Re: [PATCH] fs/proc: smaps should avoid VM_PFNMAP areas

2013-04-30 Thread Cliff Wickman
> On Tue, Apr 30, 2013 at 01:11:45PM -0500, Naoya Horiguchi wrote: > > > > /proc//smaps should not be looking at VM_PFNMAP areas. > > > > Certain tests in show_smap() (especially for huge pages) assume that the > > mapped PFN's are backed with page structures. And this is not usually true > > fo

Re: [PATCH] fs/proc: smaps should avoid VM_PFNMAP areas

2013-04-29 Thread Naoya Horiguchi
Hello Cliff, On Mon, Apr 29, 2013 at 04:25:53PM -0500, Cliff Wickman wrote: > > /proc//smaps should not be looking at VM_PFNMAP areas. > > Certain tests in show_smap() (especially for huge pages) assume that the > mapped PFN's are backed with page structures. And this is not usually true > for

[PATCH] fs/proc: smaps should avoid VM_PFNMAP areas

2013-04-29 Thread Cliff Wickman
/proc//smaps should not be looking at VM_PFNMAP areas. Certain tests in show_smap() (especially for huge pages) assume that the mapped PFN's are backed with page structures. And this is not usually true for VM_PFNMAP areas. This can result in panics on kernel page faults when attempting to addr