On Thu, Jan 11, 2018 at 1:59 PM, Eryu Guan <eguan@xxxxxxxxxx> wrote: > On Sun, Jan 07, 2018 at 08:07:19PM +0200, Amir Goldstein wrote: >> usage: >> open_by_handle -p -o <handles_file> <test_dir> [N] >> open_by_handle -p -i <handles_file> <test_dir> [N] >> >> This will be used to test decoding of file handles after various >> file systems operations including mount cycle. >> >> Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx> > > I found that it might be easier to review if the new test that takes use > of this new function is included in the same patch, so reviewer could > know how are these new functions being used without switching between > different patches. > Eryu, Thanks for your comments. I will fix them later, but let me first address this meta comment, so maybe it will help you with review of the tests. I find that the comment about posting the new functionality along with the test would have been a good thing to do for most of the options I added so far. It could be a good option for this test as well. However, the -i -o options are quite generic in nature, so it should be fine to explain them without a test, only I probably did not explain well. The basic flow of open_by_handle program is: - store file handle of parent and test files in memory array - perform operations on files - drop caches - open by file handles from array - perform tests on open files -i -o options allow to perform name_to_handle and open_by_handle with 2 different program invocations, so you don't need to rely on the program performing drop caches and you can test advances flows like: - encode files - cycle mount - decode files In my use case, the new tests in generic/467 are: - ecnode parent and files - rename parent - decode parent and files - encode files - move files to new parent - decode files - ecnode parent and files - rename grandparent - decode parent and files Cheers, Amir.
