Re: [webmin-devel] DHCPD v3 user defined options problem

[Philippe T. <phi...@m4...>]

Hi Jamie & All

I made a patch which now understands properly that format with brackets.
The parsers does OK (for the moment, I check if the bracket is following a
"=" sign, and if yes I don't start a sub, instead I consider it as simple
text.
Then, I did some small regexp changes in edit_options and save_options to
properly display and save that type. I'm now testing it a little bit, and
I'll send you a patch in a few days when it'll be OK, at least in my testing
environnement.

Bye for now.

2008/1/27, Jamie Cameron <jca...@we...>:
>
> Hi Philippe,
>
> Thanks for pointing this out - the problem here is that Webmin doesn't
> understand this new DHCPd option format with the curly brackets, causing
> it's parser to go into an infinite loop.
>
> Do you plan to work on a patch to properly handle and display these types
> of options? If so, I'd be glad to include it in a future Webmin release..
>
>  - Jamie
>
> On 25/Jan/2008 06:07 Philippe Teissier wrote ..
>
> Hi,
>
> I needed to use option 122 (DOCSIS) in my DHCPD.conf and it appears that
> Webmin had some problem with DHCPD syntax to write it.  It hangs the parser,
> giving me an almost empty page instead of my subnets and groups.
>
> Here are the lines in the dhcpd.conf :
>
> option space docsis-ccc;
> option docsis-ccc.dhcp-server-1 code 1 = ip-address;
> option docsis-ccc.dhcp-server-2 code 2 = ip-address;
>
> option docsis-ccc.provision-server code 3 = { integer 8, string };  ##
> hangs
>
> option docsis-ccc.krb-realm-name code 6 = string;
> option docsis-ccc code 122 = encapsulate docsis-ccc; ## misunderstood
>
> Two of those line are misunderstood by Webmin dhcpd's module
>
> * The 4th line is the one which hangs the parser. It defines a record (int
> 8, string) as the type of the "provision server suboption". And webmin does
> not like it.
> For the moment, I just commented out this line as I do not user the
> provision server parameter.
> * The last line, with the encapsulate type, is also misunderstood, but
> does not hang the code. Instead, webmin will parse and silently forget
> encapsulate's parameter (here docsis-ccc). After a "client optiont" change
> in webmin, the dhcpd.conf will contain "option docsis-ccc code 122 =
> encapsulate;" and then DHCPD's parser will hang (cause this is a syntax
> error).
>
> (more about this option syntax by the end of this document
> http://www.linuxmanpages.com/man5/dhcp-options.5.php)
>
> I wrote a fast hack for that 2nd problem in order to prevent users from
> crashing dhcpd.
>
> [root@localhost dhcpd]# diff edit_options.cgi.Before_patch
> edit_options.cgi
> 168,169c168,176
> <               printf "<input name=dtype_$i size=10 value='%s'>\n",
> <                       $o->{'values'}->[4];
> ---
> >               if ($o->{'values'}->[4] eq "encapsulate") {
> >                   printf "<input name=dtype_$i size=10 value='%s'>\n",
> >                   $o->{'values'}->[4]. " ".$o->{'values'}->[5];
> >
> >               } else {
> >                   printf "<input name=dtype_$i size=10 value='%s'>\n",
> >                   $o->{'values'}->[4];
> >               }
> >
> [root@localhost dhcpd]# diff save_options.cgi.Before_patch
> save_options.cgi
> 116c116
> <               $in{"dtype_$i"} =~ /^[a-z0-9\.\-\_]+$/i ||
> ---
> >               $in{"dtype_$i"} =~ /^(encapsulate |)[a-z0-9\.\-\_]+$/i ||
>
>
> I didn't found anything about this bug in the tracking system. It's the
> 1st time I look at webmin's code, so I'll try to have a deeper look at it,
> and achieve a better patch with support of encapsulate, records, and even
> arrays which are also unsupported I can imagine.
>
> --
> Philippe Teissier
> phi...@m4...<http://reply_mail.cgi?new=1&to=philippe%2Eteissier%40m4tp%2Eorg>
>
>
>


-- 
Philippe Teissier
phi...@m4...