Skip to contentSkip to conversation

sh: clk: Fix discarding const qualifier warning

PATCH In review Krzysztof Kozlowski 3 participants 6 messages Jan 7, 2020 SuperH

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sh: clk: Fix discarding const qualifier warning

ioreadX() accepts pointer to non-const memory. This fixes warnings like:

drivers/sh/clk/cpg.c: In function ‘r8’:
drivers/sh/clk/cpg.c:41:17: warning: passing argument 1 of ‘ioread8’
discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]

---
 drivers/sh/clk/cpg.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/sh/clk/cpg.c +4 -4
index eeb028b9cdb3..4f3d99d37809 100644
@@ -36,17 +36,17 @@ static void sh_clk_write(int value, struct clk *clk)
 		iowrite32(value, clk->mapped_reg);
 }
 
-static unsigned int r8(const void __iomem *addr)
+static unsigned int r8(void __iomem *addr)
 {
 	return ioread8(addr);
 }
 
-static unsigned int r16(const void __iomem *addr)
+static unsigned int r16(void __iomem *addr)
 {
 	return ioread16(addr);
 }
 
-static unsigned int r32(const void __iomem *addr)
+static unsigned int r32(void __iomem *addr)
 {
 	return ioread32(addr);
 }
@@ -55,7 +55,7 @@ static int sh_clk_mstp_enable(struct clk *clk)
 {
 	sh_clk_write(sh_clk_read(clk) & ~(1 << clk->enable_bit), clk);
 	if (clk->status_reg) {
-		unsigned int (*read)(const void __iomem *addr);
+		unsigned int (*read)(void __iomem *addr);
 		int i;
 		void __iomem *mapped_status = (phys_addr_t)clk->status_reg -
 			(phys_addr_t)clk->enable_reg + clk->mapped_reg;
Signature · 1 line
2.7.4
#1 lore raw

Hi Krzysztof,

Quoted 8 lines · Krzysztof Kozlowski

ioreadX() accepts pointer to non-const memory. This fixes warnings like:

drivers/sh/clk/cpg.c: In function ‘r8’:
drivers/sh/clk/cpg.c:41:17: warning: passing argument 1 of ‘ioread8’
discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]

Thanks for your patch!

drivers/sh/clk/cpg.c static void sh_clk_write(int value, struct clk *clk) · line 42

Quoted 11 lines
drivers/sh/clk/cpg.c +1 -1
@@ -36,17 +36,17 @@ static void sh_clk_write(int value, struct clk *clk)
                iowrite32(value, clk->mapped_reg);
 }
 
-static unsigned int r8(const void __iomem *addr)
+static unsigned int r8(void __iomem *addr)
 {
        return ioread8(addr);
 }

Isn't the real issue that some implementations of ioreadX() take const, while others don't?

Even the generic ones disagree:

include/asm-generic/io.h:static inline u8 ioread8(const volatile
void __iomem *addr)

    include/asm-generic/iomap.h:extern unsigned int ioread8(void __iomem *);

IMHO they all should take "const volatile void __iomem *".

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#2 lore raw
Quoted 30 lines · Geert Uytterhoeven

Hi Krzysztof,

Quoted 8 lines · Krzysztof Kozlowski

ioreadX() accepts pointer to non-const memory. This fixes warnings like:

drivers/sh/clk/cpg.c: In function ‘r8’:
drivers/sh/clk/cpg.c:41:17: warning: passing argument 1 of ‘ioread8’
discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]

Thanks for your patch!

drivers/sh/clk/cpg.c static void sh_clk_write(int value, struct clk *clk) · line 42

Quoted 11 lines
drivers/sh/clk/cpg.c +1 -1
@@ -36,17 +36,17 @@ static void sh_clk_write(int value, struct clk *clk)
                iowrite32(value, clk->mapped_reg);
 }
 
-static unsigned int r8(const void __iomem *addr)
+static unsigned int r8(void __iomem *addr)
 {
        return ioread8(addr);
 }

Isn't the real issue that some implementations of ioreadX() take const, while others don't?

Quoted 7 lines

Even the generic ones disagree:

include/asm-generic/io.h:static inline u8 ioread8(const volatile
void __iomem *addr)

    include/asm-generic/iomap.h:extern unsigned int ioread8(void __iomem *);

IMHO they all should take "const volatile void __iomem *".

Since this is a SuperH driver, I adjusted it to the SuperH implementation - lack of const. However iIndeed it makes sense to have them all taking "const"... Let me check, if I can fix it (without the real HW).

Best regards,
Krzysztof

#3 lore raw
Quoted 37 lines · Krzysztof Kozlowski
Quoted 36 lines · Geert Uytterhoeven

Hi Krzysztof,

Quoted 8 lines · Krzysztof Kozlowski

ioreadX() accepts pointer to non-const memory. This fixes warnings like:

drivers/sh/clk/cpg.c: In function ‘r8’:
drivers/sh/clk/cpg.c:41:17: warning: passing argument 1 of ‘ioread8’
discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]

Thanks for your patch!

drivers/sh/clk/cpg.c static void sh_clk_write(int value, struct clk *clk) · line 42

Quoted 11 lines
drivers/sh/clk/cpg.c +1 -1
@@ -36,17 +36,17 @@ static void sh_clk_write(int value, struct clk *clk)
                iowrite32(value, clk->mapped_reg);
 }
 
-static unsigned int r8(const void __iomem *addr)
+static unsigned int r8(void __iomem *addr)
 {
        return ioread8(addr);
 }

Isn't the real issue that some implementations of ioreadX() take const, while others don't?

Even the generic ones disagree:

include/asm-generic/io.h:static inline u8 ioread8(const volatile
void __iomem *addr)

    include/asm-generic/iomap.h:extern unsigned int ioread8(void __iomem *);
Quoted 6 lines
Quoted 1 line

IMHO they all should take "const volatile void __iomem *".

Since this is a SuperH driver, I adjusted it to the SuperH implementation - lack of const. However iIndeed it makes sense to have them all taking "const"... Let me check, if I can fix it (without the real HW).

That will be non-trivial because many platforms define ioreadX() with non-const. For example entire alpha with many its implementations of ioread(). Even include/asm-generic/iomap.h defines them as non-const...

Best regards,
Krzysztof

#4 lore raw

arch/powerpc/kernel/iomap.cOn Tue, Jan 7, 2020 at 2:33 PM Krzysztof

Quoted 11 lines · Kozlowski
Quoted 6 lines · Krzysztof Kozlowski

On Tue, 7 Jan 2020 at 14:00, Geert Uytterhoeven <[email protected]> wrote:

Since this is a SuperH driver, I adjusted it to the SuperH implementation - lack of const. However iIndeed it makes sense to have them all taking "const"... Let me check, if I can fix it (without the real HW).

That will be non-trivial because many platforms define ioreadX() with non-const. For example entire alpha with many its implementations of ioread(). Even include/asm-generic/iomap.h defines them as non-const...

I found these instances:

arch/alpha/include/asm/io.h
arch/alpha/kernel/io.c
arch/parisc/include/asm/io.h
arch/parisc/lib/iomap.c
arch/sh/kernel/iomap.c
arch/powerpc/kernel/iomap.c
lib/iomap.c
include/asm-generic/iomap.h

At least the last four file would have to be done at the same time as the header is shared, but the actual conversion should be trivial.

Arnd

#5 lore raw
Quoted 26 lines · Arnd Bergmann

arch/powerpc/kernel/iomap.cOn Tue, Jan 7, 2020 at 2:33 PM Krzysztof

Quoted 11 lines · Kozlowski
Quoted 6 lines · Krzysztof Kozlowski

On Tue, 7 Jan 2020 at 14:00, Geert Uytterhoeven <[email protected]> wrote:

Since this is a SuperH driver, I adjusted it to the SuperH implementation - lack of const. However iIndeed it makes sense to have them all taking "const"... Let me check, if I can fix it (without the real HW).

That will be non-trivial because many platforms define ioreadX() with non-const. For example entire alpha with many its implementations of ioread(). Even include/asm-generic/iomap.h defines them as non-const...

I found these instances:

arch/alpha/include/asm/io.h
arch/alpha/kernel/io.c
arch/parisc/include/asm/io.h
arch/parisc/lib/iomap.c
arch/sh/kernel/iomap.c
arch/powerpc/kernel/iomap.c
lib/iomap.c
include/asm-generic/iomap.h

Quoted 2 lines

At least the last four file would have to be done at the same time as the header is shared, but the actual conversion should be trivial.

Yes, assuming that I did not screw up some specific arch-implementation, it seems easy.

I have patchset ready - just need to build test it and I'll share for kbuild & company to test.

Best regards,
Krzysztof