Re: [PATCH] e1000: fix tx re-entrancy problem

[Alexander Bulekov]

On 211216 1935, Philippe Mathieu-Daudé wrote:
> On 12/16/21 16:51, Jon Maloy wrote:
> > On 12/16/21 04:36, Philippe Mathieu-Daudé wrote:
> >> Hi Jon,
> >>
> >> On 10/21/21 18:10, Jon Maloy wrote:
> >>> The fact that the MMIO handler is not re-entrant causes an infinite
> >>> loop under certain conditions:
> >>>
> >>> Guest write to TDT ->  Loopback -> RX (DMA to TDT) -> TX
> >>>
> >>> We now eliminate the effect of this problem locally in e1000, by adding
> >>> a boolean in struct E1000State indicating when the TX side is busy. This
> >>> will cause any entering new call to return early instead of interfering
> >>> with the ongoing work, and eliminates any risk of looping.
> >>>
> >>> This is intended to address CVE-2021-20257.
> >>>
> >>> Signed-off-by: Jon Maloy <jmaloy@redhat.com>
> >>> ---
> >>>   hw/net/e1000.c | 7 +++++++
> >>>   1 file changed, 7 insertions(+)
> >> I can not find the reproducer in the repository, have you sent one?
> >>
> > No, I did not add it to the repo.
> > It was referenced from the tracker BZ, but I was unable to get access
> > back then.
> > It ended up with that I had it sent by mail to me directly.
> > 
> > What is your question? Is it that it should be in the repo, or that you
> > cannot find it?
> 
> Well I'd like to reproduce the bug, but first I can not find it ;)
> Having such reproducer committed along with the fix help catching
> future regressions if we refactor code elsewhere.
> 

Blind guess, but assuming write to TDT == set_tctl, maybe this one?
https://bugs.launchpad.net/qemu/+bug/1917082