Re: [RFC,1/2] powerpc/fsl-pci: atomic get_user when pagefault_disabled

[David Hildenbrand <dahi@linux.vnet.ibm.com>]

> On Tue, Nov 25, 2014 at 12:43:25PM +0100, David Hildenbrand wrote:
> > Whenever we have pagefaults disabled, we have to use the atomic variants of
> > (set|get)_user and copy_(from|to)_user.
> > 
> > Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
> > ---
> >  arch/powerpc/sysdev/fsl_pci.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/arch/powerpc/sysdev/fsl_pci.c b/arch/powerpc/sysdev/fsl_pci.c
> > index 65d2ed4..c0af4ef 100644
> > --- a/arch/powerpc/sysdev/fsl_pci.c
> > +++ b/arch/powerpc/sysdev/fsl_pci.c
> > @@ -1025,7 +1025,7 @@ int fsl_pci_mcheck_exception(struct pt_regs *regs)
> >  	if (is_in_pci_mem_space(addr)) {
> >  		if (user_mode(regs)) {
> >  			pagefault_disable();
> > -			ret = get_user(regs->nip, &inst);
> > +			ret = __get_user_inatomic(regs->nip, &inst);
> >  			pagefault_enable();
> >  		} else {
> >  			ret = probe_kernel_address(regs->nip, inst);
> 
> Please post a non-RFC version if you're ready for this to be merged.
> 
> -Scott
> 

Hi Scott,

actually this patch was wrong. We are allowed to use the non-atomic variants
during pagefault_disable(). The semantics of get_user()/set_user() then change
- they will not sleep.

To reenable the might_sleep checks() in might_fault() makes it necessary to
count the levels of pagefault_disable() calls, to distinguish it from oridnary
preempt_disable() calls. I've got another patchset out there that deals with
this problem and is able to distinguish between them.

The interest in this feature just doesn't seem to be very high :)

So you can safely ignore this patch.

David