# Pre-RFC: Redacting parameters in back traces (SensitiveParameter Attribute)

- Thread: https://externals.io/message/116847
- Original archive: https://news-web.php.net/php.internals/116847
- Started by duesterhus@woltlab.com on 2022-01-07T18:10:36+00:00
- Messages: 2
- RFC: Redacting parameters in back traces (Implemented (PHP 8.2)) — https://externals.io/rfc/redact_parameters_in_back_traces.md
- Reply by email to internals@lists.php.net
- More for agents: search with https://externals.io/search.md?q={query} · RFCs at https://externals.io/rfc.md · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt

---

## #116847 — duesterhus@woltlab.com — 2022-01-07T18:10:36+00:00

URL: https://externals.io/message/116847#116847

Hi Internals!

PHP's stack traces in exceptions are very useful for debugging, because 
they include the original parameters for each stack frame, allowing the 
developer to see exactly what data is passed to a function call.

Unfortunately sometimes this verbosity is a drawback. Specifically when 
sensitive values, such as credentials, are passed to a function and some 
other function deep down the call stack throws an Exception.

One common "offender" is PDO which takes the database password as a 
constructor parameter and immediately attempts to connect to the 
database within the constructor, instead of having a pure constructor 
and a separate ->connect() method. Thus when the database connection 
fails the stack trace will include the database password:

> PDOException: SQLSTATE[HY000] [2002] No such file or directory in /var/www/html/test.php:3
> Stack trace:
> #0 /var/www/html/test.php(3): PDO->__construct('mysql:host=loca...', 'root', 'password')
> #1 {main}
We're developing an application that is commonly self-hosted by 
laypersons at a shared hosting provider. This application includes an 
error log, logging any uncaught exception including its stack trace. 
Depending on the configuration by the customer it might also publicly 
print error details, because this eases debugging by the layperson 
administrator.

To keep (database) credentials out of the log files - and more 
importantly hidden from any public visitors - our exception handler 
specifically scans the full stack trace for the PDO constructor to 
replace any parameters. More recently the exception handler was extended 
to also check each parameter via Reflection and redact any values for 
parameters called $password, $secret, and similar names. It also redacts 
any parameter having a `SensitiveArgument` attribute defined in our 
software.

While this works reasonably well for any code we write, it does not work 
for non-stock exception handlers, e.g. loggers such as Sentry. It also 
cannot redact sensitive arguments in third party libraries that do not 
match one of the special-cased parameter names, because they won't have 
the SensitiveArgument attribute.

For this reason we'd like to propose a *standardized* attribute to mark 
parameters as sensitive. This attribute can then be used by libraries to 
indicate their sensitive parameters.

With regards to loggers our proposal would be that PHP itself performs 
the redaction when collecting the parameters for all stack frames. 
Specifically we propose that PHP will place an object of the 
SensitiveParameter class instead of the actual parameter value. This way 
an exception handler can easily detect which parameters have been 
redacted by performing an `instanceof SensitiveParameter` check.

The `zend.exception_ignore_args` option is not an alternative in our 
opinion:
- On shared webhosters, the customer might not be able to configure it.
- The stack trace parameters are just too useful for debugging to 
completely strip them.

The `zend.exception_string_param_max_len` option is not alternative either:
- Many sensitive values might already be fully exposed before they are 
truncated. This specifically includes end-user credentials which tend to 
be low-entropy and shortish.

We have created a proof of concept implementation of our proposal at:

https://github.com/php/php-src/compare/master...WoltLab:sensitive-parameter

For the RFC and the final implementation we would need assistance by 
someone who is more experienced in PHP internals and the RFC process.

What our implementation already handles:
- Redaction of ordered arguments.
- Redaction of named arguments.
- Redaction of variadic arguments.

What's missing:
- Adding arguments to native functions / methods / classes (e.g. PDO's 
constructor)

Thank you all for reading through our proposal. We're happy to hear your 
feedback!

I already registered a Wiki account (timwolla, duesterhus@woltlab.com) 
and would require RFC karma to proceed if you feel like the proposal is 
worthwhile writing up as an RFC.

PS: It appears that the time on the ezmlm host is misconfigured. The 
email to confirm my subscription was dated 09:42:51 -0000, when it 
actually was 11:50:20 +0100 (i.e. off by 1 hour, 8 minutes).

Best regards
Tim Düsterhus
Developer WoltLab GmbH

-- 

WoltLab GmbH
Nedlitzer Str. 27B
14469 Potsdam

Tel.: +49 331 96784338

duesterhus@woltlab.com
www.woltlab.com

Managing director:
Marcel Werk

AG Potsdam HRB 26795 P

---

## #116848 — Ayesh Karunaratne — 2022-01-07T18:33:32+00:00

URL: https://externals.io/message/116847#116848 · In reply to: #116847

Hi Tim,
Thank you for opening the discussion. I personally find this feature
useful, and glancing at the diff, it looks like a rather
straightforward and minimal change.

FWIW, it is possible to selectively expose class properties in a class
object being `var_dump`-ed with the `__debugInfo` magic method, but it
is not used for stack traces.

For the RFC, someone with the permission to grant RFC karma should
grant it for you to start the RFC and open it for voting. You emailed
the right mailing list, I think someone will see this soon.
You can open a PR and start code reviews in the meantime.

Cheers,
Ayesh.

On Fri, Jan 7, 2022 at 11:40 PM Tim Düsterhus, WoltLab GmbH
<duesterhus@woltlab.com> wrote:
>
> Hi Internals!
>
> PHP's stack traces in exceptions are very useful for debugging, because
> they include the original parameters for each stack frame, allowing the
> developer to see exactly what data is passed to a function call.
>
> Unfortunately sometimes this verbosity is a drawback. Specifically when
> sensitive values, such as credentials, are passed to a function and some
> other function deep down the call stack throws an Exception.
>
> One common "offender" is PDO which takes the database password as a
> constructor parameter and immediately attempts to connect to the
> database within the constructor, instead of having a pure constructor
> and a separate ->connect() method. Thus when the database connection
> fails the stack trace will include the database password:
>
> > […]
> We're developing an application that is commonly self-hosted by
> laypersons at a shared hosting provider. This application includes an
> error log, logging any uncaught exception including its stack trace.
> Depending on the configuration by the customer it might also publicly
> print error details, because this eases debugging by the layperson
> administrator.
>
> To keep (database) credentials out of the log files - and more
> importantly hidden from any public visitors - our exception handler
> specifically scans the full stack trace for the PDO constructor to
> replace any parameters. More recently the exception handler was extended
> to also check each parameter via Reflection and redact any values for
> parameters called $password, $secret, and similar names. It also redacts
> any parameter having a `SensitiveArgument` attribute defined in our
> software.
>
> While this works reasonably well for any code we write, it does not work
> for non-stock exception handlers, e.g. loggers such as Sentry. It also
> cannot redact sensitive arguments in third party libraries that do not
> match one of the special-cased parameter names, because they won't have
> the SensitiveArgument attribute.
>
> For this reason we'd like to propose a *standardized* attribute to mark
> parameters as sensitive. This attribute can then be used by libraries to
> indicate their sensitive parameters.
>
> With regards to loggers our proposal would be that PHP itself performs
> the redaction when collecting the parameters for all stack frames.
> Specifically we propose that PHP will place an object of the
> SensitiveParameter class instead of the actual parameter value. This way
> an exception handler can easily detect which parameters have been
> redacted by performing an `instanceof SensitiveParameter` check.
>
> The `zend.exception_ignore_args` option is not an alternative in our
> opinion:
> - On shared webhosters, the customer might not be able to configure it.
> - The stack trace parameters are just too useful for debugging to
> completely strip them.
>
> The `zend.exception_string_param_max_len` option is not alternative either:
> - Many sensitive values might already be fully exposed before they are
> truncated. This specifically includes end-user credentials which tend to
> be low-entropy and shortish.
>
> We have created a proof of concept implementation of our proposal at:
>
> https://github.com/php/php-src/compare/master...WoltLab:sensitive-parameter
>
> For the RFC and the final implementation we would need assistance by
> someone who is more experienced in PHP internals and the RFC process.
>
> What our implementation already handles:
> - Redaction of ordered arguments.
> - Redaction of named arguments.
> - Redaction of variadic arguments.
>
> What's missing:
> - Adding arguments to native functions / methods / classes (e.g. PDO's
> constructor)
>
> Thank you all for reading through our proposal. We're happy to hear your
> feedback!
>
> I already registered a Wiki account (timwolla, duesterhus@woltlab.com)
> and would require RFC karma to proceed if you feel like the proposal is
> worthwhile writing up as an RFC.
>
> PS: It appears that the time on the ezmlm host is misconfigured. The
> email to confirm my subscription was dated 09:42:51 -0000, when it
> actually was 11:50:20 +0100 (i.e. off by 1 hour, 8 minutes).
>
> Best regards
> Tim Düsterhus
> Developer WoltLab GmbH
>
> --
>
> WoltLab GmbH
> Nedlitzer Str. 27B
> 14469 Potsdam
>
> Tel.: +49 331 96784338
>
> duesterhus@woltlab.com
> www.woltlab.com
>
> Managing director:
> Marcel Werk
>
> AG Potsdam HRB 26795 P
>
> --
> 
> To unsubscribe, visit: https://www.php.net/unsub.php
