| 1 |
On Thu, 08 Nov 2007 18:22:48 +0000 |
| 2 |
Steve Long <slong@××××××××××××××××××.uk> wrote: |
| 3 |
> > if [[ ${UID} -ne 0 ]]; then |
| 4 |
> > |
| 5 |
> > We've always told people not to do that. Capabilities required by |
| 6 |
> > eselect modules should be tested by attempting to perform the |
| 7 |
> > action, not by some arbitrary query done on UIDs or groups. Being |
| 8 |
> > UID 0 doesn't mean you're allowed to do something, and not being |
| 9 |
> > UID 0 doesn't mean you're not allowed to do something. |
| 10 |
> > |
| 11 |
> I've always used EUID for the root check, eg: |
| 12 |
|
| 13 |
Which is just as bad. |
| 14 |
|
| 15 |
> This won't get round capabilities (so error status should still be |
| 16 |
> checked and the script bail with appropriate output, if it can't do |
| 17 |
> something it's supposed to) but it's sufficient for root privilege |
| 18 |
> check, and is better than UID which requires login as root. This |
| 19 |
> doesn't, of course, deal with non-root users, eg where users in group |
| 20 |
> portage are allowed to carry out a task. |
| 21 |
|
| 22 |
Except you absolutely never should be checking for root. You should be |
| 23 |
checking for capabilities. |
| 24 |
|
| 25 |
> You can check for that kind of thing with a writeable test, eg: |
| 26 |
> [[ -w $PORTDIR ]] || die 'Write access to portage dir required" |
| 27 |
|
| 28 |
-w is not reliable. |
| 29 |
|
| 30 |
> Wrt signalling die, the correct way for a script to terminate on |
| 31 |
> signal is something like this code, taken from a SIG_INT handler: |
| 32 |
> trap INT |
| 33 |
> kill -INT $$ |
| 34 |
> This ensures the parent process is correctly notified. So IOW just |
| 35 |
> kill self with the appropriate signal, ensuring any traps are cleared. |
| 36 |
|
| 37 |
No it isn't. When it comes to die, from bad implementations to good |
| 38 |
implementations, the order goes: the original drobbins prefix die sucks |
| 39 |
more than the original agriffis die, which sucks more than my original |
| 40 |
signalling eselect die, which sucks more than the current signalling |
| 41 |
paludis die. As time goes on we're finding better and better tricks to |
| 42 |
work around bash's lack of exceptions; eselect is currently one |
| 43 |
generation behind the best current known solution. |
| 44 |
|
| 45 |
-- |
| 46 |
Ciaran McCreesh |
Archives