📝 General cleanup

This commit is contained in:
Felix Rieseberg 2018-01-29 11:19:43 -08:00
parent c4dc8dacbf
commit fb8606dd0c

View file

@ -1,19 +1,19 @@
# Security, Native Capabilities, and Your Responsibility # Security, Native Capabilities, and Your Responsibility
As web developers, we usually enjoy the strong security net of the browser - the As web developers, we usually enjoy the strong security net of the browser -
risks associated with the code we write are relatively small. Our websites are the risks associated with the code we write are relatively small. Our websites
granted limited powers in a sandbox, and we trust that our users enjoy a browser are granted limited powers in a sandbox, and we trust that our users enjoy a
built by a large team of engineers that is able to quickly respond to newly browser built by a large team of engineers that is able to quickly respond to
discovered security threats. newly discovered security threats.
When working with Electron, it is important to understand that Electron is not When working with Electron, it is important to understand that Electron is not
a web browser. It allows you to build feature-rich desktop applications with a web browser. It allows you to build feature-rich desktop applications with
familiar web technologies, but your code wields much greater power. JavaScript familiar web technologies, but your code wields much greater power. JavaScript
can access the filesystem, user shell, and more. This allows you to build can access the filesystem, user shell, and more. This allows you to build
high quality native applications, but the inherent security risks scale with the high quality native applications, but the inherent security risks scale with
additional powers granted to your code. the additional powers granted to your code.
With that in mind, be aware that displaying arbitrary content from untrusted With that in mind, be aware that displaying arbitrary content from un-trusted
sources poses a severe security risk that Electron is not intended to handle. sources poses a severe security risk that Electron is not intended to handle.
In fact, the most popular Electron apps (Atom, Slack, Visual Studio Code, etc) In fact, the most popular Electron apps (Atom, Slack, Visual Studio Code, etc)
display primarily local content (or trusted, secure remote content without Node display primarily local content (or trusted, secure remote content without Node
@ -34,8 +34,8 @@ contributions available today, Electron will often not be on the very latest
version of Chromium, lagging behind by either days or weeks. version of Chromium, lagging behind by either days or weeks.
We feel that our current system of updating the Chromium component strikes an We feel that our current system of updating the Chromium component strikes an
appropriate balance between the resources we have available and the needs of the appropriate balance between the resources we have available and the needs of
majority of applications built on top of the framework. We definitely are the majority of applications built on top of the framework. We definitely are
interested in hearing more about specific use cases from the people that build interested in hearing more about specific use cases from the people that build
things on top of Electron. Pull requests and contributions supporting this things on top of Electron. Pull requests and contributions supporting this
effort are always very welcome. effort are always very welcome.
@ -44,15 +44,15 @@ effort are always very welcome.
A security issue exists whenever you receive code from a remote destination and A security issue exists whenever you receive code from a remote destination and
execute it locally. As an example, consider a remote website being displayed execute it locally. As an example, consider a remote website being displayed
inside a browser window. If an attacker somehow manages to change said content inside a `BrowserWindow`. If an attacker somehow manages to change said content
(either by attacking the source directly, or by sitting between your app and (either by attacking the source directly, or by sitting between your app and
the actual destination), they will be able to execute native code on the user's the actual destination), they will be able to execute native code on the user's
machine. machine.
> :warning: Under no circumstances should you load and execute remote code with > :warning: Under no circumstances should you load and execute remote code with
Node.js integration enabled. Instead, use only local files (packaged together with Node.js integration enabled. Instead, use only local files (packaged together
your application) to execute Node.js code. To display remote content, use the with your application) to execute Node.js code. To display remote content, use
`webview` tag and make sure to disable the `nodeIntegration`. the `webview` tag and make sure to disable the `nodeIntegration`.
#### Checklist: Security Recommendations #### Checklist: Security Recommendations
@ -60,39 +60,34 @@ This is not bulletproof, but at the least, you should attempt the following:
* [Only display secure (https) content](#only-display-secure-content) * [Only display secure (https) content](#only-display-secure-content)
* [Disable the Node integration in all renderers that display remote content](#disable-node-integration-for-remote-content) * [Disable the Node integration in all renderers that display remote content](#disable-node-integration-for-remote-content)
(setting `nodeIntegration` to `false` in `webPreferences`) * [Enable context isolation in all renderers that display remote content](#enable-context-isolation-for-remote-content)
* Enable context isolation in all renderers that display remote content * [Use `ses.setPermissionRequestHandler()` in all sessions that load remote content](#handle-session-permission-requests-from-remote-content)
(setting `contextIsolation` to `true` in `webPreferences`) * [Do not disable `webSecurity`](#do-not-disable-websecurity)
* Use `ses.setPermissionRequestHandler()` in all sessions that load remote content * [Define a `Content-Security-Policy`](#define-a-content-security-policy)
* Do not disable `webSecurity`. Disabling it will disable the same-origin policy.
* Define a [`Content-Security-Policy`](http://www.html5rocks.com/en/tutorials/security/content-security-policy/)
, and use restrictive rules (i.e. `script-src 'self'`) , and use restrictive rules (i.e. `script-src 'self'`)
* [Override and disable `eval`](https://github.com/nylas/N1/blob/0abc5d5defcdb057120d726b271933425b75b415/static/index.js#L6-L8) * [Override and disable `eval`](#override-and-disable)
, which allows strings to be executed as code. , which allows strings to be executed as code.
* Do not set `allowRunningInsecureContent` to true. * [Do not set `allowRunningInsecureContent` to `true`](#do-not-set-allowRunningInsecureContent-to-true)
* Do not enable `experimentalFeatures` or `experimentalCanvasFeatures` unless * [Do not enable experimental features](#do-not-enable-enable-experimental-features)
you know what you're doing. * [Do not use `blinkFeatures`](#do-not-use-blinkfeatures)
* Do not use `blinkFeatures` unless you know what you're doing. * [WebViews: Do not use `allowpopups`](#do-not-use-allowpopups)
* WebViews: Do not add the `nodeintegration` attribute.
* WebViews: Do not use `disablewebsecurity`
* WebViews: Do not use `allowpopups`
* WebViews: Do not use `insertCSS` or `executeJavaScript` with remote CSS/JS.
* [WebViews: Verify the options and params of all `<webview>` tags](#verify-webview-options-before-creation) * [WebViews: Verify the options and params of all `<webview>` tags](#verify-webview-options-before-creation)
## Only Display Secure Content ## Only Display Secure Content
Any resources not included with your application should be loaded using a secure
protocol like `HTTPS`. Furthermore, avoid "mixed content", which occurs when the Any resources not included with your application should be loaded using a
initial HTML is loaded over an `HTTPS` connection, but additional resources secure protocol like `HTTPS`.
(scripts, stylesheets, etc) are loaded over an insecure connection.
### Why? ### Why?
`HTTPS` has three main benefits: `HTTPS` has three main benefits:
1) It authenticates the remote server, ensuring that the host is actually who it 1) It authenticates the remote server, ensuring that the host is actually who
claims to be. When loading a resource from an `HTTPS` host, it prevents an it claims to be. When loading a resource from an `HTTPS` host, it prevents
attacker from impersonating that host, thus ensuring that the computer your an attacker from impersonating that host, thus ensuring that the computer
app's users are connecting to is actually the host you wanted them to connect your app's users are connecting to is actually the host you wanted them to
to. connect to.
2) It ensures data integrity, asserting that the data was not modified while in 2) It ensures data integrity, asserting that the data was not modified while in
transit between your application and the host. transit between your application and the host.
3) It encryps the traffic between your user and the destination host, making it 3) It encryps the traffic between your user and the destination host, making it
@ -100,6 +95,7 @@ initial HTML is loaded over an `HTTPS` connection, but additional resources
the host. the host.
### How? ### How?
```js ```js
// Bad // Bad
browserWindow.loadURL('http://my-website.com') browserWindow.loadURL('http://my-website.com')
@ -118,29 +114,33 @@ browserWindow.loadURL('https://my-website.com')
<link rel="stylesheet" href="https://cdn.com/style.css"> <link rel="stylesheet" href="https://cdn.com/style.css">
``` ```
## Disable Node Integration for Remote Content
It is paramount that you disable Node integration in any renderer (`BrowserWindow`,
`BrowserView`, or `WebView`) that loads remote content. The goal of disabling Node
integration is to limit the powers you grant to remote content, thus making it
dramatically more difficult for an attacker to harm your users should they gain
the ability to execute JavaScript on your website.
Disabling Node integration does not mean that you cannot grant additional powers ## Disable Node Integration for Remote Content
to the website you are loading. If you are opening a `BrowserWindow` pointed at
`https://my-website.com`, the goal is to give that website exactly the abilities It is paramount that you disable Node integration in any renderer
it needs, but no more. (`BrowserWindow`, `BrowserView`, or `WebView`) that loads remote content. The
goal of disabling Node integration is to limit the powers you grant to remote
content, thus making it dramatically more difficult for an attacker to harm
your users should they gain the ability to execute JavaScript on your website.
Disabling Node integration does not mean that you cannot grant additional
powers to the website you are loading. If you are opening a `BrowserWindow`
pointed at `https://my-website.com`, the goal is to give that website exactly
the abilities it needs, but no more.
### Why? ### Why?
A cross-site-scripting (XSS) attack becomes dramatically more dangerous if an A cross-site-scripting (XSS) attack becomes dramatically more dangerous if an
attacker can jump out of the renderer process and execute code on the user's attacker can jump out of the renderer process and execute code on the user's
computer. Cross-site-scripting attacks are fairly common - and while an issue, computer. Cross-site-scripting attacks are fairly common - and while an issue,
their power is usually limited to messing with the website that they are executed their power is usually limited to messing with the website that they are
on. However, in a renderer process with Node.js integration enabled, an XSS attack executed on. However, in a renderer process with Node.js integration enabled,
becomes a whole different class of attack: A so-called "Remote Code Execution" an XSS attack becomes a whole different class of attack: A so-called "Remote
(RCE) attack. Disabling Node.js integration limits the power of successful XSS Code Execution" (RCE) attack. Disabling Node.js integration limits the power
attacks. of successful XSS attacks.
### How? ### How?
```js ```js
// Bad // Bad
const mainWindow = new BrowserWindow() const mainWindow = new BrowserWindow()
@ -157,13 +157,21 @@ const mainWindow = new BrowserWindow({
mainWindow.loadURL('https://my-website.com') mainWindow.loadURL('https://my-website.com')
``` ```
When disabling Node integration, you can still expose APIs to your ```html
website that do consume Node.js modules or features. Preload scripts continue to <!-- Bad -->
have access to `require` and other Node.js features, allowing developers to expose <webview nodeIntegration src="page.html"></webview>
a custom API to remotely loaded content.
In the following example preload script, the later loaded website will have access <!-- Good -->
to a `window.readConfig()` method, but no Node.js features. <webview src="page.html"></webview>
```
When disabling Node integration, you can still expose APIs to your website that
do consume Node.js modules or features. Preload scripts continue to have access
to `require` and other Node.js features, allowing developers to expose a custom
API to remotely loaded content.
In the following example preload script, the later loaded website will have
access to a `window.readConfig()` method, but no Node.js features.
```js ```js
const { readFileSync } = require('fs') const { readFileSync } = require('fs')
@ -231,6 +239,7 @@ document.addEventListener('DOMContentLoaded', () => {
}) })
``` ```
## Handle Session Permission Requests From Remote Content ## Handle Session Permission Requests From Remote Content
You may have seen permission requests while using Chrome: They pop up whenever You may have seen permission requests while using Chrome: They pop up whenever
@ -476,23 +485,27 @@ the ability to create new popups.
## Verify WebView Options Before Creation ## Verify WebView Options Before Creation
A WebView created in a renderer process that does not have Node.js integration A WebView created in a renderer process that does not have Node.js integration
enabled will not be able to enable integration itself. However, a WebView will enabled will not be able to enable integration itself. However, a WebView will
always create an independent renderer process with its own `webPreferences`. always create an independent renderer process with its own `webPreferences`.
It is a good idea to control the creation of new `WebViews` from the main process It is a good idea to control the creation of new `WebViews` from the main
and to verify that their webPreferences do not disable security features. process and to verify that their webPreferences do not disable security
features.
### Why? ### Why?
Since WebViews live in the DOM, they can be created by a script running on your Since WebViews live in the DOM, they can be created by a script running on your
website even if Node integration is otherwise disabled. website even if Node integration is otherwise disabled.
Electron enables developers to disable various security features that control Electron enables developers to disable various security features that control
a renderer process. In most cases, developers do not need to disable any of those a renderer process. In most cases, developers do not need to disable any of
features - and you should therefore not allow different configurations for newly those features - and you should therefore not allow different configurations
created `<WebView>` tags. for newly created `<WebView>` tags.
### How? ### How?
Before a `<WebView>` tag is attached, Electron will fire the Before a `<WebView>` tag is attached, Electron will fire the
`will-attach-webview` event on the hosting `webContents`. Use the event to `will-attach-webview` event on the hosting `webContents`. Use the event to
prevent the creation of WebViews with possibly insecure options. prevent the creation of WebViews with possibly insecure options.