Re: [PATCH v2 2/3] rust: dma: convert the read/write macros to return Result

[Danilo Krummrich <dakr@kernel.org>]

On Thu, Apr 10, 2025 at 10:58:10PM +0000, Benno Lossin wrote:
> On Thu Apr 10, 2025 at 5:34 PM CEST, Danilo Krummrich wrote:
> > On Thu, Apr 10, 2025 at 03:11:01PM +0000, Benno Lossin wrote:
> >> On Thu Apr 10, 2025 at 1:54 PM CEST, Danilo Krummrich wrote:
> >> > On Thu, Apr 10, 2025 at 11:58:17AM +0300, Abdiel Janulgue wrote:
> >> >> @@ -78,13 +74,14 @@ impl Drop for DmaSampleDriver {
> >> >>      fn drop(&mut self) {
> >> >>          dev_info!(self.pdev.as_ref(), "Unload DMA test driver.\n");
> >> >>  
> >> >> -        let _ = || -> Result {
> >> >> -            for (i, value) in TEST_VALUES.into_iter().enumerate() {
> >> >> -                assert_eq!(kernel::dma_read!(self.ca[i].h), value.0);
> >> >> -                assert_eq!(kernel::dma_read!(self.ca[i].b), value.1);
> >> >> -            }
> >> >> -            Ok(())
> >> >> -        }();
> >> >> +        for (i, value) in TEST_VALUES.into_iter().enumerate() {
> >> >> +            let val0 = kernel::dma_read!(self.ca[i].h);
> >> >> +            let val1 = kernel::dma_read!(self.ca[i].b);
> >> >> +            assert!(val0.is_ok());
> >> >> +            assert!(val1.is_ok());
> >> >> +            assert_eq!(val0.unwrap(), value.0);
> >> >> +            assert_eq!(val1.unwrap(), value.1);
> >> >
> >> > Maybe use if-let to avoid the unwrap?
> >> >
> >> > 	if let Ok(val0) = val0 {
> >> > 	   assert_eq!(val0, value.0);
> >> > 	}
> >> >
> >> > I know it's a bit pointless, since we know it must be ok, but the educational
> >> > message of the example should be to check and not to unwrap, so maybe that's
> >> > better.
> >> 
> >> The if-let will silently ignore any errors, so I don't think that it's
> >> fit for example code either.
> >
> > Yes, but we still have the assert!() before, so the full sequence would be:
> >
> > 	assert!(val0.is_ok());
> >
> > 	if let Ok(val0) = val0 {
> > 	   assert_eq!(val0, value.0);
> > 	}
> 
> Ah right, missed that.
> 
> > The intention would be to avoid patterns that shouldn't be used in "real" code;
> > assert!() should be obvious not to use for real code.
> 
> Yeah, I'm not sure if this is that valuable. I think having "real code"
> is better, but I don't have any idea what to do in this case.
> 
> Why does this sample do the validation in the `drop` method in the first
> place?

I assume there is no specific reason, maybe Abdiel wanted to have a bit more
lifecycle for the allocation than just probe().

I guess we could just move it to probe(). Alternatively we can also keep it in a
closure or function and only assert! once for the returned Result.

> I guess the same code on the C side would do this in `remove` or
> whatever the equivalent thing is there, but would there be the option to
> report an error? Or is `remove` an infallible operation? In that case
> `assert!` probably is still the best option.

remove() is and has to be infallible, yes.