On Wed, May 29, 2024, at 09:29, Jiri Slaby wrote:
> On 18. 04. 24, 8:18, Greg Kroah-Hartman wrote:
>>
>> This is a nice cleanup, thanks for doing it, I'll just take this one
>> change now if you don't object.
>
> Unfortunately, _IOC_NONE is 1 on some archs as noted by Arnd, and this
> commit changed the kd ioctl values in there which broke stuff as noted
> by Al.
>
> We either:
> * use _IOC(0, X, Y) in here, instead of _IO(X, Y), or
> * define KDIOC(X) as _IOC(0, KD_IOCTL_BASE, X), or
> * revert the commit which landed to -rc1 already.
I would prefer a simple revert, as the other options may
end up more confusing. Another option might be a new
global macro, if we then go an convert all plain ioctl
command numbers to that.
Arnd
