lkml.org 
[lkml]   [2021]   [Dec]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH v2 RESEND 3/5] tty: serial: samsung: Remove USI initialization
From
On 03/12/2021 17:22, Sam Protsenko wrote:
> On Wed, 1 Dec 2021 at 12:54, Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
>>
>> On Wed, Dec 1, 2021 at 12:42 AM Sam Protsenko
>> <semen.protsenko@linaro.org> wrote:
>>>
>>> USI control is now extracted to dedicated USI driver. Remove USI related
>>
>> the dedicated
>>
>>> code from serial driver to avoid conflicts and code duplication.
>>
>> Would it break run-time bisectability?
>> If so, why is it not a problem?
>>
>
> It shouldn't. This patch is [3/5], and USI driver (which takes the
> control over the USI registers) is [2/5]. As for Device Tree, the only
> platform using "samsung,exynos850-uart" right now is Exynos Auto V9
> SADK (serial node is declared in exynosautov9.dtsi). I don't have
> Exynos Auto V9 datasheet, so I can't really add the USI node properly
> there, nor I can test that. I guess it should be done separately from
> this patch series.
>
> Chanho, Krzysztof:
>
> Guys, what are your thoughts on this? Basically with this patch series
> applied, Exynos Auto V9 serial might become not functional. New USI
> node should be added for UART case in Exynos Auto V9 dtsi (providing
> correct sysreg, SW_CONF offset, clocks, etc), and serial node should
> be encapsulated inside of that USI node. Also, USI node should be
> referenced and enabled in SADK dts, providing also "clkreq-on"
> property. More details can be found in [PATCH 1/5]. Do you think it's
> ok to take this series as is, and add that later? Because otherwise we
> might need to collaborate to add that Exynos Auto V9 enablement into
> this patch series, which might take more time...

The patch in current state will probably break Exynos Auto v9 boards,
including the in-tree one, unless bootloader sets the USI to serial. The
trouble is that. Changing the Exynos Auto v9 DTSI in these series would
solve it only partially, because the kernel still won't be bisectable.

Breaking Auto v9 serial within a kernel is okay for me, because the
board was added recently, I don't expect products using it and it is
still development phase. This of course assuming that it's users agree,
so the question is to Chanho and other folks.

Best regards,
Krzysztof

\
 
 \ /
  Last update: 2021-12-04 12:23    [W:0.048 / U:2.784 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site