| Message ID | 20231004124712.3833-1-chrubis@suse.cz |
|---|---|
| Headers | show |
| Series | Add tst_iterate_fd() | expand |
Hello, Cyril Hrubis <chrubis@suse.cz> writes: > - adds tst_iterate_fd() functionality > - make use of tst_iterate_fd() in readahead01 > - add accept03 test which uses tst_iterate_fd() > > This is a prototype for how the functionality to iterate over different > file descriptors should look like it converts one tests and adds > another. There is plenty of other syscalls that can use this kind of > testing, e.g. all fooat() syscalls where we can pass invalid dir_fd, the > plan is to add these if/once we agree on the API. I imagine the results of using this with splice could be very interesting. > > Cyril Hrubis (3): > lib: Add tst_fd_iterate() > syscalls/readahead01: Make use of tst_fd_iterate() > syscalls: accept: Add tst_fd_iterate() test > > include/tst_fd.h | 39 ++++++ > include/tst_test.h | 1 + > lib/tst_fd.c | 116 ++++++++++++++++++ > runtest/syscalls | 1 + > testcases/kernel/syscalls/accept/.gitignore | 1 + > testcases/kernel/syscalls/accept/accept01.c | 8 -- > testcases/kernel/syscalls/accept/accept03.c | 46 +++++++ > .../kernel/syscalls/readahead/readahead01.c | 46 +++---- > 8 files changed, 224 insertions(+), 34 deletions(-) > create mode 100644 include/tst_fd.h > create mode 100644 lib/tst_fd.c > create mode 100644 testcases/kernel/syscalls/accept/accept03.c > > -- > 2.41.0
Hi! > > - adds tst_iterate_fd() functionality > > - make use of tst_iterate_fd() in readahead01 > > - add accept03 test which uses tst_iterate_fd() > > > > This is a prototype for how the functionality to iterate over different > > file descriptors should look like it converts one tests and adds > > another. There is plenty of other syscalls that can use this kind of > > testing, e.g. all fooat() syscalls where we can pass invalid dir_fd, the > > plan is to add these if/once we agree on the API. > > I imagine the results of using this with splice could be very interesting. Good idea, I guess that we need to figure out how to do carthesian multiplication on the different file descriptors though. Maybe we need to treat the tst_interate_fd() as an iterator so that we can advance to the next fd with each call, so that we can do: struct tst_fd fd_in = {}, fd_out = {}; while (tst_iterate_fd(&fd_in)) { while (tst_iterate_fd(&fd_out)) { ... TST_TEST(splice(fd_in.fd, 0, fd_out.fd, 0, ...)); ... } }
Hello, Cyril Hrubis <chrubis@suse.cz> writes: > Hi! >> > - adds tst_iterate_fd() functionality >> > - make use of tst_iterate_fd() in readahead01 >> > - add accept03 test which uses tst_iterate_fd() >> > >> > This is a prototype for how the functionality to iterate over different >> > file descriptors should look like it converts one tests and adds >> > another. There is plenty of other syscalls that can use this kind of >> > testing, e.g. all fooat() syscalls where we can pass invalid dir_fd, the >> > plan is to add these if/once we agree on the API. >> >> I imagine the results of using this with splice could be very interesting. > > Good idea, I guess that we need to figure out how to do carthesian > multiplication on the different file descriptors though. Maybe we need > to treat the tst_interate_fd() as an iterator so that we can advance to > the next fd with each call, so that we can do: > > struct tst_fd fd_in = {}, fd_out = {}; > > while (tst_iterate_fd(&fd_in)) { > while (tst_iterate_fd(&fd_out)) { > ... > TST_TEST(splice(fd_in.fd, 0, fd_out.fd, 0, ...)); > ... > } > } This looks promising. I think it would be good to try this sooner rather than later.
Hi! > This looks promising. I think it would be good to try this sooner rather > than later. Already rewriting the code...