| Commit message (Collapse) | Author | Age |
| |
|
|
|
|
|
|
|
|
|
|
| |
Kernel commit 99137b7 introduced an additional check of the address
length. This exposed a bug in the arp code where the address length was
being set incorrectly.
Now the length is set from the ETH_ALEN constant declared in
linux/if_ether.h
This fixes #2314
|
|
|
|
|
| |
* Add a profile for ghostwriter
|
|
|
|
|
| |
redundant check that adds defense in depth and maybe one
day can replace the other, file based check
|
|\
| |
| | |
enforce nonewprivs instead of seccomp for chroot sandboxes
|
| |
| |
| |
| |
| | |
currently users are able to specify a seccomp filter of their
choosing, leaving the real defense to nonewprivs anyway.
|
| |
| |
| |
| |
| |
| | |
* New profile supertuxkart
* review fixes
|
|/ |
|
| |
|
| |
|
|
|
|
| |
emphasize that only firecfg needs all permissions, not firejail
|
| |
|
|
|
|
|
|
|
|
|
| |
This reverts commit 93779cb9cd0d098cd3587e2f795200d98e3af1ee.
That commit removed restrictions, but also added new
inconsistencies.
Starting again from the previous state is easier than evolving
the current state, hence reverting the commit.
|
|
|
|
|
|
| |
setting the PULSE_CLIENTCONFIG environment variable to the unmounted
file is a safe fallback, use it in more cases when mounting is considered
not an option
|
| |
|
|
|
|
| |
and print more meaningful error message
|
| |
|
|
|
|
|
|
|
|
|
|
| |
setting the KDE_FORK_SLAVES environment variable removes all inconsistencies
that arise from slaves running outside the sandbox or in a different sandbox;
it also makes it slightly more difficult to abuse KIO in general and helps to
mitigate security problems due to thumbnailing, which now always happens inside
the same sandbox. The trade-off is more concurrently running slave processes.
closes #2285
|
| |
|
| |
|
|\
| |
| | |
refactor private-cache and tmpfs
|
| |
| |
| |
| |
| |
| | |
has the immediate benefit that the result of combining
--noexec and --tmpfs does not depend on the sequence of
the options
|
|\ \ |
|
| |/ |
|
| | |
|
|/ |
|
| |
|
| |
|
|
|
| |
Found using `codespell -q 3 -L shotcut,objext,als,ans,creat,varius,chage,tthe`
|
| |
|
| |
|
|\
| |
| | |
Allow prefixing colon to profile argument of --profile to for a profile search
|
| |
| |
| |
| | |
search, and disallow a directory to match as a profile file.
|
| |
| |
| |
| | |
my editor is weird sometimes
|
| | |
|
| | |
|
| | |
|
| | |
|
|\| |
|
| | |
|
| | |
|
| | |
|
| |\
| | |
| | | |
Add new config option to disable U2F in browsers, enabled by default
|
| | | |
|
|/ /
| |
| |
| |
| |
| |
| | |
run firecfg with umask 022 and print a diagnostic message if
the database is not readable.
closes #2225
|
|/ |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
* vanilla kernels before 3.15 don't expose a mount id in /proc/pid/fdinfo files.
This is still relevant on Ubuntu 14.04 with 3.13 kernel, CentOS 7 doesn't
have this problem. In this case fall back to simple a remount and print
a warning.
* drop euid switching as it doesn't really serve a purpose here (paths are not opened
in reading or writing mode, and we are not doing anything with it) and potentially
causes problems when suid programs are sandboxed
* more rigorous error handling
|
| |
|
| |
|