curl vulnerability disclosure policy

This document describes how security vulnerabilities are handled in the curl project.

Publishing Information

All known and public curl or libcurl related vulnerabilities are listed on the curl website security page.

Security vulnerabilities should not be entered in the project's public bug tracker.

Vulnerability Handling

The typical process for handling a new security vulnerability is as follows.

No information should be made public about a vulnerability until it is formally announced at the end of this process. That means, for example, that a bug tracker entry must NOT be created to track the issue since that makes the issue public and it should not be discussed on any of the project's public mailing lists. Messages associated with any commits should not make any reference to the security nature of the commit if done prior to the public announcement.

• The person discovering the issue, the reporter, reports the vulnerability on HackerOne. Issues filed there reach a handful of selected and trusted people.

• Messages that do not relate to the reporting or managing of an undisclosed security vulnerability in curl or libcurl are ignored and no further action is required.

• A person in the security team responds to the original report to acknowledge that a human has seen the report.

• The security team investigates the report and either rejects it or accepts it. See below for examples of problems that are not considered vulnerabilities.

• If the report is rejected, the team writes to the reporter to explain why.

• If the report is accepted, the team writes to the reporter to let them know it is accepted and that they are working on a fix.

• The security team discusses the problem, works out a fix, considers the impact of the problem and suggests a release schedule. This discussion should involve the reporter as much as possible.

• The release of the information should be "as soon as possible" and is most often synchronized with an upcoming release that contains the fix. If the reporter, or anyone else involved, thinks the next planned release is too far away, then a separate earlier release should be considered.

• Write a security advisory draft about the problem that explains what the problem is, its impact, which versions it affects, solutions or workarounds, when the release is out and make sure to credit all contributors properly. Figure out the CWE (Common Weakness Enumeration) number for the flaw. See SECURITY-ADVISORY for help on creating the advisory.

• Request a CVE Id for the issue. curl is a CNA (CVE Numbering Authority) and can request its own numbers.

• Update the "security advisory" with the CVE number.

• The security team commits the fix in a private branch. The commit message should ideally contain the CVE number. If the severity level of the issue is set to Low or Medium, the fix is allowed to get merged into the master repository via a normal PR - but without mentioning it being a security vulnerability.

• The monetary reward part of the bug-bounty is managed by the Internet Bug Bounty team and the reporter is asked to request the reward from them after the issue has been completely handled and published by curl.

• No more than 10 days before release, inform distros@openwall to prepare them about the upcoming public security vulnerability announcement - attach the advisory draft for information with CVE and current patch. 'distros' does not accept an embargo longer than 14 days and they do not care for Windows-specific flaws.

• No more than 48 hours before the release, the private branch is merged into the master branch and pushed. Once pushed, the information is accessible to the public and the actual release should follow suit immediately afterwards. The time between the push and the release is used for final tests and reviews.

• The project team creates a release that includes the fix.

• The project team announces the release and the vulnerability to the world in the same manner we always announce releases. It gets sent to the curl-announce, curl-library and curl-users mailing lists.

• The security webpage on the website should get the new vulnerability mentioned.

security (at curl dot se)

This is a private mailing list for discussions on and about curl security issues.

Who is on this list? There are a couple of criteria you must meet, and then we might ask you to join the list or you can ask to join it. It really is not a formal process. We basically only require that you have a long-term presence in the curl project and you have shown an understanding for the project and its way of working. You must have been around for a good while and you should have no plans of vanishing in the near future.

We do not make the list of participants public mostly because it tends to vary somewhat over time and a list somewhere only risks getting outdated.

Publishing Security Advisories

1. Write up the security advisory, using markdown syntax. Use the same subtitles as last time to maintain consistency.

2. Name the advisory file after the allocated CVE id.

3. Add a line on the top of the array in curl-www/docs/vuln.pm.

4. Put the new advisory markdown file in the curl-www/docs/ directory. Add it to the git repository.

5. Run make in your local web checkout and verify that things look fine.

6. On security advisory release day, push the changes on the curl-www repository's remote master branch.

HackerOne

Request the issue to be disclosed. If there are sensitive details present in the report and discussion, those should be redacted from the disclosure. The default policy is to disclose as much as possible as soon as the vulnerability has been published.

Bug Bounty

See BUG-BOUNTY for details on the bug bounty program.

Severity levels

The curl project's security team rates security problems using four severity levels depending how serious we consider the problem to be. We use Low, Medium, High and Critical. We refrain from using numerical scoring of vulnerabilities.

When deciding severity level on a particular issue, we take all the factors into account: attack vector, attack complexity, required privileges, necessary build configuration, protocols involved, platform specifics and also what effects a possible exploit or trigger of the issue can lead do, including confidentiality, integrity or availability problems.

Low

This is a security problem that is truly hard or unlikely to exploit or trigger. Due to timing, platform requirements or the fact that options or protocols involved are rare etc. Past example

Medium

This is a security problem that is less hard than Low to exploit or trigger. Less strict timing, wider platforms availability or involving more widely used options or protocols. A problem that usually needs something else to also happen to become serious. Past example

High

This issue in itself a serious problem with real world impact. Flaws that can easily compromise the confidentiality, integrity or availability of resources. Exploiting or triggering this problem is not hard. Past example

Critical

Easily exploitable by a remote unauthenticated attacker and lead to system compromise (arbitrary code execution) without requiring user interaction, with a common configuration on a popular platform. This issue has few restrictions and requirements and can be exploited easily using most curl configurations. Past example

Not security issues

This is an incomplete list of issues that are not considered vulnerabilities.

Small memory leaks

We do not consider a small memory leak a security problem; even if the amount of allocated memory grows by a small amount every now and then. Long-living applications and services already need to have counter-measures and deal with growing memory usage, be it leaks or just increased use. A small memory or resource leak is then expected to not cause a security problem.

Of course there can be a discussion if a leak is small or not. A large leak can be considered a security problem due to the DOS risk. If leaked memory contains sensitive data it might also qualify as a security problem.

Never-ending transfers

We do not consider flaws that cause a transfer to never end to be a security problem. There are already several benign and likely reasons for transfers to stall and never end, so applications that cannot deal with never-ending transfers already need to have counter-measures established.

If the problem avoids the regular counter-measures when it causes a never- ending transfer, it might be a security problem.

Not practically possible

If the flaw or vulnerability cannot practically get executed on existing hardware it is not a security problem.

API misuse

If a reported issue only triggers by an application using the API in a way that is not documented to work or even documented to not work, it is probably not going to be considered a security problem. We only guarantee secure and proper functionality when the APIs are used as expected and documented.

There can be a discussion about what the documentation actually means and how to interpret the text, which might end up with us still agreeing that it is a security problem.

Local attackers already present

When an issue can only be attacked or misused by an attacker present on the local system or network, the bar is raised. If a local user wrongfully has elevated rights on your system enough to attack curl, they can probably already do much worse harm and the problem is not really in curl.

Experiments

Vulnerabilities in features which are off by default (in the build) and documented as experimental, are not eligible for a reward and we do not consider them security problems.

URL inconsistencies

URL parser inconsistencies between browsers and curl are expected and are not considered security vulnerabilities. The WHATWG URL Specification and RFC 3986+ (the plus meaning that it is an extended version) are not completely interoperable.

Obvious parser bugs can still be vulnerabilities of course.

Visible command line arguments

The curl command blanks the contents of a number of command line arguments to prevent them from appearing in process listings. It does not blank all arguments even if some of them that are not blanked might contain sensitive data. We consider this functionality a best-effort and omissions are not security vulnerabilities.

• not all systems allow the arguments to be blanked in the first place
• since curl blanks the argument itself they are readable for a short moment no matter what
• virtually every argument can contain sensitive data, depending on use
• blanking all arguments would make it impractical for users to differentiate curl command lines in process listings

Busy-loops

Busy-loops that consume 100% CPU time but eventually end (perhaps due to a set timeout value or otherwise) are not considered security problems. Applications are supposed to already handle situations when the transfer loop legitimately consumes 100% CPU time, so while a prolonged such busy-loop is a nasty bug, we do not consider it a security problem.

Saving files

curl cannot protect against attacks where an attacker has write access to the same directory where curl is directed to save files.

Tricking a user to run a command line

A creative, misleading or funny looking command line is not a security problem. The curl command line tool takes options and URLs on the command line and if an attacker can trick the user to run a specifically crafted curl command line, all bets are off. Such an attacker can just as well have the user run a much worse command that can do something fatal (like sudo rm -rf /).

Terminal output and escape sequences

Content that is transferred from a server and gets displayed in a terminal by curl may contain escape sequences or use other tricks to fool the user. This is curl working as designed and is not a curl security problem. Escape sequences, moving cursor, changing color etc, is also frequently used for good. To reduce the risk of getting fooled, save files and browse them after download using a display method that minimizes risks.

NULL dereferences and crashes

If a malicious server can trigger a NULL dereference in curl or otherwise cause curl to crash (and nothing worse), chances are big that we do not consider that a security problem.

Malicious servers can already cause considerable harm and denial of service like scenarios without having to trigger such code paths. For example by stalling, being terribly slow or by delivering enormous amounts of data. Additionally, applications are expected to handle "normal" crashes without that being the end of the world.

There need to be more and special circumstances to treat such problems as security issues.