on 2022/3/19 6:12, Darrick J. Wong wrote: > You really ought to cc the xfs list for questions about longstanding > behaviors of XFS... > > [cc linux-xfs] Oh, yes, sorry for this. I have sent a RFC patch to linux-xfs and we can move discussion on that mail. Best Regards Yang Xu > > --D > > On Fri, Mar 18, 2022 at 10:37:03AM +0000, xuyang2018.jy@xxxxxxxxxxx wrote: >> Hi Darrick, Jack >> >> Petr meet a problem when running creat09 on xfs, ext4 doesn't have problem. >> >> It seems xfs will still use umask when enable default acl, but ext4 will >> not. >> >> As umask2 manpage , it said >> "Alternatively, if the parent directory has a default ACL (see acl(5)), >> the umask is ignored, the default ACL is inherited, the permission bits >> are set based on the inherited ACL, and permission bits absent >> in the mode argument are turned off. >> " >> >> It seem xfs doesn't obey this rule. >> >> the xfs calltrace as below: >> >> will use inode_init_owner(struct user_namespace *mnt_userns, >> structinode *inode) >> >> 296.760675] xfs_init_new_inode+0x10e/0x6c0 >> [ 296.760678] xfs_create+0x401/0x610 >> will use posix_acl_create(dir,&mode,&default_acl,&acl); >> [ 296.760681] xfs_generic_create+0x123/0x2e0 >> [ 296.760684] ? _raw_spin_unlock+0x16/0x30 >> [ 296.760687] path_openat+0xfb8/0x1210 >> [ 296.760689] do_filp_open+0xb4/0x120 >> [ 296.760691] ? file_tty_write.isra.31+0x203/0x340 >> [ 296.760697] ? __check_object_size+0x150/0x170 >> [ 296.760699] do_sys_openat2+0x242/0x310 >> [ 296.760702] do_sys_open+0x4b/0x80 >> [ 296.760704] do_syscall_64+0x3a/0x80 >> >> >> the ext4 calltrace as below: >> [ 296.460999] __ext4_new_inode+0xe07/0x1780 [ext4] >> posix_acl_create(dir,&inode->i_mode,&default_acl,&acl); >> [ 296.461035] ext4_create+0x106/0x1c0 [ext4] >> [ 296.461059] path_openat+0xfb8/0x1210 >> [ 296.461062] do_filp_open+0xb4/0x120 >> [ 296.461065] ? __check_object_size+0x150/0x170 >> [ 296.461068] do_sys_openat2+0x242/0x310 >> [ 296.461070] do_sys_open+0x4b/0x80 >> [ 296.461073] do_syscall_64+0x3a/0x80 >> [ 296.461077] entry_SYSCALL_64_after_hwframe+0x44/0xae >> >> I guess xfs modify its mode value instead of inode->i_mode in >> posix_acl_create by using current->umask value, so inode_init_owner >> doesn't clear no-sgid bits on created file because of missing S_IXGRP. >> >> Is it a kernel bug? >> >> Best Regards >> Yang Xu >> >>> Hi Petr >>> >>> It fails because the create file without S_IXGRP mode, then we miss >>> remove S_ISGID[1] bit. >>> >>> But I don't known why other filesystem doesn't have this problem. >>> >>> [1] >>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/inode.c#n2249 >>> >>> Best Regards >>> Yang Xu >>>> XFS fails on umask 0077: >>>> >>>> tst_test.c:1528: TINFO: Testing on xfs >>>> tst_test.c:997: TINFO: Formatting /dev/loop0 with xfs opts='' extra opts='' >>>> tst_test.c:1458: TINFO: Timeout per run is 0h 05m 00s >>>> creat09.c:61: TINFO: User nobody: uid = 65534, gid = 65534 >>>> creat09.c:62: TINFO: Found unused GID 3: SUCCESS (0) >>>> creat09.c:93: TPASS: mntpoint/testdir/creat.tmp: Owned by correct group >>>> creat09.c:97: TFAIL: mntpoint/testdir/creat.tmp: Setgid bit is set >>>> creat09.c:93: TPASS: mntpoint/testdir/open.tmp: Owned by correct group >>>> creat09.c:97: TFAIL: mntpoint/testdir/open.tmp: Setgid bit is set >>>> >>>> Thus clear the default umask. >>>> >>>> Signed-off-by: Petr Vorel<pvorel@xxxxxxx> >>>> --- >>>> testcases/kernel/syscalls/creat/creat09.c | 2 ++ >>>> 1 file changed, 2 insertions(+) >>>> >>>> diff --git a/testcases/kernel/syscalls/creat/creat09.c b/testcases/kernel/syscalls/creat/creat09.c >>>> index bed7bddb0e..70da7d2fc7 100644 >>>> --- a/testcases/kernel/syscalls/creat/creat09.c >>>> +++ b/testcases/kernel/syscalls/creat/creat09.c >>>> @@ -56,6 +56,8 @@ static void setup(void) >>>> (int)ltpuser->pw_gid); >>>> free_gid = tst_get_free_gid(ltpuser->pw_gid); >>>> >>>> + umask(0); >>>> + >>>> /* Create directories and set permissions */ >>>> SAFE_MKDIR(WORKDIR, MODE_RWX); >>>> SAFE_CHOWN(WORKDIR, ltpuser->pw_uid, free_gid); >>>
