Re: [PATCH v1 1/6] sdio: Add syntactic sugar to store a pointer in sdio_driver_id

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Wed, Apr 22, 2026 at 12:56:01PM +0200, Christian A. Ehrhardt wrote:
> On Wed, Apr 22, 2026 at 09:52:11AM +0300, Andy Shevchenko wrote:
> > > To be honest, with the involved void* this isn't really type-safe
> > > either, but at least the data keeps being a pointer which is really
> > > helpful on CHERI. FTR: The alternative would be to use uintptr_t instead
> > > of unsigned long, which also has proponents in the CHERI community and
> > > which is used in the current vendor patch stack.
> > 
> > FWIW, Linus categorically told that it has to be no uintptr_t in the
> > Linux kernel.
> 
> This is probably not the correct context to discuss this but
> I will point out that if/when CHERI support is added to the
> kernel we will need a way to distinguish between
> * A (fat) pointer converted to an integer type (in order to do
>   integer arithmetic on the address) that might be converted back
>   to a pointer in the future and
> * A plain address "extracted" from a fat pointer.
> Using `uintptr_t` for the former and `unsigned long` for the
> latter seems like a reasonable choice but there are certainly
> other options.
> 
> Sorting out the difference between the two is a very large
> part of the diff required to port Linux (or in fact any piece
> of low-level software) to a CHERI system.

Right, whatever CHERI needs is has to be discussed with the project leader(s)
first. But thanks for giving more information on this.

-- 
With Best Regards,
Andy Shevchenko