Re: [PATCH 0/2] core dump: re-purpose VM_ALWAYSDUMP to user controlled VM_DONTDUMP

[Jason Baron <jbaron@redhat.com>]

On Wed, Mar 07, 2012 at 02:30:28PM -0800, Andrew Morton wrote:
> On Wed, 7 Mar 2012 12:00:46 -0500
> Jason Baron <jbaron@redhat.com> wrote:
> 
> > Hi,
> > 
> > The motivation for this change was that I was looking at a way for a qemu-kvm
> > process, to exclude the guest memory from its core dump, which can be quite
> > large. There are already a number of filter flags in
> > /proc/<pid>/coredump_filter, however, these allow one to specify 'types' of
> > kernel memory, not specific address ranges (which is needed in this case).
> > 
> > Since there are no more vma flags available, the first patch eliminates the
> > need for the 'VM_ALWAYSDUMP' flag. The flag is used internally by the kernel to
> > mark vdso and vsyscall pages. However, it is simple enough to check if a vma
> > covers a vdso or vsyscall page without the need for this flag.
> 
> Gee, we ran out?
> 
> That makes it pretty inevitable that we will grow the vma by four
> bytes.  Once we have done that, your always_dump_vma() trickery becomes
> unneeded and undesirable, yes?  If so, we may as well recognise reality
> and grow the vma now.
> 

We could. The reason I didn't propose increasing them was because I saw
in the archives that there was resistance when this was tried before:
http://lkml.indiana.edu/hypermail/linux/kernel/1111.1/02053.html

Also, the current use of 'VM_ALWAYSDUMP' is for vdso and gate pages
exclusively. That means that VM_ALWAYSDUMP flag is 0 the majority of the
time, and thus it probably makes sense to re-purpose it, now that we are
tight on flags.

> > The second patch then replaces the 'VM_ALWAYSDUMP' flag with a new 
> > 'VM_DONTDUMP' flag, which can be set by userspace using new madvise flags:
> > 'MADV_DONTDUMP', and unset via 'MADV_DUMP'. The core dump filters continue to
> > work the same as before unless 'MADV_DONTDUMP' is set on the region.
> > 
> > The qemu code which implements this features is at:
> > http://people.redhat.com/~jbaron/qemu-dump/qemu-dump.patch
> > 
> > In my testing the qemu core dump shrunk from 383MB -> 13MB with this patch.
> > 
> > I also believe that the 'MADV_DONTDUMP' flag might be useful for security
> > sensitive apps, which might want to select which areas are dumped.
> > 
> 
> Is there any way for userspace to query the state of the flag?  
> 

A place for that might be /proc/<pid>/maps, but I don't think any of
the other vm flags are exposed.

Thanks,

-Jason