diff mbox series

[4/7] riscv: Implement xchg8/16() using Zabha

Message ID 20240528151052.313031-5-alexghiti@rivosinc.com (mailing list archive)
State Superseded
Headers show
Series Zacas/Zabha support and qspinlocks | expand

Checks

Context Check Description
conchuod/vmtest-fixes-PR fail merge-conflict

Commit Message

Alexandre Ghiti May 28, 2024, 3:10 p.m. UTC
This adds runtime support for Zabha in xchg8/16() operations.

Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
---
 arch/riscv/include/asm/cmpxchg.h | 24 ++++++++++++++++++++++--
 arch/riscv/include/asm/hwcap.h   |  1 +
 arch/riscv/kernel/cpufeature.c   |  1 +
 3 files changed, 24 insertions(+), 2 deletions(-)

Comments

Conor Dooley May 28, 2024, 3:22 p.m. UTC | #1
On Tue, May 28, 2024 at 05:10:49PM +0200, Alexandre Ghiti wrote:
	\
> diff --git a/arch/riscv/include/asm/hwcap.h b/arch/riscv/include/asm/hwcap.h
> index e17d0078a651..f71ddd2ca163 100644
> --- a/arch/riscv/include/asm/hwcap.h
> +++ b/arch/riscv/include/asm/hwcap.h
> @@ -81,6 +81,7 @@
>  #define RISCV_ISA_EXT_ZTSO		72
>  #define RISCV_ISA_EXT_ZACAS		73
>  #define RISCV_ISA_EXT_XANDESPMU		74
> +#define RISCV_ISA_EXT_ZABHA		75
>  
>  #define RISCV_ISA_EXT_XLINUXENVCFG	127
>  
> diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> index 3ed2359eae35..8d0f56dd2f53 100644
> --- a/arch/riscv/kernel/cpufeature.c
> +++ b/arch/riscv/kernel/cpufeature.c
> @@ -257,6 +257,7 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
>  	__RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
>  	__RISCV_ISA_EXT_DATA(zihpm, RISCV_ISA_EXT_ZIHPM),
>  	__RISCV_ISA_EXT_DATA(zacas, RISCV_ISA_EXT_ZACAS),
> +	__RISCV_ISA_EXT_DATA(zabha, RISCV_ISA_EXT_ZABHA),
>  	__RISCV_ISA_EXT_DATA(zfa, RISCV_ISA_EXT_ZFA),
>  	__RISCV_ISA_EXT_DATA(zfh, RISCV_ISA_EXT_ZFH),
>  	__RISCV_ISA_EXT_DATA(zfhmin, RISCV_ISA_EXT_ZFHMIN),

You're missing a dt-binding patch in this series adding zabha.

Thanks,
Conor.
Andrea Parri May 28, 2024, 6 p.m. UTC | #2
> -#define __arch_xchg_masked(prepend, append, r, p, n)			\
> +#define __arch_xchg_masked(swap_sfx, prepend, append, r, p, n)		\

This actually indicates a problem in the current (aka, no Zabha)
implementation: without your series, xchg16() gets mapped to

  lr.w     a2,(a3)
  and      a1,a2,a5
  or       a1,a1,a4
  sc.w     a1,a1,(a3)
  bnez     a1,43c <.L0^B1>

which is clearly wrong... (other "fully-ordered LR/SC sequences"
instead follow the mapping

  lr.w     a2,(a3)
  and      a1,a2,a5
  or       a1,a1,a4
  sc.w.rl  a1,a1,(a3)
  bnez     a1,43c <.L0^B1>
  fence    rw,rw  )

A similar consideration for xchg8().

  Andrea
Alexandre Ghiti May 29, 2024, 6:15 a.m. UTC | #3
Hi Conor,

On Tue, May 28, 2024 at 5:22 PM Conor Dooley <conor@kernel.org> wrote:
>
> On Tue, May 28, 2024 at 05:10:49PM +0200, Alexandre Ghiti wrote:
>         \
> > diff --git a/arch/riscv/include/asm/hwcap.h b/arch/riscv/include/asm/hwcap.h
> > index e17d0078a651..f71ddd2ca163 100644
> > --- a/arch/riscv/include/asm/hwcap.h
> > +++ b/arch/riscv/include/asm/hwcap.h
> > @@ -81,6 +81,7 @@
> >  #define RISCV_ISA_EXT_ZTSO           72
> >  #define RISCV_ISA_EXT_ZACAS          73
> >  #define RISCV_ISA_EXT_XANDESPMU              74
> > +#define RISCV_ISA_EXT_ZABHA          75
> >
> >  #define RISCV_ISA_EXT_XLINUXENVCFG   127
> >
> > diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> > index 3ed2359eae35..8d0f56dd2f53 100644
> > --- a/arch/riscv/kernel/cpufeature.c
> > +++ b/arch/riscv/kernel/cpufeature.c
> > @@ -257,6 +257,7 @@ const struct riscv_isa_ext_data riscv_isa_ext[] = {
> >       __RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
> >       __RISCV_ISA_EXT_DATA(zihpm, RISCV_ISA_EXT_ZIHPM),
> >       __RISCV_ISA_EXT_DATA(zacas, RISCV_ISA_EXT_ZACAS),
> > +     __RISCV_ISA_EXT_DATA(zabha, RISCV_ISA_EXT_ZABHA),
> >       __RISCV_ISA_EXT_DATA(zfa, RISCV_ISA_EXT_ZFA),
> >       __RISCV_ISA_EXT_DATA(zfh, RISCV_ISA_EXT_ZFH),
> >       __RISCV_ISA_EXT_DATA(zfhmin, RISCV_ISA_EXT_ZFHMIN),
>
> You're missing a dt-binding patch in this series adding zabha.

Thanks, I will add that to the v2.

>
> Thanks,
> Conor.
Alexandre Ghiti May 29, 2024, 8:04 a.m. UTC | #4
Hi Andrea,

On Tue, May 28, 2024 at 8:00 PM Andrea Parri <parri.andrea@gmail.com> wrote:
>
> > -#define __arch_xchg_masked(prepend, append, r, p, n)                 \
> > +#define __arch_xchg_masked(swap_sfx, prepend, append, r, p, n)               \
>
> This actually indicates a problem in the current (aka, no Zabha)
> implementation: without your series, xchg16() gets mapped to
>
>   lr.w     a2,(a3)
>   and      a1,a2,a5
>   or       a1,a1,a4
>   sc.w     a1,a1,(a3)
>   bnez     a1,43c <.L0^B1>
>
> which is clearly wrong... (other "fully-ordered LR/SC sequences"
> instead follow the mapping
>
>   lr.w     a2,(a3)
>   and      a1,a2,a5
>   or       a1,a1,a4
>   sc.w.rl  a1,a1,(a3)
>   bnez     a1,43c <.L0^B1>
>   fence    rw,rw  )
>
> A similar consideration for xchg8().

Ok, I will fix that separately and make it merged in -rc2.

Thanks,

Akex

>
>   Andrea
diff mbox series

Patch

diff --git a/arch/riscv/include/asm/cmpxchg.h b/arch/riscv/include/asm/cmpxchg.h
index 0789fbe38b23..43696d9e13aa 100644
--- a/arch/riscv/include/asm/cmpxchg.h
+++ b/arch/riscv/include/asm/cmpxchg.h
@@ -11,8 +11,14 @@ 
 #include <asm/fence.h>
 #include <asm/alternative.h>
 
-#define __arch_xchg_masked(prepend, append, r, p, n)			\
+#define __arch_xchg_masked(swap_sfx, prepend, append, r, p, n)		\
 ({									\
+	__label__ zabha, end;						\
+									\
+	asm goto(ALTERNATIVE("nop", "j %[zabha]", 0,			\
+			     RISCV_ISA_EXT_ZABHA, 1)			\
+			: : : : zabha);					\
+									\
 	u32 *__ptr32b = (u32 *)((ulong)(p) & ~0x3);			\
 	ulong __s = ((ulong)(p) & (0x4 - sizeof(*p))) * BITS_PER_BYTE;	\
 	ulong __mask = GENMASK(((sizeof(*p)) * BITS_PER_BYTE) - 1, 0)	\
@@ -34,6 +40,17 @@ 
 	       : "memory");						\
 									\
 	r = (__typeof__(*(p)))((__retx & __mask) >> __s);		\
+	goto end;							\
+									\
+zabha:									\
+	__asm__ __volatile__ (						\
+		prepend							\
+		"	amoswap" swap_sfx " %0, %z2, %1\n"		\
+		append							\
+		: "=&r" (r), "+A" (*(p))				\
+		: "rJ" (n)						\
+		: "memory");						\
+end:									\
 })
 
 #define __arch_xchg(sfx, prepend, append, r, p, n)			\
@@ -55,8 +72,11 @@ 
 									\
 	switch (sizeof(*__ptr)) {					\
 	case 1:								\
+		__arch_xchg_masked(".b" sfx, prepend, append,		\
+				   __ret, __ptr, __new);		\
+		break;							\
 	case 2:								\
-		__arch_xchg_masked(prepend, append,			\
+		__arch_xchg_masked(".h" sfx, prepend, append,		\
 				   __ret, __ptr, __new);		\
 		break;							\
 	case 4:								\
diff --git a/arch/riscv/include/asm/hwcap.h b/arch/riscv/include/asm/hwcap.h
index e17d0078a651..f71ddd2ca163 100644
--- a/arch/riscv/include/asm/hwcap.h
+++ b/arch/riscv/include/asm/hwcap.h
@@ -81,6 +81,7 @@ 
 #define RISCV_ISA_EXT_ZTSO		72
 #define RISCV_ISA_EXT_ZACAS		73
 #define RISCV_ISA_EXT_XANDESPMU		74
+#define RISCV_ISA_EXT_ZABHA		75
 
 #define RISCV_ISA_EXT_XLINUXENVCFG	127
 
diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
index 3ed2359eae35..8d0f56dd2f53 100644
--- a/arch/riscv/kernel/cpufeature.c
+++ b/arch/riscv/kernel/cpufeature.c
@@ -257,6 +257,7 @@  const struct riscv_isa_ext_data riscv_isa_ext[] = {
 	__RISCV_ISA_EXT_DATA(zihintpause, RISCV_ISA_EXT_ZIHINTPAUSE),
 	__RISCV_ISA_EXT_DATA(zihpm, RISCV_ISA_EXT_ZIHPM),
 	__RISCV_ISA_EXT_DATA(zacas, RISCV_ISA_EXT_ZACAS),
+	__RISCV_ISA_EXT_DATA(zabha, RISCV_ISA_EXT_ZABHA),
 	__RISCV_ISA_EXT_DATA(zfa, RISCV_ISA_EXT_ZFA),
 	__RISCV_ISA_EXT_DATA(zfh, RISCV_ISA_EXT_ZFH),
 	__RISCV_ISA_EXT_DATA(zfhmin, RISCV_ISA_EXT_ZFHMIN),