[dpdk-dev] [dpdk-stable] [PATCH] test/eal: do not scan PCI devices for memory tests

David Marchand david.marchand at redhat.com
Thu Aug 8 13:23:04 CEST 2019


On Mon, Aug 5, 2019 at 12:20 PM Thomas Monjalon <thomas at monjalon.net> wrote:
> 03/08/2019 11:51, David Marchand:
> > On Fri, Aug 2, 2019 at 10:57 PM Thomas Monjalon <thomas at monjalon.net> wrote:
> > >
> > > 01/08/2019 14:27, David Marchand:
> > > > The memory tests currently check that, for normal mode (not legacy mode),
> > > > there is no memory left behind when exiting.
> > >
> > > I think this is the real bug:
> > > we are checking a behaviour that we cannot achieve currently.
> > >
> > > > The problem is that if a ethdev port is allocated when scanning pci
> > > > devices (even if the driver probe fails like when you have a virtio
> > > > management interface attached to the kernel), on exit, dpdk won't free
> > > > the associated memory since ethdev never frees the ethdev memzone.
> > >
> > > As you said in this thread, we could think about how to free it properly
> > > in a future release.
> > > For 19.08, I would suggest to disable the test with a comment
> > > explaining the reason.
> >
> > For 19.08, as long as we test shared builds in the CI, then it just
> > "works", because the net drivers are not loaded.
> > No net driver, no ethdev leak ;-)
>
> So we keep the bug with the unit test not running with a static build
> for 19.08, and we'll try to fix it in 19.11?

Seems the more pragmatic yes.


-- 
David Marchand


More information about the dev mailing list