On 10.09.21 14:00, Mike Rapoport wrote:
On Fri, Sep 10, 2021 at 12:08:17PM +0200, David Hildenbrand wrote:
On 08.09.21 04:57, Andrew Morton wrote:
+
+ if (vma->vm_flags & (VM_WRITE | VM_EXEC)) {
+ ret = -EPERM;
+ goto out;
+ }
+
+ vma->vm_flags &= ~(VM_MAYWRITE | VM_MAYEXEC);
+ vma->vm_flags |= VM_MIXEDMAP;
+ vma->vm_ops = &kcore_mmap_ops;
+
This breaks all my efforts to sanitize /proc/kore access for virtio-mem.
Is there still a way to nack this?
Already done:
https://lore.kernel.org/mm-commits/CAHk-=wgQ+8kmczLLKCY7yDsGHQBRcZESKd1dNaKbrjUgbWeb3A@xxxxxxxxxxxxxx
and down the same thread.
Yeah, spotted Linus' reply just after I sent my reply.
... afterwards I thought about the implications fpr secretmem and
ordinary memory hotunplug and was happy that we dodged this bullet. :)
--
Thanks,
David / dhildenb
