2022-01-11 15:55:31

by Rafael J. Wysocki

[permalink] [raw]
Subject: [PATCH] ACPICA: Use uintptr_t and offsetof() in Linux kernel builds

From: Rafael J. Wysocki <[email protected]>

To avoid "performing pointer subtraction with a null pointer has
undefined behavior" compiler warnings, use uintptr_t and offsetof()
that are always available during Linux kernel builds to define
acpi_uintptr_t and the ACPI_TO_INTEGER() and ACPI_OFFSET() macros.

Based on earlier proposal from Arnd Bergmann.

Link: https://lore.kernel.org/linux-acpi/[email protected]
Signed-off-by: Rafael J. Wysocki <[email protected]>
---
include/acpi/actypes.h | 4 ++++
include/acpi/platform/aclinux.h | 5 +++++
2 files changed, 9 insertions(+)

Index: linux-pm/include/acpi/platform/aclinux.h
===================================================================
--- linux-pm.orig/include/acpi/platform/aclinux.h
+++ linux-pm/include/acpi/platform/aclinux.h
@@ -114,6 +114,11 @@
#define acpi_raw_spinlock raw_spinlock_t *
#define acpi_cpu_flags unsigned long

+#define acpi_uintptr_t uintptr_t
+
+#define ACPI_TO_INTEGER(p) ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f) offsetof(d, f)
+
/* Use native linux version of acpi_os_allocate_zeroed */

#define USE_NATIVE_ALLOCATE_ZEROED
Index: linux-pm/include/acpi/actypes.h
===================================================================
--- linux-pm.orig/include/acpi/actypes.h
+++ linux-pm/include/acpi/actypes.h
@@ -507,8 +507,12 @@ typedef u64 acpi_integer;
/* Pointer/Integer type conversions */

#define ACPI_TO_POINTER(i) ACPI_CAST_PTR (void, (acpi_size) (i))
+#ifndef ACPI_TO_INTEGER
#define ACPI_TO_INTEGER(p) ACPI_PTR_DIFF (p, (void *) 0)
+#endif
+#ifndef ACPI_OFFSET
#define ACPI_OFFSET(d, f) ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#endif
#define ACPI_PTR_TO_PHYSADDR(i) ACPI_TO_INTEGER(i)

/* Optimizations for 4-character (32-bit) acpi_name manipulation */





2022-01-12 08:45:27

by Arnd Bergmann

[permalink] [raw]
Subject: Re: [PATCH] ACPICA: Use uintptr_t and offsetof() in Linux kernel builds

On Tue, Jan 11, 2022 at 4:55 PM Rafael J. Wysocki <[email protected]> wrote:
>
> From: Rafael J. Wysocki <[email protected]>
>
> To avoid "performing pointer subtraction with a null pointer has
> undefined behavior" compiler warnings, use uintptr_t and offsetof()
> that are always available during Linux kernel builds to define
> acpi_uintptr_t and the ACPI_TO_INTEGER() and ACPI_OFFSET() macros.
>
> Based on earlier proposal from Arnd Bergmann.
>
> Link: https://lore.kernel.org/linux-acpi/[email protected]
> Signed-off-by: Rafael J. Wysocki <[email protected]>

Thanks for the follow-up. I can't easily test this at the moment, but
it looks correct to
me. I had a different approach that I had planned to eventually
submit, but yours
looks better anyway, so let's go with this.

Reviewed-by: Arnd Bergmann <[email protected]>

2022-01-26 03:10:24

by Rafael J. Wysocki

[permalink] [raw]
Subject: Re: [PATCH] ACPICA: Use uintptr_t and offsetof() in Linux kernel builds

On Wed, Jan 12, 2022 at 9:45 AM Arnd Bergmann <[email protected]> wrote:
>
> On Tue, Jan 11, 2022 at 4:55 PM Rafael J. Wysocki <[email protected]> wrote:
> >
> > From: Rafael J. Wysocki <[email protected]>
> >
> > To avoid "performing pointer subtraction with a null pointer has
> > undefined behavior" compiler warnings, use uintptr_t and offsetof()
> > that are always available during Linux kernel builds to define
> > acpi_uintptr_t and the ACPI_TO_INTEGER() and ACPI_OFFSET() macros.
> >
> > Based on earlier proposal from Arnd Bergmann.
> >
> > Link: https://lore.kernel.org/linux-acpi/[email protected]
> > Signed-off-by: Rafael J. Wysocki <[email protected]>
>
> Thanks for the follow-up. I can't easily test this at the moment, but
> it looks correct to
> me. I had a different approach that I had planned to eventually
> submit, but yours
> looks better anyway, so let's go with this.
>
> Reviewed-by: Arnd Bergmann <[email protected]>

Thank you, I've applied this patch as 5.18 material.

A pull request for the analogous change in the upstream code base has
been submitted and Bob is going to take it AFAICS.