Re: -Wconstant-conversion in drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c

[Nathan Chancellor <nathan@kernel.org>]

On Thu, Apr 16, 2020 at 05:41:20PM -0700, Nathan Chancellor wrote:
> Hi all,
> 
> I was building s390 allyesconfig with clang and came across a curious
> warning:
> 
> drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c:580:41: warning:
> implicit conversion from 'unsigned long' to 'int' changes value from
> 18446744073709551584 to -32 [-Wconstant-conversion]
>         mvpp2_pools[MVPP2_BM_SHORT].pkt_size = MVPP2_BM_SHORT_PKT_SIZE;
>                                              ~ ^~~~~~~~~~~~~~~~~~~~~~~
> drivers/net/ethernet/marvell/mvpp2/mvpp2.h:699:33: note: expanded from
> macro 'MVPP2_BM_SHORT_PKT_SIZE'
> #define MVPP2_BM_SHORT_PKT_SIZE MVPP2_RX_MAX_PKT_SIZE(MVPP2_BM_SHORT_FRAME_SIZE)
>                                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> drivers/net/ethernet/marvell/mvpp2/mvpp2.h:634:30: note: expanded from
> macro 'MVPP2_RX_MAX_PKT_SIZE'
>         ((total_size) - NET_SKB_PAD - MVPP2_SKB_SHINFO_SIZE)
>                  ~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~
> 1 warning generated.
> 
> As far as I understand it, the warning comes from the fact that
> MVPP2_BM_SHORT_FRAME_SIZE is treated as size_t because
> MVPP2_SKB_SHINFO_SIZE ultimately calls ALIGN with sizeof(struct
> skb_shared_info), which has typeof called on it.
> 
> The implicit conversion probably is fine but it would be nice to take
> care of the warning. I am not sure what would be the best way to do that
> would be though. An explicit cast would take care of it, maybe in
> MVPP2_SKB_SHINFO_SIZE since the actual value I see is 320, which is able
> to be fit into type int easily.
> 
> Any comments would be appreciated, there does not appear to be a
> dedicated maintainer of this driver according to get_maintainer.pl.

Sorry for the necrobump, I am doing a bug scrub and it seems like this
driver now has maintainers so keying them in in case they have any
comments/suggestions.

Cheers,
Nathan