Futex hash computation requires a mask operation with read-only after
init data that will be converted to a runtime constant in the subsequent
commit.
Introduce runtime_const_mask_32 to further optimize the mask operation
in the futex hash computation hot path. GCC generates a:
lui a0, 0x12346 # upper; +0x800 then >>12 for correct rounding
addi a0, a0, 0x678 # lower 12 bits
and a1, a1, a0 # a1 = a1 & a0
pattern to tackle arbitrary 32-bit masks and the same was also suggested
by Claude which is implemented here. The (__mask & val) operation is
intentionally placed outside of asm block to allow compilers to further
optimize it if possible.
__runtime_fixup_ptr() already patches a "lui + addi" sequence which has
been reused to patch the same sequence for __runtime_fixup_mask().
Assisted-by: Claude:claude-sonnet-4-5
Signed-off-by: K Prateek Nayak <kprateek.nayak@amd.com>
---
Changelog v2..v3:
o Moved the "&" operation outside the inline asm block to allow for
compilers to further optimize it if possible. (Based on David's
comment on ARM64 bits).
---
arch/riscv/include/asm/runtime-const.h | 22 ++++++++++++++++++++++
1 file changed, 22 insertions(+)
diff --git a/arch/riscv/include/asm/runtime-const.h b/arch/riscv/include/asm/runtime-const.h
index d766e2b9e6df..85efba8ecf12 100644
--- a/arch/riscv/include/asm/runtime-const.h
+++ b/arch/riscv/include/asm/runtime-const.h
@@ -153,6 +153,22 @@
__ret; \
})
+#define runtime_const_mask_32(val, sym) \
+({ \
+ u32 __mask; \
+ asm_inline(".option push\n\t" \
+ ".option norvc\n\t" \
+ "1:\t" \
+ "lui %[__mask],0x89abd\n\t" \
+ "addi %[__mask],%[__mask],-0x211\n\t" \
+ ".option pop\n\t" \
+ ".pushsection runtime_mask_" #sym ",\"a\"\n\t" \
+ ".long 1b - .\n\t" \
+ ".popsection" \
+ : [__mask] "=r" (__mask)); \
+ (__mask & val); \
+})
+
#define runtime_const_init(type, sym) do { \
extern s32 __start_runtime_##type##_##sym[]; \
extern s32 __stop_runtime_##type##_##sym[]; \
@@ -256,6 +272,12 @@ static inline void __runtime_fixup_shift(void *where, unsigned long val)
mutex_unlock(&text_mutex);
}
+static inline void __runtime_fixup_mask(void *where, unsigned long val)
+{
+ __runtime_fixup_32(where, where + 4, val);
+ __runtime_fixup_caches(where, 2);
+}
+
static inline void runtime_const_fixup(void (*fn)(void *, unsigned long),
unsigned long val, s32 *start, s32 *end)
{
--
2.34.1On Thu, Apr 2, 2026 at 7:39 PM K Prateek Nayak <kprateek.nayak@amd.com> wrote:
>
> Futex hash computation requires a mask operation with read-only after
> init data that will be converted to a runtime constant in the subsequent
> commit.
>
> Introduce runtime_const_mask_32 to further optimize the mask operation
> in the futex hash computation hot path. GCC generates a:
>
> lui a0, 0x12346 # upper; +0x800 then >>12 for correct rounding
> addi a0, a0, 0x678 # lower 12 bits
> and a1, a1, a0 # a1 = a1 & a0
>
> pattern to tackle arbitrary 32-bit masks and the same was also suggested
> by Claude which is implemented here. The (__mask & val) operation is
> intentionally placed outside of asm block to allow compilers to further
> optimize it if possible.
>
> __runtime_fixup_ptr() already patches a "lui + addi" sequence which has
> been reused to patch the same sequence for __runtime_fixup_mask().
>
> Assisted-by: Claude:claude-sonnet-4-5
> Signed-off-by: K Prateek Nayak <kprateek.nayak@amd.com>
> ---
> Changelog v2..v3:
>
> o Moved the "&" operation outside the inline asm block to allow for
> compilers to further optimize it if possible. (Based on David's
> comment on ARM64 bits).
> ---
> arch/riscv/include/asm/runtime-const.h | 22 ++++++++++++++++++++++
> 1 file changed, 22 insertions(+)
>
> diff --git a/arch/riscv/include/asm/runtime-const.h b/arch/riscv/include/asm/runtime-const.h
> index d766e2b9e6df..85efba8ecf12 100644
> --- a/arch/riscv/include/asm/runtime-const.h
> +++ b/arch/riscv/include/asm/runtime-const.h
> @@ -153,6 +153,22 @@
> __ret; \
> })
>
> +#define runtime_const_mask_32(val, sym) \
> +({ \
> + u32 __mask; \
> + asm_inline(".option push\n\t" \
> + ".option norvc\n\t" \
> + "1:\t" \
> + "lui %[__mask],0x89abd\n\t" \
> + "addi %[__mask],%[__mask],-0x211\n\t" \
Ref include/uapi/linux/reboot.h:
#define LINUX_REBOOT_CMD_CAD_ON 0x89ABCDEF
#define RUNTIME_MAGIC 0x89ABCDEF
"lui %[__mask], %%hi(RUNTIME_MAGIC)\n\t"
"addi %[__mask], %[__mask], %%lo(RUNTIME_MAGIC)\n\t"
> + ".option pop\n\t" \
> + ".pushsection runtime_mask_" #sym ",\"a\"\n\t" \
> + ".long 1b - .\n\t" \
> + ".popsection" \
> + : [__mask] "=r" (__mask)); \
> + (__mask & val); \
> +})
> +
> #define runtime_const_init(type, sym) do { \
> extern s32 __start_runtime_##type##_##sym[]; \
> extern s32 __stop_runtime_##type##_##sym[]; \
> @@ -256,6 +272,12 @@ static inline void __runtime_fixup_shift(void *where, unsigned long val)
> mutex_unlock(&text_mutex);
> }
>
> +static inline void __runtime_fixup_mask(void *where, unsigned long val)
> +{
> + __runtime_fixup_32(where, where + 4, val);
> + __runtime_fixup_caches(where, 2);
> +}
> +
> static inline void runtime_const_fixup(void (*fn)(void *, unsigned long),
> unsigned long val, s32 *start, s32 *end)
> {
> --
> 2.34.1
>
>
--
Best Regards
Guo Ren
Hello Guo,
On 4/3/2026 3:12 PM, Guo Ren wrote:
>> diff --git a/arch/riscv/include/asm/runtime-const.h b/arch/riscv/include/asm/runtime-const.h
>> index d766e2b9e6df..85efba8ecf12 100644
>> --- a/arch/riscv/include/asm/runtime-const.h
>> +++ b/arch/riscv/include/asm/runtime-const.h
>> @@ -153,6 +153,22 @@
>> __ret; \
>> })
>>
>> +#define runtime_const_mask_32(val, sym) \
>> +({ \
>> + u32 __mask; \
>> + asm_inline(".option push\n\t" \
>> + ".option norvc\n\t" \
>> + "1:\t" \
>> + "lui %[__mask],0x89abd\n\t" \
>> + "addi %[__mask],%[__mask],-0x211\n\t" \
> Ref include/uapi/linux/reboot.h:
> #define LINUX_REBOOT_CMD_CAD_ON 0x89ABCDEF
>
> #define RUNTIME_MAGIC 0x89ABCDEF
>
> "lui %[__mask], %%hi(RUNTIME_MAGIC)\n\t"
> "addi %[__mask], %[__mask], %%lo(RUNTIME_MAGIC)\n\t"
Ack! I'll clean it up in the next version while also fixing the
stuff that Sashiko reported.
Thanks a ton for taking a look at the series.
>
>
>> + ".option pop\n\t" \
>> + ".pushsection runtime_mask_" #sym ",\"a\"\n\t" \
>> + ".long 1b - .\n\t" \
>> + ".popsection" \
>> + : [__mask] "=r" (__mask)); \
>> + (__mask & val); \
>> +})
--
Thanks and Regards,
Prateek
© 2016 - 2026 Red Hat, Inc.