Hacker News Reader

Comment by infogulch | Hacker News Reader

The article describes the current format as a zip of files where the main content of presentation slides is stored in a `content.xml` file within. Yes, sqlite may have had file format vulnerabilities in the past, but I cannot imagine trusting it less than any xml library in existence.

View on HN

At work we used a library that relied on Expat XML parser. After 20 years it still gets exploitable bugs. As the software was Windows-only we patched the library to use OS API for XML parsing to delegate to Microsoft the job of fixing security bugs.

You’d think so from reading [1]. [2] though paints a very different picture.

> The attacker can submit a maliciously crafted database file to the application that the application will then open and query.

I think it’s not impossible to secure SQLite better but it takes some work (like how Chrome sandboxes it in a separate process).

[1] https://www.sqlite.org/security.html [2] https://www.sqlite.org/cves.html

I think his point is that if an attacker wants to give you bad data, there is already zero barrier to editing an XML file inside a zip.

> [2] though paints a very different picture.

I don't agree. That document states that the CVEs have historically required one of two preconditions:

> The attacker can submit and run arbitrary SQL statements.

> The attacker can submit a maliciously crafted database file to the application that the application will then open and query.

If you look at the actual list of CVEs, all but one start with ‘Malicious SQL statement’. The single one that doesn't is suffixed with ‘The bug never appeared in any official SQLite release’.

In other words, there has never been an official release of SQLite which was vulnerable when presented with a crafted database file.

> If you look at the actual list of CVEs, all but one start with ‘Malicious SQL statement’. The single one that doesn't is suffixed with ‘The bug never appeared in any official SQLite release’.

The list on the page is partial, and the language on that page is disturbingly dismissive. Reading leaves me more concerned that they're too arrogant to take security researchers seriously.

I'm sure they aren't perfect, but their test suite and extensive fuzz testing is far better than any XML or JSON parser I've ever seen

https://www.sqlite.org/testing.html

I do not agree about JSON. As a format it is so trivial one can cover it with few test cases. SQL dialect that SQLite uses is vastly more complex.

You'd think so, but in practice no two JSON parsers have the same bugs, so it's more complex than you might think:

http://seriot.ch/json/parsing.html#all_results

Think about it if even JSON which is kinda simple has that many problems with parsers how likely is it that something multiple degrees of complexity higher which wasn't developed with untrusted database fiels in mind is secure?

Besides that the large majority of thinks tested in your link are irrelevant for the security in this use-case. They do matter if you need fully stable de- and re-serialization or e.g. have a fully trusted validator. E.g. for certain security token use-cases it matters. Many of the thinks which are marked as failure could even be considered as a "more robust" or "more secure" (through not fully standard conform) parser. E.g. parsing tailing commas (disallowed by JSON), rejecting certain unescaped control/special characters (which JSON allows), rejecting to long field values (which JSON allows) or similar. What matters is that a) it has a protection about to deep recursion and as such no stack overflow can happen, b) it makes sure the text it outputs is correctly encoded (independent of weather or not it was well-formed in the JSON blob) and c) it doesn't crash in a way which leads to security problems (which the test suite you linked doesn't differentiate from "acceptable" crashes).

As long as one uses a single parser to read a particular input it matters little if it produces a result that is different from what some other parser generates as long as the parser has no security bugs. And JSON is simple enough to cover a particular interpretation of its spec with test suits to make a security vulnerability extremely unlikely.

Surely if one uses one parser to verify the payload and another to use it, a disaster comes as was with IPhone verification bug.

I seem to remember really great interaction with lcamtuf (of AFL fame): https://lcamtuf.blogspot.com/2015/04/finding-bugs-in-sqlite-...

I feel like there should be a way to make a maliciously crafted database file execute arbitrary SQL... SQLite doesn't do stored procedures, but there's gotta be something in that surface area.

if select is all you need then a view would suffice

I expect the majority of the vulnerabilities were in the command parser. Which attack vector is gone from the database file.

AFAIK, SQLite stores database schema as text, so certain subset of parser still has to be invoked on untrusted DDL.

> SQLite doesn't do stored procedures

Link: https://www.sqlite.org/gencol.html

Example:

CREATE TABLE t1(

   a INTEGER PRIMARY KEY,

   b INT,

   c TEXT,

   d INT GENERATED ALWAYS AS (a*abs(b)) VIRTUAL,

   e TEXT GENERATED ALWAYS AS (substr(c,b,b+1)) STORED

);

Note how you can call functions in the columns. Thats one feature which a hardened sqlite for less trusted documents would need to not have.

The proposal to use sqlite database files instead of zip files falls under the precondition that an attacker can submit a maliciously crafted database file. So the security concerns apply.

SQLitte is mostly used in situations where neither the database nor the SQL statements can be provided by the attacker. If SQLite is used to exchange data both of these attack vectors are available.

https://www.sqlite.org/cves.html

18 CVEs

http://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=xml

> There are 1776 CVE Records that match your search.

Do I win?

Probably better to compare Apples and Apples[1].

> There are 126 CVE Records that match your search.

[1] https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=sqlite

That's fair. Though the disparity is still an order of magnitude...

More seriously, my point is that SQLite is very unlikely to be the most vulnerable piece of software in your stack. It's probably the most stable and bug free piece of software that you use on a day to day basis that hasn't been formally proven correct. Even openoffice itself has more CVEs: https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=openoffice

And some of them are due to the very issue I brought up above...

> CVE-2012-0037 Redland Raptor (aka libraptor) before 2.0.7, as used by OpenOffice 3.3 and 3.4 Beta, LibreOffice before 3.4.6 and 3.5.x before 3.5.1, and other products, allows user-assisted remote attackers to read arbitrary files via a crafted XML external entity (XXE) declaration and reference in an RDF document.

> My point is that SQLite is very unlikely to be the most vulnerable piece of software in your stack.

Sqlite is very high quality, but I am rather doubtful this is true if you actually care about having a secure system to start with. I can think of three main reasons an office suite would have serious security vulnerabilities:

1. It's written in an unsuitable, memory-unsafe, programming language (as most office suites sadly are).

2. It includes an unsuitable programming language for scripting purposes (Word Macros are of course infamous for this).

3. Vulnerabilities in one of its dependencies (like XML parsers)

If you care to, 1 and 2 are very easy to avoid and 3 will be pretty hard to avoid completely, but you can limit your exposure. You need a GUI toolkit at the minimum, but you can completely avoid parsing as an attack vector by either not using a braindamaged format like XML in the unlikely case you don't need compatibility, or if you must, use an xml (zip etc.) parser written in a memory safe language. If you followed all other steps, but chose sqlite instead, I'd say chances a very good it's now your main vulnerability.

You have crafted a nigh-impossible tower of requirements that you admit are way beyond what is commonly practiced in the software under question in order to disclaim the statement "SQLite is very unlikely to be the most vulnerable piece of software in your stack". Right now on the front page of HN is an article about how Microsoft was also a victim of what may come to be labeled the most impactful hack ever, a list which includes most (all?) US government agencies. The root vuln? A crap admin password used by an IT monitoring software company which has software deployed to an alarming percentage of the industry.

I dunno man, I feel like my original statement is pretty reasonable. Let's please not move the goalposts to the moon. :)

I don't think your statement is unreasonable at all coming from the common perspective that it is part of the natural order of things that opening a text document would expose your computer to a non-negligible risk of hostile takeover.

Personally I find this a very unreasonable thing to accept. Secure (single user) text processing (etc.) is not a hard problem at all and there is zero reason it should create any security risks.

Crap passwords (and security practices) are not pertinent at all here, unless you also argue that Boeing shouldn't really bother to make airplanes with non-lethal autopilot interactions because being killed by a mis-designed airplane is the least of risks to the typical US citizen's life give how obese he or she is.

The pertinent question is: if you are to design a wordprocessor, should you avoid baking sqlite into the design because of security concerns? I think you should. Sqlite is written and tested very well, but it is written in not formally verified C, and using a memory unsafe language for reading potentially hostile databases poses a completely unnecessary and avoidable security risk.

Or run the IO in a sandbox where it is streamed out in a simpler/safer format. Or run the entire suite in a sandbox like we do with parts of web browsers.

How is this better than just avoiding the problem in the first place?

Because SQLite, as a general purpose database system, can be used for loads of things.

A two-piece component (DB + sandbox) that you use for 100 different things is going to end up much more secure (overall) and require less human effort than maintaining 100 different systems for each use case.

It's like "don't roll your own crypto". That is the advice not because it's impossible to roll your own secure crypto, but because it is just much more likely that you will repeat bugs that other people have already figured out, so it is better if everyone is using one library so that when an issue is found and resolved, it applies to everyone. It is still good advice even though that 3rd party crypto lib is probably going to have a lot of unneeded functionality that actually increases the overall attack surface area than if you made your own for whatever your specific purpose is. I think the same logic applies here.

No, the article just replaces the container with SQL but not the components (e.g. a slide). Which without extra changes are still XML, stored as text in the database.

The change from storing the inner document format in something which is not XML is completely independent of changing the outer container to sqlite.

SQL isn't really suited as a markup language so storing styles "flowing" text in it isn't that good of an idea. The article only recommends storing all separate blobs of styled text in tables (it also sidesteps the whole problem about flow text across multiple pages by simple using slides as an example ;=) ).

Lastly I'm against using a "stock"/unhardened sqlite version, not sqlite in general. Ironically on of the features I would disable for now is related to text search but I don't remember the name of it so I can't really use it as an argument :=/.

So if you ask me:

- Change containers to a different format then zipped sql files, consider sqlite but use a hardened feature limited sqlite version.

- Change the inner representation away from XML(1)

(1): Ok, there are ways to safely use XML but this isn't useful for this discussion and they require a bunch of thinks which all reduce usability (of the XML library) and performance and only work for certain use-cases where you e.g. do not need fully stable serialization and de-serialization and might be limited to a subset of XML and don't rely on the verifier for anything "safety" realated and preferably don't use C/C++ to write a parser... so it's easier to just not do it tbh.

I think the GP is more an statement of how bad security is with XML than how good it is with SQLite. See for example https://owasp.org/www-community/vulnerabilities/XML_External....

Reply on news.ycombinator.com