[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Mode of emacs executable

Sorry about my earlier message; apparently, /tmp on our machine became
unwritable for a while. 

> > > Another quick question.  Why is the mode of the emacs executable 1755,
> > > i.e., rwxr-xr-t, in debian 0.90?  Should I, or should I not change it
> > > to mode 755?  I am worried about possible security problems, if any.
> > 
> > The sticky bit is to do with the program remaining 'in memory' when it
> > has finished executing so it will (theoretically) start quicker next
> > time around.
> But doesn't Linux already do that, with the demand loading and
> sharable code pages?  From what you say, the `sticky bit' appears to
> be unnecessary.  Am I right?
Well, no.  With text and clean data sharing another concurrent invocation
of a sticky program will load faster.  However, once all instances of Emacs
have been quit, no Emacs pages hang around (except in the buffer cache).
The sticky bit insures that Emacs hangs around in swap.  Linux doesn't
honor the sticky bit on executables, which is really an obsolete hack
for machines like the PDPs that had very slow disks.  Furthermore, Linux
is structured a bit differently from systems on which the sticky bit
means something.

Don't confuse this with the sticky bit on directories, which is another
hack to let multiple people write into a directory (such as /tmp) but
not be able to affect each other's files.  I wish POSIX would hurry
up on their ACL proposal, although POSIX hasn't been exactly revolutionary
in their work.  Most of the POSIX drafts could be summarized as a simple
"[]SYSV []BSD - check one" for each issue.

Reply to: