I changed the Chromium browser (as a masters project) to intercept suspicious extension actions like inserting elements etc and to alert users of what the extension is attempting to do. Using this proof-of-concept browser would have helped you debug your ad injection problem!
One of the reasons the Chrome browser uses one process per tab is so that memory between web-pages is isolated. By doing this, an attacker in control of a renderer process cannot read data from another renderer process (web-page).
It seems like Firefox has made a security / memory tradeoff here since renderer processes can render multiple (4 by default) web-pages at a time.
Chrome doesn't strictly follow the one-process-per-tab rule either. If you browse for a while with a bunch of tabs open, and then check Chrome's Task Manager, you'll see that it's quite common for a single process to render tabs from multiple different domains.
Correction: Firefox doesn't use n content processes which render 4 tabs each. By default, it uses 4 content processes which share the task of rendering your n tabs. (AFAIK, at creation time, each tab is assigned one of those rendering processes.)
> One of the reasons the Chrome browser uses one process per tab
Chrome does not use a process per tab. It uses a process per domain (mostly), unless you middle click a link in which case it always shares the same process as the original tab.
No, that's not quite right. What they do really is closer to proces-per-tab (with some complications around cross-site navigation) unless you have more than some number of tabs, in which case they will just have them share processes. See https://www.chromium.org/developers/design-documents/process... and note that the default is to put multiple "independent" instances of the same site in different processes, even though they're same-domain. What you're describing is the non-default "Process-per-site" model.
> unless you middle click a link in which case it always shares the same process as the original tab
I believe they changed that behavior starting with Chrome 60. See https://codereview.chromium.org/2680353005/ which talks about ctrl-click, but I would assume (watch me turn out to be wrong!) that middle-click takes the same codepath.
Pretty sure that wouldn't be enough to write the HDD. You'd need sudo and theres a couple more double checks. In modern versions of Linux/Unix at least.
But the cost of incorporating is really tiny isn't it (£15 in the UK where I live)? And I'm thinking about what users think seeing an app created by a person vs a company
The cost of incorporating in the UK is cheap but you're creating a legal entity so there's more to it than just coming up with a name and parting with the cash. I'd have a read of the following to get an idea of what it entails and what other options you have:
Most people won't care if its made by a person or company, as long as it solves their problem.
There are plenty of apps that are run by solo founder...I think you're thinking into it too much...always better to save as much money as you can and put it into the product at first.
https://cypher.codes/writing/intercepting-suspicious-chrome-...
- Note: my project specifically tries to protect users from Facebook hijacking and ad injection attacks - the two most common attacks on the CWS!