On Fri, May 27, 2022 at 12:37:19PM +0200, Jan Beulich wrote:
> As per [1] the expansion of the pirq_dpci() macro causes a -Waddress
> controlled warning (enabled implicitly in our builds, if not by default)
> tying the middle part of the involved conditional expression to the
> surrounding boolean context. Work around this by introducing a local
> inline function in the affected source file.
> 
> Reported-by: Andrew Cooper <[email protected]>
> Signed-off-by: Jan Beulich <[email protected]>
> 
> [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102967
> ---
> This is intended to replace an earlier patch by Andrew [2], open-coding
> and then simplifying the macro in the one problematic place.
> 
> Note that, with pirq_dpci() presently used solely in the one file being
> changed here, we could in principle also remove the #define and use just
> an inline(?) function in this file. But then the macro would need
> reinstating as soon as a use elsewhere would become necessary.
> 
> As to the inline - I think it's warranted here, but it goes against our
> general policy of using inline only in header files. Hence I'd be okay
> to drop it to avoid controversy.
> 
> [2] https://lists.xen.org/archives/html/xen-devel/2021-10/msg01635.html
> 
> --- a/xen/drivers/passthrough/x86/hvm.c
> +++ b/xen/drivers/passthrough/x86/hvm.c
> @@ -25,6 +25,18 @@
>  #include <asm/hvm/support.h>
>  #include <asm/io_apic.h>
>  
> +/*
> + * Gcc12 takes issue with pirq_dpci() being used in boolean context (see gcc
> + * bug 102967). While we can't replace the macro definition in the header by 
> an
> + * inline function, we can do so here.
> + */
> +static inline struct hvm_pirq_dpci *_pirq_dpci(struct pirq *pirq)
> +{
> +    return pirq_dpci(pirq);
> +}
> +#undef pirq_dpci
> +#define pirq_dpci(pirq) _pirq_dpci(pirq)

That's fairly ugly.  Seeing as pirq_dpci is only used in hvm.c, would
it make sense to just convert the macro to be a static inline in that
file? (and remove pirq_dpci() from irq.h).

Thanks, Roger.

Reply via email to