| Commit message (Collapse) | Author | Age |
|---|
| | |
|
| | |
|
| |
|
|
|
| |
This fixes ln5's test case. Couldn't reproduce arma's, but I'll admit I
didn't try very hard since the other page was consistently crashing.
|
| | |
|
| |
|
|
|
| |
This make the dynlib process more resilient to ld.so.conf having
oddities in it, and more importantly makes this work on my laptop again.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |\ |
|
| | |
| |
| |
| |
| |
| | |
Depending on something that's system specific (recent Ewwbuntu, Arch,
but not F26), obfs4proxy calls mmap slightly differently, and dies
because the call is getting `ENOSYS`ed by seccomp.
|
| | |
| |
| |
| | |
https://gitweb.torproject.org/builders/tor-browser-build.git/commit/projects/tor-browser/Bundle-Data/PTConfigs/bridge_prefs.js?id=5eb4c84fd87c80496dc522f44f26928f1330b814
|
| |/ |
|
| |
|
|
|
|
|
| |
This is more fallout from #20773, that only manifests on systems running
kernels that pre-date 3.17 (Debian oldstable). The "better" fix might
be to only engage the workaround if `SECCOMP_FILTER_FLAG_TSYNC` isn't
supported.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
| |
Apparently tabs crash without this in 7.5a5, and according to the report
this is the first thing it complains about before crashing deep in IPC
land.
At a minimum this shuts the error up, and a fresh install appears to
work...
|
| |
|
|
|
| |
I don't care what the people whining on trac or bugzilla say, I like
the existing behavior, so I'll flip the pref now to keep it.
|
| |
|
|
|
|
| |
It's essentially mostly harmless since even if a file is created, it
won't be persisted, but the browser shouldn't be keeping track of
recently used files to begin with.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Non-portable but this is cleaner.
|
| |
|
|
|
|
|
|
| |
This reverts commit 8ccea0137d87d049c0f09ece7428942af78b7570.
I need to think about this more. While I still think the call is
basically only used to make tracking people easier, if mainline Tor
Browser doesn't disable it, this will impact the user anonymity set.
|
| |
|
|
|
|
| |
The container doesn't have anything to "Open with", nor does it have
D-Bus access. Neither of these things will change, so the dialog can be
suppressed.
|
| |
|
|
|
| |
Why Tor Browser doesn't do this, I have no idea, but it's basically part
of the cancer that is killing the internet.
|
| |
|
|
|
|
|
|
|
|
| |
Updating addons is disabled by default and will not work, even though it
may appear to (#23265). Instead of checking and then failing to apply
the update, by removing the URLs from the prefs, the check will fail
immediately.
This applies to everything but HTTPSE, because HTTPSE provides it's own
updateURL in the install.rdf file.
|
| |
|
|
| |
attacks.
|
| |
|
|
| |
The Tor Browser developers have not took action on this, so I will.
|
| | |
|
| | |
|
| |
|
|
|
| |
The browser people apparently didn't update SelfRando, so the workaround
for the syscall stupiditiy is still needed.
|
| |
|
|
| |
Honestly, bluring the line between CA signed HTTPS and onions is terrible.
|
| | |
|
| |
|
|
|
| |
Since crashdumps are developed as a GSOC project, ensure that if/when
this makes it to production, that it will be force disabled.
|
| |
|
|
|
| |
Since no action is being taken by either upstream, unilaterally fix
this.
|
| |
|
|
|
|
| |
Leaking information to Mozilla is bad, the container setup takes step
not to allow non-standard extensions, and it's totally unsupported
anyway.
|
| | |
|
| | |
|
| |
|
|
| |
Fix some comments, no functional changes.
|
| |
|
|
|
|
|
|
|
|
|
| |
This is currently an advanced config option as it is still experimental
and needs more testing. When enabled it will copy the Firefox profile
directory into a tmpfs mount when the container is created, that is
discarded when the container is torn down.
The following directories are read-only bind mounted as usual:
* profile.default/preferences
* profile.default/extensions
|
| |
|
|
|
|
| |
Foot + gun options are bad, the "Get Addons" pane is force disabled via
a locked pref by virtue of being Mozilla + Google botnet bullshit, and
this commit also switches to an extension whitelist.
|
| |
|
|
|
| |
This seems to be the solution that the Tor Browser devs are going for,
which is sensible.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
treated as such.
Loading Google Analytics as part of an IFRAME that implements an
internal `about:` URL, without being explicitly opt-in is the total
antithesis of privacy respecting.
|
| |
|
|
|
| |
I should have done this to begin with, and part of me wants to deprecate
all of these options entirely, but people will complain.
|
| |
|
|
|
|
| |
Only enable the workaround for the alpha series, and only if the version
is 7.5a2 or older, on the assumption that since it's fixed upstream it
will be reflected in the next Tor Browser alpha build.
|
