Newbie Security Questions

Hello!

I've got some misc. questions about PHP and its usage with MySQL.


The following web page:

http://www.freewebmasterhelp.com/tutorials/phpmysql/3

shows that it is normal to include mysql database usernames and
passwords in the php file. Is this good programming practice? I'm
worried that people would be able to read my php file through a web
browser or through other nefarious means.

This is the statement that must be in the source file to connect to a
database:

mysql_connect(localhost,$username,$password);

with $username and $password defined elsewhere in the source file.
This seems scary to me!


How to properly defend against an injection attack? Wikipedia has the
following code as for how to defend:

$query_result = mysql_query
(
"select * from users where name = '"
.
mysql_real_escape_string($user_name, $dbh)
.
"'"
);

If this is all it takes to defend against the attacks why is such a
big deal made about them? Is there something more that you need to
defend against?


Also one more question on how to keep track of people who are
submitting information on a website. How to set a time limit to how
often people can submit information? This is easy to do on the client
side, just disable the button for a set amount of time, but if they
went hunting through my html and found the php script they could
easily whip up a program to POST information willy nilly as fast as
they wanted.


Also any more information or websites that would contain useful
information for newcomers to PHP and MySQL would be grand!

Thanks a lot!

Re: Newbie Security Questions

Dan wrote:
> Hello!
>
> I've got some misc. questions about PHP and its usage with MySQL.
>
>
> The following web page:
>
> http://www.freewebmasterhelp.com/tutorials/phpmysql/3
>
> shows that it is normal to include mysql database usernames and
> passwords in the php file. Is this good programming practice? I'm
> worried that people would be able to read my php file through a web
> browser or through other nefarious means.
>
> This is the statement that must be in the source file to connect to a
> database:
>
> mysql_connect(localhost,$username,$password);
>
> with $username and $password defined elsewhere in the source file.
> This seems scary to me!
>
>
> How to properly defend against an injection attack? Wikipedia has the
> following code as for how to defend:
>
> $query_result = mysql_query
> (
> "select * from users where name = '"
> .
> mysql_real_escape_string($user_name, $dbh)
> .
> "'"
> );
>
> If this is all it takes to defend against the attacks why is such a
> big deal made about them? Is there something more that you need to
> defend against?
>
>
> Also one more question on how to keep track of people who are
> submitting information on a website. How to set a time limit to how
> often people can submit information? This is easy to do on the client
> side, just disable the button for a set amount of time, but if they
> went hunting through my html and found the php script they could
> easily whip up a program to POST information willy nilly as fast as
> they wanted.
>
>
> Also any more information or websites that would contain useful
> information for newcomers to PHP and MySQL would be grand!
>
> Thanks a lot!

Here is the simple answer: they cannot see your PHP script./ The PHP
script resides on the server. It generates html as output and it is the
html that is sent to the browser. Try looking at a "Page source" in a
browser for a page with a php suffix. All you will see is the resultant
html.

--
Shelly

Re: Newbie Security Questions

Dan wrote:
> Hello!
>
> I've got some misc. questions about PHP and its usage with MySQL.
>
>
> The following web page:
>
> http://www.freewebmasterhelp.com/tutorials/phpmysql/3
>
> shows that it is normal to include mysql database usernames and
> passwords in the php file. Is this good programming practice? I'm
> worried that people would be able to read my php file through a web
> browser or through other nefarious means.

Only on a misconfigured server. Files with a .php extension will ALWAYS
be executed rather than downladed by te web serve.

And even if they do, its a strange Mysql server that is sitting on the
internet accepting requests from all and sundry..normally you run them
(php/apache/mysql)on a local network, or the same box and set mysql to
only accept requests FROM the apache server..

Of course if your server machine itself is hacked, all bets are off
anyway..never mind password access to Mysql, just strip all the database
files out and run them on YOUR mysql setup..

In essence, on a properly configured server, php sources are private.
End of story.


>
> Also one more question on how to keep track of people who are
> submitting information on a website. How to set a time limit to how
> often people can submit information? This is easy to do on the client
> side, just disable the button for a set amount of time, but if they
> went hunting through my html and found the php script they could
> easily whip up a program to POST information willy nilly as fast as
> they wanted.
>

Should be able to use a cookie or session thing to keep track of
individual users..but no absolute certainty. One of essences of web
acess is there is intrinsically no notion of a connected user. You have
to layer that over the top using cookies and user logins if you want it.
But that relies on co-operation from the remote user.

So, unless you enforce some kind of user login, you cant distinguish
between loads of different people doing stuff, and one person doing lots
of stuff.

IP address stuff doesn't work either as you may be dealing with a proxy
server.

Re: Newbie Security Questions

Dan wrote:
> Hello!
>
> I've got some misc. questions about PHP and its usage with MySQL.
>
>
> The following web page:
>
> http://www.freewebmasterhelp.com/tutorials/phpmysql/3
>
> shows that it is normal to include mysql database usernames and
> passwords in the php file. Is this good programming practice? I'm
> worried that people would be able to read my php file through a web
> browser or through other nefarious means.
>
> This is the statement that must be in the source file to connect to a
> database:
>
> mysql_connect(localhost,$username,$password);
>
> with $username and $password defined elsewhere in the source file.
> This seems scary to me!
>

You need to define the userid and password somewhere in the source file
so that you can access MySQL. If your server is properly configured, no
one will be able to see your PHP source code (as long as it's in .php
files, anyway).

But for additional safety, just put the information in an include file
outside the web server's document root. Then it will be accessible via
PHP (which uses the file system) but not through the web server.

>
> How to properly defend against an injection attack? Wikipedia has the
> following code as for how to defend:
>
> $query_result = mysql_query
> (
> "select * from users where name = '"
> .
> mysql_real_escape_string($user_name, $dbh)
> .
> "'"
> );
>
> If this is all it takes to defend against the attacks why is such a
> big deal made about them? Is there something more that you need to
> defend against?
>

That's part of it. Always process strings with
mysql_real_escape_string(), something too many people don't do. But
also validate your numeric values to ensure they are numeric. And if
there are specific values which can be passed (i.e. from a list), ensure
the value passed is in that list.
>
> Also one more question on how to keep track of people who are
> submitting information on a website. How to set a time limit to how
> often people can submit information? This is easy to do on the client
> side, just disable the button for a set amount of time, but if they
> went hunting through my html and found the php script they could
> easily whip up a program to POST information willy nilly as fast as
> they wanted.
>

You need to do it server side. Every time the user submits information,
check against the time they previously submitted. If it's not too
quick, save the new time and allow the submission.

>
> Also any more information or websites that would contain useful
> information for newcomers to PHP and MySQL would be grand!
>
> Thanks a lot!
>

How many years do you want to spend? There are millions of pages about
this on the internet.

--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attglobal.net
==================

Re: Newbie Security Questions

Thanks for all the help all!

Yeah the users are anonymous when submitting information so there
really is no way to keep track of them, as there are no usernames and
passwords to go around.

I was toying with the idea of maybe creating a buffer of submissions,
with their IP address, and maybe checking the buffer every 5 seconds
or so and seeing if there was an inordinate amount of submissions from
the same IP address. I'd then flush the buffer and process the
requests if there was no bad juju detected. This is out of my league
at the moment though, and wouldn't really solve the problem of proxy
servers, or say if a botnet decided to focus it's attention on my
amateurish website. :P

But once again thanks for the help! :D

Re: Newbie Security Questions

Dan wrote:
> Thanks for all the help all!
>
> Yeah the users are anonymous when submitting information so there
> really is no way to keep track of them, as there are no usernames and
> passwords to go around.
>
> I was toying with the idea of maybe creating a buffer of submissions,
> with their IP address, and maybe checking the buffer every 5 seconds
> or so and seeing if there was an inordinate amount of submissions from
> the same IP address. I'd then flush the buffer and process the
> requests if there was no bad juju detected. This is out of my league
> at the moment though, and wouldn't really solve the problem of proxy
> servers, or say if a botnet decided to focus it's attention on my
> amateurish website. :P
>
> But once again thanks for the help! :D
>

And also not accurate. A person may come from different IP addresses on
different requests (i.e. a round-robin proxy). AOL does this, for
instance. Or you might have 1000 users behind one corporate proxy and
all using the same IP address.

If you want to control submissions, you can use cookies, but that's not
foolproof. If you need to prevent it, you'll need userid's and passwords.

Which you will probably need anyway, or you will be deluged with spammers.

--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attglobal.net
==================

Re: Newbie Security Questions

Post removed (X-No-Archive: yes)

Re: Newbie Security Questions

Michael Vilain wrote:
> In article <13kg7mvjssbnac8@corp.supernews.com>,
> "Shelly" wrote:
>
>> Dan wrote:
>>> Hello!
>>>
>>> I've got some misc. questions about PHP and its usage with MySQL.
>>>
>>>
>>> The following web page:
>>>
>>> http://www.freewebmasterhelp.com/tutorials/phpmysql/3
>>>
>>> shows that it is normal to include mysql database usernames and
>>> passwords in the php file. Is this good programming practice? I'm
>>> worried that people would be able to read my php file through a web
>>> browser or through other nefarious means.
>>>
>>> This is the statement that must be in the source file to connect to a
>>> database:
>>>
>>> mysql_connect(localhost,$username,$password);
>>>
>>> with $username and $password defined elsewhere in the source file.
>>> This seems scary to me!
>>>
>>>
>>> How to properly defend against an injection attack? Wikipedia has the
>>> following code as for how to defend:
>>>
>>> $query_result = mysql_query
>>> (
>>> "select * from users where name = '"
>>> .
>>> mysql_real_escape_string($user_name, $dbh)
>>> .
>>> "'"
>>> );
>>>
>>> If this is all it takes to defend against the attacks why is such a
>>> big deal made about them? Is there something more that you need to
>>> defend against?
>>>
>>>
>>> Also one more question on how to keep track of people who are
>>> submitting information on a website. How to set a time limit to how
>>> often people can submit information? This is easy to do on the client
>>> side, just disable the button for a set amount of time, but if they
>>> went hunting through my html and found the php script they could
>>> easily whip up a program to POST information willy nilly as fast as
>>> they wanted.
>>>
>>>
>>> Also any more information or websites that would contain useful
>>> information for newcomers to PHP and MySQL would be grand!
>>>
>>> Thanks a lot!
>> Here is the simple answer: they cannot see your PHP script./ The PHP
>> script resides on the server. It generates html as output and it is the
>> html that is sent to the browser. Try looking at a "Page source" in a
>> browser for a page with a php suffix. All you will see is the resultant
>> html.
>
> I read in an article a neat trick--store the username and password for
> your MySQL database as environment variables in an INCLUDE to the
> startup file for your Apache server. This way the file can be protected
> with appropriate permissions and is run as root when Apache starts. I'm
> lucky. My web host was willing to do this for my site. Yours may not.
>
> http://shiflett.org/articles/shared-hosting
>

Which also means anyone else on your host can also see your userid and
password with a simple phpinfo()....

--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attglobal.net
==================

Re: Newbie Security Questions

On Nov 24, 11:31 am, Dan wrote:
> Hello!
>
> I've got some misc. questions about PHP and its usage with MySQL.
>
> The following web page:
>
> http://www.freewebmasterhelp.com/tutorials/phpmysql/3
>
> shows that it is normal to include mysql database usernames and
> passwords in the php file. Is this good programming practice? I'm
> worried that people would be able to read my php file through a web
> browser or through other nefarious means.
>
> This is the statement that must be in the source file to connect to a
> database:
>
> mysql_connect(localhost,$username,$password);
>
> with $username and $password defined elsewhere in the source file.
> This seems scary to me!
>
> How to properly defend against an injection attack? Wikipedia has the
> following code as for how to defend:
>
> $query_result = mysql_query
> (
> "select * from users where name = '"
> .
> mysql_real_escape_string($user_name, $dbh)
> .
> "'"
> );
>
> If this is all it takes to defend against the attacks why is such a
> big deal made about them? Is there something more that you need to
> defend against?
>
> Also one more question on how to keep track of people who are
> submitting information on a website. How to set a time limit to how
> often people can submit information? This is easy to do on the client
> side, just disable the button for a set amount of time, but if they
> went hunting through my html and found the php script they could
> easily whip up a program to POST information willy nilly as fast as
> they wanted.
>
> Also any more information or websites that would contain useful
> information for newcomers to PHP and MySQL would be grand!
>
> Thanks a lot!

I would be inclined to put the username, password, etc in a seperate
file. There are a number of advantages, the main one being if you
accidentally delete the ' be automatically displayed.

Unlikely I know, but might as well be safe.

Rob.

Re: Newbie Security Questions

>
> This is the statement that must be in the source file to connect to a
> database:
>
> mysql_connect(localhost,$username,$password);
>
> with $username and $password defined elsewhere in the source file.
> This seems scary to me!

Indeed. And I've been working on solutions to that problem lately.
Typically, the reason you want to worry about this is an attack
gaining access to the server who can just obtain your passwords and
such from scripts to access databases. It doesn't usually have
anything to do with someone downloading the PHP source from a
misconfigured server, as you'd be a fool to have your database server
open to the internet. I'll cover more on my idea in a moment.



> How to properly defend against an injection attack? Wikipedia has the
> following code as for how to defend:
>
> $query_result = mysql_query
> (
> "select * from users where name = '"
> .
> mysql_real_escape_string($user_name, $dbh)
> .
> "'"
> );
>
> If this is all it takes to defend against the attacks why is such a
> big deal made about them? Is there something more that you need to
> defend against?

A lot of outsourced work and "out of college" developers don't take
security into consideration. In some cases, it's companies going to
pre-packages solutions or source snippets from the web which are just
examples to do something; which never get fixed.

You can even defend against attacks by using addslashes(). Situations
where addslashes() and mysql_real_escape_string() is useless is
something like this:

$nAccountID = /*some user input..*/ "0; DROP users;"
$nAccountID = mysql_real_escape_string($nAccountID);
$sSQL = "SELECT * from users where account_id = {$nAccountID}";

mysql_real_escape_string will not change $nAccountID from the value of
"0; DROP users;".. there's nothing to be escaped in the string. The
proper way to protect against SQL injection is usually a combination
of sprintf() or intval(), and mysql_real_escape_string(). For example:

$nAccountID = intval("0; DROP users;"); // comes out 0

$sSQL = sprintf('SELECT * FROM users WHERE account_id = %u',
$nAccountID)

Note here we don't need mysql_real_escape_string() because sprintf()
is going to error if $nAccountID is not a number.

$sSQL = sprintf('SELECT * FROM users WHERE user_name = "%s"',
mysql_real_escape_string($sUsername));

This will escape the string to remove SQL escape attempts.

Now, mysql_query() has a protection built-in.. it will only execute
one query. So if $sUsername included multiple statements, only the
first would be executed.. which would be enough to complete the
statement you started, then the second statement would be the harmful
one.


> Also one more question on how to keep track of people who are
> submitting information on a website. How to set a time limit to how
> often people can submit information? This is easy to do on the client
> side, just disable the button for a set amount of time, but if they
> went hunting through my html and found the php script they could
> easily whip up a program to POST information willy nilly as fast as
> they wanted.

JavaScript methods are like car keys. They keep honest people honest.
Anyone who really wants to, or knows how can just bypass it. That's
where you implement something server side with a database. This would
usually be done by creating a table in a database to track recent
submissions.. usually tracking IP or account ID and submission
timestamp.

When you get a submission request, check

sprintf('SELECT COUNT(*) FROM thetable WHERE ip = %u AND submit_stamp
= >= %u',
ip2long($_REQUEST['REMOTE_ADDR']), time() - (60 * 5))

To check for how many submissions were made in the last five minutes.


In securing a site and protecting your databases.. it comes down to,
NEVER trust user input! Even if you're writing an internal tool where
employees of your own company will be using them.. DON'T pass
unescaped data to the database! Even if they aren't trying to cause
problems, eventually something is going to happen. Also, don't give
more rights to a script than it needs. I have a lot of user accounts
in my databases. When a page only needs to execute a function or issue
a SELECT, there's no reason to give it UPDATE, INSERT, or DROP
privileges. In some situations, I don't even give it SELECT.. I only
give it EXECUTE, and write a procedure which issues the SELECT, which
the PHP page will do a "CALL getAccountName(%u)" where %u is the
provided account ID on and process the results. The procedure in MySQL
is written with minimal security needed. For example, if my user on
the PHP page only has rights for a EXECUTE but they need to SELECT or
UPDATE data.. you need to define your procedure or function in MySQL
with SQL SECURITY DEFINER, and provide a DEFINER which has rights to
update or edit the table in question.

Really, I could go on all day about this.. security has always been a
big thing for me, having turned ethical hacker many years ago.

As for the solution I'm working on, it'll work with Apache and PHP.
It's an application written in PHP, compiled with Roadsend PHP
compiler to an application. It gets the MD5 checksum of scripts and
all theirs parents, up to the init() process. It makes sure that
nothing has been edited or tampered with. To get a DB connection,
you'd provide the DB hostname and user name; and the script itself
would return the password. The password is actually generated off the
hostname, user name and script MD5.. so not even the local developers
would know it. If the page trying to get the password for a DB
connection is modified.. say, to add a "echo $sPassword;" then it
wouldn't pass the sanity check and would never get a password back.
The way this would work, is that a PHP script is created that contains
a wrapper to this application.. that's the primary script we're making
sure hasn't been changed. It returns a link to the database. This
allows changes to other scripts which call the wrapper, without making
an administrator re-run the password generation tools to sync with
MySQL every time scripts are changed on a web site.

But whee.. security. Fun stuff.


Regards,
Michael Martinek

Re: Newbie Security Questions

..oO(Michael Martinek)

>mysql_real_escape_string will not change $nAccountID from the value of
>"0; DROP users;".. there's nothing to be escaped in the string. The
>proper way to protect against SQL injection is usually a combination
>of sprintf() or intval(), and mysql_real_escape_string().

Or prepared statements.

Micha