Skip to contentSkip to conversation

ACPI: avoid NULL pointer arithmetic

PATCH Applied Arnd Bergmann 4 participants 7 messages Sep 27, 2021 – Sep 30, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ACPI: avoid NULL pointer arithmetic

From: Arnd Bergmann <[email protected]>

There are some very old macros for doing an open-coded offsetof() and cast between pointer and integer in ACPI headers. clang-14 now complains about these:

drivers/acpi/acpica/tbfadt.c:86:3: error: performing pointer subtraction with a null pointer has undefined behavior [-Werror,-Wnull-pointer-subtraction]

         ACPI_FADT_OFFSET(pm_timer_block),
         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actbl.h:376:47: note: expanded from macro 'ACPI_FADT_OFFSET'
 #define ACPI_FADT_OFFSET(f)             (u16) ACPI_OFFSET (struct acpi_table_fadt, f)
                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:511:41: note: expanded from macro 'ACPI_OFFSET'
 #define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
                                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:505:79: note: expanded from macro 'ACPI_PTR_DIFF'
 #define ACPI_PTR_DIFF(a, b)             ((acpi_size) (ACPI_CAST_PTR (u8, (a)) - ACPI_CAST_PTR (u8, (b))))
                                                                              ^ ~~~~~~~~~~~~~~~~~~~~~~~

Convert them to the modern equivalents.

---
 include/acpi/actypes.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
include/acpi/actypes.h +2 -2
index 92c71dfce0d5..285bc7b73de3 100644
@@ -507,8 +507,8 @@ typedef u64 acpi_integer;
 /* Pointer/Integer type conversions */
 
 #define ACPI_TO_POINTER(i)              ACPI_CAST_PTR (void, (acpi_size) (i))
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
 #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
 #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
 
Signature · 1 line
2.29.2
#1 lore raw

Bob, this is ACPICA material.

Would it be possible to apply this to the upstream from the patch or do you need  a PR for this?

Quoted 40 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

There are some very old macros for doing an open-coded offsetof() and cast between pointer and integer in ACPI headers. clang-14 now complains about these:

drivers/acpi/acpica/tbfadt.c:86:3: error: performing pointer subtraction with a null pointer has undefined behavior [-Werror,-Wnull-pointer-subtraction]

          ACPI_FADT_OFFSET(pm_timer_block),
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actbl.h:376:47: note: expanded from macro 'ACPI_FADT_OFFSET'
  #define ACPI_FADT_OFFSET(f)             (u16) ACPI_OFFSET (struct acpi_table_fadt, f)
                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:511:41: note: expanded from macro 'ACPI_OFFSET'
  #define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
                                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:505:79: note: expanded from macro 'ACPI_PTR_DIFF'
  #define ACPI_PTR_DIFF(a, b)             ((acpi_size) (ACPI_CAST_PTR (u8, (a)) - ACPI_CAST_PTR (u8, (b))))
                                                                               ^ ~~~~~~~~~~~~~~~~~~~~~~~

Convert them to the modern equivalents.

---
  include/acpi/actypes.h | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
include/acpi/actypes.h +2 -2
index 92c71dfce0d5..285bc7b73de3 100644
@@ -507,8 +507,8 @@ typedef u64 acpi_integer;
  /* Pointer/Integer type conversions */
 
  #define ACPI_TO_POINTER(i)              ACPI_CAST_PTR (void, (acpi_size) (i))
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
  #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
  #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
 
#2 lore raw

I can take this patch as-is, I think. I'll try for the next acpica release later this week.

-----Original Message-----
From: Wysocki, Rafael J <rafael.j.wysocki(a)intel.com>
Sent: Tuesday, September 28, 2021 10:44 AM
To: Arnd Bergmann <arnd(a)kernel.org>; Moore, Robert <robert.moore(a)intel.com>
Cc: Arnd Bergmann <arnd(a)arndb.de>; Nathan Chancellor <nathan(a)kernel.org>; Nick Desaulniers <ndesaulniers(a)google.com>; Erik Kaneda <erik.kaneda(a)intel.com>; linux-acpi(a)vger.kernel.org; devel(a)acpica.org; linux-kernel(a)vger.kernel.org; llvm(a)lists.linux.dev; Len Brown <lenb(a)kernel.org>; Rafael J. Wysocki <rafael(a)kernel.org>
Subject: Re: [PATCH] ACPI: avoid NULL pointer arithmetic

Bob, this is ACPICA material.

Would it be possible to apply this to the upstream from the patch or do you need  a PR for this?

Quoted 42 lines · Arnd Bergmann

From: Arnd Bergmann <arnd(a)arndb.de>

There are some very old macros for doing an open-coded offsetof() and cast between pointer and integer in ACPI headers. clang-14 now complains about these:

drivers/acpi/acpica/tbfadt.c:86:3: error: performing pointer subtraction with a null pointer has undefined behavior [-Werror,-Wnull-pointer-subtraction]

          ACPI_FADT_OFFSET(pm_timer_block),
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

include/acpi/actbl.h:376:47: note: expanded from macro 'ACPI_FADT_OFFSET'
#define ACPI_FADT_OFFSET(f) (u16) ACPI_OFFSET (struct acpi_table_fadt, f)

^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/acpi/actypes.h:511:41: note: expanded from macro 'ACPI_OFFSET'
#define ACPI_OFFSET(d, f) ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)

^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/acpi/actypes.h:505:79: note: expanded from macro 'ACPI_PTR_DIFF'
#define ACPI_PTR_DIFF(a, b) ((acpi_size) (ACPI_CAST_PTR (u8, (a)) - ACPI_CAST_PTR (u8, (b))))

^ ~~~~~~~~~~~~~~~~~~~~~~~ Convert them to the modern equivalents.

Signed-off-by Arnd Bergmann <arnd(a)arndb.de>
---
  include/acpi/actypes.h | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
include/acpi/actypes.h index renamed from include/acpi/actypes.h +0 -0

92c71dfce0d5..285bc7b73de3 100644

include/acpi/actypes.h +2 -2
@@ -507,8 +507,8 @@ typedef u64 acpi_integer;
  /* Pointer/Integer type conversions */
 
  #define ACPI_TO_POINTER(i)              ACPI_CAST_PTR (void, (acpi_size) (i))
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
  #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
  #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
 
#3 lore raw
Quoted 2 lines · Robert

I can take this patch as-is, I think. I'll try for the next acpica release later this week.

Thanks!

Quoted 56 lines

-----Original Message-----
From: Wysocki, Rafael J <rafael.j.wysocki(a)intel.com>
Sent: Tuesday, September 28, 2021 10:44 AM
To: Arnd Bergmann <arnd(a)kernel.org>; Moore, Robert <robert.moore(a)intel.com>
Cc: Arnd Bergmann <arnd(a)arndb.de>; Nathan Chancellor <nathan(a)kernel.org>; Nick Desaulniers <ndesaulniers(a)google.com>; Erik Kaneda <erik.kaneda(a)intel.com>; linux-acpi(a)vger.kernel.org; devel(a)acpica.org; linux-kernel(a)vger.kernel.org; llvm(a)lists.linux.dev; Len Brown <lenb(a)kernel.org>; Rafael J. Wysocki <rafael(a)kernel.org>
Subject: Re: [PATCH] ACPI: avoid NULL pointer arithmetic

Bob, this is ACPICA material.

Would it be possible to apply this to the upstream from the patch or do you need a PR for this?

Quoted 42 lines · Arnd Bergmann

From: Arnd Bergmann <arnd(a)arndb.de>

There are some very old macros for doing an open-coded offsetof() and cast between pointer and integer in ACPI headers. clang-14 now complains about these:

drivers/acpi/acpica/tbfadt.c:86:3: error: performing pointer subtraction with a null pointer has undefined behavior [-Werror,-Wnull-pointer-subtraction]

          ACPI_FADT_OFFSET(pm_timer_block),
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

include/acpi/actbl.h:376:47: note: expanded from macro 'ACPI_FADT_OFFSET'
#define ACPI_FADT_OFFSET(f) (u16) ACPI_OFFSET (struct acpi_table_fadt, f)

^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/acpi/actypes.h:511:41: note: expanded from macro 'ACPI_OFFSET'
#define ACPI_OFFSET(d, f) ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)

^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/acpi/actypes.h:505:79: note: expanded from macro 'ACPI_PTR_DIFF'
#define ACPI_PTR_DIFF(a, b) ((acpi_size) (ACPI_CAST_PTR (u8, (a)) - ACPI_CAST_PTR (u8, (b))))

^ ~~~~~~~~~~~~~~~~~~~~~~~ Convert them to the modern equivalents.

Signed-off-by Arnd Bergmann <arnd(a)arndb.de>
---
  include/acpi/actypes.h | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
include/acpi/actypes.h index renamed from include/acpi/actypes.h +0 -0

92c71dfce0d5..285bc7b73de3 100644

include/acpi/actypes.h +2 -2
@@ -507,8 +507,8 @@ typedef u64 acpi_integer;
  /* Pointer/Integer type conversions */
 
  #define ACPI_TO_POINTER(i)              ACPI_CAST_PTR (void, (acpi_size) (i))
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
  #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
  #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
 
maintainer Applied
#4 lore raw
In reply to Arnd Bergmann
Quoted 32 lines · Arnd Bergmann

From: Arnd Bergmann <arnd(a)arndb.de>

There are some very old macros for doing an open-coded offsetof() and cast between pointer and integer in ACPI headers. clang-14 now complains about these:

drivers/acpi/acpica/tbfadt.c:86:3: error: performing pointer subtraction with a null pointer has undefined behavior [-Werror,-Wnull-pointer-subtraction]

         ACPI_FADT_OFFSET(pm_timer_block),
         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actbl.h:376:47: note: expanded from macro 'ACPI_FADT_OFFSET'
 #define ACPI_FADT_OFFSET(f)             (u16) ACPI_OFFSET (struct acpi_table_fadt, f)
                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:511:41: note: expanded from macro 'ACPI_OFFSET'
 #define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
                                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:505:79: note: expanded from macro 'ACPI_PTR_DIFF'
 #define ACPI_PTR_DIFF(a, b)             ((acpi_size) (ACPI_CAST_PTR (u8, (a)) - ACPI_CAST_PTR (u8, (b))))
                                                                              ^ ~~~~~~~~~~~~~~~~~~~~~~~

Convert them to the modern equivalents.

Signed-off-by Arnd Bergmann <arnd(a)arndb.de>
---
 include/acpi/actypes.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
include/acpi/actypes.h +0 -0
index 92c71dfce0d5..285bc7b73de3 100644
@@ -507,8 +507,8 @@ typedef u64 acpi_integer;
 /* Pointer/Integer type conversions */

include/acpi/actypes.h typedef u64 acpi_integer;

Quoted 10 lines
include/acpi/actypes.h +2 -3
 
 #define ACPI_TO_POINTER(i)              ACPI_CAST_PTR (void, (acpi_size) (i))
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
 #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
 #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
 
--

Queued up as 5.16 material, converted into an upstream ACPICA pull request and submitted, thanks!

#5 lore raw
Quoted 43 lines · Rafael J. Wysocki
Quoted 41 lines · Arnd Bergmann

From: Arnd Bergmann <arnd(a)arndb.de>

There are some very old macros for doing an open-coded offsetof() and cast between pointer and integer in ACPI headers. clang-14 now complains about these:

drivers/acpi/acpica/tbfadt.c:86:3: error: performing pointer subtraction with a null pointer has undefined behavior [-Werror,-Wnull-pointer-subtraction]

         ACPI_FADT_OFFSET(pm_timer_block),
         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actbl.h:376:47: note: expanded from macro 'ACPI_FADT_OFFSET'
 #define ACPI_FADT_OFFSET(f)             (u16) ACPI_OFFSET (struct acpi_table_fadt, f)
                                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:511:41: note: expanded from macro 'ACPI_OFFSET'
 #define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
                                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/acpi/actypes.h:505:79: note: expanded from macro 'ACPI_PTR_DIFF'
 #define ACPI_PTR_DIFF(a, b)             ((acpi_size) (ACPI_CAST_PTR (u8, (a)) - ACPI_CAST_PTR (u8, (b))))
                                                                              ^ ~~~~~~~~~~~~~~~~~~~~~~~

Convert them to the modern equivalents.

Signed-off-by Arnd Bergmann <arnd(a)arndb.de>
---
 include/acpi/actypes.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
include/acpi/actypes.h +2 -2
index 92c71dfce0d5..285bc7b73de3 100644
@@ -507,8 +507,8 @@ typedef u64 acpi_integer;
 /* Pointer/Integer type conversions */
 
 #define ACPI_TO_POINTER(i)              ACPI_CAST_PTR (void, (acpi_size) (i))
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
 #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
 #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
 
Quoted 4 lines
Quoted 1 line

--

Queued up as 5.16 material, converted into an upstream ACPICA pull request and submitted, thanks!

And reverted from there, because it introduced build issues.

Can we use alternative definitions that don't depend on uintptr_t and offsetof()?

#6 lore raw
Quoted 18 lines · Rafael J. Wysocki
Quoted 11 lines · Rafael J. Wysocki
Quoted 9 lines · Arnd Bergmann

#define ACPI_TO_POINTER(i) ACPI_CAST_PTR (void, (acpi_size) (i))

include/acpi/actypes.h +2 -2
-#define ACPI_TO_INTEGER(p)              ACPI_PTR_DIFF (p, (void *) 0)
-#define ACPI_OFFSET(d, f)               ACPI_PTR_DIFF (&(((d *) 0)->f), (void *) 0)
+#define ACPI_TO_INTEGER(p)              ((uintptr_t)(p))
+#define ACPI_OFFSET(d, f)               offsetof(d, f)
 #define ACPI_PHYSADDR_TO_PTR(i)         ACPI_TO_POINTER(i)
 #define ACPI_PTR_TO_PHYSADDR(i)         ACPI_TO_INTEGER(i)
Quoted 4 lines
Quoted 1 line

--

Queued up as 5.16 material, converted into an upstream ACPICA pull request and submitted, thanks!

Quoted 4 lines

And reverted from there, because it introduced build issues.

Can we use alternative definitions that don't depend on uintptr_t and offsetof()?

It's a bit tricky, as both were introduced to avoid portability issues.

For uintptr_t, we could use 'unsigned long', which works on everything that Linux can run on, but wouldn't work if this code can be compiled for 64-bit Windows. 'size_t' probably works, but likely has the same problem as 'uintptr_t' because they require and additional #include. I see that some code uses acpi_uintptr_t, which looks like it is meant to replace uintptr_t, this is defined as 'void *' in include/acpi/actypes.h, so that probably wouldn't avoid the warning.

For offsetof(), we could use __builtin_offsetof(), which would work with any gcc-compatible compiler, if the goal is to avoid including <stddef.h>. If it has to work on other compilers, there is no portable way that doesn't rely on standard headers. The best idea I'd have would be to use "#ifdef offsetof" to choose between the trivial implementation I had and the old one that works for non-standard C but may invoke undefined behavior.

Arnd