On Wed, Oct 2, 2024, at 16:04, Greg Kroah-Hartman wrote:
> On Wed, Oct 02, 2024 at 03:45:08PM +0000, Arnd Bergmann wrote:
>> On Wed, Oct 2, 2024, at 14:23, Christian Brauner wrote:
>>
>> Here, the 64-bit 'old' has the same size as the 32-bit 'new',
>> so if we try to handle them in a shared native/compat ioctl
>> function, this needs an extra in_conmpat_syscall() check that
>> adds complexity and is easy to forget.
>
> Agreed, "extending" ioctls is considered a bad thing and it's just
> easier to create a new one. Or use some flags and reserved fields, if
> you remember to add them in the beginning...
>
> Anyway, this is all great, but for now, I'll take this series in my tree
> and we can add onto it from there. I'll dig up some sample code that
> uses this too, so that we make sure it works properly. Give me a few
> days to catch up before it lands in my trees...
Sounds good to me, it's clear we don't get a quick solution and
there is nothing stopping us from revisiting this after we have a
couple of drivers using ioctl.
Arnd
