Re: [PATCH 00/19] amba: store owner from modules with amba_driver_register()

From: Krzysztof Kozlowski
Date: Tue Apr 02 2024 - 05:48:29 EST


On 02/04/2024 10:56, Russell King (Oracle) wrote:
> On Sat, Mar 30, 2024 at 01:18:30PM +0100, Krzysztof Kozlowski wrote:
>> On 26/03/2024 21:23, Krzysztof Kozlowski wrote:
>>> Merging
>>> =======
>>> All further patches depend on the first amba patch, therefore please ack
>>> and this should go via one tree.
>>>
>>> Description
>>> ===========
>>> Modules registering driver with amba_driver_register() often forget to
>>> set .owner field.
>>>
>>> Solve the problem by moving this task away from the drivers to the core
>>> amba bus code, just like we did for platform_driver in commit
>>> 9447057eaff8 ("platform_device: use a macro instead of
>>> platform_driver_register").
>>>
>>> Best regards,
>>
>> I tried to submit this series to Russell patch tracker and failed. This
>> is ridiculous. It's 2024 and instead of normal process, like every other
>> maintainer, so b4 or Patchwork, we have some unusable system rejecting
>> standard patches.
>
> Sorry but no. Stop being offensive.
>
>> First, it depends some weird, duplicated signed-off-by's.
>
> Eh? There is no such logic in there.

In the web system there is - read the error message I pasted. It wants
another SoB from the unrelated email account, the one used purely for
registering in some web system, not the one used for code handling.

>
>> Second it > submitting patch-by-patch, all with clicking on some web
>> (!!!) interface.
>
> Again, no it doesn't, and you're just throwing crap out because you
> failed. Unlike most of the "normal" processes, the patch system allows
> you to submit both by *email* and also by *web* for those cases where

The email one requires additional steps, so this is unnecessary work
confusing submitters. I submit dozens or hundreds of patches every
release cycle. That's the only subsystem which is odd to use.

> the emails get screwed up by ones company mail server. That's why the
> web interface exists - to give people *flexibility*.

No, they are not. None of my emails are screwed by my company system.

>
> The fact is, the web interface is merely a front end interface that
> generates an email and submits it in the usual way by email - an
> email that you can perfectly well generate that is *very* close to
> the standard email that git format-patch generates.
>
> The *only* difference is that the patch system wants a KernelVersion:
> tag in the email _somewhere_ and it doesn't matter where it appears.
> Git even has support to do this.
>
> git format-patch --add-header="KernelVersion: $foo"
>
> Why does it want the kernel version? Because when we were running 2.4
> and 2.5 kernel versions in parallel, it was important to know which
> tree the patch was being submitted for. It has continued to be required

Which is absolutely ridiculous now. Expecting submitters to adhere to
some rule for 20 year old kernel is not reasonable.

> because it means when there's problems applying a patch, it gives me
> the additional information about the base used for the patch (and it
> keeps on being useful to have.)
>
>> That's the response:
>> -------------
>> Your patch has not been logged because:
>>
>> Error: Please supply a summary subject line briefly describing
>> your patch.
>>
>>
>> Error: Please supply a "KernelVersion: " tag after "PATCH FOLLOWS" or
>> "---".
>>
>> Error: the patch you are submitting has one or more missing or incorrect
>> Signed-off-by lines:
>>
>> - author signoff <krzkreg@xxxxxxxxx> is missing.

^^^ here you have additional SoB expectation.

>>
>> Please see the file Documentation/SubmittingPatches, section 11
>> for details on signing off patches.
>
> Lots of people use it without a problem. I've just run the parser
> through its offline tests, and it parses email content correctly.
> I've no idea what you're doing wrong, but it looks like something
> pretty serious if it didn't parse the subject line.
>
> Rather than getting stressed about it, why don't you send me an email
> the first time something goes wrong so I can investigate, turn on
> debugging to capture the problem email?

I don't know any person who enjoyed working with your patch workflow.
>From few people I talked, it was always "now I have to learn this weird
system" or "I need to get through this process which is different than
everything in the kernel".

Plus you entirely ignored poor usability of this system which:
1. Allows submitting patches only 1-by-1, so 19 useless steps in my case.
2. Accepts the first/second/all patches without problem encouraging me
to submit the rest... and then tells me via email they were wrong and
could not be accepted.

This is the poorest user-experience one can imagine. If you put effort
into some web form, make it at least helpful so reject the patch if it
does not match your expectations.

Best regards,
Krzysztof