On Wed, Apr 01, 2020 at 08:57:42PM -0500, Eric Sandeen wrote:
> On 4/1/20 7:00 PM, Luis Chamberlain wrote:
> > On commit 6ac93117ab00 ("blktrace: use existing disk debugfs directory")
> > Omar fixed the original blktrace code for multiqueue use. This however
> > left in place a possible crash, if you happen to abuse blktrace in a way
> > it was not intended.
> >
> > Namely, if you loop adding a device, setup the blktrace with BLKTRACESETUP,
> > forget to BLKTRACETEARDOWN, and then just remove the device you end up
> > with a panic:
>
> Weird, I swear I tested that and didn't hit it, but ...
The real issue was also the dangling block device, a loop device proves
easy to test that. I'll note that another way to test this as well would
be to have a virtual machine with a block devices that goes in and out
via whatever virsh shenanigans to make a block device appear / disappear.
> > This issue can be reproduced with break-blktrace [2] using:
> >
> > break-blktrace -c 10 -d
>
> + -s, right?
Yeap, sorry about that.
> > This patch fixes this issue. Note that there is also another
> > respective UAF but from the ioctl path [3], this should also fix
> > that issue.
> >
> > This patch then also contends the severity of CVE-2019-19770 as
> > this issue is only possible using root to shoot yourself in the
> > foot by also misuing blktrace.
> >
> > [0] https://bugzilla.kernel.org/show_bug.cgi?id=205713
> > [1] https://nvd.nist.gov/vuln/detail/CVE-2019-19770
> > [2] https://github.com/mcgrof/break-blktrace
>
> I verified that this does reproduce the exact same KASAN splat on
> kernel 4.19.83 as reported in the original bug, thanks!
I just codified what Nicolai suspected, we should thank him :)
Luis
