Security/Sandbox/2018-01-25

From MozillaWiki
Jump to: navigation, search

« previous week | index | next week »

Alex_Gaynor

  • bug 1428055 - Removed access to macOS sandbox properties that are on-by-default. Landed, no awful fallout (yet).
    • Sent a PI request for additional QA on a variety of hardware
  • bug 1432811 - Small code cleanup to GeckoChildProcessHost
  • bug 1426807 - Landed fix for a11y + printing, huge thanks to :jwatt for figuring out the fix
  • bug 1407693 - CrashReporter file writing, tracked down one of the failing tests finally. Caused by different code paths for sandboxing vs. non-sandboxed process spawning + xpcshell not using the sandbox.
  • Heard back from Apple about the security report on (deny dynamic-code-generation) I sent them. They said it's not a bug, so I'll probably do a post (on my personal blog) at some point comparing this with MS's Arbitrary Code Guard.

gcp


bobowen

  • bug 1423628 - Stop processing native events in the content process
    • bug 1396984 - Scrollbar becomes black on first connection of second screen
      • Trying to pass down correct DPI to child process.
    • bug 1430744 - Stop processing native events in the content process on Windows in Nightly
      • Landed
  • Lots of reviews
  • Probably breaking release with the aid of some AVs.

handyman

  • bug 1382251 - Brokering https in NPAPI process
  • bug 1429643 - Limit SSL brokering to 64-bit
  • bug 1430586 - Intermittent crash @ Checker::StartReadOp
    • fixed new mingw + perma-orange issues blocking landing
  • bug 1358372 - sndvol.exe shows multiple volume sliders for browser
    • Attempt to dispatch to main at shutdown
    • Also, MS is apparently working on it
  • bug 1415160 - Set process mitigations on NPAPI proc
    • Getting started with Japanese IME
  • Updates to graphics doc

haik

  • bug 1393259 - [Mac] Remote access to fonts from custom directories, font managers
    • did some more tests and landed on Nightly
  • bug 1432567 - [Mac] Add a test that renders fonts from non-standard directories
    • Got python code working to register/unregister font
  • Needinfos
  • bug 1417005 - [Mac] Allow disabling sandboxing
    • 1-line change to make Mac respect MOZ_DISABLE_GMP_SANDBOX
  • Sandboxing Flash plugin process on Mac

tjr

  • **This Week:**
  • Triaged ~200 [tor] and [fingerprinting] bugs.
  • bug 1230910 - [MinGW] Working on uplifting the MinGW Sandbox patches with bobowen, going quick!
  • bug 1432213 - [MinGW] Fix MinGW build on Beta branch
  • bug 1432009, bug 1432668 - [MinGW] Fix two MinGW perma-fails and messed up David :)
  • bug 1409973 - [Fingerprinting] Add tests for Locale/Timezone JS Date Formatting in Resist Fingerprinting Mode. Blocked on bug 1358653
  • bug 1425462 - [Spectre] Longstanding work on Timer Fuzzing, moving steadily but slowly =/
  • bug 1431842 - [Spectre] Fix intermittent (maybe?) for timer rounding
  • **Rest/Next Week**
  • bug Something - Investigate /d2guardswtablesuppress (thanks gcp)
  • bug 1425462 - [Spectre] More Timer Fuzzing
  • bug 1230910 - [MinGW] Finish up the MinGW Sandbox stuff
  • bug 1418162 - [Fingerprinting] Fix the ESR detection/version spoofing
  • bug 1336208 - [Fingerprinting] Pick up font bug, see if I can carry it across finish line
  • bug 1235982 - [Hardening] Hopefully work on Control Flow Guard
  • bug 1376819 - [Hardening] Hopefully look into JIT Constant Blinding
  • More P1 fingerprinting bugs
  • Need to update spreadsheet for Jim
  • If anyone is interested in: Private Mac Graphics APIs, Ancient Bugs, and Massive User Experience Annoyances, bug 440895 has had some recent interest.
    • Comment from irc: There's one more hunch regarding SkyLight.framework that I'm investigating atm, but I'm not very hopeful that I'll find the solution in there either.

jld

  • The patches I've been talking about for months are finally landing in 60.
  • Landed:
    • bug 1401062 - clone(), as mentioned last week
    • bug 1430756 - stop testing for unshare(); was causing assertion failure with containers' external sandboxing
    • bug 1126437 - blocking socket/connect in content
      • This is the important one — raises the bar for sandbox escapes.
      • Potential problem: Nvidia GL has an extra connection to the X server, which is now blocked, but this doesn't seem to affect performance.
      • Sandbox level 4 now exists for this. Level is capped at 3 on Linux if audio remoting is turned off.
      • Probably "fixes" any remaining bind() issues, because they'll quietly fail to socket() instead.
      • Any surprise DBus calls will also start breaking now.
    • bug 1386019 - misc audio stuff that content doesn't need anymore
  • Almost landed:
    • bug 1430949 - Network namespace
      • Yak shaving for detecting "remote" X, which includes TCP to localhost.
      • I learned some new things about X11 display names.
    • bug 1213998 - chroot; literally one line
  • Wrote:
    • bug 1376910 - The Revenge of SysV IPC Blocking
      • Did telemetry queries to try to figure out how to detect fglrx; actual testing would be nice but I don't have hardware

Roundtable

Retrieved from "https://wiki.mozilla.org/index.php?title=Security/Sandbox/2018-01-25&oldid=1190252"