casaroli opened a new pull request, #4002:
URL: https://github.com/apache/nuttx-apps/pull/4002

   depends-on: apache/nuttx/pull/21613
   
   ## Summary
   
   No test in nuttx-apps covers `open()` of a directory without O_DIRECTORY, or 
`open()` of a file with O_DIRECTORY. `testing/testsuites` has one case for the 
first, and it is disabled with `#if 0` (`fs_open_test.c`, "fatfs will fail"). 
This PR adds `testing/fs/diropen`, a test of the POSIX rules:
   
   - For a directory D with an entry E: `open(D, O_WRONLY)` and `open(D, 
O_RDWR)` fail with EISDIR. `open(D, O_RDONLY)` gives a descriptor: `fstat()` 
says it is a directory, `openat()` opens E relative to it, `fdopendir()` and 
`readdir()` find E. `open(D, O_RDONLY | O_DIRECTORY)` with `fdopendir()` and 
`readdir()` also finds E.
   - For a file F: `open(F, O_RDONLY | O_DIRECTORY)` and `opendir(F)` fail with 
ENOTDIR.
   - In a writable directory, the test makes D and the regular file D/a, and 
checks that D/a still opens for read and for append.
   
   The test runs the cases on the pseudo file system (`/dev` with `null`, `/` 
with `dev`, and a new directory made with `mkdir()`), on a tmpfs that it mounts 
when the kernel has one, and in each directory given as an argument, for 
example a hostfs, v9fs or rpmsgfs mount: `diropen /host`. It prints one PASS or 
FAIL line per case and returns EXIT_FAILURE when a case fails.
   
   It is a standalone program in `testing/fs`, as `fatgap`, not a 
`testing/testsuites` case: the cases must run on several file systems in one 
run, and the testsuites run their file system cases only in 
`CONFIG_TESTS_TESTSUITES_MOUNT_DIR`. The program also needs no cmocka, so it 
builds in kernel builds and on small boards.
   
   ## Impact
   
   Test only. New option `TESTING_DIROPEN`, default n, `depends on 
!DISABLE_MOUNTPOINT` (the test calls `mount()` for its tmpfs). No defconfig 
sets it, so CI builds and runs nothing new.
   
   ## Testing
   
   `diropen`, nuttx master vs apache/nuttx#21613, pass / fail:
   
   | Configuration | Argument | nuttx master | with apache/nuttx#21613 | ostest 
|
   |---|---|---|---|---|
   | sim:nsh (flat, macOS host), make | hostfs `/host` | 23 / 15 | 47 / 0 | 
status 0 / 0 |
   | qemu-armv8a:nsh (flat), make, HVF | v9fs `/system` | 21 / 17 | 47 / 0 | 
status 0 / 0 |
   | qemu-armv8a:knsh (kernel), make, HVF | v9fs `/system` | 21 / 17 | 47 / 0 | 
status 0 / 0 |
   | mps2-an500:knsh (protected), make, QEMU TCG | none | 15 / 11 | 35 / 0 | 
not run |
   | qemu-armv8a:rpserver + rpproxy over ivshmem, QEMU TCG on a Linux host | 
rpmsgfs `/srv` | 24 / 14 | 47 / 0 | not run |
   | qemu-armv8a:nsh, CMake | - | build only, 0 warnings | build only, 0 
warnings | - |
   | qemu-armv8a:nsh, make, `DISABLE_MOUNTPOINT=y` (and `FS_PROCFS` off: the 
board mounts procfs) | - | - | `TESTING_DIROPEN` drops out, build OK, 0 
warnings | - |
   
   On master the pseudo-fs directories fail with ENXIO, `/dev/null` opens with 
O_DIRECTORY, and the hostfs, v9fs and rpmsgfs directories fail as 
apache/nuttx#21613 describes. The tmpfs cases pass with both. The mps2-an500 
and rpmsgfs rows ran on an earlier version of the two PRs with the same cases.
   
   `tools/checkpatch.sh -c -u -m -g` passes.
   
   Not run: FAT, littlefs, romfs and other file systems as the argument; real 
hardware.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to