* Re: [PATCH] dma-mapping: Allow use of DMA_BIT_MASK(64) in global scope
2025-10-30 14:05 ` [PATCH] dma-mapping: Allow use of DMA_BIT_MASK(64) in global scope James Clark
@ 2025-10-30 18:47 ` Nathan Chancellor
2025-11-05 12:45 ` Marek Szyprowski
1 sibling, 0 replies; 3+ messages in thread
From: Nathan Chancellor @ 2025-10-30 18:47 UTC (permalink / raw)
To: James Clark
Cc: Marek Szyprowski, Robin Murphy, Nick Desaulniers, Bill Wendling,
Justin Stitt, Stephen Rothwell, Mark Brown, Vinod Koul,
Arnd Bergmann, Guodong Xu, iommu, linux-kernel, llvm
On Thu, Oct 30, 2025 at 02:05:27PM +0000, James Clark wrote:
> Clang doesn't like that (1ULL<<(64)) overflows when initializing a
> global scope variable, even if that part of the ternary isn't used when
> n = 64. The same initialization can be done without warnings in function
> scopes, and GCC doesn't mind either way.
>
> The build failure that highlighted this was already fixed in a different
> way [1], which also has detailed links to the Clang issues. However it's
> not going to be long before the same thing happens again, so it's better
> to fix the root cause.
Yeah, it already has since the instance you bring up below:
https://git.kernel.org/gregkh/usb/c/93741bd104ce07a790519ecee3a331ee7cf61ae3
> Fix it by using GENMASK_ULL() which does exactly the same thing, is much
> more readable anyway, and doesn't have a shift that overflows.
>
> [1]: https://lore.kernel.org/all/20250918-mmp-pdma-simplify-dma-addressing-v1-1-5c2be2b85696@riscstar.com/
>
> Signed-off-by: James Clark <james.clark@linaro.org>
Thanks, this should hopefully be fixed in clang relatively soon [1] but
you are right that there is nothing to stop this from occuring again,
especially since older clang versions will still be impacted by this
behavior. My basic testing shows that this should be equivalent so:
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
[1]: https://github.com/llvm/llvm-project/pull/163885
> ---
> include/linux/dma-mapping.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
> index 8248ff9363ee..2ceda49c609f 100644
> --- a/include/linux/dma-mapping.h
> +++ b/include/linux/dma-mapping.h
> @@ -90,7 +90,7 @@
> */
> #define DMA_MAPPING_ERROR (~(dma_addr_t)0)
>
> -#define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1))
> +#define DMA_BIT_MASK(n) GENMASK_ULL(n - 1, 0)
>
> struct dma_iova_state {
> dma_addr_t addr;
>
> ---
> base-commit: e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6
> change-id: 20251030-james-fix-dma_bit_mask-624dbeb89afa
>
> Best regards,
> --
> James Clark <james.clark@linaro.org>
>
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] dma-mapping: Allow use of DMA_BIT_MASK(64) in global scope
2025-10-30 14:05 ` [PATCH] dma-mapping: Allow use of DMA_BIT_MASK(64) in global scope James Clark
2025-10-30 18:47 ` Nathan Chancellor
@ 2025-11-05 12:45 ` Marek Szyprowski
1 sibling, 0 replies; 3+ messages in thread
From: Marek Szyprowski @ 2025-11-05 12:45 UTC (permalink / raw)
To: James Clark, Robin Murphy, Nathan Chancellor, Nick Desaulniers,
Bill Wendling, Justin Stitt, Stephen Rothwell, Mark Brown,
Vinod Koul, Arnd Bergmann, Guodong Xu
Cc: iommu, linux-kernel, llvm
On 30.10.2025 15:05, James Clark wrote:
> Clang doesn't like that (1ULL<<(64)) overflows when initializing a
> global scope variable, even if that part of the ternary isn't used when
> n = 64. The same initialization can be done without warnings in function
> scopes, and GCC doesn't mind either way.
>
> The build failure that highlighted this was already fixed in a different
> way [1], which also has detailed links to the Clang issues. However it's
> not going to be long before the same thing happens again, so it's better
> to fix the root cause.
>
> Fix it by using GENMASK_ULL() which does exactly the same thing, is much
> more readable anyway, and doesn't have a shift that overflows.
>
> [1]: https://lore.kernel.org/all/20250918-mmp-pdma-simplify-dma-addressing-v1-1-5c2be2b85696@riscstar.com/
>
> Signed-off-by: James Clark <james.clark@linaro.org>
Thanks, applied to dma-mapping-fixes branch.
> ---
> include/linux/dma-mapping.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
> index 8248ff9363ee..2ceda49c609f 100644
> --- a/include/linux/dma-mapping.h
> +++ b/include/linux/dma-mapping.h
> @@ -90,7 +90,7 @@
> */
> #define DMA_MAPPING_ERROR (~(dma_addr_t)0)
>
> -#define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1))
> +#define DMA_BIT_MASK(n) GENMASK_ULL(n - 1, 0)
>
> struct dma_iova_state {
> dma_addr_t addr;
>
> ---
> base-commit: e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6
> change-id: 20251030-james-fix-dma_bit_mask-624dbeb89afa
>
> Best regards,
Best regards
--
Marek Szyprowski, PhD
Samsung R&D Institute Poland
^ permalink raw reply [flat|nested] 3+ messages in thread