[PATCH] avr32: fix integer overflow in ELF_ET_DYN_BASE

[PATCH] avr32: fix integer overflow in ELF_ET_DYN_BASE

[Andrey Ryabinin]

Almost all arches define ELF_ET_DYN_BASE as 2/3 of TASK_SIZE.
Though it seems that some architectures do this in a wrong way.
The problem is that 2*TASK_SIZE may overflow 32-bits so
the real ELF_ET_DYN_BASE becomes wrong.
Fix this overflow by dividing TASK_SIZE prior to multiplying:
	 (TASK_SIZE / 3 * 2)

Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
---
 arch/avr32/include/asm/elf.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/avr32/include/asm/elf.h b/arch/avr32/include/asm/elf.h
index d232888..0388ece 100644
--- a/arch/avr32/include/asm/elf.h
+++ b/arch/avr32/include/asm/elf.h
@@ -84,7 +84,7 @@ typedef struct user_fpu_struct elf_fpregset_t;
    the loader.  We need to make sure that it is out of the way of the program
    that it will "exec", and that there is sufficient room for the brk.  */
 
-#define ELF_ET_DYN_BASE         (2 * TASK_SIZE / 3)
+#define ELF_ET_DYN_BASE         (TASK_SIZE / 3 * 2)
 
 
 /* This yields a mask that user programs can use to figure out what
-- 
2.3.3

[PATCH] cris: fix integer overflow in ELF_ET_DYN_BASE

[Andrey Ryabinin]

Almost all arches define ELF_ET_DYN_BASE as 2/3 of TASK_SIZE.
Though it seems that some architectures do this in a wrong way.
The problem is that 2*TASK_SIZE may overflow 32-bits so
the real ELF_ET_DYN_BASE becomes wrong.
Fix this overflow by dividing TASK_SIZE prior to multiplying:
	(TASK_SIZE / 3 * 2)

Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
---
 arch/cris/include/asm/elf.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/cris/include/asm/elf.h b/arch/cris/include/asm/elf.h
index 30ded8f..c2a394f 100644
--- a/arch/cris/include/asm/elf.h
+++ b/arch/cris/include/asm/elf.h
@@ -71,7 +71,7 @@ typedef unsigned long elf_fpregset_t;
    the loader.  We need to make sure that it is out of the way of the program
    that it will "exec", and that there is sufficient room for the brk.  */
 
-#define ELF_ET_DYN_BASE         (2 * TASK_SIZE / 3)
+#define ELF_ET_DYN_BASE         (TASK_SIZE / 3 * 2)
 
 /* This yields a mask that user programs can use to figure out what
    instruction set this CPU supports.  This could be done in user space,
-- 
2.3.3

Re: [PATCH] cris: fix integer overflow in ELF_ET_DYN_BASE

[Jesper Nilsson]

On Tue, Mar 24, 2015 at 06:31:23PM +0300, Andrey Ryabinin wrote:
> Almost all arches define ELF_ET_DYN_BASE as 2/3 of TASK_SIZE.
> Though it seems that some architectures do this in a wrong way.
> The problem is that 2*TASK_SIZE may overflow 32-bits so
> the real ELF_ET_DYN_BASE becomes wrong.
> Fix this overflow by dividing TASK_SIZE prior to multiplying:
> 	(TASK_SIZE / 3 * 2)

Good catch, thanks, added to the CRIS-tree.

> Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>

/^JN - Jesper Nilsson
-- 
               Jesper Nilsson -- jesper.nilsson@axis.com

[PATCH] x86, UML: fix integer overflow in ELF_ET_DYN_BASE

[Andrey Ryabinin]

Almost all arches define ELF_ET_DYN_BASE as 2/3 of TASK_SIZE.
Though it seems that some architectures do this in a wrong way.
The problem is that 2*TASK_SIZE may overflow 32-bits so
the real ELF_ET_DYN_BASE becomes wrong.
Fix this overflow by dividing TASK_SIZE prior to multiplying:
	(TASK_SIZE / 3 * 2)

Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
---
 arch/x86/um/asm/elf.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/um/asm/elf.h b/arch/x86/um/asm/elf.h
index 25a1022..0a656b7 100644
--- a/arch/x86/um/asm/elf.h
+++ b/arch/x86/um/asm/elf.h
@@ -210,7 +210,7 @@ extern int elf_core_copy_fpregs(struct task_struct *t, elf_fpregset_t *fpu);
 
 #define ELF_EXEC_PAGESIZE 4096
 
-#define ELF_ET_DYN_BASE (2 * TASK_SIZE / 3)
+#define ELF_ET_DYN_BASE (TASK_SIZE / 3 * 2)
 
 extern long elf_aux_hwcap;
 #define ELF_HWCAP (elf_aux_hwcap)
-- 
2.3.3

Re: [PATCH] avr32: fix integer overflow in ELF_ET_DYN_BASE

[Hans-Christian Egtvedt]

Around Tue 24 Mar 2015 18:31:22 +0300 or thereabout, Andrey Ryabinin wrote:
> Almost all arches define ELF_ET_DYN_BASE as 2/3 of TASK_SIZE.
> Though it seems that some architectures do this in a wrong way.
> The problem is that 2*TASK_SIZE may overflow 32-bits so
> the real ELF_ET_DYN_BASE becomes wrong.
> Fix this overflow by dividing TASK_SIZE prior to multiplying:
> 	 (TASK_SIZE / 3 * 2)

Thank you for fixing (-:

> Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
>

Acked-by: Hans-Christian Egtvedt <egtvedt@samfundet.no>

Added to for-linus branch in
git://git.kernel.org/pub/scm/linux/kernel/git/egtvedt/linux-avr32.git
queued for Linux 4.1.

> ---
>  arch/avr32/include/asm/elf.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/avr32/include/asm/elf.h b/arch/avr32/include/asm/elf.h
> index d232888..0388ece 100644
> --- a/arch/avr32/include/asm/elf.h
> +++ b/arch/avr32/include/asm/elf.h
> @@ -84,7 +84,7 @@ typedef struct user_fpu_struct elf_fpregset_t;
>     the loader.  We need to make sure that it is out of the way of the program
>     that it will "exec", and that there is sufficient room for the brk.  */
>  
> -#define ELF_ET_DYN_BASE         (2 * TASK_SIZE / 3)
> +#define ELF_ET_DYN_BASE         (TASK_SIZE / 3 * 2)
>  
>  
>  /* This yields a mask that user programs can use to figure out what
-- 
Best regards,
Hans-Christian Egtvedt