Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(65)

Side by Side Diff: docs/linux_sandbox_ipc.md

Issue 1309473002: WIP: Migrate Wiki content over to src/docs (Closed) Base URL: https://chromium.googlesource.com/chromium/src.git@master
Patch Set: Created 5 years, 3 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
« no previous file with comments | « docs/linux_proxy_config.md ('k') | docs/linux_sandboxing.md » ('j') | no next file with comments »
Toggle Intra-line Diffs ('i') | Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
OLDNEW
(Empty)
1 The Sandbox IPC system is separate from the 'main' IPC system. The sandbox IPC i s a lower level system which deals with cases where we need to route requests fr om the bottom of the call stack up into the browser.
2
3 The motivating example is Skia, which uses fontconfig to load fonts. In a chroot ed renderer we cannot access the user's fontcache, nor the font files themselves . However, font loading happens when we have called through WebKit, through Skia and into the SkFontHost. At this point, we cannot loop back around to use the m ain IPC system.
4
5 Thus we define a small IPC system which doesn't depend on anything but <tt>base< /tt> and which can make synchronous requests to the browser process.
6
7 The zygote (LinuxZygote) starts with a UNIX DGRAM socket installed in a well kno wn file descriptor slot (currently 4). Requests can be written to this socket wh ich are then processed on a special "sandbox IPC" process. Requests have a magic <tt>int</tt> at the beginning giving the type of the request.
8
9 All renderers share the same socket, so replies are delivered via a reply channe l which is passed as part of the request. So the flow looks like:
10 1. The renderer creates a UNIX DGRAM socketpair.
11 1. The renderer writes a request to file descriptor 4 with an SCM\_RIGHTS cont rol message containing one end of the fresh socket pair.
12 1. The renderer blocks reading from the other end of the fresh socketpair.
13 1. A special "sandbox IPC" process receives the request, processes it and writ es the reply to the end of the socketpair contained in the request.
14 1. The renderer wakes up and continues.
15
16 The browser side of the processing occurs in <tt>chrome/browser/renderer_host/re nder_sandbox_host_linux.cc</tt>. The renderer ends could occur anywhere, but the browser side has to know about all the possible requests so that should be a go od starting point.
17
18 Here is a (possibly incomplete) list of endpoints in the renderer:
19
20 ### fontconfig
21
22 As mentioned above, the motivating example of this is dealing with fontconfig fr om a chrooted renderer. We implement our own Skia FontHost, outside of the Skia tree, in <tt>skia/ext/SkFontHost_fontconfig**</tt>.**
23
24 There are two methods used. One for performing a match against the fontconfig da ta and one to return a file descriptor to a font file resulting from one of thos e matches. The only wrinkle is that fontconfig is a single-threaded library and it's already used in the browser by GTK itself.
25
26 Thus, we have a couple of options:
27 1. Handle the requests on the UI thread in the browser.
28 1. Handle the requests in a separate address space.
29
30 The original implementation did the former (handle on UI thread). This turned ou t to be a terrible idea, performance wise, so we now handle the requests on a de dicated process.
OLDNEW
« no previous file with comments | « docs/linux_proxy_config.md ('k') | docs/linux_sandboxing.md » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698