Re: [git pull] coredump infoleak fix

From: Ingo Molnar
Date: Thu May 28 2020 - 03:44:49 EST

* Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:

> On Thu, May 28, 2020 at 09:02:55AM +0200, Ingo Molnar wrote:
> > Looks good to me.
> >
> > I'm wondering, shouldn't we also zero-initialize the dump data to
> > begin with? See the patch below (untested).
> Note that this hides the bug from KASAN, though ;-) And the bug
> is not just infoleak - not all components are "all zeroes" in the
> init state.

Yeah, but is zero-init really a problem though? Wouldn't it be
'better' to have all zeroes if the dump doesn't fit? But I might be
missing something ...