on 2022/4/8 9:20, xuyang2018.jy@xxxxxxxxxxx wrote: > on 2022/4/7 20:55, Christian Brauner wrote: >> On Thu, Apr 07, 2022 at 08:09:30PM +0800, Yang Xu wrote: >>> Since we plan to increase setgid test covertage, it will find new bug >>> , so add a new test group test-setgid is better. >>> >>> Also add a new test case to test test-setgid instead of miss it. >>> >>> Signed-off-by: Yang Xu<xuyang2018.jy@xxxxxxxxxxx> >>> --- >>> src/idmapped-mounts/idmapped-mounts.c | 19 +++++++++++++++---- >>> tests/generic/999 | 26 ++++++++++++++++++++++++++ >>> tests/generic/999.out | 2 ++ >> >> I actually didn't mean to split out the existing setgid tests. I mean >> adding new ones for the test-cases you're adding. But how you did it >> works for me too and is a bit nicer. I don't have a strong opinion so as >> long as Dave and Darrick are fine with it then this seems good to me. > Ok, let's listen .. When I write v3, I add mknodat patch as 1st patch and tmpfile as 2nd patch(by using a file doesn't under DIR1 directory, so I don't need to concern about xfs_irix_sgid_inherit_enabled), errno reset to 0 as 3st patch. It seems this way can't introduce the new failure for generic/633. So I will add a new group for umask and acl and add new case for them instead of split setgid case from test-core group. ps: I doubt whether I need to send two patch sets(one is about mknodat,tmpfile,errno, the another is about umask,acl,new case). What do you think about this? Best Regards Yang Xu >> >> One note about the test name/numbering though. It seems you haven't >> added the test using the provided xfstest infrastructure to do that. >> Instead of manually adding the test you should run the "new" script. >> >> You should run: >> >> ~/src/git/xfstests$ ./new generic >> >> Next test id is 678 >> Append a name to the ID? Test name will be 678-$name. y,[n]: >> Creating test file '678' >> Add to group(s) [auto] (separate by space, ? for list): auto quick attr idmapped mount perms >> Creating skeletal script for you to edit ... >> >> that'll automatically figure out the correct test number etc. > Thanks, TBH, I don't know this usage. I don't name to 678 because > fstests patchwork has others new case(in reviewing), so I add a biger > number. > > Best Regards > Yang Xu
