Re: [PATCH] kprobes: Introduce is_kprobe_fault()

[Harvey Harrison <harvey.harrison@gmail.com>]

On Thu, 2008-01-03 at 12:35 +0530, Ananth N Mavinakayanahalli wrote:
> On Wed, Jan 02, 2008 at 08:05:14PM -0800, Harvey Harrison wrote:
> 
> Thanks for the cleanup...
> 
> ...
> 
> > diff --git a/arch/x86/mm/fault_32.c b/arch/x86/mm/fault_32.c
> > index a2273d4..f2e909b 100644
> > --- a/arch/x86/mm/fault_32.c
> > +++ b/arch/x86/mm/fault_32.c
> > @@ -33,28 +33,6 @@
> > 
> >  extern void die(const char *,struct pt_regs *,long);
> > 
> > -#ifdef CONFIG_KPROBES
> > -static inline int notify_page_fault(struct pt_regs *regs)
> > -{
> > -	int ret = 0;
> > -
> > -	/* kprobe_running() needs smp_processor_id() */
> > -	if (!user_mode_vm(regs)) {
> 		^^^^^^^^^^^
> For x86_32, this check is important. See commit
> 6444541671bd821b950dbaafee70d65188198aa6 (Never allow int3 traps
> from V8086 mode to enter the kprobes handler) for precise reason why its
> user_mode_vm() and not user_mode() for x86_32.
> 
> You'll need to make room for this check in the generic macro below...

Or just keep X86_32 using its existing notify_page_fault, everyone else
can use the new is_kprobe_fault.  I'll wait for a bit and see if
the s390 or avr32 people respond before resending.

Harvey