[PATCH v3 0/5] Fix a whole host of nvmem registration/cleanup issues

Hector Martin marcan at marcan.st
Tue Jan 3 17:15:14 PST 2023


On 04/01/2023 06.15, Srinivas Kandagatla wrote:
> Hi Russell,
> 
> 
> On 03/01/2023 16:58, Russell King (Oracle) wrote:
>> Hi,
>>
>> This series fixes a whole host of nvmem registration/error cleanup
>> issues that have been identified by both Hector and myself. It is a
>> substantial rework of my original patch fixing the first problem.
>>
>> The first most obvious problem is the race between nvmem registration
>> and use, which leads to sporadic failures of drivers to probe at boot
>> time.
>>
>> While fixing this, it has been noticed that a recent fix to check the
>> return value of dev_set_name() introduced a new bug where wp_gpio was
>> not being put in that newly introduced error path.
>>
>> Then there's a fix for a previous fix which itself purports to fix
>> another bug, but results in the allocated ID being leaked. Fix for a
>> fix for a fix is not good!
>>
>> Then there's an error in the docbook documentation for wp_gpio (it's
>> listed as wp-gpio instead) but as nothing seems to set wp_gpio, we
>> might as well get rid of it - which also solves the issue that we
>> call gpiod_put() on this whether we own it or not.
>>
>> Lastly, there's a fix for yet another spurious white-space in this
>> code, one of what seems to be a long history of past white-space
>> fixes.
>>
>> These patches have been individually build-tested in the order of
>> posting, but not run-time tested except for the entire series.
> 
> 
> thanks for fixing these issues, I have applied these after fixing the 
> subject on all the patches, as it ended up with email ids in subject.

Right. I see none of the issues you two lectured me about actually
mattered, it was all for show, and this isn't getting backported anyway.

Good job you two. The day I finally rage quit the kernel after enough of
this nonsense and make a big dossier of just how fucked up the kernel
maintainer community's attitude is, I will be sure to use this as an
example.

That day is not today though. But I certainly won't be upstreaming any
more patches to nvmem.

- Hector



More information about the linux-arm-kernel mailing list