Closed
      
        Bug 1270228
      
      
        Opened 9 years ago
          Closed 5 years ago
      
        
    
  
about:debugging spawning child process in non-e10s mode
Categories
(DevTools :: about:debugging, defect, P3)
        DevTools
          
        
        
      
        
    
        about:debugging
          
        
        
      
        
    Tracking
(Not tracked)
        RESOLVED
        INVALID
        
    
  
People
(Reporter: bkelly, Unassigned)
References
(Blocks 1 open bug)
Details
STR:
1) Launch nightly in non-e10s mode.
2) Open about:memory and measure.  Verify there is only a Main Process.
3) Open about:debugging in a tab.  Switch to the worker panel.
4) Measure in about:memory again.  Notice there is now a Web Content process with a single about:blank window.
If you happen to have a service worker registered you may also see one of those works running in the child process in step 4.
about:debugging should not force a child process in non-e10s mode.
| Updated•9 years ago
           | 
          tracking-e10s:
          --- → -
| Comment 1•9 years ago
           | ||
Alex, it looks like about:debugging's Workers panel accidentally creates a Web Content process upon being loaded, but I'm not sure why.
Is it possible that calling `client.mainRoot.listProcesses()` [0] or `client.getProcess()` creates a Web Content process if there are non?
Or maybe somehow loading a process script like in [1] is to blame?
[0] https://dxr.mozilla.org/mozilla-central/source/devtools/client/aboutdebugging/modules/worker.js#58
[1] https://dxr.mozilla.org/mozilla-central/source/devtools/server/actors/worker.js#332
Component: Developer Tools → Developer Tools: about:debugging
Flags: needinfo?(poirot.alex)
Priority: -- → P2
| Comment 2•9 years ago
           | ||
I don't think listProcesses nor getProcess create a process.
But the loadProcessScript might be, did you tried commenting this line?
Flags: needinfo?(poirot.alex)
| Updated•9 years ago
           | 
Flags: needinfo?(janx)
| Updated•9 years ago
           | 
Flags: needinfo?(janx)
Priority: P2 → P3
| Updated•9 years ago
           | 
Flags: needinfo?(janx)
| Comment 3•9 years ago
           | ||
Benoit, while you work on spawning or observing child processes in about:debugging, could you please keep an eye on this bug as well? (In non-e10s, we shouldn't spawn a child process, but apparently we do.)
Flags: needinfo?(janx) → needinfo?(bchabod)
| Comment 4•9 years ago
           | ||
(In reply to Jan Keromnes [:janx] from comment #3)
> Benoit, while you work on spawning or observing child processes in
> about:debugging, could you please keep an eye on this bug as well? (In
> non-e10s, we shouldn't spawn a child process, but apparently we do.)
Sure! I will try to reproduce, and see if I can isolate some code responsible for this bug.
Flags: needinfo?(bchabod)
| Comment 5•9 years ago
           | ||
Ben, sorry for the delay, I've been busy with other bugs.
However, on latest Nightly, I can't manage to reproduce this bug.
I tried different variations of the STR, with workers registered/running, without workers, with a lot of content pages opened, with about:debugging only, and in about:memory there is only the main process at each measure.
Can you please confirm that you're able reproduce this bug, and maybe provide a more precise scenario if it's the case?
Flags: needinfo?(bkelly)
| Reporter | ||
| Comment 6•9 years ago
           | ||
I see this:
1) Launch nightly
2) Disable e10s in options
3) Restart browser
4) Open example.com in a tab
5) Open about:memory in a tab
6) Minimize and measure memory
7) Observe a content process containing a single about:blank window.
Eventually this process goes away.  It seems unrelated to devtools in any case.
Flags: needinfo?(bkelly)
| Updated•7 years ago
           | 
Product: Firefox → DevTools
|   | ||
| Comment 7•5 years ago
           | ||
Closing as non-e10s is gone
          You need to log in
          before you can comment on or make changes to this bug.
        
Description
•