newToLinux: chmod permission denied

Hi, I have windows xp with a VFAT partition as D: in windows and files on that. I have it mounted in Fed Hat linux 9.0 as /mnt/winD. As a root I can create/modify files on /mnt/winD, but group does not have a write permission.

PROBLEM-1: As a root, when I give a command

#: chmod 774 /mnt/winD it says 'Operation not permitted'.

If root does not have permission then who has it? How do I solve this problem. It also does not let me change group of this filesystem.

PROBLEM-2: Where do I find the program 'linuxconf'? I could not find it using 'find / -name linuxconf', neither was it in any of the 'bin' or 'sbin' directories. command! I have installed every package in the linux, except some kernel stuff.

This is because the FAT filesystem does not have owners or permissions. To allow users to write to it you have to specify the group and permissions at mount time. A sample fstab entry:

/dev/hda3 /windows vfat noauto,gid=100,umask=002 0 0

This gives all users in group number 100 write permission to all files in the filesystem.

Regards, Ian

Of course it doesn't - dos filesystems don't have unix groups and owners and permissions and such! You know that!

Anywhere you like - but please don't. Why do you want it? It was deprecated ages ago.

]> I have windows xp with a VFAT partition as D: in windows and files on ]> that. I have it mounted in Fed Hat linux 9.0 as /mnt/winD. As a root I ]> can create/modify files on /mnt/winD, but group does not have a write ]> permission.

VFAT does not have any permissions. The linux Vfat driver will by default allow only root to write to a vfat partition. HOwever, if you want to allow other users to do so, put the word users as an option in /etc/fstab for that partition. If you want only certain users, eg a specific group, then umount the mnt/winD, and do a chgrp group_name /mnt/winD and chmod g+rw,o-rwx /mnt/wind. Then remount it. This will ( at least it used to) not allow any but someone in that group to enter that directory.

When the get-group system call (whatever its proper name is) is issued against a file in a DOS filesystem, the code responsible for the mounting is going to return *something*. Since the native filesystem has nothing comparable, it will come up with a value on its own.

It's plausible that the code might remember a separate value for each file, and furthermore that it might respond to 'chgrp' by changing the value in its memory.

Apparently, it does not do this, and instead it provides the same fixed value (determined by /etc/fstab) for all files on that filesystem. But that doesn't mean it *couldn't* do this, nor does it mean it's dumb to think that it might do this. It's not dumb - merely uninformed.

There you go again, Peter. He doesn't know HOW! Do you really fail to understand that? (I suspect that, the more you get irritated by whatever the OP wrote, the more you lean toward irritating them back, and the less you care about being helpful.)

The OP might want to append '-print' to his command, and see if that works. Even though linuxconf was deprecated ages ago

'Remembering' a separate value for every file on a file system is what a file system does, not what an operating system does. It's dumb to think that an o/s would do that. It goes against all principles of design that I can readily think of ... if one wants to unixify an msdos fs, one can use something like umsdos, which fs adds a special file to each directory of the dos fs, containing the permissions and owners and such things.

Well, he asked where, not how. If he asked how, maybe I'd have suggested 'locate', or 'google'.

It's because the -print is default in gnu find.

That would count. Adding a special file outside the dos fs (possibly discarding it on unmount) would also count. Depending on circumstances, it might be the only way to accomplish such a thing (e.g. mounting a NTFS fs; read-only is safe but read-write is not), or at least the preferable way (e.g. mounting a nearly-full floppy).

Hmm, I suppose there is somewhat of an ambiguity between 'am I using 'find' wrong and missing it?' and 'where do I download it?'. Still, I cannot fathom any interpretation of the OP's words other than 'where do I *successfully* find it?', to which 'anywhere you like' is not a valid answer. (He can look under his Aunt Bertha's couch, but I don't think it's very likely to be there.)

That's not quite what I asked. Clarification: Why would you use 'find' with -print turned off?

Ah, return codes. I figured it was something simple.

I understood him to mean: where should I go to look for it. It's the kind of question that makes one throw up ones hands ... the answer is, 'the universe'. It's not being hidden. It'll either be on his install medium or already on his disk or out there on the internet, in all the usual places. There is no sensible answer. It's like asking 'where do I find a picture of the queen'.

Then he would be foolish to like to look there.

I didn't know you could. For the return result, I suppose.

(man find)

If the expression contains no actions other than -prune, -print is performed on all files for which the expression is true.

Que? So do they mean that find -type f -prune prints, or don't they?

It prints.