Re: Error: freeing invalid memtype

["Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com>]

On Mon, 2009-03-23 at 12:48 -0700, Ingo Molnar wrote:
> * Pallipadi, Venkatesh <venkatesh.pallipadi@intel.com> wrote:
> 
> > On Sat, Mar 21, 2009 at 04:05:48PM -0700, Kevin Winchester wrote:
> > > Pallipadi, Venkatesh wrote:
> > > > Kevin,
> > > > 
> > > > Can you please send me the output of
> > > > # cat /debug/x86/pat_memtype_list
> > > > 
> > > > (with debugfs mounted at /debug) as soon as you login into X and also
> > > > when you start seeing these errors with etracer and glxinfo.
> > > > 
> > > 
> > > Here is the output before attempting to run glxinfo:
> > > : :
> > > : :
> > > 
> > > Does that help track down the problem?  I am about to try disabling PAT
> > > in my config to see if that fixes things.
> > > 
> > 
> > Can you please try the patch below and let me whether it helps.
> > 
> > Thanks,
> > Venki
> > 
> > x86, PAT: Change vma prot in pci_mmap to reflect inherited prot
> >     
> > While looking at the issue in the thread
> > http://marc.info/?l=dri-devel&m=123606627824556&w=2
> > noticed a bug in pci PAT code and memory type setting.
> >     
> > pci mmap code did not set the proper protection in vma, when it
> > inherited protection in reserve_memtype. This bug only affects
> > the case where there exists a WC mapping before X does an mmap
> > with /proc or /sys pci interface. This will cause X userlevel
> > mmap from /proc or /sysfs to fail on fork.
> >     
> > Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
> > Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
> > 
> > diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c
> > index 5ead808..f234a37 100644
> > --- a/arch/x86/pci/i386.c
> > +++ b/arch/x86/pci/i386.c
> > @@ -319,6 +319,9 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
> >  			return -EINVAL;
> >  		}
> >  		flags = new_flags;
> > +		vma->vm_page_prot = __pgprot(
> > +			(pgprot_val(vma->vm_page_prot) & ~_PAGE_CACHE_MASK) |
> > +			flags);
> 
> If it solves the problem it will be for -stable too, right?
> 
> It could be done a bit cleaner i think: is the ~_PAGE_CACHE_MASK 
> really needed? Does ->vm_page_prot ever have page frame bits set?
> 

Yes. This is a candidate for stable. And yes. vm_page_prot should not
have cacheability bits on, so ~PAGE_CACHE_MASK can be avoided here. I
had that, just wanting to be careful with and avoid further silly bugs
in this code. I will test version without ~PAGE_CACHE_MASK and resend
it.

Thanks,
Venki