On Thu, Jul 28, 2022 at 09:26:16PM +0200, David Marchand wrote: > On Thu, Jul 28, 2022 at 6:57 PM Bruce Richardson > <bruce.richard...@intel.com> wrote: > > > > On Thu, Jul 28, 2022 at 05:26:35PM +0200, David Marchand wrote: > > > Make rte_bus opaque for non internal users. > > > This will make extending this object possible without breaking the ABI. > > > > > > Introduce a new driver header and move rte_bus definition and helpers. > > > Update drivers and library to use the internal header. > > > > > > Some applications may have been dereferencing rte_bus objects, mark > > > this object's accessors as stable. > > > > > > Signed-off-by: David Marchand <david.march...@redhat.com> > > > --- > > > Changes since RFC v2: > > > - updated release notes, > > > - marked accessors as stable, > > > > > > Changes since RFC v1: > > > - update all existing users of the public header to use the internal one, > > > > > > > Acked-by: Bruce Richardson <bruce.richard...@intel.com> > > > > One small comment below... > > > > > --- > > > app/test/test_devargs.c | 2 +- > > <snip> > > > lib/pcapng/rte_pcapng.c | 2 +- > > > 38 files changed, 343 insertions(+), 319 deletions(-) > > > create mode 100644 lib/eal/include/bus_driver.h > > > > > > diff --git a/app/test/test_devargs.c b/app/test/test_devargs.c > > > index ac5bc34c18..0a4c34a1ad 100644 > > > --- a/app/test/test_devargs.c > > > +++ b/app/test/test_devargs.c > > > @@ -9,7 +9,7 @@ > > > #include <rte_common.h> > > > #include <rte_devargs.h> > > > #include <rte_kvargs.h> > > > -#include <rte_bus.h> > > > +#include <bus_driver.h> > > > #include <rte_class.h> > > > > > > > This seems a strange file to need to access bus internals. Maybe something > > to investigate in future. > > Yeah... I am not sure it is worth fixing. > > This code dereferences a bus object to check the ring driver > registered (by calling vdev_bus->parse()). > We could make it possible to list the drivers (maybe per bus, or even > globally). > Or we may expose something like bus_can_parse() / > rte_bus_find_by_device_name(). > > But I did not find applications needing this so I let the code as is. > Agree it may not be worth fixing.
However, if we do look to fix, how about just dropping the runtime check completely and just using a build-time one for the RTE_NET_RING macro. I don't consider it valid for the tests to run with some of the built drivers deleted. If it was enabled in the build, I think the tests are ok to assume it is present at runtime. /Bruce
