On Fri, Mar 10, 2023 at 04:06:52PM -0800, Nicolin Chen wrote: > On Fri, Mar 10, 2023 at 04:15:33PM -0400, Jason Gunthorpe wrote: > > On Wed, Mar 08, 2023 at 06:25:59AM -0800, Nicolin Chen wrote: > > > Add a new IOMMU_TEST_OP_ACCESS_SET_IOAS to allow setting access->ioas > > > individually, corresponding to the iommufd_access_set_ioas() helper. > > > > > > Reviewed-by: Kevin Tian <kevin.tian@xxxxxxxxx> > > > Signed-off-by: Nicolin Chen <nicolinc@xxxxxxxxxx> > > > --- > > > drivers/iommu/iommufd/iommufd_test.h | 4 +++ > > > drivers/iommu/iommufd/selftest.c | 26 +++++++++++++++---- > > > tools/testing/selftests/iommu/iommufd_utils.h | 22 ++++++++++++++-- > > > 3 files changed, 45 insertions(+), 7 deletions(-) > > > > I'd prefer we keep it so that the IOAS can be setup with an argument, > > this will greatly help syzkaller > > > > Lets have it so a 0 ioas will avoid the setup so the second call can > > happen > > I assume that you mean the iommufd_access_set_ioas() call and > the "unsigned int ioas_id" input of iommufd_test_create_access? I changed it to keep the id in iommufd_test_create_access(). Instead, I renamed the IOMMU_TEST_OP_ACCESS_SET_IOAS ioctl to IOMMU_TEST_OP_ACCESS_REPLACE_IOAS. So an access can be created by the original ioctl, and then be replaced using the new one. Thanks Nic
