Re: [PATCH] include all vmas with unbacked pages in ELF core dumps

[Hugh Dickins]

On Wed, 20 Oct 2004, Linus Torvalds wrote:
> 
> In general, I'd much prefer just adding a VM_DIRTY flag, and making the
> COW logic set it. That should guarantee that we write out the minimal 
> required sections.

VM_DIRTY flag would be problematic, since you need mmap_sem exclusively
to modify vm_flags, whereas faulting holds mmap_sem non-exclusively.

But if I understand you aright, testing vma->anon_vma should do it.

> So "maydump()" might sanely look something like
> 
> 	/* Shared or IO mappings are never written out */
> 	if (vma->vm_flags & (VM_IO | VM_SHARED | VM_SHM))
> 		return 0;

VM_SHM is a relic, there might be a driver or two still setting it,
but we should have deleted it long ago: you mean VM_RESERVED, I think.

> 	/*
> 	 * Was the mapped backing store opened for writing?
> 	 * This really only happens for VM_SHARED (which we
> 	 * don't write out anyway), but it's conceptually
> 	 * the right thing to do. Some future internal use
> 	 * might end up doing this.
> 	 */
> 	if (!(vma->vm_flags & VM_MAYWRITE))
> 		return 0;

Conceptually the right thing to do?  I think it's best left out until
that future internal use makes sense of it.  As it stands, I can't work
out what it's up to; and VM_MAYWRITE doesn't say if the backing store
was opened for writing, but whether the memory may be mprotected writable.

> 	/* We really should add this flag */
> 	if (!(vma->vm_flags & VM_DIRTY))
> 		return 0;

	if (!vma->anon_vma)
		return 0;

> 	/* otherwise, write it out */
> 	return 1;

Hugh

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/