Chromium's multi-process architecture

Summary

The script from a Chrome Multi-Process Architecture presentation delves into the design goals of creating a stable, fast, and secure browser through multi-process division. It discusses the importance of asynchronous communication, the architecture's IPC (Inter-Process Communication) model, and the challenges of rendering, painting, and scrolling in a multi-process environment. The talk also touches on sandboxing for security and the handling of plug-ins and tabs, providing insights into the complexities of browser development.

Takeaways

• 🌟 The main goal of Chrome's Multi-Process Architecture is to enhance browser stability, speed, and security by dividing the browser into multiple processes.
• 🔒 Security was significantly improved by sandboxing the renderer process, limiting the damage in case of corruption, and employing a team of sandboxing experts.
• 🖥️ The architecture separates the browser into different threads for UI, I/O, file operations, and database management to prevent blocking and ensure responsiveness.
• 🔄 Asynchronous communication was crucial for stability, leading to the development of an IPC (Inter-Process Communication) system that avoids hanging the browser UI.
• 🖼️ The use of backing stores in RenderView hosts allows Chrome to display a recent representation of web pages even if the renderer is unresponsive.
• 🖱️ Input events on Windows take precedence over painting, which led to the development of an acknowledgment-based system for updating the display without causing UI jank.
• 🔄 For performance, Chrome uses optimized scrolling techniques like ScrollDC and ScrollWindow, but certain web page content can defeat these optimizations.
• 🎨 Painting, scrolling, and resizing operations are managed through a combination of IPC, backing stores, and acknowledgments to ensure a smooth user experience.
• 📊 The script highlights the complexity of managing multi-process architecture, especially in handling欣欣向荣欣欣向荣欣欣向荣同步IPCs and avoiding deadlocks with plugins.
• 📑 The architecture avoids using hWnds for renderers to prevent message view interlocking and potential browser hangs, instead relying on an in-memory approach.
• 🚀 Chrome can quickly close tabs by terminating child processes if there are no unload handlers, showcasing the benefits of a multi-process design.

Q & A

What is the primary goal of Chrome's Multi-Process Architecture?

The primary goal of Chrome's Multi-Process Architecture is to create a browser that is stable, fast, and secure by dividing the browser into multiple processes.

How does the Multi-Process Architecture contribute to browser stability?

The Multi-Process Architecture contributes to stability by ensuring that if a renderer process crashes, it doesn't affect the entire browser, and by preventing the renderer from causing the browser to hang or lock up.

What is the significance of asynchronous communication in the Multi-Process Architecture?

Asynchronous communication is significant because it prevents the user interface from becoming janky or unresponsive, thereby enhancing the overall user experience.

How does Chrome utilize multi-core systems to improve performance?

Chrome can take advantage of multi-core systems by allowing different processes to run independently, which can improve performance, especially when handling multiple applications or complex web pages.

What is the role of sandboxing in Chrome's security strategy?

Sandboxing plays a crucial role in Chrome's security strategy by isolating the renderer process from the operating system, limiting the potential harm if the renderer were ever corrupted.

Can you explain the concept of RenderView and RenderView host in the context of Chrome's architecture?

In Chrome's architecture, RenderView is a wrapper around the WebView in the renderer process, responsible for IPC communication. The RenderView host is in the browser process and corresponds to a native widget, with one for each tab, facilitating the communication between the renderer and the browser.

Why did Chrome decide against using COM for IPC?

Chrome decided against using COM for IPC because it had fundamental limitations, including synchronous IPC which could block the browser UI thread, leading to poor performance and stability.

What is the purpose of having multiple threads in Chrome's browser and renderer processes?

Multiple threads in Chrome's processes serve different purposes: the UI thread handles native widgetry, the I/O thread manages IPC traffic and network loading, the file thread handles blocking I/O operations, and the render thread is where all WebKit actions happen.

Why is it problematic to have synchronous IPCs terminate at the UI thread?

Synchronous IPCs terminating at the UI thread can lead to deadlocks, especially when dealing with plug-ins that also require synchronous communication with the renderer.

How does Chrome handle painting and scrolling operations to ensure smooth performance?

Chrome uses a backing store in the RenderView host to store the last rendered version of the page for painting. For scrolling, it shifts pixels in the backing store and fills in the exposed region, allowing for optimized scrolling with low CPU usage.

What is the advantage of Chrome's multi-process architecture when it comes to closing tabs?

The multi-process architecture allows Chrome to potentially kill the child process associated with a tab when it is closed, especially if the web page has no unload handlers, leading to faster tab closures.