Re: [POWERPC] Mask 32-bit system call arguments to 32 bits on PPC64 in audit code

Re: [POWERPC] Mask 32-bit system call arguments to 32 bits on PPC64 in audit code

[Heiko Carstens]

On Thu, Feb 08, 2007 at 06:59:16PM +0000, Linux Kernel Mailing List wrote:
> Commit:     cfcd1705b61ecce1ab102b9593cf733fef314a19
> Parent:     0e47e3cca100e7c8e8124378e4e44969c2e042fd
> Author:     David Woodhouse <dwmw2@infradead.org>
> AuthorDate: Sun Jan 14 09:38:18 2007 +0800
> Committer:  Paul Mackerras <paulus@samba.org>
> CommitDate: Wed Jan 24 21:13:58 2007 +1100
>  
>     [POWERPC] Mask 32-bit system call arguments to 32 bits on PPC64 in audit code
>     
>     The system call entry code will clear the high bits of argument
>     registers before invoking the system call; don't report whatever noise
>     happens to be in the high bits of the register before that happens.
>     
>     Signed-off-by: David Woodhouse <dwmw2@infradead.org>
>     Signed-off-by: Paul Mackerras <paulus@samba.org>
> ---
>  arch/powerpc/kernel/ptrace.c |   24 +++++++++++++++---------
>  1 files changed, 15 insertions(+), 9 deletions(-)
>  
> diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
> index 975102a..cc44c7b 100644
> --- a/arch/powerpc/kernel/ptrace.c
> +++ b/arch/powerpc/kernel/ptrace.c
> @@ -532,16 +532,22 @@ void do_syscall_trace_enter(struct pt_regs *regs)
>  	    && (current->ptrace & PT_PTRACED))
>  		do_syscall_trace();
>  
> -	if (unlikely(current->audit_context))
> -		audit_syscall_entry(
> -#ifdef CONFIG_PPC32
> -				    AUDIT_ARCH_PPC,
> -#else
> -				    test_thread_flag(TIF_32BIT)?AUDIT_ARCH_PPC:AUDIT_ARCH_PPC64,
> +	if (unlikely(current->audit_context)) {
> +#ifdef CONFIG_PPC64
> +		if (!test_thread_flag(TIF_32BIT))
> +			audit_syscall_entry(AUDIT_ARCH_PPC64,
> +					    regs->gpr[0],
> +					    regs->gpr[3], regs->gpr[4],
> +					    regs->gpr[5], regs->gpr[6]);
> +		else
>  #endif
> -				    regs->gpr[0],
> -				    regs->gpr[3], regs->gpr[4],
> -				    regs->gpr[5], regs->gpr[6]);
> +			audit_syscall_entry(AUDIT_ARCH_PPC,
> +					    regs->gpr[0],
> +					    regs->gpr[3] & 0xffffffff,
> +					    regs->gpr[4] & 0xffffffff,
> +					    regs->gpr[5] & 0xffffffff,
> +					    regs->gpr[6] & 0xffffffff);
> +	}
>  }
>  
>  void do_syscall_trace_leave(struct pt_regs *regs)

Why is this needed? The audit code is not supposed to look at the high bits
if the audited task is a compat task.
It does that at at least one place in kernel/auditsc.c:

	case 4: /* socketcall */
		return ((mask & AUDIT_PERM_WRITE) && ctx->argv[0] == SYS_BIND);

So does the audit code assume that the high bits are cleared or is this
a bug in the audit code?

-- 
Heiko Carstens
Linux on System z Development

IBM Deutschland Entwicklung GmbH
Vorsitzender des Aufsichtsrats: Johann Weihen
Geschaeftsfuehrung : Herbert Kircher
Sitz der Gesellschaft: Boeblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294

Re: [POWERPC] Mask 32-bit system call arguments to 32 bits on PPC64 in audit code

[Heiko Carstens]

On Thu, Feb 22, 2007 at 12:20:40PM +0100, Heiko Carstens wrote:
> On Thu, Feb 08, 2007 at 06:59:16PM +0000, Linux Kernel Mailing List wrote:
> > Commit:     cfcd1705b61ecce1ab102b9593cf733fef314a19
> > Parent:     0e47e3cca100e7c8e8124378e4e44969c2e042fd
> > Author:     David Woodhouse <dwmw2@infradead.org>
> > AuthorDate: Sun Jan 14 09:38:18 2007 +0800
> > Committer:  Paul Mackerras <paulus@samba.org>
> > CommitDate: Wed Jan 24 21:13:58 2007 +1100
> >  
> >     [POWERPC] Mask 32-bit system call arguments to 32 bits on PPC64 in audit code
> >     
> >     The system call entry code will clear the high bits of argument
> >     registers before invoking the system call; don't report whatever noise
> >     happens to be in the high bits of the register before that happens.
> >     
> >     Signed-off-by: David Woodhouse <dwmw2@infradead.org>
> >     Signed-off-by: Paul Mackerras <paulus@samba.org>
> > ---
> >  arch/powerpc/kernel/ptrace.c |   24 +++++++++++++++---------
> >  1 files changed, 15 insertions(+), 9 deletions(-)
> >  
> > diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
> > index 975102a..cc44c7b 100644
> > --- a/arch/powerpc/kernel/ptrace.c
> > +++ b/arch/powerpc/kernel/ptrace.c
> > @@ -532,16 +532,22 @@ void do_syscall_trace_enter(struct pt_regs *regs)
> >  	    && (current->ptrace & PT_PTRACED))
> >  		do_syscall_trace();
> >  
> > -	if (unlikely(current->audit_context))
> > -		audit_syscall_entry(
> > -#ifdef CONFIG_PPC32
> > -				    AUDIT_ARCH_PPC,
> > -#else
> > -				    test_thread_flag(TIF_32BIT)?AUDIT_ARCH_PPC:AUDIT_ARCH_PPC64,
> > +	if (unlikely(current->audit_context)) {
> > +#ifdef CONFIG_PPC64
> > +		if (!test_thread_flag(TIF_32BIT))
> > +			audit_syscall_entry(AUDIT_ARCH_PPC64,
> > +					    regs->gpr[0],
> > +					    regs->gpr[3], regs->gpr[4],
> > +					    regs->gpr[5], regs->gpr[6]);
> > +		else
> >  #endif
> > -				    regs->gpr[0],
> > -				    regs->gpr[3], regs->gpr[4],
> > -				    regs->gpr[5], regs->gpr[6]);
> > +			audit_syscall_entry(AUDIT_ARCH_PPC,
> > +					    regs->gpr[0],
> > +					    regs->gpr[3] & 0xffffffff,
> > +					    regs->gpr[4] & 0xffffffff,
> > +					    regs->gpr[5] & 0xffffffff,
> > +					    regs->gpr[6] & 0xffffffff);
> > +	}
> >  }
> >  
> >  void do_syscall_trace_leave(struct pt_regs *regs)
> 
> Why is this needed? The audit code is not supposed to look at the high bits
> if the audited task is a compat task.
> It does that at at least one place in kernel/auditsc.c:
> 
> 	case 4: /* socketcall */
> 		return ((mask & AUDIT_PERM_WRITE) && ctx->argv[0] == SYS_BIND);
> 
> So does the audit code assume that the high bits are cleared or is this
> a bug in the audit code?

Extended the cc list with a few people that recently worked on the audit code,
maybe somebody could answer my question above?