On Sat, 04 Jul 2009, Jean-Pierre André wrote: > I went through the patches and the tests, and I can > announce that ntfs-3g is the first fuse-based file system > which passes the PJD Posix test fully, including the > Posix ACL parts. Thanks for testing. > I used kernel 2.6.29 with your proposed patches applied > to the fuse client (fuse.ko), and a simplified version of > the fuse server (fuse-lite), which I upgraded manually > according to your proposed patches. > > There were only three modifications needed not mentioned > in your proposal, probably consequences of different > reference versions : > - new fields in attr_out and write_in > - umask not copied from low level context to high level. This should be fixed in latest CVS. > I checked the client-server compatibility in the four > possible associations (new client / old server, etc.), > but of course both new client and new server are required > to get the correct permissions on file creation. > > For releasing a file system source compatible with both > old and new fuse.h, I check whether FUSE_CAP_DONT_MASK > is defined. From now on, I can release code which will compile > and run with current fuse (server and file system ready) > and will benefit from permission fixes when kernels with > updated fuse client will be released. > > There is just a point which worries me : I cannot understand > how your client patch deactivates the old code which was > applying the umask. I would prefer not to discover it the > hard way... The sb->s_flags |= MS_POSIXACL; line tells the VFS that it should leave umask applying to the filesystem. The flag is misnamed, it really should be "MS_NOMASK" or something. Thanks, Miklos -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
