[PATCH] riscv: Less inefficient gcc tishift helpers (and export their symbols)

From: Olof Johansson <olof@lixom.net>
To: Paul Walmsley <paul.walmsley@sifive.com>,
	Palmer Dabbelt <palmer@dabbelt.com>,
	Albert Ou <aou@eecs.berkeley.edu>
Cc: Olof Johansson <olof@lixom.net>,
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: [PATCH] riscv: Less inefficient gcc tishift helpers (and export their symbols)
Date: Mon, 16 Dec 2019 20:06:31 -0800	[thread overview]
Message-ID: <20191217040631.91886-1-olof@lixom.net> (raw)
In-Reply-To: <20191217003057.39300-1-olof@lixom.net>

The existing __lshrti3 was really inefficient, and the other two helpers
are also needed to compile some modules.

Add the missing versions, and export all of the symbols like arm64
already does.

This fixes a build break triggered by ubsan:

riscv64-unknown-linux-gnu-ld: lib/ubsan.o: in function `.L2':
ubsan.c:(.text.unlikely+0x38): undefined reference to `__ashlti3'
riscv64-unknown-linux-gnu-ld: ubsan.c:(.text.unlikely+0x42): undefined reference to `__ashrti3'

Signed-off-by: Olof Johansson <olof@lixom.net>
---
 arch/riscv/include/asm/asm-prototypes.h |  4 ++
 arch/riscv/lib/tishift.S                | 71 +++++++++++++++++++++++++--------
 2 files changed, 59 insertions(+), 16 deletions(-)

diff --git a/arch/riscv/include/asm/asm-prototypes.h b/arch/riscv/include/asm/asm-prototypes.h
index dd62b691c443d..27e005fca5849 100644
--- a/arch/riscv/include/asm/asm-prototypes.h
+++ b/arch/riscv/include/asm/asm-prototypes.h
@@ -5,4 +5,8 @@
 #include <linux/ftrace.h>
 #include <asm-generic/asm-prototypes.h>
 
+long long __lshrti3(long long a, int b);
+long long __ashrti3(long long a, int b);
+long long __ashlti3(long long a, int b);
+
 #endif /* _ASM_RISCV_PROTOTYPES_H */
diff --git a/arch/riscv/lib/tishift.S b/arch/riscv/lib/tishift.S
index 15f9d54c7db63..64a7b6abd38fb 100644
--- a/arch/riscv/lib/tishift.S
+++ b/arch/riscv/lib/tishift.S
@@ -4,34 +4,73 @@
  */
 
 #include <linux/linkage.h>
+#include <asm-generic/export.h>
 
 ENTRY(__lshrti3)
 	beqz	a2, .L1
 	li	a5,64
 	sub	a5,a5,a2
-	addi	sp,sp,-16
 	sext.w	a4,a5
 	blez	a5, .L2
 	sext.w	a2,a2
-	sll	a4,a1,a4
 	srl	a0,a0,a2
-	srl	a1,a1,a2
+	sll	a4,a1,a4
+	srl	a2,a1,a2
 	or	a0,a0,a4
-	sd	a1,8(sp)
-	sd	a0,0(sp)
-	ld	a0,0(sp)
-	ld	a1,8(sp)
-	addi	sp,sp,16
-	ret
+	mv	a1,a2
 .L1:
 	ret
 .L2:
-	negw	a4,a4
-	srl	a1,a1,a4
-	sd	a1,0(sp)
-	sd	zero,8(sp)
-	ld	a0,0(sp)
-	ld	a1,8(sp)
-	addi	sp,sp,16
+	negw	a0,a4
+	li	a2,0
+	srl	a0,a1,a0
+	mv	a1,a2
 	ret
 ENDPROC(__lshrti3)
+EXPORT_SYMBOL(__lshrti3)
+
+ENTRY(__ashrti3)
+	beqz	a2, .L3
+	li	a5,64
+	sub	a5,a5,a2
+	sext.w	a4,a5
+	blez	a5, .L4
+	sext.w	a2,a2
+	srl	a0,a0,a2
+	sll	a4,a1,a4
+	sra	a2,a1,a2
+	or	a0,a0,a4
+	mv	a1,a2
+.L3:
+	ret
+.L4:
+	negw	a0,a4
+	srai	a2,a1,0x3f
+	sra	a0,a1,a0
+	mv	a1,a2
+	ret
+ENDPROC(__ashrti3)
+EXPORT_SYMBOL(__ashrti3)
+
+ENTRY(__ashlti3)
+	beqz	a2, .L5
+	li	a5,64
+	sub	a5,a5,a2
+	sext.w	a4,a5
+	blez	a5, .L6
+	sext.w	a2,a2
+	sll	a1,a1,a2
+	srl	a4,a0,a4
+	sll	a2,a0,a2
+	or	a1,a1,a4
+	mv	a0,a2
+.L5:
+	ret
+.L6:
+	negw	a1,a4
+	li	a2,0
+	sll	a1,a0,a1
+	mv	a0,a2
+	ret
+ENDPROC(__ashlti3)
+EXPORT_SYMBOL(__ashlti3)
-- 
2.11.0


