On Fri, Jul 29, 2022 at 01:13:49PM +0200, David Marchand wrote:
> On Fri, Jul 29, 2022 at 12:20 PM Bruce Richardson
> <bruce.richard...@intel.com> wrote:
> > > +if not is_windows and dpdk_conf.has('RTE_LIB_TELEMETRY')
> > > + test_args = [dpdk_test]
> > > + test_args += test_no_huge_args
> > > + if get_option('default_library') == 'shared'
> > > + foreach drv:dpdk_drivers
> > > + test_args += ['-d', drv.full_path().split('.a')[0] + '.so']
> > > + endforeach
> >
> > Rather than looping for each driver and building a huge cmdline, we should
> > just be able to pass in the path to the drivers directory, and have DPDK
> > auto-load all .so files there. Passing in "meson.project_build_root() +
> > '/drivers'" should probably work. If we want a more correct way to get the
> > drivers build directory, we can call meson.current_build_dir() when
> > processing the drivers and save off that value to be reused here.
>
> I copied this loop from an existing block added with 207b1c813f39
> ("test: fix build without ring PMD").
> On the need for this loop at the time... I think it was to avoid
> loading non .so stuff from a build directory.
> And you probably handled this with 49b536fc3060 ("eal: load only
> shared libs from driver plugin directories").
>
> I'll try your suggestion and, if it works, I'll update the other existing
> test.
>
Many thanks. If it doesn't work and needs some investigation, let me know
and maybe I can take a look at it for a bit.
/Bruce
