-Wshift-count-overflow in drivers/misc/mic/card/mic_x100.c

[tpimh]

drivers/misc/mic/card/mic_x100.c:297:27: warning: shift count >= width of type [-Wshift-count-overflow]
static u64 mic_dma_mask = DMA_BIT_MASK(64);
                          ^~~~~~~~~~~~~~~~
./include/linux/dma-mapping.h:142:54: note: expanded from macro 'DMA_BIT_MASK'
#define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1))
                                                     ^ ~~~
drivers/misc/mic/card/mic_x100.c:305:24: warning: shift count >= width of type [-Wshift-count-overflow]
                .coherent_dma_mask = DMA_BIT_MASK(64),
                                     ^~~~~~~~~~~~~~~~
./include/linux/dma-mapping.h:142:54: note: expanded from macro 'DMA_BIT_MASK'
#define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1))
                                                     ^ ~~~
2 warnings generated.

[nickdesaulniers]

This is tricky; Clang is complaining about the clearly false case. The kernel has this pattern of relying on optimization to run before semantic analysis occurs in a few places. These instances tightly couple the codebase to gcc. We could simply initialize mic_dma_mask to ~0ULL, but maybe there's a more clever expression we could use? (There might be other instances of DMA_BIT_MASK(64).)

[nickdesaulniers]

Duplicate: #92