* [PATCH RFC] riscv: introduce asm/swab.h
@ 2025-03-10 22:03 Ignacio Encinas
2025-03-17 16:36 ` Alexandre Ghiti
0 siblings, 1 reply; 3+ messages in thread
From: Ignacio Encinas @ 2025-03-10 22:03 UTC (permalink / raw)
To: Paul Walmsley, Palmer Dabbelt, Alexandre Ghiti
Cc: Eric Biggers, linux-riscv, linux-kernel, linux-kernel-mentees,
skhan, Zhihang Shao, Björn Töpel, Ignacio Encinas
Implement endianness swap macros for RISC-V.
Use the rev8 instruction when Zbb is available. Otherwise, rely on the
default mask-and-shift implementation.
Signed-off-by: Ignacio Encinas <ignacio@iencinas.com>
---
Motivated by [1]. A couple of things to note:
We need a default implementation to fall back on, but there isn't any in
`asm-generic/swab.h`. Should I introduce a first patch moving
___constant_swab<XX> into include/uapi/asm-generic/swab.h?
I don't particularly like the ARCH_SWAB macro but I can't think of
anything better that doesn't result in code duplication.
Tested with crc_kunit as pointed out here [2]. I can't provide
performance numbers as I don't have RISC-V hardware yet.
Ccing everyone involved with [1].
[1] https://lore.kernel.org/all/20250302220426.GC2079@quark.localdomain/
[2] https://lore.kernel.org/all/20250216225530.306980-1-ebiggers@kernel.org/
---
arch/riscv/include/asm/swab.h | 81 +++++++++++++++++++++++++++++++++++++++++++
1 file changed, 81 insertions(+)
diff --git a/arch/riscv/include/asm/swab.h b/arch/riscv/include/asm/swab.h
new file mode 100644
index 0000000000000000000000000000000000000000..8f8a13b343f6ffbefbb3c7747ab4e14243852014
--- /dev/null
+++ b/arch/riscv/include/asm/swab.h
@@ -0,0 +1,81 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef _ASM_RISCV_SWAB_H
+#define _ASM_RISCV_SWAB_H
+
+#include <linux/types.h>
+#include <linux/compiler.h>
+#include <asm/alternative-macros.h>
+#include <asm/hwcap.h>
+
+#if defined(CONFIG_RISCV_ISA_ZBB) && !defined(NO_ALTERNATIVE)
+
+/*
+ * FIXME, RFC PATCH: This is copypasted from include/uapi/linux/swab.h
+ * should I move these `#defines` to include/uapi/asm-generic/swab.h
+ * and include that file here and in include/uapi/linux/swab.h ?
+ */
+#define ___constant_swab16(x) ((__u16)( \
+ (((__u16)(x) & (__u16)0x00ffU) << 8) | \
+ (((__u16)(x) & (__u16)0xff00U) >> 8)))
+
+#define ___constant_swab32(x) ((__u32)( \
+ (((__u32)(x) & (__u32)0x000000ffUL) << 24) | \
+ (((__u32)(x) & (__u32)0x0000ff00UL) << 8) | \
+ (((__u32)(x) & (__u32)0x00ff0000UL) >> 8) | \
+ (((__u32)(x) & (__u32)0xff000000UL) >> 24)))
+
+#define ___constant_swab64(x) ((__u64)( \
+ (((__u64)(x) & (__u64)0x00000000000000ffULL) << 56) | \
+ (((__u64)(x) & (__u64)0x000000000000ff00ULL) << 40) | \
+ (((__u64)(x) & (__u64)0x0000000000ff0000ULL) << 24) | \
+ (((__u64)(x) & (__u64)0x00000000ff000000ULL) << 8) | \
+ (((__u64)(x) & (__u64)0x000000ff00000000ULL) >> 8) | \
+ (((__u64)(x) & (__u64)0x0000ff0000000000ULL) >> 24) | \
+ (((__u64)(x) & (__u64)0x00ff000000000000ULL) >> 40) | \
+ (((__u64)(x) & (__u64)0xff00000000000000ULL) >> 56)))
+
+#define ___constant_swahw32(x) ((__u32)( \
+ (((__u32)(x) & (__u32)0x0000ffffUL) << 16) | \
+ (((__u32)(x) & (__u32)0xffff0000UL) >> 16)))
+
+#define ___constant_swahb32(x) ((__u32)( \
+ (((__u32)(x) & (__u32)0x00ff00ffUL) << 8) | \
+ (((__u32)(x) & (__u32)0xff00ff00UL) >> 8)))
+
+
+#define ARCH_SWAB(size) \
+static __always_inline unsigned long __arch_swab##size(__u##size value) \
+{ \
+ unsigned long x = value; \
+ \
+ asm goto(ALTERNATIVE("j %l[legacy]", "nop", 0, \
+ RISCV_ISA_EXT_ZBB, 1) \
+ :::: legacy); \
+ \
+ asm volatile (".option push\n" \
+ ".option arch,+zbb\n" \
+ "rev8 %0, %1\n" \
+ ".option pop\n" \
+ : "=r" (x) : "r" (x)); \
+ \
+ return x >> (BITS_PER_LONG - size); \
+ \
+legacy: \
+ return ___constant_swab##size(value); \
+}
+
+#ifdef CONFIG_64BIT
+ARCH_SWAB(64)
+#define __arch_swab64 __arch_swab64
+#endif
+
+ARCH_SWAB(32)
+#define __arch_swab32 __arch_swab32
+
+ARCH_SWAB(16)
+#define __arch_swab16 __arch_swab16
+
+#undef ARCH_SWAB
+
+#endif /* defined(CONFIG_RISCV_ISA_ZBB) && !defined(NO_ALTERNATIVE) */
+#endif /* _ASM_RISCV_SWAB_H */
---
base-commit: 2014c95afecee3e76ca4a56956a936e23283f05b
change-id: 20250307-riscv-swab-b81b94a9ac1b
Best regards,
--
Ignacio Encinas <ignacio@iencinas.com>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH RFC] riscv: introduce asm/swab.h
2025-03-10 22:03 [PATCH RFC] riscv: introduce asm/swab.h Ignacio Encinas
@ 2025-03-17 16:36 ` Alexandre Ghiti
2025-03-17 20:29 ` Ignacio Encinas Rubio
0 siblings, 1 reply; 3+ messages in thread
From: Alexandre Ghiti @ 2025-03-17 16:36 UTC (permalink / raw)
To: Ignacio Encinas, Paul Walmsley, Palmer Dabbelt
Cc: Eric Biggers, linux-riscv, linux-kernel, linux-kernel-mentees,
skhan, Zhihang Shao, Björn Töpel
Hi Ignacio,
On 10/03/2025 23:03, Ignacio Encinas wrote:
> Implement endianness swap macros for RISC-V.
>
> Use the rev8 instruction when Zbb is available. Otherwise, rely on the
> default mask-and-shift implementation.
>
> Signed-off-by: Ignacio Encinas <ignacio@iencinas.com>
> ---
> Motivated by [1]. A couple of things to note:
>
> We need a default implementation to fall back on, but there isn't any in
> `asm-generic/swab.h`. Should I introduce a first patch moving
> ___constant_swab<XX> into include/uapi/asm-generic/swab.h?
Yes, that or something else to avoid the code duplication.
>
> I don't particularly like the ARCH_SWAB macro but I can't think of
> anything better that doesn't result in code duplication.
>
> Tested with crc_kunit as pointed out here [2]. I can't provide
> performance numbers as I don't have RISC-V hardware yet.
>
> Ccing everyone involved with [1].
>
> [1] https://lore.kernel.org/all/20250302220426.GC2079@quark.localdomain/
> [2] https://lore.kernel.org/all/20250216225530.306980-1-ebiggers@kernel.org/
> ---
> arch/riscv/include/asm/swab.h | 81 +++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 81 insertions(+)
>
> diff --git a/arch/riscv/include/asm/swab.h b/arch/riscv/include/asm/swab.h
> new file mode 100644
> index 0000000000000000000000000000000000000000..8f8a13b343f6ffbefbb3c7747ab4e14243852014
> --- /dev/null
> +++ b/arch/riscv/include/asm/swab.h
> @@ -0,0 +1,81 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +#ifndef _ASM_RISCV_SWAB_H
> +#define _ASM_RISCV_SWAB_H
> +
> +#include <linux/types.h>
> +#include <linux/compiler.h>
> +#include <asm/alternative-macros.h>
> +#include <asm/hwcap.h>
> +
> +#if defined(CONFIG_RISCV_ISA_ZBB) && !defined(NO_ALTERNATIVE)
NO_ALTERNATIVE does not exist to my knowledge, and you don't need to
check for ALTERNATIVE as alternative-macros.h will do the right thing in
that case.
> +
> +/*
> + * FIXME, RFC PATCH: This is copypasted from include/uapi/linux/swab.h
> + * should I move these `#defines` to include/uapi/asm-generic/swab.h
> + * and include that file here and in include/uapi/linux/swab.h ?
> + */
> +#define ___constant_swab16(x) ((__u16)( \
> + (((__u16)(x) & (__u16)0x00ffU) << 8) | \
> + (((__u16)(x) & (__u16)0xff00U) >> 8)))
> +
> +#define ___constant_swab32(x) ((__u32)( \
> + (((__u32)(x) & (__u32)0x000000ffUL) << 24) | \
> + (((__u32)(x) & (__u32)0x0000ff00UL) << 8) | \
> + (((__u32)(x) & (__u32)0x00ff0000UL) >> 8) | \
> + (((__u32)(x) & (__u32)0xff000000UL) >> 24)))
> +
> +#define ___constant_swab64(x) ((__u64)( \
> + (((__u64)(x) & (__u64)0x00000000000000ffULL) << 56) | \
> + (((__u64)(x) & (__u64)0x000000000000ff00ULL) << 40) | \
> + (((__u64)(x) & (__u64)0x0000000000ff0000ULL) << 24) | \
> + (((__u64)(x) & (__u64)0x00000000ff000000ULL) << 8) | \
> + (((__u64)(x) & (__u64)0x000000ff00000000ULL) >> 8) | \
> + (((__u64)(x) & (__u64)0x0000ff0000000000ULL) >> 24) | \
> + (((__u64)(x) & (__u64)0x00ff000000000000ULL) >> 40) | \
> + (((__u64)(x) & (__u64)0xff00000000000000ULL) >> 56)))
> +
> +#define ___constant_swahw32(x) ((__u32)( \
> + (((__u32)(x) & (__u32)0x0000ffffUL) << 16) | \
> + (((__u32)(x) & (__u32)0xffff0000UL) >> 16)))
> +
> +#define ___constant_swahb32(x) ((__u32)( \
> + (((__u32)(x) & (__u32)0x00ff00ffUL) << 8) | \
> + (((__u32)(x) & (__u32)0xff00ff00UL) >> 8)))
> +
> +
> +#define ARCH_SWAB(size) \
> +static __always_inline unsigned long __arch_swab##size(__u##size value) \
I don't like the ARCH_SWAB macro neither but I don't have another
solution too, maybe someone will come up with some macro magic.
> +{ \
> + unsigned long x = value; \
> + \
> + asm goto(ALTERNATIVE("j %l[legacy]", "nop", 0, \
> + RISCV_ISA_EXT_ZBB, 1) \
> + :::: legacy); \
> + \
> + asm volatile (".option push\n" \
> + ".option arch,+zbb\n" \
> + "rev8 %0, %1\n" \
> + ".option pop\n" \
> + : "=r" (x) : "r" (x)); \
> + \
> + return x >> (BITS_PER_LONG - size); \
> + \
> +legacy: \
> + return ___constant_swab##size(value); \
> +}
> +
> +#ifdef CONFIG_64BIT
> +ARCH_SWAB(64)
> +#define __arch_swab64 __arch_swab64
> +#endif
> +
> +ARCH_SWAB(32)
> +#define __arch_swab32 __arch_swab32
> +
> +ARCH_SWAB(16)
> +#define __arch_swab16 __arch_swab16
> +
> +#undef ARCH_SWAB
> +
> +#endif /* defined(CONFIG_RISCV_ISA_ZBB) && !defined(NO_ALTERNATIVE) */
> +#endif /* _ASM_RISCV_SWAB_H */
>
> ---
> base-commit: 2014c95afecee3e76ca4a56956a936e23283f05b
> change-id: 20250307-riscv-swab-b81b94a9ac1b
>
> Best regards,
It would be nice to have some perf numbers too, maybe Bjorn will chime in :)
Thanks for your patch, looking forward the v2,
Alex
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH RFC] riscv: introduce asm/swab.h
2025-03-17 16:36 ` Alexandre Ghiti
@ 2025-03-17 20:29 ` Ignacio Encinas Rubio
0 siblings, 0 replies; 3+ messages in thread
From: Ignacio Encinas Rubio @ 2025-03-17 20:29 UTC (permalink / raw)
To: Alexandre Ghiti, Paul Walmsley, Palmer Dabbelt
Cc: Eric Biggers, linux-riscv, linux-kernel, linux-kernel-mentees,
skhan, Zhihang Shao, Björn Töpel
On 17/3/25 17:36, Alexandre Ghiti wrote:
>> We need a default implementation to fall back on, but there isn't any in
>> `asm-generic/swab.h`. Should I introduce a first patch moving
>> ___constant_swab<XX> into include/uapi/asm-generic/swab.h?
>
> Yes, that or something else to avoid the code duplication.
Thanks for double checking. I figured as much but I ended up getting a
bit confused with the include directories.
>> +#if defined(CONFIG_RISCV_ISA_ZBB) && !defined(NO_ALTERNATIVE)
>
> NO_ALTERNATIVE does not exist to my knowledge, and you don't need to check for ALTERNATIVE as alternative-macros.h will do the right thing in that case.
It's a define hardcoded into drivers/firmware/efi/libstub/Makefile. It
was introduced in [1]. It helps avoid these types of warnings [2]:
riscv64-unknown-linux-gnu-ld: warning: orphan section
`.init.alternative' from `./drivers/firmware/efi/libstub/riscv.stub.o'
being placed in section `.init.alternative'
Let me know if you think I should drop it.
[1] https://lore.kernel.org/all/20231031064553.2319688-3-xiao.w.wang@intel.com/
[2] https://lore.kernel.org/all/DM8PR11MB5751DA69DEEAAC9B06623E0BB80FA@DM8PR11MB5751.namprd11.prod.outlook.com/
> It would be nice to have some perf numbers too, maybe Bjorn will chime in :)
>
> Thanks for your patch, looking forward the v2,
Thank you!
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2025-03-17 20:29 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-03-10 22:03 [PATCH RFC] riscv: introduce asm/swab.h Ignacio Encinas
2025-03-17 16:36 ` Alexandre Ghiti
2025-03-17 20:29 ` Ignacio Encinas Rubio
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®