[PATCH v3 0/2] riscv: enable lockless lockref implementation

[PATCH v3 0/2] riscv: enable lockless lockref implementation

[Jisheng Zhang]

This series selects ARCH_USE_CMPXCHG_LOCKREF to enable the
cmpxchg-based lockless lockref implementation for riscv. Then,
implement arch_cmpxchg64_{relaxed|acquire|release}.

After patch1:
Using Linus' test case[1] on TH1520 platform, I see a 11.2% improvement.
On JH7110 platform, I see 12.0% improvement.

After patch2:
on both TH1520 and JH7110 platforms, I didn't see obvious
performance improvement with Linus' test case [1]. IMHO, this may
be related with the fence and lr.d/sc.d hw implementations. In theory,
lr/sc without fence could give performance improvement over lr/sc plus
fence, so add the code here to leave performance improvement room on
newer HW platforms.

Link: http://marc.info/?l=linux-fsdevel&m=137782380714721&w=4 [1]

Since v2:
  - rebase on v6.8-rc1
  - collect Reviewed-by tag

Since v1:
  - only select ARCH_USE_CMPXCHG_LOCKREF if 64BIT

Jisheng Zhang (2):
  riscv: select ARCH_USE_CMPXCHG_LOCKREF
  riscv: cmpxchg: implement arch_cmpxchg64_{relaxed|acquire|release}

 arch/riscv/Kconfig               |  1 +
 arch/riscv/include/asm/cmpxchg.h | 18 ++++++++++++++++++
 2 files changed, 19 insertions(+)

-- 
2.43.0

[PATCH v3 1/2] riscv: select ARCH_USE_CMPXCHG_LOCKREF

[Jisheng Zhang]

Select ARCH_USE_CMPXCHG_LOCKREF to enable the cmpxchg-based lockless
lockref implementation for riscv.

Using Linus' test case[1] on TH1520 platform, I see a 11.2% improvement.
On JH7110 platform, I see 12.0% improvement.

Link: http://marc.info/?l=linux-fsdevel&m=137782380714721&w=4 [1]
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Reviewed-by: Andrea Parri <parri.andrea@gmail.com>
---
 arch/riscv/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
index fdd1a595ebd8..248cea3e2749 100644
--- a/arch/riscv/Kconfig
+++ b/arch/riscv/Kconfig
@@ -51,6 +51,7 @@ config RISCV
 	select ARCH_SUPPORTS_PAGE_TABLE_CHECK if MMU
 	select ARCH_SUPPORTS_PER_VMA_LOCK if MMU
 	select ARCH_SUPPORTS_SHADOW_CALL_STACK if HAVE_SHADOW_CALL_STACK
+	select ARCH_USE_CMPXCHG_LOCKREF if 64BIT
 	select ARCH_USE_MEMTEST
 	select ARCH_USE_QUEUED_RWLOCKS
 	select ARCH_USES_CFI_TRAPS if CFI_CLANG
-- 
2.43.0

[PATCH v3 2/2] riscv: cmpxchg: implement arch_cmpxchg64_{relaxed|acquire|release}

[Jisheng Zhang]

After selecting ARCH_USE_CMPXCHG_LOCKREF, one straight futher
optimization is implementing the arch_cmpxchg64_relaxed() because the
lockref code does not need the cmpxchg to have barrier semantics. At
the same time, implement arch_cmpxchg64_acquire and
arch_cmpxchg64_release as well.

However, on both TH1520 and JH7110 platforms, I didn't see obvious
performance improvement with Linus' test case [1]. IMHO, this may
be related with the fence and lr.d/sc.d hw implementations. In theory,
lr/sc without fence could give performance improvement over lr/sc plus
fence, so add the code here to leave performance improvement room on
newer HW platforms.

Link: http://marc.info/?l=linux-fsdevel&m=137782380714721&w=4 [1]
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Reviewed-by: Andrea Parri <parri.andrea@gmail.com>
---
 arch/riscv/include/asm/cmpxchg.h | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/arch/riscv/include/asm/cmpxchg.h b/arch/riscv/include/asm/cmpxchg.h
index 2f4726d3cfcc..6318187f426f 100644
--- a/arch/riscv/include/asm/cmpxchg.h
+++ b/arch/riscv/include/asm/cmpxchg.h
@@ -360,4 +360,22 @@
 	arch_cmpxchg_relaxed((ptr), (o), (n));				\
 })
 
+#define arch_cmpxchg64_relaxed(ptr, o, n)				\
+({									\
+	BUILD_BUG_ON(sizeof(*(ptr)) != 8);				\
+	arch_cmpxchg_relaxed((ptr), (o), (n));				\
+})
+
+#define arch_cmpxchg64_acquire(ptr, o, n)				\
+({									\
+	BUILD_BUG_ON(sizeof(*(ptr)) != 8);				\
+	arch_cmpxchg_acquire((ptr), (o), (n));				\
+})
+
+#define arch_cmpxchg64_release(ptr, o, n)				\
+({									\
+	BUILD_BUG_ON(sizeof(*(ptr)) != 8);				\
+	arch_cmpxchg_release((ptr), (o), (n));				\
+})
+
 #endif /* _ASM_RISCV_CMPXCHG_H */
-- 
2.43.0