On 11/06/12 13:58, Alejandro López wrote:


On Mon, Jun 11, 2012 at 2:30 PM, Len Lawrence <[email protected] <mailto:[email protected]>> wrote:

    On 11/06/12 12:28, Doug Laidlaw wrote:

        On Mon, 11 Jun 2012 11:33:46 +0100
        Len Lawrence<[email protected] <mailto:[email protected]>>
         wrote:

            After a warm reboot this morning I found that I no longer
            had the
            ability to run my own commands even though the permissions
            are correct
            and ownership is lcl (uid=500).  User system commands are
            OK so
            running a script via ruby for instance works but trying to
            execute the
            script by itself fails even though it is fully executable.
             This
            applies to all my local bin commands.  Command aliases
            however do work
            as long as they do not involve running any of my bin files.

            Even root cannot execute these bin commands; same message
            "Permission
            denied".

            In addition the system has switched me to autologin.
             Trying to run
            mcc I was told it cannot be run in console mode (??).  If
            I login as
            su mcc comes up in console mode, which I am not inclined
            to use.

            The hostname on this machine is belexeuli; this does not
            appear in
            the command prompt: [lcl@localhost ~]$

            After su: [root@belexeuli lcl]#

            This may all have something to do with my adding groups
            and changing
            group ids yesterday in my attempts to implement a viable
            sudoers
            command.  It worked and I could log out and in again
            without any
            problems.  I even managed a reboot without trouble but
            today is
            another story.

            I suspect that solving these multiple problems is beyond
            my technical
            skill even with help so a full reinstall is probably the
            best bet.
            However I will await any comments.

            Len

        You say that the permissions are correct, but do they include
        execute
        permissions?  The prompt difference may be simply that root's
        prompt is
        no longer the same as a user's prompt.  It is set by a config
        file for
        each user.  You can see the code for it by typing "echo $PS1"
        In my
        case, that gives "[\u@\h \W]\$"  The \h puts in the hostname.
         You can
        change it for the current session by typing at the user prompt:

            PS1="[\u@\h \W]\$"

        You can make that permanent by putting it in your
        .bash_profile, where
        it should override the other at your next login, but really,
        it is only
        a workaround.

        With so many issues, I would do a full reinstall, but more
        knowledgeable
        people tell me it is the easy way out.

        HTH,

        Doug.

    Yes, all the commands have execute permission.  I have been using
    my local bin directory for years and I have never had execute
    refused so this must reflect some deep system level screwup
    relating to lcl and maybe something in pam.d.  That is unknown
    country for me.
    Until yesterday there was no lcl group, only user lcl.  The group
    for lcl was live, which I have
    removed from my group list.  live was my primary group, now it is
    lcl which I added yesterday.  Ownership of my files is now lcl:lcl
    and in /home/lcl/bin the permissions are nearly all 755.
    Note that I can chmod -x <file> and chmod +x <file> but that does
    not change anything.

    I notice that home now contains a "live" directory: /home/live,
    ownership lcl:lcl, containing tmp and nothing else.  Now that is
    weird.

    The difference in the root and user prompts is probably related to
    the fact that root cannot access the user's X display.  I have
    seen that sort of thing before when the two have been using
    different hostnames.  I think that root is now looking at
    belexeuli:0 whereas the user has for some reason reverted to
    localhost:0.  Attempts at using the gui by root throw up protocol
    errors.

    As you say, a reinstall looks like the best way out.  More
    knowledgeable people would probably know just where to look for
    the root of the problem(s) but even after 21 years experience of
    Unix and Linux I know next to nothing about access and security
    policies.



Maybe it was somehow mounted with the -noexec flag?

Alejandro.


mtab contains this entry:

/dev/sda6 /home ext4 rw,nosuid,nodev,noexec,relatime,user_xattr,barrier=1,data=ordered 0 0

fstab had:
# Entry for /dev/sda6 :
UUID=341956e4-fddb-45a6-a191-4c912328ec7a /home ext4 user,defaults 1 2
none /proc proc defaults 0 0

I have removed the "user," because it does not tally with my other mga2 workstation.
That does not have noexec against /home.

About to reboot.

Thanks for the pointer.





Reply via email to