Skip to contentSkip to conversation

RISC-V: Remove udivdi3

PATCH In review Palmer Dabbelt 4 participants 6 messages Aug 7, 2019 RISC-V

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 RISC-V: Remove udivdi3

This should never have landed in the first place: it was added as part of 64-bit divide support for 32-bit systems, but the kernel doesn't allow this sort of division. I must have forgotten to remove it.

This patch removes the support. Since this routine only worked on 64-bit platforms but was only built on 32-bit platforms, it's essentially just nonsense anyway.

---
 arch/riscv/lib/Makefile  |  2 --
 arch/riscv/lib/udivdi3.S | 32 --------------------------------
 2 files changed, 34 deletions(-)
 delete mode 100644 arch/riscv/lib/udivdi3.S
arch/riscv/lib/Makefile +0 -2
index 8e364ebf37de..267feaa10f6a 100644
@@ -5,5 +5,3 @@ lib-y	+= memset.o
 lib-y	+= uaccess.o
 
 lib-$(CONFIG_64BIT) += tishift.o
-
-lib-$(CONFIG_32BIT) += udivdi3.o
arch/riscv/lib/udivdi3.S deleted +0 -32
deleted file mode 100644
index 3f07476a91a9..000000000000
@@ -1,32 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0-only */
-/*
- * Copyright (C) 2016-2017 Free Software Foundation, Inc.
- */
-
-#include <linux/linkage.h>
-
-ENTRY(__udivdi3)
-	mv	a2, a1
-	mv	a1, a0
-	li	a0, -1
-	beqz	a2, .L5
-	li	a3, 1
-	bgeu	a2, a1, .L2
-.L1:
-	blez	a2, .L2
-	slli	a2, a2, 1
-	slli	a3, a3, 1
-	bgtu	a1, a2, .L1
-.L2:
-	li	a0, 0
-.L3:
-	bltu	a1, a2, .L4
-	sub	a1, a1, a2
-	or	a0, a0, a3
-.L4:
-	srli	a3, a3, 1
-	srli	a2, a2, 1
-	bnez	a3, .L3
-.L5:
-	ret
-ENDPROC(__udivdi3)
Signature · 7 lines
2.21.0


_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
#1 lore raw
Quoted 8 lines · Palmer Dabbelt

This should never have landed in the first place: it was added as part of 64-bit divide support for 32-bit systems, but the kernel doesn't allow this sort of division. I must have forgotten to remove it.

This patch removes the support. Since this routine only worked on 64-bit platforms but was only built on 32-bit platforms, it's essentially just nonsense anyway.

It would be more complete if we add "Reported-by: Eric Lin <[email protected]>" here.

Quoted 64 lines
---
 arch/riscv/lib/Makefile  |  2 --
 arch/riscv/lib/udivdi3.S | 32 --------------------------------
 2 files changed, 34 deletions(-)
 delete mode 100644 arch/riscv/lib/udivdi3.S
arch/riscv/lib/Makefile +0 -2
index 8e364ebf37de..267feaa10f6a 100644
@@ -5,5 +5,3 @@ lib-y	+= memset.o
 lib-y	+= uaccess.o
 
 lib-$(CONFIG_64BIT) += tishift.o
-
-lib-$(CONFIG_32BIT) += udivdi3.o
arch/riscv/lib/udivdi3.S deleted +0 -32
deleted file mode 100644
index 3f07476a91a9..000000000000
@@ -1,32 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0-only */
-/*
- * Copyright (C) 2016-2017 Free Software Foundation, Inc.
- */
-
-#include <linux/linkage.h>
-
-ENTRY(__udivdi3)
-	mv	a2, a1
-	mv	a1, a0
-	li	a0, -1
-	beqz	a2, .L5
-	li	a3, 1
-	bgeu	a2, a1, .L2
-.L1:
-	blez	a2, .L2
-	slli	a2, a2, 1
-	slli	a3, a3, 1
-	bgtu	a1, a2, .L1
-.L2:
-	li	a0, 0
-.L3:
-	bltu	a1, a2, .L4
-	sub	a1, a1, a2
-	or	a0, a0, a3
-.L4:
-	srli	a3, a3, 1
-	srli	a2, a2, 1
-	bnez	a3, .L3
-.L5:
-	ret
-ENDPROC(__udivdi3)
Signature · 7 lines
2.21.0


_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

#2 lore raw
Quoted 5 lines · Alan Kao
Quoted 4 lines · Palmer Dabbelt

This should never have landed in the first place: it was added as part of 64-bit divide support for 32-bit systems, but the kernel doesn't allow this sort of division. I must have forgotten to remove it.

Quoted 10 lines
Quoted 4 lines

This patch removes the support. Since this routine only worked on 64-bit platforms but was only built on 32-bit platforms, it's essentially just nonsense anyway.

It would be more complete if we add "Reported-by: Eric Lin <[email protected]>" here.

Quoted 2 lines

Thanks, queued for v5.3-rc with Eric's Reported-by.

- Paul

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

#3 lore raw
Quoted 12 lines · Paul Walmsley
Quoted 10 lines · Alan Kao
Quoted 8 lines · Palmer Dabbelt

This should never have landed in the first place: it was added as part of 64-bit divide support for 32-bit systems, but the kernel doesn't allow this sort of division. I must have forgotten to remove it.

This patch removes the support. Since this routine only worked on 64-bit platforms but was only built on 32-bit platforms, it's essentially just nonsense anyway.

Quoted 7 lines
Quoted 5 lines

It would be more complete if we add "Reported-by: Eric Lin <[email protected]>" here.

Quoted 2 lines

Thanks, queued for v5.3-rc with Eric's Reported-by.

This patch breaks the rv32_defconfig build, since udelay() still relies on __udivdi3(). I'll queue "riscv: delay: use do_div() instead of __udivdi3()" along with it:

https://lore.kernel.org/linux-riscv/[email protected]/T/#u

- Paul

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

#4 lore raw

Review

Quoted 13 lines · Paul Walmsley

This patch breaks the rv32_defconfig build, since udelay() still relies on __udivdi3(). I'll queue "riscv: delay: use do_div() instead of __udivdi3()" along with it:

https://lore.kernel.org/linux-riscv/[email protected]/T/#u

(quoted diff) +4 -1
+       u64 n;
+       u32 rem;
 
        if (unlikely(usecs > MAX_UDELAY_US)) {
-               __delay((u64)usecs * riscv_timebase / 1000000ULL);
+               n = (u64)usecs * riscv_timebase;
+               rem = do_div(n, 1000000);

You may omit rem and ignore the return value here.

Nicolas

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

#5 lore raw
Quoted 11 lines · Nicolas Pitre
Quoted 9 lines · Paul Walmsley

This patch breaks the rv32_defconfig build, since udelay() still relies on __udivdi3(). I'll queue "riscv: delay: use do_div() instead of __udivdi3()" along with it:

https://lore.kernel.org/linux-riscv/[email protected]/T/#u

(quoted diff) +2 -0
+       u64 n;
+       u32 rem;
Quoted 6 lines

Review

Quoted 4 lines
        if (unlikely(usecs > MAX_UDELAY_US)) {
(quoted diff) +2 -1
-               __delay((u64)usecs * riscv_timebase / 1000000ULL);
+               n = (u64)usecs * riscv_timebase;
+               rem = do_div(n, 1000000);

You may omit rem and ignore the return value here.

Thanks Nico. Updated patch below

- Paul

From: Paul Walmsley <[email protected]>
Date: Tue, 6 Aug 2019 18:28:33 -0700
Subject: [PATCH] riscv: delay: use do_div() instead of __udivdi3()

In preparation for removing __udivdi3() from the RISC-V architecture-specific files, convert its one user to use do_div(). This avoids breaking the RV32 build after __udivdi3() is removed.

This second version removes the assignment of the remainder to an unused temporary variable. Thanks to Nicolas Pitre <[email protected]> for the suggestion.

---
 arch/riscv/lib/delay.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
arch/riscv/lib/delay.c +5 -1
index 87ff89e88f2c..f51c9a03bca1 100644
@@ -81,9 +81,13 @@ EXPORT_SYMBOL(__delay);
 void udelay(unsigned long usecs)
 {
 	u64 ucycles = (u64)usecs * lpj_fine * UDELAY_MULT;
+	u64 n;
 
 	if (unlikely(usecs > MAX_UDELAY_US)) {
-		__delay((u64)usecs * riscv_timebase / 1000000ULL);
+		n = (u64)usecs * riscv_timebase;
+		do_div(n, 1000000);
+
+		__delay(n);
 		return;
 	}
 
Signature · 7 lines
2.22.0


_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv