On 01/01/2017 02:48 PM, Stefan Bruens wrote:
On Montag, 12. Dezember 2016 11:04:34 CET you wrote:
On 12/04/2016 05:52 PM, Stefan Brüns wrote:
From: Stefan Brüns stefan.bruens@rwth-aachen.de
The following checks are currently implemented:
- listing a directory
- verifying size of a file
- veryfying md5sum for a file region
- reading the beginning of a file
- Doesn't mounting/unmounting require root? Nothing else in test/py
does. It'd be useful to put a comment at the top of the file detailing which command one might want to add to /etc/sudoers to avoid having to run the whole thing as root, and use sudo within the test where necessary.
It already uses sudo, the run command is visible from the log. Currently, I just run "sudo true" prior to executing the test, the password is cached for 5 minutes.
I don't see sudo used anywhere in test/py; can you point out where you see it using sudo at present?
+class FsImage:
...
- def mount(self, log):
if not os.path.exists(self.mountpath):os.mkdir(self.mountpath)Use os.makedirs(path) instead; you can avoid the if statement, and it'll handle parent directories too.
exist_ok only exists in python >= 3.2.
There's no need to use exist_ok; do this (example from conftest.py):
try: os.makedirs(path) except OSError as exc: if exc.errno == errno.EEXIST and os.path.isdir(path): pass else: raise
+@pytest.fixture(scope='module', params=['fat', 'ext4']) +def fsimage(prereq_commands, u_boot_config, u_boot_log, request):
- """Filesystem image instance."""
- datadir = u_boot_config.result_dir + '/'
Wouldn't it be better to put this into u_boot_config.persistent_data_dir, plus avoid creating the image file if it already exists? See u_boot_utils.py's PersistentRandomFile() as an example. I wonder if that could be expanded to create files not just of size n, but with sparse layout specs like this test uses?
See above. I don't think PersistentRandomFile() does fit here, as the files are created *inside* an image. Maybe it could be used after the image and filesystem is created and mounted ...
You missed my point. PersistentRandomFile() is existing code that creates a persistent data file. You can create a new function/... that uses the same techniques to create the data file, but make it create a filesystem image rather than random data.
- fstype = request.param
- imagepath = datadir + '3GB.' + fstype + '.img'
- mountpath = datadir + 'mnt_' + fstype
- with u_boot_log.section('Create image "{0}"'.format(imagepath)):
fsimage = FsImage(fstype, imagepath, mountpath)fsimage.mkfs(u_boot_log)- yield fsimage
- fsimage.unmount(u_boot_log)
Unmounting seems to happen in a lot of different places. Can we isolate it to just one place?
The image is mounted/unmounted for two different reasons - creating/populating the image, and when accessing it using the hostfs commands.
IIRC, there's more duplication than that, but I'll look again when this is reposted.
Also, what happens if the code throws an exception after obtaining an fsimage from this generator; I'm not sure that any cleanup happens in that case. Should there be "try: ... finally: unmount()" somewhere to clean up even in the case of an error? Alternatively, perhaps class FsImage should have a destructor that does the unmount (at least if it hasn't happened already)?
Its a fixture and will be torn down by pytest.
How does that work? Once the fixture function has yielded the image, surely if an exception is thrown, it'll be "thrown through" the generator function and hence prevent the rest of the function body from running? Or is there some special magic that lets the generator complete, even if the yield effectively threw an exception?
return output[1:3]No error checking for output[0]? I suppose if u_boot_console_base.py's bad_pattern_defs[] included error patterns that "readcmd" was expected to emit, that'd be fine, but it doesn't currently. Maybe we expect that the other command can't possibly succeed if the read doesn't. Similar comment for run_sizecmd() below, and perhaps elsewhere.
filesize is only set if the read succeeds, and "env print filesize" matches bad_pattern_defs in case of an error.
What pattern is printed that matches bad_pattern_defs? Nothing in bad_pattern_defs obviously would get printed if $filesize wasn't set.
Also, what if $filesize is still set from an earlier test (e.g. in a different Python file), and a later test fails, yet this isn't detected since that relies on $filesize not being set? I see this script doing "env set filesize" /after/ its own tests which should prevent this, but not /before/ the first test.