[PATCH 1/1] sh: add support for cmpxchg on u8 and u16 pointers

[PATCH 1/1] sh: add support for cmpxchg on u8 and u16 pointers

[Liam Beguin]

The kernel test bot reported[1] that using set_mask_bits on a u8 causes
the following issue on SuperH:

    >> ERROR: modpost: "__cmpxchg_called_with_bad_pointer" [drivers/phy/ti/phy-tusb1210.ko] undefined!

Add support for cmpxchg on u8 and u16 pointers.

[1] https://lore.kernel.org/patchwork/patch/1288894/#1485536

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Liam Beguin <liambeguin@gmail.com>
---

Hi,

This was reported by the kernel test bot on an architecture I can't
really test on. I was only able to make sure the build succeeds, but
nothing more.
This patch is based on the __cmpxchg_u32 impletmentation and seems
incomplete based on the different cmpxchg headers I can find.

Do these function need to be impletmented in each header
simulataneously?

Thanks,
Liam

 arch/sh/include/asm/cmpxchg-irq.h | 27 +++++++++++++++++++++++++++
 arch/sh/include/asm/cmpxchg.h     |  5 +++--
 2 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/arch/sh/include/asm/cmpxchg-irq.h b/arch/sh/include/asm/cmpxchg-irq.h
index 07d3e7f08389..918c4153a930 100644
--- a/arch/sh/include/asm/cmpxchg-irq.h
+++ b/arch/sh/include/asm/cmpxchg-irq.h
@@ -51,4 +51,31 @@ static inline unsigned long __cmpxchg_u32(volatile int *m, unsigned long old,
 	return retval;
 }
 
+static inline unsigned long __cmpxchg_u16(volatile u16 *m, unsigned long old,
+	unsigned long new)
+{
+	u16 retval;
+	unsigned long flags;
+
+	local_irq_save(flags);
+	retval = *m;
+	if (retval == old)
+		*m = new;
+	local_irq_restore(flags);
+	return (unsigned long)retval;
+}
+
+static inline unsigned long __cmpxchg_u8(volatile u8 *m, unsigned long old,
+	unsigned long new)
+{
+	u8 retval;
+	unsigned long flags;
+
+	local_irq_save(flags);
+	retval = *m;
+	if (retval == old)
+		*m = new;
+	local_irq_restore(flags);
+	return (unsigned long)retval;
+}
 #endif /* __ASM_SH_CMPXCHG_IRQ_H */
diff --git a/arch/sh/include/asm/cmpxchg.h b/arch/sh/include/asm/cmpxchg.h
index e9501d85c278..7d65d0fd1665 100644
--- a/arch/sh/include/asm/cmpxchg.h
+++ b/arch/sh/include/asm/cmpxchg.h
@@ -56,8 +56,9 @@ static inline unsigned long __cmpxchg(volatile void * ptr, unsigned long old,
 		unsigned long new, int size)
 {
 	switch (size) {
-	case 4:
-		return __cmpxchg_u32(ptr, old, new);
+	case 4: return __cmpxchg_u32((int *)ptr, old, new);
+	case 2: return __cmpxchg_u16((u16 *)ptr, old, new);
+	case 1: return __cmpxchg_u8((u8 *)ptr, old, new);
 	}
 	__cmpxchg_called_with_bad_pointer();
 	return old;
-- 
2.27.0

Re: [PATCH 1/1] sh: add support for cmpxchg on u8 and u16 pointers

[Geert Uytterhoeven]

Hi Liam,

On Wed, Aug 19, 2020 at 5:07 AM Liam Beguin <liambeguin@gmail.com> wrote:
> The kernel test bot reported[1] that using set_mask_bits on a u8 causes
> the following issue on SuperH:
>
>     >> ERROR: modpost: "__cmpxchg_called_with_bad_pointer" [drivers/phy/ti/phy-tusb1210.ko] undefined!
>
> Add support for cmpxchg on u8 and u16 pointers.
>
> [1] https://lore.kernel.org/patchwork/patch/1288894/#1485536
>
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Liam Beguin <liambeguin@gmail.com>
> ---
>
> Hi,
>
> This was reported by the kernel test bot on an architecture I can't
> really test on. I was only able to make sure the build succeeds, but
> nothing more.
> This patch is based on the __cmpxchg_u32 impletmentation and seems
> incomplete based on the different cmpxchg headers I can find.

Indeed. This version is suitable for non-SMP machines only.
BTW, it looks like this version can be replaced by the one in asm-generic?

>
> Do these function need to be impletmented in each header
> simulataneously?

Yes, we need them for all variants.

>  arch/sh/include/asm/cmpxchg-irq.h | 27 +++++++++++++++++++++++++++
>  arch/sh/include/asm/cmpxchg.h     |  5 +++--
>  2 files changed, 30 insertions(+), 2 deletions(-)
>
> diff --git a/arch/sh/include/asm/cmpxchg-irq.h b/arch/sh/include/asm/cmpxchg-irq.h
> index 07d3e7f08389..918c4153a930 100644
> --- a/arch/sh/include/asm/cmpxchg-irq.h
> +++ b/arch/sh/include/asm/cmpxchg-irq.h
> @@ -51,4 +51,31 @@ static inline unsigned long __cmpxchg_u32(volatile int *m, unsigned long old,
>         return retval;
>  }
>
> +static inline unsigned long __cmpxchg_u16(volatile u16 *m, unsigned long old,
> +       unsigned long new)
> +{
> +       u16 retval;
> +       unsigned long flags;
> +
> +       local_irq_save(flags);
> +       retval = *m;
> +       if (retval == old)
> +               *m = new;
> +       local_irq_restore(flags);
> +       return (unsigned long)retval;
> +}
> +
> +static inline unsigned long __cmpxchg_u8(volatile u8 *m, unsigned long old,
> +       unsigned long new)
> +{
> +       u8 retval;
> +       unsigned long flags;
> +
> +       local_irq_save(flags);
> +       retval = *m;
> +       if (retval == old)
> +               *m = new;
> +       local_irq_restore(flags);
> +       return (unsigned long)retval;
> +}
>  #endif /* __ASM_SH_CMPXCHG_IRQ_H */
> diff --git a/arch/sh/include/asm/cmpxchg.h b/arch/sh/include/asm/cmpxchg.h
> index e9501d85c278..7d65d0fd1665 100644
> --- a/arch/sh/include/asm/cmpxchg.h
> +++ b/arch/sh/include/asm/cmpxchg.h
> @@ -56,8 +56,9 @@ static inline unsigned long __cmpxchg(volatile void * ptr, unsigned long old,
>                 unsigned long new, int size)
>  {
>         switch (size) {
> -       case 4:
> -               return __cmpxchg_u32(ptr, old, new);
> +       case 4: return __cmpxchg_u32((int *)ptr, old, new);
> +       case 2: return __cmpxchg_u16((u16 *)ptr, old, new);
> +       case 1: return __cmpxchg_u8((u8 *)ptr, old, new);
>         }
>         __cmpxchg_called_with_bad_pointer();
>         return old;

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: [PATCH 1/1] sh: add support for cmpxchg on u8 and u16 pointers

[Liam Beguin]

Hi Geert,

On Wed Aug 19, 2020 at 5:09 AM EDT, Geert Uytterhoeven wrote:
> Hi Liam,
>
> On Wed, Aug 19, 2020 at 5:07 AM Liam Beguin <liambeguin@gmail.com>
> wrote:
> > The kernel test bot reported[1] that using set_mask_bits on a u8 causes
> > the following issue on SuperH:
> >
> >     >> ERROR: modpost: "__cmpxchg_called_with_bad_pointer" [drivers/phy/ti/phy-tusb1210.ko] undefined!
> >
> > Add support for cmpxchg on u8 and u16 pointers.
> >
> > [1] https://lore.kernel.org/patchwork/patch/1288894/#1485536
> >
> > Reported-by: kernel test robot <lkp@intel.com>
> > Signed-off-by: Liam Beguin <liambeguin@gmail.com>
> > ---
> >
> > Hi,
> >
> > This was reported by the kernel test bot on an architecture I can't
> > really test on. I was only able to make sure the build succeeds, but
> > nothing more.
> > This patch is based on the __cmpxchg_u32 impletmentation and seems
> > incomplete based on the different cmpxchg headers I can find.
>
> Indeed. This version is suitable for non-SMP machines only.
> BTW, it looks like this version can be replaced by the one in
> asm-generic?
>

Thanks for your feedback I'll have a look at the asm-generic functions
and try to use those instead.

> >
> > Do these function need to be impletmented in each header
> > simulataneously?
>
> Yes, we need them for all variants.
>

Okay, I'll look into that. Would you recommend a good way to test these
changes?

> >  arch/sh/include/asm/cmpxchg-irq.h | 27 +++++++++++++++++++++++++++
> >  arch/sh/include/asm/cmpxchg.h     |  5 +++--
> >  2 files changed, 30 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/sh/include/asm/cmpxchg-irq.h b/arch/sh/include/asm/cmpxchg-irq.h
> > index 07d3e7f08389..918c4153a930 100644
> > --- a/arch/sh/include/asm/cmpxchg-irq.h
> > +++ b/arch/sh/include/asm/cmpxchg-irq.h
> > @@ -51,4 +51,31 @@ static inline unsigned long __cmpxchg_u32(volatile int *m, unsigned long old,
> >         return retval;
> >  }
> >
> > +static inline unsigned long __cmpxchg_u16(volatile u16 *m, unsigned long old,
> > +       unsigned long new)
> > +{
> > +       u16 retval;
> > +       unsigned long flags;
> > +
> > +       local_irq_save(flags);
> > +       retval = *m;
> > +       if (retval == old)
> > +               *m = new;
> > +       local_irq_restore(flags);
> > +       return (unsigned long)retval;
> > +}
> > +
> > +static inline unsigned long __cmpxchg_u8(volatile u8 *m, unsigned long old,
> > +       unsigned long new)
> > +{
> > +       u8 retval;
> > +       unsigned long flags;
> > +
> > +       local_irq_save(flags);
> > +       retval = *m;
> > +       if (retval == old)
> > +               *m = new;
> > +       local_irq_restore(flags);
> > +       return (unsigned long)retval;
> > +}
> >  #endif /* __ASM_SH_CMPXCHG_IRQ_H */
> > diff --git a/arch/sh/include/asm/cmpxchg.h b/arch/sh/include/asm/cmpxchg.h
> > index e9501d85c278..7d65d0fd1665 100644
> > --- a/arch/sh/include/asm/cmpxchg.h
> > +++ b/arch/sh/include/asm/cmpxchg.h
> > @@ -56,8 +56,9 @@ static inline unsigned long __cmpxchg(volatile void * ptr, unsigned long old,
> >                 unsigned long new, int size)
> >  {
> >         switch (size) {
> > -       case 4:
> > -               return __cmpxchg_u32(ptr, old, new);
> > +       case 4: return __cmpxchg_u32((int *)ptr, old, new);
> > +       case 2: return __cmpxchg_u16((u16 *)ptr, old, new);
> > +       case 1: return __cmpxchg_u8((u8 *)ptr, old, new);
> >         }
> >         __cmpxchg_called_with_bad_pointer();
> >         return old;
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 --
> geert@linux-m68k.org
>
> 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

Thanks for your time,
Liam

Re: [PATCH 1/1] sh: add support for cmpxchg on u8 and u16 pointers

[Geert Uytterhoeven]

Hi Liam,

On Wed, Aug 19, 2020 at 3:34 PM Liam Beguin <liambeguin@gmail.com> wrote:
> On Wed Aug 19, 2020 at 5:09 AM EDT, Geert Uytterhoeven wrote:
> > On Wed, Aug 19, 2020 at 5:07 AM Liam Beguin <liambeguin@gmail.com>
> > wrote:
> > > The kernel test bot reported[1] that using set_mask_bits on a u8 causes
> > > the following issue on SuperH:
> > >
> > >     >> ERROR: modpost: "__cmpxchg_called_with_bad_pointer" [drivers/phy/ti/phy-tusb1210.ko] undefined!
> > >
> > > Add support for cmpxchg on u8 and u16 pointers.
> > >
> > > [1] https://lore.kernel.org/patchwork/patch/1288894/#1485536
> > >
> > > Reported-by: kernel test robot <lkp@intel.com>
> > > Signed-off-by: Liam Beguin <liambeguin@gmail.com>
> > > ---
> > >
> > > Hi,
> > >
> > > This was reported by the kernel test bot on an architecture I can't
> > > really test on. I was only able to make sure the build succeeds, but
> > > nothing more.
> > > This patch is based on the __cmpxchg_u32 impletmentation and seems
> > > incomplete based on the different cmpxchg headers I can find.
> >
> > Indeed. This version is suitable for non-SMP machines only.
> > BTW, it looks like this version can be replaced by the one in
> > asm-generic?
> >
>
> Thanks for your feedback I'll have a look at the asm-generic functions
> and try to use those instead.
>
> > >
> > > Do these function need to be impletmented in each header
> > > simulataneously?
> >
> > Yes, we need them for all variants.
> >
>
> Okay, I'll look into that. Would you recommend a good way to test these
> changes?

That's gonna be harder, I'm afraid.
Who has suitable hardware?

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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