On 7/14/2017 10:12 PM, Jan Blunck wrote: > Since the scan-mode of the bus is now based on the bus configuration it > isn't possible to have blacklisted and whitelisted devices existing for > the same bus. This fixes the unittest to reflect that. > > Signed-off-by: Jan Blunck <jblu...@infradead.org> > --- > test/test/test_devargs.c | 8 +++++--- > 1 file changed, 5 insertions(+), 3 deletions(-) > > diff --git a/test/test/test_devargs.c b/test/test/test_devargs.c > index 18f54edc1..02fec8b1f 100644 > --- a/test/test/test_devargs.c > +++ b/test/test/test_devargs.c > @@ -68,13 +68,15 @@ test_devargs(void) > goto fail; > if (rte_eal_devargs_add(RTE_DEVTYPE_WHITELISTED_PCI, "0000:5:00.0") < 0)
[1] see below. > goto fail; > - if (rte_eal_devargs_add(RTE_DEVTYPE_BLACKLISTED_PCI, "04:00.0,arg=val") > < 0) > + if (rte_eal_devargs_add(RTE_DEVTYPE_BLACKLISTED_PCI, "04:00.0,arg=val") > + == 0) Although as comment said, bus scan policy only can be whitelist or blacklist, and previous call sets it to whitelist [1], this API call doesn't return error. So changing its return type will fail the unittest. Fix can be changing type to whitelist, and update count to 4 in below check [2]. And keep blacklist count 0 [3]. > goto fail; > - if (rte_eal_devargs_add(RTE_DEVTYPE_BLACKLISTED_PCI, "0000:01:00.1") < > 0) > + if (rte_eal_devargs_add(RTE_DEVTYPE_BLACKLISTED_PCI, "0000:01:00.1") > + == 0) > goto fail; > if (rte_eal_devargs_type_count(RTE_DEVTYPE_WHITELISTED_PCI) != 2) > goto fail; [2] > - if (rte_eal_devargs_type_count(RTE_DEVTYPE_BLACKLISTED_PCI) != 2) > + if (rte_eal_devargs_type_count(RTE_DEVTYPE_BLACKLISTED_PCI) != 0) > goto fail; [3] > if (rte_eal_devargs_type_count(RTE_DEVTYPE_VIRTUAL) != 0) > goto fail; >
