Re: [PATCH] tty: serial: convert atomic_* to refcount_* APIs

Re: [PATCH] tty: serial: convert atomic_* to refcount_* APIs

[Greg Kroah-Hartman]

On Thu, Dec 22, 2022 at 01:27:40AM +0530, Deepak R Varma wrote:
> The refcount_* APIs are designed to address known issues with the
> atomic_t APIs for reference counting. They protect the reference
> counters from overflow/underflow, use-after-free errors, provide
> improved memory ordering guarantee schemes, are neater and safer.
> Hence, replace the atomic_* APIs by their equivalent refcount_t
> API functions.
> 
> This patch proposal address the following warnings generated by
> the atomic_as_refcounter.cocci coccinelle script
> 	atomic_add_return(-1, ...)
> 
> 
> Signed-off-by: Deepak R Varma <drv@mailo.com>
> ---
> Note: The patch is compile tested using dec_station.defconfig for
>       MIPS architecture.

Do you have this hardware?  If not, please just do
one-variable-at-a-time so that if there are real problems, we can revert
the offending change easier.  And it makes it simpler to review.

But, are you sure this is correct:

> -	irq_guard = atomic_add_return(1, &mux->irq_guard);
> -	if (irq_guard != 1)
> +	refcount_inc(&mux->irq_guard);
> +	if (refcount_read(&mux->irq_guard) != 1)

That is now different logic than before, why?  Are you sure this is ok?

I stopped reviewing here...

thanks,

greg k-h

Re: [PATCH] tty: serial: convert atomic_* to refcount_* APIs

[Deepak R Varma]

On Thu, Dec 22, 2022 at 06:55:01AM +0100, Greg Kroah-Hartman wrote:
> On Thu, Dec 22, 2022 at 01:27:40AM +0530, Deepak R Varma wrote:
> > The refcount_* APIs are designed to address known issues with the
> > atomic_t APIs for reference counting. They protect the reference
> > counters from overflow/underflow, use-after-free errors, provide
> > improved memory ordering guarantee schemes, are neater and safer.
> > Hence, replace the atomic_* APIs by their equivalent refcount_t
> > API functions.
> >
> > This patch proposal address the following warnings generated by
> > the atomic_as_refcounter.cocci coccinelle script
> > 	atomic_add_return(-1, ...)
> >
> >
> > Signed-off-by: Deepak R Varma <drv@mailo.com>
> > ---
> > Note: The patch is compile tested using dec_station.defconfig for
> >       MIPS architecture.
>
> Do you have this hardware?  If not, please just do
> one-variable-at-a-time so that if there are real problems, we can revert
> the offending change easier.  And it makes it simpler to review.

Hi Greg,
I do not have the hardware. I will make a one variable per patch change as
suggest and send in a patch set.

>
> But, are you sure this is correct:
>
> > -	irq_guard = atomic_add_return(1, &mux->irq_guard);
> > -	if (irq_guard != 1)
> > +	refcount_inc(&mux->irq_guard);
> > +	if (refcount_read(&mux->irq_guard) != 1)
>
> That is now different logic than before, why?  Are you sure this is ok?

This is not correct. I read it wrong. I will correct this in the patch set.

>
> I stopped reviewing here...
Thanks. I will review the rest of the changes per your advise and make the
corrections.

regards,
./drv

>
> thanks,
>
> greg k-h