In arch/sandbox/include/asm/types.h we have Therefore for 32 bit Sandbox build BITS_PER_LONG turns out to be 32 as CONFIG_PHYS64 is not set
This messes up the current logic of GENMASK macro due to mismatch b/w size of unsigned long (64 bit) and that of BITS_PER_LONG. Fix this by using CONFIG_SANDBOX_BITS_PER_LONG which is set to 64/32 based on the host machine on which its being compiled.
Without this patch: GENMASK(14,0) => 0x7fffffffffff After this patch: GENMASK(14,0) => 0x7fff
Signed-off-by: Vignesh R vigneshr@ti.com ---
I marked it as RFC because, I am not really sure if I am running Sandbox tests properly. My host machine is x86_64 (Ubuntu 18.04). Running below command[1] with some of my yet to be upstreamed patches that use GENMASK macro shows this issue.
[1]: $ make check
include/linux/bitops.h | 5 +++++ 1 file changed, 5 insertions(+)
diff --git a/include/linux/bitops.h b/include/linux/bitops.h index a47f6d17bb5f..259df43fb00f 100644 --- a/include/linux/bitops.h +++ b/include/linux/bitops.h @@ -21,8 +21,13 @@ * position @h. For example * GENMASK_ULL(39, 21) gives us the 64bit vector 0x000000ffffe00000. */ +#ifdef CONFIG_SANDBOX +#define GENMASK(h, l) \ + (((~0UL) << (l)) & (~0UL >> (CONFIG_SANDBOX_BITS_PER_LONG - 1 - (h)))) +#else #define GENMASK(h, l) \ (((~0UL) << (l)) & (~0UL >> (BITS_PER_LONG - 1 - (h)))) +#endif
#define GENMASK_ULL(h, l) \ (((~0ULL) << (l)) & (~0ULL >> (BITS_PER_LONG_LONG - 1 - (h))))