| 1 |
On Wed, 28 Sep 2005 15:02:17 +0000 |
| 2 |
Pedro Venda <pjvenda@××××××××××××.org> wrote: |
| 3 |
|
| 4 |
> On Wednesday 28 September 2005 07:16, Julius Loman wrote: |
| 5 |
> > On Wed, Sep 28, 2005 at 12:08:24AM -0700, Anthony Gorecki |
| 6 |
> <agorecki@××××××××××.com> wrote: |
| 7 |
> > > On Tuesday, September 27, 2005 23:57, Julius Loman wrote: |
| 8 |
> > > > For security reasons I want to have php scripts running |
| 9 |
> > > > under file owner UID. |
| 10 |
> > > |
| 11 |
> > > PHP's "safe mode" might offer the security that you're looking |
| 12 |
> > > for. |
| 13 |
> > |
| 14 |
> > I don't think so, at least files created by php scripts will be |
| 15 |
> > created with UID that php is running. I don't want this, so do |
| 16 |
> > users. |
| 17 |
> |
| 18 |
> This is indeed a serious issue and it deserves attention from careful |
| 19 |
> sysadmins. |
| 20 |
> |
| 21 |
> On my previous sysadmin job, me and my team discussed this issue |
| 22 |
> exactly. we were a team of sysadmins that (between many other things) |
| 23 |
> handled a sun solaris server with about 8000 user accounts. |
| 24 |
> |
| 25 |
> I can't remember all the arguments, but: |
| 26 |
> - suphp was abandoned due to poor testing and not supporting solaris |
| 27 |
> well; |
| 28 |
> - we didn't know fastcgi, but I assume it works by running php |
| 29 |
> scripts through a shell interpreter called as a cgi. it's an option, |
| 30 |
> and it doesn't require fastcgi... still it caused us trouble because |
| 31 |
> the interpreter was being called with the apache owner. the suexec |
| 32 |
> module has some limitations on this, since the php scripts didn't |
| 33 |
> have .cgi extensions. |
| 34 |
> |
| 35 |
> our solution involved patching apache's suexec to handle .php scripts |
| 36 |
> just like .cgi ones, running the interpreter on them. |
| 37 |
> |
| 38 |
> the guy who patched the suexec (and worked with me then) is on this |
| 39 |
> list and I believe he'll give it to you, if it looks pretty enough. |
| 40 |
> (CC'd him) |
| 41 |
> |
| 42 |
> as for performance, it surely dropped but we didn't really measure it |
| 43 |
> with numbers. that wasn't really the main concern. |
| 44 |
> |
| 45 |
> regards, |
| 46 |
> pedro venda. |
| 47 |
Hi, |
| 48 |
Maybe you all know this, but just to mention it - for quite a few of |
| 49 |
the daemons there're ready to use (selinux) policy configurations. |
| 50 |
SELinux policy ebuilds. Check all them by:"#eix selinux | grep policy". |
| 51 |
Unfortunately there's none for PHP, so somebody might want to write |
| 52 |
one and submit it for all the others to use. |
| 53 |
Think that using an MAC system by itself is quite a defence, all the |
| 54 |
other solutions will be weaker. Just my point here. |
| 55 |
PS: sorry but can't (easy) do this myself as i use RSBAC not SELinux. |
| 56 |
HTH.Rumen |
Archives