|
in reply to Re: Hacker Proofing My Script
in thread Hacker Proofing My Script
Placeholders are not the same as "automatic quoting rules". The $dbh->quote() method applies quoting to the values and then those are *inserted into the SQL statement* and that SQL statement is passed to the RDBMS. OTOH, for an RDBMS that supports placeholder, and when placeholders are used instead of $dbh->quote(), the values are not quoted, *are not inserted into the SQL statement*, and are passed separately to the RDBMS along with the SQL statement which still has placeholders marks in it. The RDBMS than operates on the statement + the values without ever needing to create a SQL statement that contains the values and therefore without the danger of having unkown SQL statements (injected as values) executed. So if the RDBMS supports placeholders (as distinct from a DBD that emulates them), placeholders are much more secure than quoting, even quoting with $dbh->quote(). |
Re^3: Hacker Proofing My Script
by CountZero (Bishop) on Oct 04, 2004 at 20:15 UTC
|
Quite right, thanks for refreshing my memory! But I was not entirely wrong here: what is OK on one DB might be terribly insecure on another.In that respect, MySQL supports prepared statements only since its version 4.1 and unless DBD::MySQL is updated to take advantage of it (I didn't think so), could it be that the placeholder-magic is faked by DBD/DBD::MYSLQ and that it simply relies on quoting and interpolating the placeholders? That would of course be a Bad Thing.
CountZero "If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law
| [reply] |
|
|
could it be that the placeholder-magic is faked by DBD/DBD::MYSLQ and that it simply relies on quoting and interpolating the placeholders? That would of course be a Bad Thing.
Not necessarily. The DBD::MySQL driver is likely to have been
written by someone competent, who understands how to do the correct quoting to make it injection proof; this is in contrast
to typical user-level code, which has a good chance of getting it wrong. So even faked placeholders buy you security.
Dave.
| [reply] |
Re^3: Hacker Proofing My Script
by awohld (Hermit) on Oct 04, 2004 at 21:02 UTC
|
jZed -
So you say if my database, which is mySQL 3.22.32, supports placeholders, then I don't have to worry if the DBD supports or emulates them?
Thanks
Adam | [reply] |
|
|
I don't know a heck of a lot about MySQL, but according to gmax's excellent New twist for DBD::mysql, DBD::mysql currently does emulate placeholders but in such a way as to preserve security. I'd suggest reading his node for further details.
| [reply] |
|
|
It doesn't seem to be totally secure:
As for LIMIT ?,?. The reason why it was not supported since 2.9002 is that it allowed for sql injection attacks, and it is not trivial to fix, in fact, I *just* scanned over Patrick's code and found a bug in the LIMIT handling code as can be read in Re: New twist for DBD::mysql.
CountZero "If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law
| [reply] |
|
