable simplefb support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <55F2857B.4020800@osg.samsung.com>
Date:	Fri, 11 Sep 2015 09:40:43 +0200
From:	Javier Martinez Canillas <javier@....samsung.com>
To:	Krzysztof Kozlowski <k.kozlowski@...sung.com>,
	linux-kernel@...r.kernel.org
Cc:	Michael Turquette <mturquette@...libre.com>,
	Kukjin Kim <kgene@...nel.org>,
	Thierry Reding <treding@...dia.com>,
	Anand Moon <linux.amoon@...il.com>,
	Russell King <linux@....linux.org.uk>,
	linux-samsung-soc@...r.kernel.org,
	Lukasz Majewski <l.majewski@...sung.com>,
	linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH] ARM: exynos_defconfig: Disable simplefb support

Hello Krzysztof,

On 09/11/2015 09:16 AM, Krzysztof Kozlowski wrote:
> On 11.09.2015 16:07, Javier Martinez Canillas wrote:
>> Hello Krzysztof,
>>
>> On 09/11/2015 07:01 AM, Krzysztof Kozlowski wrote:
>>> On 10.09.2015 22:42, Javier Martinez Canillas wrote:
>>>> The simplefb driver allows the kernel to render on a pre-allocated
>>>> buffer that's been initialized by firmware before the kernel boots.
>>>>
>>>> This option was enabled to have display working on the Exynos5250
>>>> Snow Chromebook by commit da9d0fbf5e9a ("ARM: exynos: defconfig
>>>> update") since proper DRM/KMS support did not exist at that time.
>>>>
>>>> But now that the Exynos DRM driver has support for this hardware,
>>>> there is no need to have simplefb enabled. In fact, if a user has
>>>> a u-boot that injects the simplefb dev node to the FDT before pass
>>>> it to the kernel, display won't be properly initialized and only a
>>>> blank screen will be shown since there isn't a proper handoff from
>>>> the simplefb driver to the Exynos DRM driver.
>>>>
>>>> Signed-off-by: Javier Martinez Canillas <javier@....samsung.com>
>>>>
>>>> ---
>>>>
>>>>  arch/arm/configs/exynos_defconfig | 1 -
>>>>  1 file changed, 1 deletion(-)
>>>
>>> Seems logical. None of the boards use simple-framebuffer compatible
>>> anyway. I understand that on Snow simplefb was needed along with change
>>> in Uboot like this one:
>>> https://chromium.googlesource.com/chromiumos/third_party/u-boot/+/refs/changes/58/49358/2
>>>
>>
>> Exactly but you won't see the dev node with the "simple-framebuffer"
>> compatible string in the DTS since is the bootloader that adds this
>> device node to the FDT before passing it to the kernel.
>>
>> The bootloader shouldn't mangle the FDT (with the exception of the
>> memory and choosen/bootargs nodes) but simplefb is just a hack to
>> re-use the display HW initialization made by the bootloader.
>>
>>> and now none of Exynos boards use simplefb anymore?
>>>
>>
>> Yes, there are no other Exynos boards using simplefb besides Snow
>> that I'm aware of but since Exynos DRM is working well on this board
>> from v4.0, there is no need for it anymore.
> 
> OK,
> Reviewed-by: Krzysztof Kozlowski <k.kozlowski@...sung.com>
>

Thanks!
 
>>
>> In fact, as explained in the commit message, it could do more harm
>> than good since users that are still booting with a u-boot that adds
>> the simplefb device node, only get a blank screen since the simplefb
>> driver is probed, creates a console and later the Exynos DRM probes
>> and re-initializes the HW creating its own console, causing this issue.
>>
>> I got several reports of users that says that mainline stop booting for
>> them but is just that they didn't get display working. Disabling simplefb
>> makes display to work again so maybe this is even -rc material and should
>> go to stable # v4.0+
> 
> You know, it is only a defconfig. The issue is there regardless of
> change in defconfig. I am not convinced that defconfig problems are
> worth backporting. multi_v7 has it enabled anyway.
>

Yes, that's why I said "maybe" :)
 
> Maybe the EXYNOS_DRM should have some anti-dependency on SIMPLE_FB? But

hrmm I don't know, I think this issue happens with any DRM driver. Hopefully
there should be soon a nice hand off from simplefb to DRM drivers like early
console does to the real console drivers for UARTs.

> on the other hand the issue is actually caused by hacks in bootloader...
>

Yeah and in fact this does not happen with the stock bootloader that comes
shipped on these devices, is with a special build that has this hack enabled.
 
> 
> Best regards,
> Krzysztof
> 
>

Thanks a lot for your feedback.

Best regards,
-- 
Javier Martinez Canillas
Open Source Group
Samsung Research America
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/