[v2] ARM: alignment: Make SIGBUS sent to userspace POSIXly correct

With the UM_SIGNAL alignment fault mode, no siginfo structure is
passed to userspace.

POSIX specifies how siginfo_t should be populated for alignment
faults, so this patch does just that:

  * si_signo = SIGBUS
  * si_code = BUS_ADRALN
  * si_addr = misaligned data address at which access was attempted

Signed-off-by: Dave Martin <dave.martin@linaro.org>
---
v2: si_addr in siginfo_t changed to contain the faulting data
    access address, not the address of the faulting instruction.

    This behaviour is consistent with the way other fault signals
    such as SIGSEGV are already reported, as well as matching the
    apparent intent of the POSIX sigaction interface.

 arch/arm/mm/alignment.c |   14 +++++++++++---
 1 files changed, 11 insertions(+), 3 deletions(-)

On Wed, 27 Jul 2011, Dave Martin wrote:

> With the UM_SIGNAL alignment fault mode, no siginfo structure is
> passed to userspace.
> 
> POSIX specifies how siginfo_t should be populated for alignment
> faults, so this patch does just that:
> 
>   * si_signo = SIGBUS
>   * si_code = BUS_ADRALN
>   * si_addr = misaligned data address at which access was attempted
> 
> Signed-off-by: Dave Martin <dave.martin@linaro.org>
> ---
> v2: si_addr in siginfo_t changed to contain the faulting data
>     access address, not the address of the faulting instruction.
> 
>     This behaviour is consistent with the way other fault signals
>     such as SIGSEGV are already reported, as well as matching the
>     apparent intent of the POSIX sigaction interface.

Makes sense.  ACK.

> 
>  arch/arm/mm/alignment.c |   14 +++++++++++---
>  1 files changed, 11 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm/mm/alignment.c b/arch/arm/mm/alignment.c
> index 724ba3b..65ed9c6 100644
> --- a/arch/arm/mm/alignment.c
> +++ b/arch/arm/mm/alignment.c
> @@ -22,6 +22,7 @@
>  #include <linux/sched.h>
>  #include <linux/uaccess.h>
>  
> +#include <asm/system.h>
>  #include <asm/unaligned.h>
>  
>  #include "fault.h"
> @@ -883,9 +884,16 @@ do_alignment(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
>  	if (ai_usermode & UM_FIXUP)
>  		goto fixup;
>  
> -	if (ai_usermode & UM_SIGNAL)
> -		force_sig(SIGBUS, current);
> -	else {
> +	if (ai_usermode & UM_SIGNAL) {
> +		siginfo_t si;
> +
> +		si.si_signo = SIGBUS;
> +		si.si_errno = 0;
> +		si.si_code = BUS_ADRALN;
> +		si.si_addr = (void __user *)addr;
> +
> +		force_sig_info(si.si_signo, &si, current);
> +	} else {
>  		/*
>  		 * We're about to disable the alignment trap and return to
>  		 * user space.  But if an interrupt occurs before actually
> -- 
> 1.7.4.1
>

On Wed, Jul 27, 2011 at 01:10:36PM +0100, Dave Martin wrote:
> With the UM_SIGNAL alignment fault mode, no siginfo structure is
> passed to userspace.
> 
> POSIX specifies how siginfo_t should be populated for alignment
> faults, so this patch does just that:
> 
>   * si_signo = SIGBUS
>   * si_code = BUS_ADRALN
>   * si_addr = misaligned data address at which access was attempted
> 
> Signed-off-by: Dave Martin <dave.martin@linaro.org>
> ---
> v2: si_addr in siginfo_t changed to contain the faulting data
>     access address, not the address of the faulting instruction.
> 
>     This behaviour is consistent with the way other fault signals
>     such as SIGSEGV are already reported, as well as matching the
>     apparent intent of the POSIX sigaction interface.

Acked-by: Kirill A. Shutemov <kirill@shutemov.name>